blind sequential lineups - PDF document

Double - blind sequential lineups 1 Double - Blind Sequential Police Lineup Procedures: Toward an Integrated Laboratory & Field Practice Perspective Final Report Grant # 2004 - IJ - CX - 0044 March 31, 2007 Nancy K. Steblay Augsburg College Double - blind sequential lineups 2 Abstract The p roject purpose was to join behavio ral data from scientific research, current field experience, and new laboratory investigation to advance knowledge of best police lineup practice for law enforcement and research communities. Th e project wa s a collaborati ve effort between the Hennepin Cou nty (Minnesota) Attorney’s Office and the P I’s research laboratory . T hree data collection and analysis c omponents were completed : (1) H ennepin County’s pilot implementation of double - blind sequential lineup procedures , includ ing 2 80 field lineups ; (2) a la boratory evaluation of the quality of the Hennepin County lineups; and (3) a n experimental l aboratory test of how revisions to prescribed lineup protocol affect eye witness lineup decisions . The Hennepin County (HC) results indicate a successful applicati on of double - blind sequential lineups to street investigations. Double - blind sequential lineups are now established county - wide, providing a standardized scientifically - based lineup procedure that has been demonstrated to be practicable in real cases. H C field data and laboratory test data converged to demonstrate increased misidentifications when a witness is allowed to view the lineup more than once. The lab study also revealed how reduced lineup size — attrition due to the witness’s recognition of fille rs — can negatively affect eyewitness identification accuracy . Completed grant o bjectives include d : (1) D escriptive data providing t he first available baseline measure for blind sequential field lineup practice; (2) S ummary of the field lineup implemen tation process ; (3) A laboratory test of the impact on eyewitness decisions of an opportunity for repeated viewing of the sequential lineup (4) A laboratory test of the effect on eyewitness decisions of a reduction in lineup size through witness Double - blind sequential lineups 3 familiar ity with fillers ; (5) Integration of laboratory and field data to generate practical, empirical, and theoretical knowledge of effective lineup procedure ; and (6) P ractical and scholarly presentations and publications as appropriate to law enforcement profe ssionals , the psycho - legal research community , and the NIJ Data Resource Program . Double - blind sequential lineups 4 Blind/Sequential Police Lineup Procedures: Toward an Integrated Laboratory & Field Practice Perspective The purpose of th is project was to join data from past scientific eyewitness research , current field experience, and new laboratory investigation to advance k nowledge of best police lineup practice for law enforcement and research communities. Th e project include d two components : (1) a pilot program of double - blind sequ ential lineups in Hennepin County, Minnesota, and (2) a laboratory investigation of eyewitness memory under revised lineup pro cedures . Th ese combined sources of data – field and laboratory – have generated valuable new information about successful field practice as well as scientifically sound answers to questions regarding the effects of specific changes to prescribed lineup

protocol . Such knowledge can contribute enormously to policy reform efforts . Objectives The data generated by the Hennepin Co unty (HC) pilot lineup pro gram address ed two primary objectives : (1 ) c ollection and analysis of descriptive data detailing field lineup administration, situational context, and witness response , to form the first available scientific baseline measure for double - blind sequential lineup practice ; and (2) a cohesive and instructive summary of the HC lineup implementation process. Two questions prompted the HC program and data analysis: Can double - blind sequential lineup procedures be effectively implemente d in the field? Do the number and quality of identifications change with the blind - sequential procedure? T he laboratory objectives involved controlled testing of two specific deviations from prescribed lineup protocol : (3) t o d etermine the impact on eye witness lineup decisions of a witness’s second (or third) viewing of a lineup in which the photos were presented Double - blind sequential lineups 5 sequentially , and (4) t o assess the impact on eyewitness decisions of a naturally - occurring reduction of lineup size due to a witness’s familia rity with lineup foils . The collaboration between the County Attorney’s Office and the research lab also facilitated the final project objective , (5) the i ntegration of laboratory and field data to generate practical, empirical, and theoretical knowledge o f effective lineup procedure , including i dentification of lessons as to how more effective lineup procedures can best be brought to practice . Brief r ationale for the project Past scientific research has led to a cohesive lineup prototype that promises a s ignificant improvement in eyewitness accuracy (Well, Small, Penrod, Malpass, Fulero, & Brimacombe, 1998) . A dvancement has been achieved through creative application of insightful theory and rigorous laboratory testing. One compelling feature of th e resea rch endeavor is the relative speed at which lab results have been usefully applied to police practice (Wells, Malpass, Lindsay, Fisher, Turtle, & Fulero, 2000) . The l aboratory research has resulted in guidelines for law enforcement aimed at prevent ing wit ness errors in police lineup identification s (Technical Working Group for Eyewitness Accuracy, 1999, hereinafter The NIJ Guide ) . The recommendations include refinements in both lineup structur e and procedure . More recently, scientists have advised police to also in corporate double - blind administration and a sequential photo presentation format into their lineup procedures (Wells, et al., 2000) . Versions of th is newer protocol now are being introduced into practice in a number of jurisdictions ; one such i nstance is a year - long lineup pilot project in Hennepin County, Minnesota. As scientific research is translated to public policy, as is the case in Hennepin County, it is desirable to maintain a timely exchange of information among scientists, law Double - blind sequential lineups 6 enforc ement professionals, and legal practitioners. L aw enforcement officials are in a good position to identify gaps in laboratory - generated knowledge , and laboratory researchers can use field research to inform their efforts to achieve desirable levels of au thenticity and ecological validity in the lab (see e.g., Diamond, in press) . At the same time, f ield practice prompts questions that compel subsequent laboratory examination . Q uestions of interest usually center on practical issues,

but hold implications for broader theoretical issues as well. Of particular relevance is the matter of how adjustments in police lineup procedure that meet convenience or practical needs of the local jurisdiction might compromise or enhance witness accuracy. W ith its randomi zed controlled trials, t he laboratory offers the most objective means to address such question s . Legal rationale and review Eyewitness identification is persuasive evidence of criminal wrongdoing . O n the witness stand a confident eyewitness is likely to deliver compelling testimony (Greene 1988; Rattner 1988) . However memory is fallible , and even a well - intentioned and confident eyewitness may bring flawed recall to a police lineup and false ly incriminating evidence to court. Recognizing this problem , the United States Supreme Court in the 1 9 60s began to put into place safeguards to protect criminal defendants from wrongful convictions through misidentification. For example, i n United States v. Wade (1967) the Court held that the Sixth Amendment right to counsel applies to critical stages of pretrial proceedings includ ing the physical lineup procedure. The Court recognized the “vagaries of eyewitness identification” and the “innumerable dangers and variable factors which might seriously, even cruciall y, derogate from a fair trial” (p. 218). The United States Supreme Court ruled in Stovall v. Denno (1967) that an unduly suggestive lineup Double - blind sequential lineups 7 constitute s a due process violation if it could lead to an irreparably mistaken identification. Therefore a defenda nt could move to suppress prejudicial identification testimony depending on the “totality of the circumstances” surrounding the testimony (p.302). The next year, in Simmons v. United States , the Court ruled that each potential due process violation during a lineup must be examined on the facts of the individual case. Lineups would be excluded from trial if the “procedure was so impermissibly suggestive as to give rise to a very substantial likelihood of irreparable misidentification” (p. 384). In the 1970 s, however, the Court began retreating from the broader safeguards guaranteed in Wade , Stovall and Simmons . In United States v. Ash (1973) the Court refused to extend the protection of a Sixth Amendment right to counsel to photographic lineups, reasoning that a photo display did not involve such risk so as to require a safeguard . The Court also found that even extremely biased lineups were not per se exclusionary. Instead, it was necessary to determine whether an admittedly suggestive lineup was nonethel ess reliable. In Neil v. Biggers (1972) the court considered five factors for determining the dependability of an eyewitness identification: the witness’s opportunity to view the perpetrator during the crime, the witness’s attention to the perpetrator at the time of the crime , the accuracy of the witness’s initial description of the perpetrator, the witness’s certainty at the lineup, and the length of time between the crime and the identification. In Manson v. Brathwaite (1977) the Court concluded, “reli ability is the linchpin in determining the admissibility of identification testimony” (p. 114). This even more firmly emphasized that the important question was not whether the identification procedure was prejudicial to the criminal defendant, but whether the identification itself Double - blind sequential lineups 8 was reliable. To varying degrees, these cases sought to remedy the effects of suggestive lineups, but did little to dis courage

use of prejudicial procedures. The subsequent introduction of DNA testing allowed the exonerati ons of many wrongfully convicted individuals , to da te num bering 18 3 ( The Innocence Project), and drew substantial attention to eyewitness error as the major contributor to these unjust conviction s (Wells et al., 1988) . Investigators, attorneys, and testif ying witnesses who ha ve helped to prosecute a later - exonerated individual realize with extreme regret that even well - intentioned “by the book” procedures can end very badly. Along with the horrific effects on the li v e s of the violated innocent person and his or her loved ones , a wrongful c onviction leaves the true perpetrator on the streets to commit additional offenses. The reality of wrongful conviction also has the potential to substantially erode public confidence in the justice system and citizens’ sense of security. As the Court considered eyewitness evidence problems , scientists over four decades developed and reported findings that shed light on the reasons for memory errors in the eyewitness experience (Wells, et al., 1998). Principles of huma n perception, memory, and social influence illuminated not only the causes of faulty memory but preventive measures to preclude eyewitness failure. Scientific recommendations about improved lineup structure and procedure provide one means to make it less l ikely tha t innocent suspects are prosecuted and more likely that true perpetrator s are held responsible for the ir crime s . DNA exonerations became the catalyst for lineup reform in some jurisdictions . For example, in the wake of the New Jersey Cromedy d ecisio n — an eyewitness evidence case in which a DNA test of biological evidence collected from the victim exonerated the defendan t — Attorn ey General John Farmer turned to the lineup reforms recommended by Double - blind sequential lineups 9 researchers (Doyle, 2005). Farmer approved new lineu p procedure s with safeguards exceeding those recommended by the National Institute of Justice . Using the unique authority granted the Attorney General in that state, Farmer implemented mandatory statewide guidelines, making New Jersey the first state to u niformly adopt double - blind sequential lineup procedures (State of New Jersey, 2002) . Other initiatives have been undertaken as well. I n 2002 , Governor Ryan’s Commission on Capital Punishment, charged with ensuring the accuracy and justness of capital pun ishment in Illinois, recommended the implementation of eyewitness identification reforms (Report of the Governor’s Commission on Capitol Punishment). The North Carolina Actual Innocence Commission created a series of recommendations in 2003 for state law enforcement officers, including a comprehensive lineup protocol . In early 2005, the Avery Task Force wrote model recommendations for the Wisconsin criminal justice system , r ecently affirmed by the State of Wisconsin Office of the Attorney General (2006) . The Virginia General Assembly also instructed the Virginia State Crime Commission (2005) to create guidelines for improving lineup procedures in the commonwealth . Individual jurisdictions in a handful of states also have im plemented improved procedures, Massachusetts, Virginia, Washington, Minnesota and California among them (see e.g, Northhampton Police Department, 2005; Police Chief’s Association of Santa Clara County, 2002). Scientific rational e and review The U.S. Depa rtment of Justice published Eyewitness E vidence: A G uide for L aw E nforcement in 1999 . This document followed from concerns about unjust convictions revealed through DNA - exoneration cases , it s recommendations based on the growing

Double - blind sequential lineups 10 collection of sound scientific laboratory research. I n the majority of DNA - exoneration cases, mistaken eyewitness es were the primary evidence leading to conviction (Wells, et al., 2000). The p sychological research has shown that eye witness re ports are often un reliable and that unintentional police influenc e can exacerbate witness tendencies tow ard inaccurate lineup choices. The NIJ G uide is clear about remedia tion of this problem: The eyewitness should be given unbiased lineup instruction (“The perpetrator may or may not be in this lineup”), lineups should be constructed fairly ( e.g., foils matched to perpetrator description and the perpetrator not stand ing out in the lineup ) , and officers should record results in a prescribed manner . A lso included in the NIJ Guide is a recognition that “ a dvances in social science and technology will, over time, affect procedures used to gather and preserve eyewitness evidence” (p. 8). Three examples were specified as areas for future exploration: the s equential method of lineup presentation; b lind lineup administration p rocedure; and computer - based imaging systems to facilitate the use of improved procedures. Sequential method of lineup presentation . A sequential lineup presentation attempts to remedy a troublesome aspect of the eyewitness decision - making process , rel ativ e — v ersus absolut e — j udgment (Wells, 1984) . Standard police lineups present the eyewitness with all lineup members ( e.g., six persons) at one time. Under th i s s imultaneous format , eyewitnesses tend to compare lineup members to each other to determine w hich most closely resembles the offender in memory , a process of relative judgment . If the witness was able to encode a vivid memory of the perpetrator and this person is in the lineup (a target - present array), the likelihood of a positive and correct ide ntification is increased . The concern , however, is whether the witness will recognize the absence of the offender when in fact the suspect is not the perpetrator. Th e DNA - exoneration cases – the majority Double - blind sequential lineups 11 of which were instances in which the actual offe nder was not in the lineup -- illustrate exactly this p roblem: witness in ability to correctly reject a target - absent lineup (The Innocence Project, 2006) . The results of c ontrolled experiments predict a negative outcome when police unknowingly place an in nocent suspect in a lineup. The impact of relative judgment when the offender is absent from the lineup was demonstrated convincingly by Wells ( 19 93 ) . Pa rticipant - witnesses to a staged crime were shown one of two versions of a lineup. When the perpetrato r was present in a six - person lineup, 54% of the witnesses selected him. All witnesses had been given an unbiased instruction (“the perpetrator may or may not be in the lineup”) and 21% opted not to choose from the lineup. Now, the key question: What w ill happen to the second group of witnesses who view the same lineup minus the perpetrator? If 54% of witnesses truly recognize the offender when he is present , th i s 54% -- who would have identified the offender had he been in the lineup -- should join the 21% who reject the lineup, producing a 75% “no - choice” response. What happen ed is quite different: Only 32% of the witnesses correctly rejected this target - absent lineup , only a small gain from the 21% expected in this response category. Sixty - eight percent of the witnesses chose from the lineup, most of the

filler identifications falling on the photo that was the next - best match to the offender , placing this innocent suspect in jeopardy. This result has been dubbed the “target - to - foils shift” (Cla rk & Davey, 2005). Even in a perpetrator - absent lineup, it is likely that one lineup member will provide a better relative match to memory than the others , thereby drawing the attention of the eyewitness and increasing the risk of false identification. To address this problem, t he sequential procedure presents the eyewitness with one lineup member at a time and requires the witness to decide whether or not that person is the Double - blind sequential lineups 12 perpetrator before moving to the next photo . The witness does not know how man y photos will be shown, thereby decreasing the pressure to choose any one photo in the sequence. This one - at - a - time procedure is intended to discourage the eyewitness from simply deciding who most resembles the perpetrator , thus forcing a more absolute de cision criterion. The NIJ Guide only suggested the sequential method for practice, presumably because the available research in 1999 did not yet confirm the sequential lineup’s benefit beyond the traditional simultaneous format. Since that time, a meta - an alytic review has demonstrated reliable positive laboratory outcomes from use of a sequential procedure (Steblay, Dysart, Fulero, & Lindsay, 2001). S imultaneous versus sequential formats produce dramatically different choice and accuracy outcomes . Witnes ses who view a simultaneous lineup array are more likely to choose a photo from the lineup. When the perpetrator is p resent, this higher choosing rate may boost correct identification s, likely aided by relative judgment . In a target - absent s imultaneous d isplay , the increased tendency to choose translates into greater risk of false identification. The reverse is true for the sequential format. Particularly when the perpetrator is absent from the lineup, t he sequential format is preferable , reducing false alarms by 23% and identification of a designate d innocent suspect by 2/3 (27% to 9%). Some investigators may be concerned about the reduction of correct identifications for sequential lineup s compared to simultaneous outcomes (3% and 15% average dec reases, for the subgroup of “choosers” and for all witnesses, respectively). It is not known if this drop represents a loss of accurate eyewitness identifications, a reduction o f lucky guesses, or both. Penrod (2004) has presented a compelling argument t hat guessing is a significant component of eyewitness decisions. This is likely to be particularly so with Double - blind sequential lineups 13 the simultaneous format. Why? In simultaneous lineups, witnesses with weak memories can pick the lineup member who looks most like the culprit – a shallow and relative judgment. With a sequential lineup, the witness with a weak memory is inhibited by the greater difficulty of the task and cannot simply compare photos to arrive at a best guess (Wells, 2006). It is also important to note that the s equential lineup is more diagnostic of guilt when the witness does make a choice , y ield ing odds of guilt almost twice that of the simultaneous lineup. For police, the critical question is: Is the identification a good predictor of guilt? The blind - seque ntial lineup procedure improves the odds that a suspect, if identified, is the actual culprit (Wells, 2006 a ). R esearchers use the phrase “ s equential lineup” as shorthand for what is actually a collection of ru

les that represent best practice for conducting eyewitness identifications. For example, t he sequential procedure assumes a single - suspect model (only one suspect in the array) and that the lineup task is the first identification attempt by the witness . Furthermore, a n effective sequential procedure in cludes the following (see e.g., Lindsay & Wells, 1985 ; Cutler & Penrod, 1988 ) :  Effective lineup construction , designed to detect the unreliable witness and to protect an innocent suspect by drawing guesses to filler photos .  A lineup of at least six memb ers, five of whom are foils unknown to the eyewitness.  Each lineup member chosen to match the witness’s description of the perpetrator.  The suspect’s position in the lineup determined in an (approximately) random manner. Double - blind sequential lineups 14  Instruction to the witness t hat t he perpetrator may or may not be in the collection of photos to be displayed ( an “unbiased ” or “c autionary ” i nstruction).  Instruction to the witness that the sequence must be completed even if/after an identification is made , and the procedure conducted in this manner .  The witness unaware of how many photos are in the sequence.  Photos presented one at a time, with a decision made before examining the next.  The witness not allowed to “go back” over the sequence.  The witness not allowed to place photos next to one another. Blind lineup administration . Double - blind experimental procedure , in which neither experimenter nor subject know the subject’s treatment condition , is a n essential element of sound scientific method used to prevent inadvertent contaminatio n of research results . Interpersonal expectancy effects occur across a broad set of human interactions , necessitating a double - blind method for data accuracy (Harris & Rosenthal, 1985; Rosenthal, 1988; Rosenthal & Rubin, 2002) . First noted as essential f or lineup administration by Wells in 1988 and reinforced by a broader group of scientists in recommendations for lineups in 1998 (Wells, et al, 1998) there is wide agreement among eyewitness scientists that the double - blind lineup procedure is crucial in e yewitness procedures (see e.g, Wells, 2006 b ; McQuiston - Surrett, Malpass, & Tredoux, 2006; Haw & Fisher, 2004). A double - blind lineup protocol (often simply referred to as “blind”) helps to manage the inherently suggestive nature of the situation (Douglas s, Smith, & Fraser - Thill, in press; Phillips, McAuliff, Kovera, & Cutler, 1999). A lineup administrator who does not know the identity of the suspect is unlikely to l ead the witness to the suspect through verbal or Double - blind sequential lineups 15 nonverbal cues. A supplement to this pr ocedur e — no tification to the witness that the officer does no t know which is the suspec t — af fords the additional advantage that the witness is less likely to seek or infer cues from the officer’s behavior. Specifically, then, the blind lineup procedure inclu des the following key features :  The officer displaying the photos does not know which photo depicts the suspect (“blind administration”).  The witness is informed that the lineup administrator doesn’t know which photo, if any, is the suspect (“blind instruc tion”)  An assessment of witness confidence is taken at the time of the identification and before feedback from police or others (Wells, et al., 1998). T he NIJ G uide identified blind procedure as a direction for future exploration and field testing . The ex

pressed concern was that this desirable feature was likely to be impractical in police field use. Focused r esearch conducted si nce publication of the NIJ Guide indicates that the double - blind procedure essential to scientific integrit y more broadly is al so critical to the task of securing accurate eyewitness accounts (Douglass, et al., in press) . Any operational difficulties in the blind procedure must be seriously weighed against the inherent defects of the non - blind lineup. In the lineup situation, there are multiple avenues for administrator influence (Wells & Seelau, 1994) Mo st obvious are the unknowing leaks and “tells” that c ue the witness as to the suspect’s position in the lineup . These include verbal comment, facial expression, posture chan ge, hesitancy, enthusiasm, a different handling of the photo , and any number of such unintentional signal s . T he r e is no need to assume intentional ity of influence on the part of the lineup administrator for these influences to be dangerous to an innocent suspect . Double - blind sequential lineups 16 T wo additional conduits for influence are the lineup administrator’s selective attention to witness comments -- which can significantly shap e the witness’ s decision -- and selective recording of fillers especially in case of low - confidence fill er selection s . Examples may help to portray these threats. Co nsider the lineup scenario , using either a simultaneous or sequential format , in which the witness pauses at Photo #3 , the suspect’s photo . Note that even a seemingly conscientious response f rom an attentive but non - blind lineup administrator may p ush the witness toward the suspect : Witness : “Hmmm … ” (long pause) “Maybe… t hat’s the look …” (shrug) . Non - blind Administrator : “Tell me more…What’s familiar about #3?” Witness : “The eyes are familiar. The hair was different” … (more conversation ensues ) Non - blind Administrator : “Okay, should I indicate you chose #3?” The conversation may be quite different if the lineup administrator knows photo #3 to be a filler . The investigator may unintentionally push the witness away from the filler, a photo pre sumed to be not directly helpful to the investigation. Witness : “Hmmm..” (long pause) “Maybe… That’s the look …” (shrug). Non - blind Administrator : “Okay, not exactly? Look carefully, you don’t have choose unless you’re certain.” … Should we go to the next photo?” Consider a final scenario, again with the witness slowing down to examine P hoto # 3 . This scenario reveals the easy error of selective recording. Witness : “Hmmm..” (long pause) “Could be… I’m just not entirely sure…” Non - blind Administrator : (If #3 is the s uspect) r ecord s as tentative ID of suspect ( If #3 is a f iller) r ecord s as failure to ID the suspect. Double - blind sequential lineups 17 The double - blind procedure counters the above problems b y as sur ing equal attention to all witness comments and non - selective interpretation of witness responses. The witness must depend on memory and t he admin istrator records an objective appraisal of the lineup procedure and results , thereby protecting the in tegrity of the lineup . Blind administration is essential . A l ack of administrator masking confounds interpretation of lineup outcomes because such a flaw changes the investigative question from “Is this suspect the perpetrator?” to “ Did the witnes

s pick t he person police thought was the suspect? ” When double - blind administration of the lineup is not used, there is potential not only for erroneous identification, but for the inves tigator ’s behavior to influence the eye witness ’s subsequent confidence in her or her decision . Research demonstrates a mode st correlation between witness confidence and accuracy (Sporer, Penrod, Read, & Cutler, 1995) and this relation ship is easily corrupted. Factors that produce witness accuracy ( e.g., good viewing conditions, len gthy exposure, distinct perpetrator features) are not the same as those that may affect confidence ( e.g., confirmatory feedback from police, post - event information, or influence from other witnesses). A dangerous erosion of the confidence - accuracy relatio nship occurs when an eyewitness is exposed to factors that inflate confidence but have little relationship to accuracy (Wells & Bradfield, 1998). A recent meta - analytic review underscores the need for lineup administrators to assess eyewitness confidence before any feedback is provided (Douglass & Steblay, in press). A nalysis of 20 available laboratory tests demonstrated that confirming post - identification feedback received by the witness immediately after the identification (i.e., “Good. You identified t he actual suspect.”) inflated significantly the participant - witnesses’ retrospective confidence reports when compared with a control group told nothing about the accuracy of Double - blind sequential lineups 18 the identification. (Witnesses indicated how certain they had been at the time of their identification.) Perhaps more alarming is that an extensive range of variables was inflated in conjunction with retrospective certainty, including witness reports of the quality of their view of the perpetrator, how much attention was paid, ease of the identification, and basis for the identification. Participant witnesses who received confirming feedback were also more willing to testify about their identification and reported a greater ability to remember strangers. These outcomes demonstrate the reliability and robustness of distortions produced by post - identification feedback and reinforce the need for double - blind lineup administration and prompt full recording of eyewitness certainty comments. The Hennepin County pilot program incorporated bo th double - blind and sequential components, offering the first objective baseline measure of eyewitness responses under these conditions in the field. Study 1 Hennepin County Field Pilot Pro gram In November 2003, t he Hennepin County Attorney’s Office embarked on a year - long pilot study during which investigators used a revised lineup procedure in the city of Minneap olis and three suburban communities . The goal of the Hennepin County pilot wa s to “implement an eyewitness identification pilot program w ith H ennepin C ounty law enforcement agencies to reduce the possibility of incorrect identifications from photo lineup s ” (Klobuchar, 2004). The strength of scientific laboratory results underlying the new lineup protocol ha d convinced HC leaders that the n ew method wa s superior . Thus, t he purpose of the pilot wa s not a comparative evaluation between new and old technique s . Instead , the program was an attempt to identify and remedy implementation issues , t o determine how recommended lineup procedures can b est be brought into practice. T he task Double - blind sequential lineups 19 was t o see if “ the techniques suggested by lab investigation really work in real police investigation s ” (Scoggin, 2004).

Two questions framed the project: Can the procedures be effectively implemented in the field? Do the number and quality of identifications change with the blind - sequential procedure? (Scoggin, 2005). Method Sample The Hennepin County pilot pro gram focused on felony cases in four municipal police departments, including both stranger and familiar perpetrator lineups. The cities represent ed four levels of population and include d both urban and suburban locales. In Minneapolis, the largest of the four cities, the protocol was used exclusively by Central Investigations, which handles violent crimes. T he project involved 280 lineups from 117 cases, representing 206 eyewitnesses over a twelve month period ending in November 2004. (See Table 1 ) The Augsburg College Institutional Review Board approved the research protocol. Hennepin County Judge Kevin Burke signed the appropriate documentation on behalf of Hennepin County to allow PI access to the confidential case files . Judge Burke prescribed that the data be extracted from these files without personal identifiers and that it be reported only in aggr egate form. The case files were gathered by the H ennepin County Attorney’s Office and made available to the PI for the collection and analysis of descriptive data. Data were drawn directly from investigators’ lineup reports which provided information reg arding lineup structure and administration, lineup context ( e.g. , crime type), and eyewitness responses Double - blind sequential lineups 20 to the lineup task. 1 Incomplete reports were supplemented with data from the complete police file. Most i nvestigators continued their prior reporting procedure, wherein each wrote a narrative summary of the lineup process. The Minnetonka Police Department also developed a supplementary form for the witness that provided instructions , including the cautionary instruction that the perpetrator may or may not be in the collection of photos to be displayed . Minnetonka witnesses were requested to check one of two options — “I am unable to select any photo as the suspect in this case” or “I have selected photograph # ___ from the group” — and were offered space to write comments. Procedure HC employed single - suspect lineups , and all lineups involved each witness ’ s first identification attempt. The following five basic principles for blind sequential lineup s were employed , the first three part of police proced ures prior to the pilot program. ( Additional detail is provided in Table 2 .)  Effective lineup construction . A six - member lineup included one suspect and five fillers, with fillers chosen to match the witness’s description of the perpetrator.  Cautionary ( unbiased) instruction . The witness was instructed that the perpetrator may or may not be in the lineup.  Confidence statement . A statement of witness confidence, in the witness’s own words, was recorded at the time of the identification and before any fe edback.  Blind Administration . The lineup administrator did not know who the suspect was, and the witness was instructed that the administrator did not know which lineup member was the suspect. Double - blind sequential lineups 21  Sequential presentation . Lineup photos were presented one at a time . T he witness was not allowed to compare photos side - by - side at any time. The full sequence was completed even if an early identification was made, and the witness was informed that this completion is required by the procedure. E ach witness was in structed that he or she would view an entire photo series. The objective

of the procedure was to neither inhibit nor demand a witness response after each photo. T herefore, the protocol did not require a specific “yes” or “no” answer to each individual ph oto. The witness generated the pace through the photos, commenting or signaling readiness to view the next photo. Under this procedure it is entirely appropriate if a witness makes an identification before seeing all the photos; such a witness has made a positive identification without needing to see the entire array. However, t he protocol also requires that the witness view s all the photos even if a choice i s made before the lineup display i s completed. When a witness identified a photo during the pre sentation of the array, the officer documented this, and then reminded the witness that the procedure required all the photos to be shown , and the remaining photos were shown. If the witness had not made any identification by the end of the photo array, the officer then solicited a response, asking the witness if any of the lineup members had been familiar . 2 This practice allowed the benefits of a full lineup display and the blind recording of any witness response s to each lineup member. One benefit of this may be seen, for example, if identification of a filler occurred early in the lineup; the investigating detective will certainly want to know how the witness reacted to the suspect’s photo later in the lineup. Hennepin County adjusted the recommend ed protocol s o as to allow the witness to view the entire sequential lineup display as many times as desired. Review of the lineup Double - blind sequential lineups 22 occurred only at the witness’s request. If a review of the photos was requested by the witness, all photos were shown in the same order for the second “lap. ” Results Research question 1: Can the laboratory procedures be effectively implemented in the field? Hennepin County ’s situation was different from many jurisdictions that have implemented lineup reform . T he Cromedy case in New Jersey prompted reform, the Avery case in Wisconsin, the Ryan Commission in Illinois. I n Hennepin County, however, police chiefs could claim that existing procedures were working well. With the exception of one misidentification c aught before trial , there had been no apparent problems with lineups (Klobuchar & Knight, 2005) . N othing seemed to be broken . County Attorney Klobuchar reports that d iscussions and training sessions about the new protocol nevertheless convince d four depa rtment chiefs to move forward with a pilot test as a means to reduce the risk of false identification and to increase confidence in eyewitness evidence (Klobuchar, Steblay, & Caligiuri, 2006) . The question of i mplementation effectiveness i s largely infor med by qualitative and anecdotal information . Midway through the pilot, a group of thirteen investigators conveyed their concerns and impressions at a meeting with the researcher and a n attorney from the Hennepin County Attorney’s Office. Additional feed back from police agencies was fielded by the C ounty A ttorney’s O ffice. Interviews with police chiefs and law enforcement personnel were conducted by HC attorneys and are reported more fully in a recent article by Klobuchar, Steblay, and Caligiuri (2006). I ssues that surfaced through Double - blind sequential lineups 23 these meetings and interviews are summarized below. Th e introduction of double - blind administration and sequential presentation to lineup procedures initially posed problems, but proved to be less trouble s

ome than originally presumed . The final action taken after the pilo t — to roll out lineup reform county - wid e — is the strongest endorsement of the new procedures (Scoggin, 2005). Sequential Presentation . The problems experienced with implementation of sequential lineups were primarily in the simple details of orchestrating the new physical format of the photos. Prior to the pilot project, lineups consisted of “six - packs,” six photographs presented on a single sheet of paper . Filler pictures were selected from online photograph repositories of arrested persons by using culprit description parameters . It took some experimentation and innovation to determine how to best turn th e downloaded single sheet of paper into a viewable sequence of six individual photos. In vestigators had to separate the six pictures and enlarge them, while maintaining consistency of background and color. “A lot of people spent a lot of time at the Xerox machine,” remarked Bloomington’s police chief Laux (Klobuchar, et al., 2006, p. 406). Within a week, however, one department created a new photo template to remedy the problem. In an attempt to make the process more efficient, ongoing efforts are being made in the county to create software and adapt the online photo repositories to the requirements of the new protocol. The introduction of the sequential lineup presentation made apparent f or many investigators the strong desire of eyewitnesses to aid their lineup decisions by comparing ph otos side by side . Even after being instructed about the new lineup f ormat , witnesses still would ask to see two photographs simultaneously . Now realizing the increased Double - blind sequential lineups 24 potential for misidentifications when eyewitnesses engage in relative judgment, investigators say they better understand and appreciate the new protocol. One investigator explained, “I like the format better. I like that the person is studying one picture. It’s larger, and you can see them concentrating on a single picture and reflecting back to whatever event they had. From that alone, I think it’s a su ccess.” (p. 406). The larger size of photos in the sequential display was considered a collateral benefit of the sequential format. Ultimately, th e change to a sequential lineup format caused few problems, none of them serious or enduring. Although New Hope investigators were initially reluctant to implement the sequential presentation requirement, Chief Gary Link remarked, “it took maybe two or three lineups before they realized that it wasn’t that big of a deal.” (p. 406). Double - Blind Administration . Police chiefs and investigators were significantly more concerned about blind administrat ion of the lineup s . With a traditional lineup, an investigator conduct ed the array as part of the case investigation with no a dditional staff required. A number of operational challenges were anticipated with the new blind requirement and some problems emerged . In smaller departments with few investigators, all officers may be focused on a single case , so a blind administrator can be difficult to find. Witnesses l ocated at odd hours (middle of the night) or those in transient populations can make the coordination of a second investigator at the scene inconvenient. Most municipal departments operate independently and a lthough municipalities have mutual aid packs fo r patrol response, cooperative arrangements are rare for follow - up investigation s . Even in larger departments, difficulties surfaced . Some departments bring a near full force of officers to a Doub

le - blind sequential lineups 25 significant crime just after it happens. This “all - hands” pol icy constrains the pool of uninformed officers available to meet the requirement of a blind administration. A related circumstance occurs with investigations of great urgency ( e.g. , an Amber Alert) or very high profile crimes. A concern specific to Minnea polis police was the problem of chronic offenders, whose mere presence in a lineup would reveal the identity of the suspect to a blind administrator (Scoggin, 2005) . In H ennepin County , 34% of the pilot cases involved multiple witnesses . In these situatio ns , o ne witness’s identification of a lineup member can compromise the blind status of the administrator for the next witness. A separate blind administrator for each witness may be prohibitively cumbersome and expensive. Finally, s ince a blind administr ator must sometimes travel with the main case investigator to meet witnesses, the administrator loses time that would otherwise be spent working in a larger role on other cases . Investigators also worried that the r apport and trust that develop between a n investigator and witness may be jeopardized when a n officer unknown to the witness is introduced for the viewing of the lineup. The lack of prior relationship between investigator and witness also was anticipated to be troublesome when the case involved an uncooperative witness . Crime often takes place between not - so - innocent bystanders, thus witness motives may compromise any lineup procedure, sequential or simultaneous. The “no choice” category (and possibl y filler choices of witness responses ) in th is s tudy may hide some of these uncooperative witnesses. R eal problems with the blind administrator requirement were less serious than anticipated. To solve problem s related to limited availability of investigators to serve as blind administrators, smalle r departments turned to staff, such as patrol officers, captains, Double - blind sequential lineups 26 and sergeants. New Hope, the smallest department, reported no problems with the double - blind procedure despite employing only two investigators. Without greatly hindering collaboration, one larger department used property crime investigators as blind administrators for investigations dealing with crimes against persons and vice versa. The introduction of a second o fficer to a witness for lineup administration was not found to hinder inve stigations in any significant way. No witness refused to view the lineup with the new unknown officer. On the contrary, some witnesses reported appreciat ion of the procedure and an understand ing of the reasoning behind it. Investigators report ed no drop in effective suspect identifications or in their ability to “get the job done , ” as they compared their perception of blind sequential lineups to past simultaneous lineup outcomes (Klobuchar, et al, 2006, p. 408) . A gencies also reported positive effects on lineup results due to the new procedures. Investigators noticed that witnesses were now less likely to make a misidentification of a filler . “We’re not having a lot of people pick fillers,” Chief Rikala observed (p. 410) . To address concerns about re peat offenders and multiple w itnesses, as well as to explore options for investigators to deliver lineups for their own cases, the Minneapolis Police Department has been working with the Hennepin County Attorney’s Office to develop laptop computer lineup a dministration. Laptops can ran domly order six photos for viewing by a witness and be positioned across the table from the administering officer (dubbed the “Battleship” position, i n reference to the well - known table game) . In cases with

multiple witnesse s, the laptop will randomly shuffle the six pictures for each administration. This new procedure will allow in vestigators to conduct a lineup w ithout the aid of a second officer, dispelling lingering concerns about the double - blind procedure. As noted ea rlier, the Double - blind sequential lineups 27 NIJ guidelines were positioned as a “framework for innovation” (NIJ Guide, p. 9) in anticipation of technological developments such as computer - based imaging to add effectiveness and efficiency to lineup procedures. Sequential lineup presentati on as well as double - blind procedure can be facilitated through laptop delivery of the lineup, as researchers do in the lab . Although computer - based lineups were not a formal part of the HC p ilot , the lab technology may bec o me the solution to some impleme ntation issues. I nvestigators sometimes follow ed o ther jurisdictions in the c reat ion of functional equivalents to blind lineup administration for situations in which a true blind administrator is unavailable. The most common substitute method is the fold er (or envelope ) technique. S ix lineup photos are randomly placed into six numbered folders by a person uninvolved in the case, shuffled, and then provided to the witness by the investigating officer. The witness is asked to respond to each photo without displaying the photo to the officer. This helps to maintain a blind status of the officer. In Hennepin County’s neighbor ing Ramsey County , a wood - crafted 6 - door “lineup box” provides a similar functional equivalent when a second officer slides the rand omly positioned 6 - pack photos into the box and the lineup administrator is positioned in the room so as not to view the photos. The witness opens, examines, and closes each of the six doors, one at a time. Overall, police chiefs and investigators found th e pilot project to be much easier to implement than anticipated. In smaller jurisdictions, investigators had the whole project underway in less than two weeks. In the larger jurisdictions, the process took less than a month. Initial skepticism and uneas e faded. “By the end of the project,” Minneapolis Police Chief William McManus reported, “the burden on investigators was far less than my department had anticipated” (Klobuchar, et al., 2006, p. 409). The pilot project also Double - blind sequential lineups 28 involved minimal cost. From an administrative perspective, the police chiefs had wondered if the need for blind administrators would significantly increase man - hours. However, as Minnetonka Police Chief Joy Rikala noted, “There [are] no cost implications of this. It’s negligible . ” (p. 409). The double - blind sequential lineup procedure is well within the reach of police agencies. Yet - u nmeasured p ractical advantages of the new procedure are likely in the future. Better lineup screening devices allow police to move more quickly in th eir work, to find the right perpetrator without wasting time on false leads, and to remove perpetrators from the streets before additional offenses are committed. And, t he potential for damaging cross - examination at trial regarding cues consciously or in advertently sent to witnesses is effectively removed by the use of a blind administrator. Chief Rikala’s report of low er numbers of filler picks deserves comment. A low filler selection rate means that HC prosecutors and investigators can spend less time tracking down and clearing the filler to avoid defense challenges at trial. I t is also important to recognize that a witness who identifies a filler (perhaps when the true offe

nder is not in the lineup) is a “burned” witness , one who will not be seen as c redible for the continuing investigation or for court testimony. Investigators do not want to spend their witnesses with lineup procedures that may prompt filler selections . Finally, f iller identifications may imperil convictions even when the actual per petrator is ultimately identified, as the mistaken identification will undercut the correct one. The biggest hurdle in implementation seemed to be a general resistance to change . Thus, even fewer problems are expected the longer the protocol is used. N ew investigators will be trained in the reformed procedure, “It was simple to pick up.” (p. 410). HC has Double - blind sequential lineups 29 produced an effective DVD for training and review. A similar positive experience has been reported by Lt. Kenneth Patenaude (2006) of the Northampto n Police Department (MA), where his department ultimately came to s trongly prefer the new procedure. In 2003, the Northampton Police Department made an “additional modification to the policy, by making the blind administration of photo arrays the mandator y, rather than preferred, method of presentation.” (p. 418). Research Question 2 : Do the Number and Quality of Identi fications Change with the Double - Blind Sequential Procedure? T he sequential format is expected to lower eyewitness choosing rates. 3 Resea rchers believe this is due to the witness’s movement from relative to absolute judgment, a process that also may involve an upward criterion shift (Wells & Olson, 2003). The witness knows that there could be another person, a better match to memory, comi ng later in the sequence , and at any given time the witness does not know how many photos remain in the array (Lindsay & Wells, 1985) . Thus, he or she is forced to dig a bit more deeply into memory. If a match between memory and a photo is compelling , a choice from the lineup is likely ; otherwise, the witness should reject the lineup. Guessing — lucky or otherwis e — should be reduced. The sequential procedure is seemingly a more conservative test of memory, and, if this technique is working well, a low rat e of filler selections is likely. In the best of situations, the blind sequential lineup would maintain a reasonable level of accurate identifications, but an appreciable drop in false identifications as guessing is reduced. One of three outcomes will o ccur as the endpoint of a witness’s decision for a police lineup in which a single suspect is present: a suspect identification of the person who Double - blind sequential lineups 30 the police believe is the offender ; a filler (known error) selection; or a rejection of the lineup (no choice from the lineup). W itness performance in the field is measured , then, through percentage rates of suspect identifications, filler selections, and no choice responses . In the field, however, we cannot assess ground truth — memory accuracy or error — so suspe ct identifications and filler selections become proxy measures . Unfortunately, t hese measures alone cannot tell us much about memory accuracy. A t the simplest level, a new lineup procedure that produce s a plummet ing drop in suspect identifications, a peri lous rise in filler selections , or overwhelming c onfusion in witnesses , will and should raise eyebrows and concerns . Such outcomes signal that a procedure may not be conducive to tapping witness memory or perhaps that earlier rates of field eyewitness per formance were dramatically misleading. Simila rly, a sharp rise in suspect

identification rates should give pause ; while this outcome may indicate a better lineup procedure , it may al ternatively indicate l ineups with very poor fillers. A mbiguity is inheren t in field measures, so an absolute level of goodness is not established. I n a subsequent section of this report , I will return to the issue of how field data might best be interpreted . Of importance at present is a cautious and nuanced understanding of f iller identification s. Filler selections in the field, as in the lab , are counted as eyewitness mistakes. Filler selections are not dangerous errors (they typically do not lead to a prosecution of the filler) , but the rate of filler selections, under t he right conditions, can provide information about the risk for an innocent suspect in the lineup. Wells (2006b) has termed filler identifications “conditional proxy measures” that can work well under some conditions but not others. The key condition fo r use of filler selections as an indication of Double - blind sequential lineups 31 the risk of false identification (not dangerous false identification itself ) is present in this Hennepin County pilot: Administrator masking allows filler identifications to work as a conditional proxy measure . Assessment of specific change in lineup outcomes resulting from use of the new lineup procedure would require suitable data from concurrent blind simultaneous field lineups against which to compare blind sequential lineup performance. There are no such comparative baseline data for blind simultaneous lineups in Hennepin County or elsewhere . However, existing sources of relevant information from other venues can be examined to provide a picture, albeit limited, of (presumably non - blind) simultaneous line up field performance . The reader should keep in mind that the HC field data include some witnesses who elected multiple laps through the lineup and some lineups in which the perpetrator was familiar to the witness due to interactions prior to the crime. Also, i n any field test, suspect identifications may include some unknown proportion of false identifications. Comparative f ield d ata from s imultaneous l ineups . Behrman and Davey (2001) , in a Cal ifornia field study , reported eyewitness decisions for fift y - eight simultaneous live lineups, with a suspect identification rate of 50%, filler choice of 24%, and no choice at 26%. In 284 photo arrays, 48% of the identifications were of the suspect; foil identifications were not recorded. Behrman and Richards (2 005) later report ed archival police data (a portion of the cases formed part of the database for the earlier 2001 study) for 424 photographic lineups and 37 physical lineups. These were single - suspect simultaneous lineups and all were first identification attempts by the witnesses (none were preceded by Double - blind sequential lineups 32 identification procedures involving the same suspect). Fifty - two percent identified the su spect ; 15% identified a foil. Three teams of researchers from England (Valentine , Pickering & Darling , 2003 ) re corded the responses of over 30 0 0 eyewitnesses to simultaneous live lineups, including situations of suspects known and unknown to witnesses. As summarized by Valentine et al., results of these studies are quite consistent: approximately 40% of witnesses identified the suspect; approximately 20% identified a f iller, thereby making a known mistaken identification; and approximately 40% made no identification . Comparative l aboratory d ata from s imultaneous and s equential l ineups . Steblay and c

olleagues (2001 ) summarized thirty laboratory comparisons of simultaneous and sequential lineup performance. This review, representing 4145 test witnesses, demonstrated that the sequential procedure reduced eyewitness choosing rates, with significant positive effects on accuracy (see Table 3 ). Regardless of whether the criminal was in the lineup ( perpetrator - present condition) or not ( perpetrator - absent ), the simultaneous format produced a n almost even split between filler choices and no choice responses. There were si gnificantly more errors in simultaneous versus sequential lineups when the perpetrator was not in the lineup, 51% versus 28% filler selection rate , respectively . Witness decisions . For comparison purposes, the Hennepin County data (HC) is presented in th e upper section of Table 3 alongside simultaneous (SIM) field data sets . Hennepin County generated a suspect identification rate of 54% , filler selections at 8%, and 38% of witnesses made no choice from the lineup. The 54% suspect identification rate is comparable to that achieved with simultaneous lineups in the field and higher than Double - blind sequential lineups 33 laboratory sequential rates, and the filler selection rate (8%) much lower than other procedures . These rates change somewhat as the context of the crime shifts, particula rly the relationship between witness and perpetrator, as discussed below. Witness performance on sequential repetitions (“laps”) . Ideal blind sequential protocol calls for only one viewing of the lineup per witness . Researchers reason that a second la p through the sequential lineup may produce a de facto simultaneous array. That is, subjects may begin to mentally compare photos and lapse into relative judgment. However, HC brought a practical worry to the considerations – that some number of good ide ntifications would be lost due to overly cautious eyewitnesses . A revision to the procedure was determined: witnesses in the pilot project were allowed multiple repetitions, or laps, through the lineup, and lineup administrators were to record the details of this process. The details were not as thoroughly recorded as hoped, with 46% of lineup reports specifically indicating the number of repetitions. The results in Table 4 are based on these 128 lineups, summarizing witness decisions for those who viewe d a lineup just once , twice , three times , or more . Repeated viewing of the lineup was associated with significantly increased likelihood of filler choices ( known errors), X 2 (10) = 31.23 , p .001 , r s = .13 . 4 For crimes involving perpetrators familiar to the witness, this is somewhat less evident. With familiar perpetrators, the suspect identification rate was 92% for witnesses who took either one or two laps through the lineup, and 50% for the few witnesses who requested a third viewing, X 2 (4) = 7.34 , p = .06 . For lineups in which the perpetrator was a stranger to the eyewitness, a risk of additional laps was most apparent in the filler selection rate, which increased from 3% to 29%, X 2 (10) = 22.00 , p = .007 . Double - blind sequential lineups 34 Returning to the subset of 128 lineups, over half of the witnesses (53%) viewed the lineup just once. Another way to look at the impact of repeated viewing is to examine witness decisions in the remaining 47% of lineups, 60 lineups in which witnesses requested additional laps. In 15 of 36 lin eups in which complete information was available, witnesses indicated recognition of a photo during the first showing of the lineup, with nine jump - outs ( i.e. , the witness made an

immediate choice or made comments such as “that’s the guy,”) and six tentati ve identifications. The tentative identifications included comments such as “hold that one” and “that looks like him.” The second lap (for two witnesses, a third lap) was the point at which the witness confirmed his or her choice; 100% , all 15 , selected the suspect. Although the additional 21 witnesses selected a photo from the lineup, they did not indicate recognition until at least after the second showing of the lineup. In these cases, climbing error levels are apparent. Thirteen identifications made after two lineup showings produced 62% suspect and 38% filler choices; eight witnesses made a decision after three or more showings and generated more filler identifications (50%) than suspect identifications (38%). In summary, the evidence shows that a witness who has not made even a tentative identification at the end of the first lineup viewing presents a subsequent risk for misidentification. “ Jump - out” identifications . A concern of investigators was that sequential lineups might diminish the like lihood of “jump - out” identification s , inhibiting desirable witness expressions of absolute certainty. For this analysis, jump - outs were taken to be those lineups for which the investigator recorded an “immediate choice” , and/or witness comments such as “th at’s the guy!” Of 175 choosers in th e data set (those who actually Double - blind sequential lineups 35 selected a photo from the lineup), ninety - six (55%) were jump - outs. The resulting choices produced 99% suspect identifications and only one filler choice. The “immediate choice” criterion is similar to that used by Behrman and Richards (2005) in their field research as a signal of automatic processing versus eliminative memory strategies of witnesses. Six percent of their witnesses were reported by police to use such an automatic memory r esponse. In this HC data, we use a broader criterion , including not just officer interpretation but also strong positive witness responses , and 34% of witnesses fall into this “jump - out” category. Stranger vs. familiar perpetrators . The Hennepin Co unty program required blind sequential lineups for all felony cases, regardless of familiarity between eyewitness and perpetrator. In 66% of the lineups, witnesses reported that the offender was a stranger. The following analyses explore eyewitness respo nse as a function of familiarity and of the witness’s opportunity to view the culprit. First is a comparison between situations in which the perpetrator was a stranger to the witness, based on the best knowledge of the case investigator, and those in whic h the culprit was at least familiar to the witness and sometimes known quite well. Not surprisingly, suspect identification rates were significantly lower for stranger crimes (35% as opposed to 90% for known perpetrators), X 2 (2) = 74.68 , p .001 , with l ower choosing rates (47% versus 94% for known perpetrators). Filler rates were low in both categories (see top frame of Table 5) . The lower suspect identification rate in the stranger set of lineups may elicit concern from investigators. A finer dist inction is perhaps useful (Table 6 ). Crimes of brief duration committed by strangers (estimated as only a few minutes) produced 32% suspect identifications and 11% filler choices. Crimes in which the witness viewed a stranger for a Double - blind sequential lineups 36 longer time (more than ten minutes) generated a 59% suspect identification and 14% filler choice rate. We do not know the accuracy of suspect identifications, of course . T

hese rates suggest that witnesses are more willing to choose from the lineup when they have had longer ex posure to the culprit. Lower suspect identification rates occur in situations where one might expect weak memory. An interesting ancillary finding is that 43% of the witnesses who made a lineup selection after viewing a very brief stranger crime were rep orted to have expressed some qualifications to their identification (see table 6 ). A similar breakdown of the familiar (known) perpetrator category also shows an outcome which is at least intuitively coherent . Witnesses who reported some familiarity with the perpetrator ( e.g. , a familiar face seen on multiple prior occasions) chose from the lineup at a very high rate (91%), selecting the suspect in 84% of the lineups recorded, fillers in only 7%, and making no choice in 9% of the lineups. And , those witne sses who knew the perpetrator , often by a street name , made suspect identifications at 96% with 4% m aking no lineup choice. It should be noted that, in this latter group, persons making lineup choices included not only observers and victims of crim e; 25% of the witnesses were those involved through indirect knowledge of the crime. Thus, this category included “confirmatory” lineups that are not typically part of laboratory studies. Given the many factors that co - occur in each of the four viewing categorie s, it is risky to infer cause and effect between any set of them . Table 6 describe s the correlates of these eyewitness experiences . The patterns of eyewitness response make sense, given the context for the identification. For example, brief interaction with a stranger produces the lowest choosing rates, jump - outs, and suspect identifications of the four groups. Also, Double - blind sequential lineups 37 greater levels of qualifiers to the choice occur with stranger crimes than with familiar perpetrators. Witness c onfidence and decision out comes . Lineup administrators were asked to record verbatim any eyewitness comments regarding confidence. However, this requirement led to specific comments in only 15% of lineup reports ( n = 42 ). Assuming that jump - out identifications also indicate a mea ningful level of certainty and can therefore be added to the analysis, a total of 125 lineups afford information regarding witness confidence (45% of the overall lineups, and 71% of lineups in which the witness made a lineup selection ). Witnesses expresse d confidence in a decision not to choose from the lineup in only two of the 125 cases (1%). For analysis, the memorialized comments were sorted into four categories of decreasing certainty: (1) jump - outs ; (2) high confidence — those not included as jump - out s but involving statements of 80%+ certainty or phrasing such as “quite certain,” “sure,” and “positive”; (3) moderate confidence , involving “pretty sure,” “fairly sure” and estimates between 50% and 80%; and (4) low confidence , with such phrasing as “not sure,” “not very,” or “not too” and “low.” Within this subset of lineup data, witness confidence and decision outcomes were significantly related, with greater confidence associated with higher levels of suspect identifications, X 2 (9) = 83.73 , p = .0001 , and r s = .74 . This significant statistical relationship was largely due to the impact of jump - out decisions, 99% of which were suspect identifications. If the jump - out category is removed from analysis (leaving n = 26 ), a significant relationship no long er appears. In each of the remaining three confidence categories (high, moderate, and low), witnesses selected fillers at a slightly higher rate than suspects (filler identification rates of 58%, 67%, and 63%, respectively), r s = .08 .

Double - blind sequential lineups 38 A similar relationsh ip between confidence and decision outcome exists when only stranger identifications are examined: X 2 (9) = 31.08, p = .0001 and r s = .63 , when jump - outs are included ( n = 50); r s = - .07 when jump - outs are eliminated from the analysis ( n = 24). Again, fi ller rates are higher than suspect ID rates in the three remaining confidence categories: 75%, 60%, and 63%, respectively. Confidence and witness decisions are significantly related in this sample when jump - out identifications are included in the analysi s . Beyond jump - out decisions, however, even witnesses who express high confidence in their decision choose a filler more often than the suspect. Three caveats to these results are important. First, the data do not directly address the confidence - accurac y relationship, as ground truth (actual guilt or innocence) is impossible to ascertain in this archival field data. Second, only a very small number of lineup reports included a clear statement of witness confidence, therefore this sample may not well rep resent the broader data set. Finally, it is difficult in some cases (particularly the “jump - outs”) to determine whether the confidence rating is directly that of the witness or is the officer’s assessment of witness confidence. Given these difficulties, no further data regarding between confidence and other witness behaviors are reported. Cross - rac e and same - race identifications . The research literature has established a reliable same - race identification advantage such that faces of one’s own race are better remembered when compared wit h faces of another, less familiar race (Meissner & Brigham, 2001) . The field data here do not offer clear information, particularly because accuracy of identification is impossible to ascertain in the field. In additio n, while 96% of lineup reports ascribed the race of the offender, only 44% of reports included mention of the witness’s race. Finally, it is appropriate to eliminate from analysis the lineups that Double - blind sequential lineups 39 involved perpetrators who were well known to the witness. The remaining data leave a small er subset of 109 lineups. In this group , cross - race identification decisions resulted in 53% suspect choices, and 11% filler identifications (a 64% choosing rate); same - race decisions generated a similar level of suspect c hoices, 55%, and 4% filler identifications (59% choosing rate). The 7% difference in known errors (fillers) is not significantly different, z = 1.40, p = .08 . S ummary of data analysis Hennepin County double - blind sequential field tests produced accept able suspect identification rates relatively comparable to those in prior laboratory and field tests. Repeated viewing of the lineup was associated with increased filler selections ( known errors). The new procedure do es not appear to have sacrificed jump - out identifications. Patterns of eyewitness response to stranger and familiar perpetrators were reasonable, with stranger suspect identifications at a lower level. Confidence and suspect identifications were significantly related, particularly for jump - out identifications. For other categories of expressed confidence (even high ), confidence and decision outcome were not significantly related. A positive outcome of the project was the low filler identification rate, which suggests increased protection for innocent suspects. However, additional attention to the fillers is necessary. Could there be an alternative explanation for the HC results : poor fillers ? Functional size of the lineu p — the number of lineup members who effectively match the de

scrip tion of the perpetrato r — is an important consideration for all lineups. Lineup fairness depends on adequate lineup construction as well as appropriate administration procedures, and lineups are vulnerable to bias in both structure and Double - blind sequential lineups 40 procedure. A faulty lineup structure prior to the blind - sequential procedure may jeopardize the usefulness of this new protocol. A blind - sequential protocol should lower the filler selection rate by inhibiting witness guessing , thereby reducing eyewitness error and danger f or the innocent suspect . The HC data sug gest that th i s objective was achieved. However, w hen the HC results were released, critics reasonably argued that the low filler selection rate (8%) could be due to poor lineup construction, i.e., perhaps the res ults are due not exclusively to the merits of the double - blind sequential procedure, but to poor construction of the lineups prior to the identification process . A low filler selection rate could occur if the lineup is too “easy”: fillers are not plausible options , so eyewitnesses can arrive at the suspect through inference rather than memory. Taking this concern very seriously , we developed the means to analyze quality of the lineup structure through a laboratory mock witness procedure (Doob & Kirshenbau m, 1973) applied to a sample of the HC lineups. Study 2: Evaluation of H ennepin C ounty lineup filler quality Assessing lineup fairness: The mock witness method The evaluation of lineup fairness involves two related qualities: lineup bias and lineup s ize (Malpass & Lindsay, 1999). F irst, the lineup must not be biased against the suspect, i.e., the suspect should not be identifiable on the basis of position in the lineup, photo quality, or other non - memory cue. Second, all members of the lineup must be reasonably plausible suspects based on the description provided by the witness. Lineup evaluation is accomplished, then, by determining whether the suspect stands out in the lineup in a manner not attributable to the witness’s memory of the event (bias), and Double - blind sequential lineups 41 whether lineup fillers provide effective alternatives for the unreliable witness (size). This assessment is accomplished using the mock witness method . The mock witness paradigm at first appears somewhat counterintuitive: Persons who did not wi tness the crime -- in fact have no knowledge of it – view the lineup and are asked to posit who the suspect is in the array (“Which person is the accused?” ) . The mock witness is provided one clue : the real witness’s pre - lineup verbal description of the pe rpetrator. Armed with this information, the mock witness is hazarding a best guess based not on memory, but on inference derived from the physical description of the perpetrator. As recommended by Wells and Bradfield (1999) , use of the specific form of t he question “Which person is the accused?” instead of “Which person best fits the description?” also serves to capture the mock witness’s use of any other non - memory information also available to the eyewitness (e.g., subtle aspects of lineup display that make the suspect stand out.) The reasoning from this point is straightforward in regard to determination of bias. The real witness’s identification is interpreted in the context of mock witnesses’ ability to identify the suspect in the lineup. That is , the mock witnesses serve as a type of control condition against which to compare the actual witness’s lineup choice (Wells, Luus, & Windschitl, 1994). If mock witnesses show preference for the suspect despit

e their absence of any memory of the crime, t he real witness’s selection of the suspect may also be due to non - memory factors rather than true recognition ; a ny conclusion about accuracy of the real witness is confounded. As Valentine and Heaton (1999) explain , “A mock witness simulates the worst pos sible scenario of a witness who has no memory of the culprit beyond that which they gave in a verbal description…” (p. 61). A maximally fair lineup Double - blind sequential lineups 42 would limit the damage of such a witness in a real identification task, in that all of the lineup members w ould fit the witness’s description to the same extent; therefore, the suspect would be at no greater risk of identification than any one filler. Method Sample 104 college students each evaluated a sample of 37 lineups . The study was approved by Augsb urg’s Institutional Review Board and by the Hennepin County Court. Participants were recruited with posters and through the campus research participant pool and compensated with class credit or cash. P articipants were required to sign a consent form and treated in accordance with APA ethical principles. Lineup bias is not a property of a lineup per se, but rather a result of the interaction between the lineup and the verbal description provided by a specific witness (Wells & Bradfield, 1999) . In order t o use the mock witness protocol, it was necessary to find lineups in which the police report included the witness’s prior verbal description of the perpetrator. Only one lineup per case was allowed for this analysis, reducing the available 280 HC blind - se quential lineups to 117. A sample of 50 lineups was requested and 52 provided by HC. Of these, 37 were appropriate for this research task. Lineups were eliminated from the sample of 52 for the following reasons: 1) the police report failed to indicate w hich lineup member was the suspect; 2) the perpetrator was familiar (known) to the witness, so a physical description was not provided; 3) a non - witness source implicated the suspect (e.g., license plate, weapon, citizen tip); or 4) it was not clear as to which eyewitness provided the description used to construct the lineup. Procedure Double - blind sequential lineups 43 The testing was done by two experimenters . The first was blind to which lineup member was the suspect; the second experimenter (the PI) was not blind, but left the subject ’s cubicle during the rating of the photos. Neither experiment er interact ed with the subject during the rating task. Presentation order of the 37 lineups was counterbalanced across subjects. The participants were instructed as follows: In this packet is a series of real police lineups. Connected to each one is a description of the perpetrator that was provided by the real witness to the crime. Some of these descriptions are very brief, others provide more detailed information. The entire description provided by the witness is given, even if some features cannot be seen in the photo. For example, an attribute of the perpetrator’s body, not visible in the picture, may be part of the witness’s description. Also, just as would a real witness, you ca n assume that clothing and sometimes appearance will have changed between the witnessed event and the time of the lineup. Your task for each lineup is to read the description, view the lineup, and make a choice, as best you can, as to which of the lineup m embers you think is the accused . That is, who do you think the suspect is? Results and Discussion To evaluate the quality of HC lineups, three statistics were calculated: the proportion of mock

witnesses who choose the suspect, functional size of the lineup, and Double - blind sequential lineups 44 effective size of the lineup (see Brigham, Meissner, & Wasserman , 1999 , for a useful discussion of the strengths and weaknesses of each index of lineup fairness). Average proportion of mock witnesses who choose the suspect . Lineup bias is most simply calculated as the proportion of mock witnesses who choose the suspect (Doob & Kirshenbaum, 1973) . This proportions technique h as been noted to be the most useful estimate of lineup bias (Brigham, et al., 1999) . Also, the proportion of witness es who choose the suspect is a significant predictor of the proportion of witnesses who identify the lineup member in a perpetrator - absent lineup (Lindsay, Smith, & Pryke, 1999). The median proportion of mock witnesses who chose the subject was 22% , a no n - significant difference from the rate of chance at 17% . A compliment to this figure is more relevant to the current question of whether fillers were effective in drawing mock witness responses : An average 78% of mock witnesses chose a lineup member othe r than the suspect. All 6 photos drew choices (median across the 37 lineups = 6). Average functional size of the lineup . Because lineup size directly contributes to fairness, a supplement t o this proportion index of lineup fairness is a measure of fu nctional size . The number of members appearing in the lineup regardless of their physical appearance or fit to witness description is referred to as nominal size . Functional size , on the other hand, is the number of lineup members that are physically sim ilar to the witness’s description of the perpetrator (Wells & Turtle, 1986). Wells and Bradfield (1999) explain that the lineup task must control for simple recall of the description by ensuring that all lineup members fit the eyewitness’s verbal descripti on equally well Functional size represents the number of plausible choices for the witness – the suspect and some number of viable alternatives. Double - blind sequential lineups 45 The lineup’s functional size was calculated as the reciprocal of the proportion of mock witnesses choosing th e suspec t, 4.54 (Wells, Leippe, & Ostrom, 1980) . This also corresponds to the median f unctional size across all 37 lineups. Average effective size of the lineup . Effective size is an index of the number of plausible lineup members, without sp ecif ic reference to the suspect (Malpass, 1981). This measure is useful as an indicator of the spread of choices across the lineup (operable alternatives). Median e ffective size of th e s e lineup s was 3. 96 . In essence, the calculation reduces the nominal siz e of the lineup for each filler photo that fails to draw its proportionate level of identifications. The specific calculation is offered by Valentine and Heaton (1999). In summary, t he fillers were found to be functioning well, drawing mock witness resp onses in appropriate p roportions. Brigham, et al. , (1999) contend that a lineup should contain a minimum of three viable alternatives to the suspect in a 6 - person lineup, a criterion met by the statistics of this HC sample. Importantly, w itnesses did not avoid the fillers, thus an alternative interpretation for low filler rat e — in effective filler s — can be ruled out in the H ennepin County study . Discussion : Hennepin County Pilot Pro gram Overall Summary : What has been gained? The Hennepin County pilo t project has been judged a successful application of double - blind sequential lineups in

the field . Blind sequential procedures are no w county - wide , and adjacent Ramsey County has rolled out the new protocol . The new procedure Double - blind sequential lineups 46 has established a standardi zed scientifically - based practice that has been demonstrated to work in real street investigations. The key objective of the new blind sequential procedure i s to secure better quality identifications based on what eyewitnesses actually remember. How do we know these HC lineup results are “better”? In the field, w e can not know identification accuracy in a hard factual manner . Instead we must look to procedures that have been demonstrated to protect witness memory and therein engender accuracy. What we do know are the following: (1) Laboratory evidence provides a compelling foundation for belief in the superiority of sequential over simultaneous format as a means to acquire a more trustworthy rendition of eyewitness memory and thereby reduc e the risk o f misidentification . (2) T he double - blind procedure facilitates the witness’s independent judgment and the investigator’s objective documentation of th at judgment. (3) Hennepin County i nvestigators perceived no drop in suspect identification s or in the ir ability to do their jobs . T he preservation of an effective suspect identification rate with in the safeguards of the blind - sequential procedure is a very positive development. And, (4) H ennepin County lineups provid ed appropriate risk management for inno cent suspects through effective blind sequential procedure . This is evidenced by the low filler rate , a conclusion reinforced by the results of the mock witness test of HC lineup fillers . The new lineup t echnique reduce s concerns that administrator influ ence and relative judgment ha ve affected witness decision s . Because of these safeguards, Hennepin County prosecutors and investigators anticipate benefits in court. Most officers recognize the potential for harsh cross - examination should lineup evidence g o to trial . Properly conducted, t he blind sequential lineup can be expected to better withstand challenges to the Double - blind sequential lineups 47 validity of identification evidence. The County Attorney’s Office notes: “ The procedures give us more information about what an eyewitness really means when an identification is made and, in the end, a clearer view of the truth” (Scoggin, 2005, p. 5). The Hennepin County pilot project also brought the first available information regarding the effect of repeated viewing of the lineup. One can extrapolate from theory and existing laboratory data to surmise that a repeated sequential lineup will ease the witness into relative judgment, with a predictable reduction in performance accuracy. Now, Hennepin County has provided field data about t he effects of lineup repetition: identifications are likely to be more reliable when the witness has made a decision after a single lap. While it is impossible to know the extent to which suspect identifications include false identification s of innocent pe rsons, repeated lineup laps are shown to be associated with increased likelihood of error in the form of filler selections . Interestingly, witnesses who made an initial comment of recognition during the first lap were highly likely to identify the suspect in the second lap. Perhaps these are cases of witnesses who have reasonably strong memories of the perpetrator, but simply need to be sure with a second reinforcing look at the lineup . It remains important that lineup administrators record procedural d

e tails and witness comments carefully and methodically, particularly if a witness opts for a repeated viewing of the lineup. Laboratory studies of blind - sequential lineups typically deal with stranger crimes of short duration, which might be considered the most difficult test of memory. The subset of Hennepin County data that involved stranger crimes provides evidence that laboratory principles generalize well and productively to the field. In addition, the Hennepin County data examin ed eyewitness response s to lineup s with familiar perpetrator s . Patterns of Double - blind sequential lineups 48 eyewitness response to stranger versus familiar perpetrators square well with what is known regarding eyewitness memory, with stranger suspect identifications at a lower level. The Hennepin County data also indicate that blind sequential lineups work well in situations of familiar perpetrators and confirmatory lineups. Given that even a confirmatory lineup is still a test of memory (and the investigating officer is unlikely to know just how familiar th e perpetrator is to the witness), sequential lineups provide an appropriate protocol. The purpose of the project was to determine how recommended lineup procedures can best be brought into practice. The experience of the pilot project indicates that the b lind sequential protocol is workable for police in both large and small departments without undercutting the ability to solve cases. At the same time, the protocol elicits valuable new information for the effective investigation and prosecution of crimina l cases. Lessons from the field: Implications for policy and practice The Hennepin County experience offer s specific suggestions for successful field use of the new lineup protocol beyond the recommended components of the double - blind sequential techni que itself. Many of these recommendations have previously been put forth by other researchers and policy - makers, but deserve to be underscored as lineup reform efforts move ahead. (1) B ackground information: T ransparent and close adherence to the under lying operating assumptions of double - blind sequential protocol . O perating principles for effective lineup protocol include : a single - suspect lineup with properly chosen fillers ; isolation of each witness from the potential influence of other s; and an ide ntification task that is the witness’s first view of the apprehended suspect . These components of the lineup Double - blind sequential lineups 49 process occur prior to the lineup proper, demand ing early documentation of the witness’s description of the offender along with a detailed summar y of lineup preparation and all interactions with the witness preceding the lineup. From that point, m emorializing the lineup itself is essential, as recommended by the NIJ Guide. Particularly i n jurisdictions where the witness may have attempted an earli er identification, recording of the witness’s history in the investigation is crucial – including specific results from each identification task and the suspect’s history of appearance in each of the earlier identification arrays. The witness’s relationsh ip with the offender -- stranger or familiar – should be recorded as well . (2) Productive information gathering from the witness during the lineup procedure . P rocedure and instructions to the witness a r e tightly prescribed in the new protocol , and the bl ind lineup administrator is not to volunteer information beyond the protocol . However, the interaction between the blind administrator and the witness is not intended to be rigid or the conversation stifle

d . Th e lineup interview is an opportunity to lear n from the witness, and a conversational flow focused on the witness ’s experience and memory is desirable. T he bl ind - sequential interview may allow for but not demand comments or q ualifiers from the witness for each photo . “ Y e s ,” “ No , ” “ I ’m not sure , ” “ n ex t picture ” and such phrases are all reasonable responses from the witness as a means for him or her to advance the lineup to the next photo. A forced “yes” or “no” may push the witness into a response that does not truly capture his or her position. ( 3 ) E ffective r ecord ing of witness response . The lineup administrator should keep a careful record of verbatim witness comments about all photos ( preferably with videotape supplement ) , including qualifiers to the identification and comments about photo s not chosen . The witness should be asked about his or her certainty of identification immediately Double - blind sequential lineups 50 when a photo is chosen, or of non - identification at the end of the array. W itness comment s should be recorded as stated (“I’m positive”) rather than as the o fficer ’s assessment of the witness ’s position (“He was positive”). Alternatively, the witness may write a statement in his or her own words. Clarification and precise recording of witness r esponses is critical . For example, “ the witness failed to identi fy the suspect” is ambiguous; did the witness identify a filler (which one?), indicate that the culprit is not in the lineup ( “ he’s not there” ) or express indecision ( “ I don’t know” ) ? Filler identification s should be noted , including multiple identificati ons, as th e s e may be d iagnostic regarding the quality of the lineup and the identity of the suspect as perpetrator . Just a s with a witness, feedback, forgetting and post - event information can intrude upon the administrator’s memory of the lineup interactio n. T he blind administrator should document the results before conversation or feedback from the investigator , and d elayed recording of the lineup should be avoided . ( 4 ) Standardize d training, procedure, and reporting . Until the procedure becomes habit for all officers, some mechanism for memory prompts is likely to be necessary. The latter could be achieved with a scripted reporting form and perhaps a simple laminated wallet card. Training should include a brief but clear explanation of the underlyi ng rationale for the double - blind and sequential components. Understanding of basic principles of memory and the potential benefits to police investigations is likely to ease acceptance of the new protocol in the field and to spur the enthusiasm and creati vity essential to genera ting solutions to the specific operational challenges of each local jurisdiction. In particular, investigators must be assured that the new procedure is not an assault on their professional integrity or a hindrance to securing tim ely and accurate identifications . Double - blind sequential lineups 51 The core rationale for these information - gathering and documentation requirements is illustrated in the HC written protocol for investigators: “Sequential identification and the following documentation requirements do a better job of conveying what a witness really means. In some ways this is Dr. Wells’ real point. The witness didn’t mean this is the guy – he meant this might be the guy, the photo looks a lot like him. We’re learning that eyewitness identification is n’t all or

nothing. Identification takes place on a sliding scale. Some identifications “jump out” – the witness is absolutely certain …who committed the crime. Others are less sure or place conditions on the identification (“it’s him except he’s older” – thinner, or has different hair). The procedure…encourages the investigator to document what those judgments are. This gives the rest of us (prosecutors, judges, and jurors) a clearer view of the truth – and much more confidence in the result.” (Scoggin , 2005, p. 4 - 5) (5) C aution with repeated viewings of the lineup . A second lap should occur only at the witness’s request and the witness should not be told beforehand that a second viewing is an option. T he lineup should end with a maximum of two repet itions and a record kept of the eyewitness’s response to each viewing of the lineup. Given the HC data, it must be presumed that an identification made on the second viewing, particularly with no recognition of a photo on the first lap, incorporates some increased level of relative judgment. ( 6 ) Develop ment of functional equivalents . F or situations in which a blind administrator is unavailable , the practical need for emergency substitutions should be met with convenient but valid alternatives. It is desirable that researchers follow up in the lab to establish the effectiv eness of proposed equivalents. Double - blind sequential lineups 52 Lessons from the field: The interpretation of field studies As lineup reform continues, it is important that additional evaluative field data be col lected and reviewed. Both s cientists and non - scientists then will attempt to draw conclusions about the meaning of field data. T he following points a re offered to guide this potentially confusing endeavor. Interpretation of the HC data illustrates thre e imp ortant considerations regarding lineup field studies. First, it is essential to be clear about what is being measured; this is particularly true when simple aggregate figures will represent the study to policy - makers and laypersons. Second, field mea sures are imperfect . T here are no established absolute levels of “goodness , ” and the measures by themselves tell us very little about eyewitness memory accuracy. Instead, l ineup outcomes must be evaluated within the context of the study design and the ty pe of data that contributed to the statistical outcome , as well as estimated gains or losses in accuracy that are likely from the procedures employed. Finally, good field practice and tests must e mploy procedures that rule out alternative (non - memory) in fluences on eye witness decisions . Be very clear about what is measured . Any hope of comprehending field study results will fade quickly if researcher s, field administrators , and intended audience have not shared a specified and detailed protocol for lineu p procedure. Prior to lineup reforms, investigators have long - seated habits for conducting lineups, sometimes following a written departmental procedure, often without. It is important to ascertain w hich conventions of the status quo are intentionally o r inadvertently transferred to the new lineup routine. HC had the benefit of established procedures that fed easily into lineup reform, e.g., one - suspect lineups of uniform size, lineup construction using a statewide photo repository, and Double - blind sequential lineups 53 a policy of usin g only single identification attempts . A n effective scripting of the new protocol pulled together previous and new procedures into a coherent package . An example from other jurisdictions that sh

ows a more problematic scenario is the evidentiary requiremen t in New York and Illinois that demands a live lineup identification by the eyewitness. The practice that flows from this is that the eyewitness will have first successfully select ed the suspect from a photo lineup. This introduces the possibility that t he witness’s recognition of the suspect in the live lineup stems from the previous procedure rather than the crime or that the witness can infer the suspect’s identity due to his presence as a common denominator across two lineup tasks. Furthermore, unsu ccessful witnesses will have been screened out in earlier tasks, distorting overall identification rates. F iller selection rates can be expected to be tamped down and suspect identification rates elevated for reasons u nrelated to memory accuracy . Behrman and Davey’s 2001 documentation of traditional simultaneous field lineups in California showed that the practice of using a second identification task increased suspect ID rates 17% for the later identification. Interpretation of d ata from a recent pilot project in Illinois (Mecklenburg, 2006) was problematic due to this and o ther problems with a non - sta ndardized protocol (see e.g., Diamond, 2006; Doyle, Penrod, Kovera, Bull, & Dysart, 2006; O’Toole, 2006; Steblay, 2006; Well s , 2006b). At the same time th at clarity of initial protocol is essential , it is also critical to pay close a ttention to subsequent deviation from prescribed protocol . Blind - se quential procedure is a package of components that together form best practice. It is important to track ho w these components are implemented in the field , how revisions to the recommended protocol are employed , and their impact on the outcome data . One example Double - blind sequential lineups 54 is the use of repeated laps in the sequential procedure. In HC, a witness’s additional viewings of the sequential lineup generated more known errors (fillers). Would filler rates decline if witnesses were held to a single viewing of the lineup? Now we know: yes. Some variability in lineup protocol is not a problem, particularly as it may represent na tural field practice and if it is well documented . The challenge is to evaluate lineup outcomes when procedural changes violate prescribed pr otocol or when they confound interpretation of the outcome . For example, placement of more than one suspect in the lineup runs counter to recommended protocol ; b lind sequential protocol is very clear about a - priori definition of one suspect and at least five fillers. While i t may be reasonably argued that the sequential format may help to minimize effects from some a spects of faulty lineup construction (Lindsay, Lea, Nosworthy, Fulford, Hector, LeVan, & Seabrook, (1991), post - hoc estimation of such effects will be difficult . Finally, field data may be parsed many ways for analysis, with subsequent effects on the ove rall aggregate figures that become the simple summary of the more complex data. For example, HC chose to include every lineup conducted for felony crimes, including crimes in which the witness was familiar with the offender. Most HC lineups followed crim es of very short duration for which the witness did not know the perpetrator; not surprisingly, this subset resulted in lower suspect ID rates and somewhat higher filler selection rates compared to situations in which the witness had longer or even multipl e exposures to the perpetrator. Of course, the aggregate statistics change if the mix of stranger v ersus familiar perpetrator percentages that make up the data set is adjusted. Clarity of results and subsequent comparisons and recommendations demand that the

Double - blind sequential lineups 55 audience and policy - makers know the particulars of both the witnesses’ exposure to the offender and of the identification experience and how these affect the overall statistics. Be clear and realistic about the meaning of e yewitness performance measu res. A primary benefit of the laboratory test is that the data analyst can tally precise and meaningful percentages of correct identifications , filler selections, and false rejections in a target - present lineup. Knowledge of c o rrect lineup rejections and identification errors in target - absent arrays likewise are quite straightforward because t he lab researcher has control led the offender’s presence or absence in the lab lineup. It is somewhat more challenging to simulate and evaluate the risk to an inno cent suspect in laboratory studies in as much as there are many ways in which an innocent suspect may find his way into a lineup in the real world. However, it is likely that an innocent suspect has at least modest physical similarity to the true offender in most cases. In the lab, the researcher therefore typically plants a suspect who looks similar to the perpetrator in a target - absent lineup. This procedure simulates the worst possible scenario that may occur in a real police investigation – i.e., a lo ok - alike to the offender has become a suspect. An alternative method is to calculate the risk to any one innocent suspect as 1/6 of the overall filler identification rate in a (6 - person) target - absent lineup. Either way, the lab researcher is certain tha t the “filler” category has captured all mistaken identifications. F ield lineup results can be much more challenging to evaluate . There is no parallel in the field to the laboratory’s “target - present” and “target - absent” lineups, because the true status of the s uspect as offender or innocent is unknown. While laboratory protocol is established to assess the witness’s memory, witness responses in the field have implications for both the witness and the lineup . The purpose of the field lineup is to gain ad ditional Double - blind sequential lineups 56 knowledge about the likelihood that the perpetrator and the suspect are one and the same. T he suspect in the lineup represents the investigator’s belief about who the perpetrator is , and the witness’s response may provide diagnostic information a bout whether the investigator has in fact placed the true perpetrator in the lineup (Wells & Olson, 2002) . The lineup and its procedure must be set up in a manner that will challenge the witness not to simply confirm the investigator’s belief, but to t es t memory f or the offender. In the field , then, Suspect IDs cannot be directly equated to Accurate IDs , because any false identification of an innocent suspect (dangerous error) is buried within Suspect ID s . A Suspect ID may be an accurate identificati on of the offender or a false i dentification of an innocent suspect . A f iller selection is a known error but should not be confused with a dangerous false identification . There is no direct danger to th e selected filler , but this choice by the witness is a signal to investigators that the witness has a poor memory of the offender, is guessing or uncooperative, or perhaps that the filler is a better match to the perpetrator than is the suspect . No choice responses will in clude those witnesses unable or unw il ling to select from the lineup, or, particularly when a witness respon d s with “he’s not there” will indicate that the suspect is not the offender. Left with f ield measures that are inherently open to multiple interpretation s, researchers must be very c autious as th

ey determine the meaning of the data. The appropriate and effective evaluation of a study must occur at the nexus of witness response and the method used to generate th e data. A powerful example of the confusion that ensues when this point i s missed is provided by O’Toole (2006) in his critique of the Illinois pilot program . Double - blind sequential lineups 57 “…There was one other glaring defect in the report’s methodology: It equated a pick of the police suspect with a correct identification [italics in original]. To put a fine point on it, this means that the Mecklenburg Report would have counted every single one of the DNA exonerations as “correct” identifications. Even worse, the benchmark picked by the Mecklenburg Report actually rewards suggestive police procedures by skewing the results in favor of any method that encouraged witnesses to make more “suspect” selections regardless of accuracy.” (p. 19 - 20 ). Employ procedures that rule out alternative (non - memory) interpretations of witnesses responses . The objecti ve of a lineup is to accurately access the eyewitness’s memory of the perpetrator, circumventing non - memory factors that may otherwise influence the witness’s decision. Examples of non - memory factors include expectations and beliefs of the witness, inferen ces that the witness may draw from the lineup construction or presentation, the relative judgment process, and influences (imagined by the witness or real) from the lineup administrator. Challenges to eyewitness identification evidence typically claim th at the witness’s decision can be attributed to non - memory factors . Assertions of s ound eyewitness evidence must be able to demonstrate that the identification is of the highest quality – that extraneous influences on memory have been avoided . To that end , f i eld practice m ust rely on procedures that have been devised to secure memory accuracy , and evaluation of field data must recogni ze the extent to which the outcomes represent procedural safeguards against memory error . One requisite safeguard , a s noted above , is double - blind administration of the lineup (Wells, 2006b) ; a second is reflected in the mock witness test of filler quality conducted by HC. Both Double - blind sequential lineups 58 procedures – one a method to prevent suggestibility effects stemming from the investigator and the second a n assessment of how well lineup construction has constrained suggestive presentation – bring greater confidence in the accuracy of witness decisions. In the field we must depend on the design of the study to help rule out non - memory explanations o f witness responses . An understanding of field lineup outcomes demands that those who design, undertake and analyze field research understand central principles about human memory, lineup research, sound scientific method, and common police practice. T h e usefulness of laboratory lineup research is to methodically identify and strip away factors that reduce witness accuracy and so to devise technique s that add precision to our attempts to access witness memory. The new double - blind sequential lineup proc edure successfully employed by Hennepin County was designed to limit the impact of non - memory factors that may occur in traditional lineups. The fruitful collaboration of HC and lab is extended now in Study 3, in a further investigation of best practices . Study 3 : Laboratory tests of lineup protocol revisions Underlying recommended lineup protocol is a theoretical model that explains how eyewitnesses process information, h

ow memory can be effectively tapped or unfortunately contaminated, and the conditions under which better memory products may be expected. Empirical work has further demonstrated that relatively simple changes in lineup structure or pro cedure (Steblay, 1997) c an dramatically influence witness accuracy. Therefore, even seemingly benign changes to lineup protocol may require empirical examination . Study 3 addresses two questions regarding adjustments to recommended lineup procedure. Double - blind sequential lineups 59 Repeated laps of the sequential lineup . H ennepin C ounty has structured a lineup pro cedure that a llows a smooth logistical process and make s theoretical and practical sense. All key components of a blind sequential method are included except one: Instead of limiting the eyewitness to the recommended single viewing of the lineup, HC allow ed the witne ss to repeat the lineup . The HC field data revealed that repeated laps of the lineup w ere associated with increased numbers of decision errors as revealed by filler selections . In the laboratory this investigation can be extended to measure identificatio n accuracy . The research question is : What is the effect of a witness’s repeated viewing of the lineup on witness accuracy ? Th e decision to allow a hesitant witness a second chance at the lineup was a revision aimed at securing a greater number of true p ositive identification s . However, one could extrapolate from theory and research to surmise that this deviation w ill move the lineup format in the direction of a simultaneous presentation and give the witness greater opportunity to engage in relative judg ment. Based on this logic, it is hypothesized that t he performance of eyewitnesses on a repeated sequential lineup will not improve identification accuracy, but rather will approximate that of witnesses who view a simultaneous lineup: more choosing, incre ased correct identifications in the target - present condition, and increased false identifications in the target - absent condition. Impact of familiar foils . The second research question relates to lineup structure . HC investigators recognize that in som e instances lineup foils are familiar to the witness , a fact un known to the lineup administrator until the lineup is underway . This circumstance is particularly likely to occur with gang - related crimes in minority populations , wherein the photo pool of sa me - race, same - age persons available for lineup foils is constrained. The Double - blind sequential lineups 60 ideal lineup – with five unfamiliar foils -- is unrealized. The research question is : How does familiarity with lineup foils affect the decision of the eyewitness? A lineup size o f at least six is prescribed as a safeguard to protect the innocent suspect. Errors of memory should be distributed across f illers in a manner that reduces the likelihood of false identification. A reduction in number of f illers is worrisome, as there ar e fewer foils to absorb errors. Researchers refer to the number of photos in the lineup as its nominal size. A lineup with fillers th at are d isparate from the perpetrator description or that can be easily dismissed by the witness for reasons other than a lack of match to memory is unfair. In such a case the functional size of the lineup (real choices for the witness) becomes smaller than its nominal size. A reduction in number of f iller s changes the chance distribution of choices, thus should logically e xacerbate the effects of differential choosing that occur in

simultaneous versus sequential lineups. If a high er choosing rate is main tained for simultaneous lineups despite reduced functional size , witness choices will be spread over a smaller number of photos. The result will be an elevation of correct identifications when the target is present but also an increase in identification errors for target - absent lineups compared to the sequential lineup. Recent research provides evidence for an other conc ern , that the addition of very weak alternatives ( duds ) to a judgment task increases the perceived likelihood of the focal outcome (Windschitl & Chambers, 2004). Although the investigators of this dud phenomenon did not use a lineup task in their research , they posit ed that confidence judgments of lineup witnesses are a real - world example for which the dud - alternative effect is relevant. More specifically, their work suggests that the inclusion of ill - fitting lineup distract e rs may boost the eyewitness’s confidence about who ever is chosen from the lineup and about how well the Double - blind sequential lineups 61 selected person matches memory for the culprit. The current study will examine both accuracy and confidence as a means to determine whether the dud - alternative effect extends to fam iliar foils in lineups. It is predicted that w ithin the simultaneous lineup condition , decreases in the number of real f iller choices will be associated with a higher level of choosing and a greater number of both false identifications in target - absent l ineups and correct identification s in target present lineups . Given the sequential lineup’s tendency to hold down choosing rates, it is expected that decreases in the number of real foil choices will not affect choosing or accuracy rate s . Foil familiarity is expected to be associated with higher levels of confidence in both types of lineups. Study 3 was structured to explore th ese two d eviations from prescribed lineup protoco l — r epeated viewing of the sequential lineup and reduction in lineup size due to foil familiarity — while also testing two variables common to lineup lab investigation , s imultaneous v ersus sequential presentation format and presence versus absence of the perpetrator in the lineup. Th e inclusion of standard conditions allow s comparison of o utcomes to past research and integration of findings into the theoretic al foundations of the research, as well as comparative evaluation of the sequential versus simultaneous lineup format. The procedures for both simultaneous and sequential lineup fo rmats include d unbiased instructions, blind administration, and fair lineup construction . T he sequential lineup differ ed from the simultaneous in its sequential positioning of the photos, in instructions specific to sequential lineup process, and in the op tion for a dditional review s (lap s ) . Method Double - blind sequential lineups 62 The stud y design is a 2 X 2 factorial with independent variables of lineup presentation (simultaneous or sequential) , and target presence in the lineup (present or absent) . Number of lineup viewings and fun ctional size of the lineup are ex post facto (non - manipulated ) variable s , varying naturally depending on each subject’s previous exposure to the foils and the subject’s decision as to whether to repeat the lineup viewing. The inclusion of these two varia bles in an ex post facto configuration is intentional ; as a naturally occurring phenomenon, the behavior affords greater ecological validity to the laboratory results. Key dependent mea sures are witness choic

e , accuracy, and confidence . For target - pres ent lineups, aggregate outcome measures are recorded as frequenc ies and percentages of correct identification s , filler selections , and incorrect rejection s of the lineup ; for target - absent lineups, outcome measures are correct rejection s and error identifi cation s of any foil. Confidence is assessed through subject report on a 1 - 6 scale. Participants 3 7 2 undergraduate students took part in the study, male and female, ages 18 to 65 , with a median age of 20 years . Eighty - seven percent of participants ide ntified themselves as Caucasian. Participants were treated in accordance with American Psychological Association Ethical Standards for Treatment of Human Subjects. The experimental protocol was approved by Augsburg’s Institutional Review Board . In order to achieve appropriate levels of statistical power (.70 to .80, depending on the specific analysis), the study require d 370 subjects , with a greater proportion in the sequential condition . Participants were recruited through t he existing college research participant pool , recruit ment signs posted on campus , and announcements in classes . Participants were provided the choice of course Double - blind sequential lineups 63 credit (for participant pool members) or a $5 payment. S tudent actors featured in the video and lineup segments of the l aboratory materials provided written consent for their images to be used for laboratory purposes. Stimulus material s The experiment was delivered to subjects via laptop computer within an Authorware software program . A video of a short ( 3 0 - second) purse - snatching incident serve d as the stimulus incident. Th e event wa s filmed in color with audio, shot from the victim’s perspective, and the male Caucasian perpetrator , age 21, is visible for approximately 10 seconds. A 6 - person lineup provide d the memory t est for subject eyewitnesses. The lineup consist ed of Caucasian males of approximately 18 - 23 years of age and was constructed to capture moderate physical similarity between lineup members and the perpetrator. In a prior pilot study, 17 participants watch ed the video and were asked to describe the perpetrator first via free recall, followed by prompts for clothing, race, hair/eye color, height and age. The modal description , excluding clothing , was used to find n ine potential lineup members (a fit - to - desc ription method ) . Next, 11 additional participants who had not seen the crime video were presented with a 10 - photo perpetrator - present lineup and asked to decide “Which one of these photos best meets the description?” The participants were allowed multipl e, rank - ordered answers. The same 11 participants also were shown a side - by - side comparison of each filler photo with the perpetrator and asked to rate the similarity of each of the nine (one at a time) foils to the perp etrator , on a 1 - 10 scale. This prod uced a range of scores for the nine photos from 2.09 - 5.18 ( low to moderate similarity). Finally, b ased on these measures, 6 fillers were selected. No one filler stood out as an obvious “innocent” suspect nor did the Double - blind sequential lineups 64 perpetrator stand out as the best fit to his own description. Each photo was a full frontal head - and - shoulders view with a gray background. All l ineup members including the offender wore their street clothes for the lineup, none matching the clothing worn by the perpetrator at the time of t he crime. For each subject, t he computer program generat

e d a 6 - person lineup of sequential or simultaneous format , target - absent or target - present. The six fillers rotated in and out of the target - present lineup. The p osition of filler s in all l ineups and the position of the perp etrator in the target - present array w ere balanced, with the exception that the offender never appeared in position 1. The stimulus crime video and subsequent lineup feature d current or recently graduated Augsburg Coll ege students, a point central to this project . S timulus materials were prepared in April of 2003. The currency of the video allow s the natural occurrence of subject s’ previous familiarity with lineup foils and the perpetrator , and subsequent evaluation o f lineup performance for student subjects who recognize d lineup members from outside the lab . A pilot study demonstrat ed familiarity percentages of 45%, 15%, 20%, 15%, and 5% for subjects who recognize 0 , 1, 2, 3, and 4 foils, respectively . T he current st udy found familiarity percentages of 35%, 20%, 14%, 12%, 7%, 5%, and 7%, for subjects who recognized 0, 1, 2, 3, 4, 5, and all lineup members, respectively. Procedure Each subject wa s run individually with instr uctions provided through the computer program and by the experimenter during a 15 - miniute laboratory interaction . The experimenter maintain ed a blind status by remaining across the room from the subject, with the laptop computer facing the subject. The e xperimenter also positioned herself Double - blind sequential lineups 65 perpendicular to the subject, rather than in a direct sight line. A minor deception was employed to establish the blind status of the experimenter for witness - participants (see instruction below) by informing the witnes s that the specific perpetrator in the witness’s film was one of many employed in multiple versions of the film , thus not knowable to the experimenter. The computer prompt ed the subject to wait for specific instructions from the experimenter at one key po int in the process. Subjects were informed that the study is funded by the National Institute of Justice and results were to be used to inform police practice nationally, therefore to “please pay serious and close attention to the procedures.” After a b rief introduction, subjects worked through a computer application that requested demographic information and showed the crime scenario. The experimenter was out of the room during this time. At the conclusion of the crime video, the experimenter re - enter ed the room and provided additional instructions. The script for the experimenter emphasize d five key points:  The film segment you saw is one of a number we use. It was randomly generated by the computer, so that I don’t know who the perpetrator was in y our display.  In the lineup you’re about to see, we are interested in whether you recognize one of the lineup members to be the perpetrator of the crime in the film. However, the lineup may or may not include the perpetrator you saw in the film.  Sometim es the appearance of the offender will change between the event and the lineup. Please keep this in mind .  This procedure works best if you view all of the photos – So the program will allow you time to examine each photo in the full lineup . Double - blind sequential lineups 66  I am here to record any comments that you may have about the photos, but I cannot answer questions about the crime video or the lineup. The computer maintain ed a data file that capture d aspects of the method and subject responses, including de

mographics (age, sex, rac e), target presence/absence, lineup format (simultaneous or sequential), position of the perpetrator (randomly determined, with the exception that the perpetrator never appear ed as the first photo), number of viewings of the sequential lineup, subject choi ce, and subject rating of confidence. The subjects were also prompted to report their familiarity with any of the crime scene and lineup participants , and these responses were recorded through the computer application. In addition, t he experimenter kep t a written record of any subject comments and questions, in cluding verbatim comments about familiarity with any of the men in the photos , q ualifiers offered , and expressions regarding confidence . 5 Results Overall eyewitness responses Tables 7, 8, and 9 document witness performance in simultaneous lineups (column 1) and sequential lineups (columns 2 - 5). The direct comparison of simultaneous to sequential lineup performance is available from columns 1 and 2; these figures , representing data from all wit nesses, represent performance of the simultaneous array (Sim) and a sequential procedure a fter one showing of the lineup (Seq1). Column three (SeqFinal) indicates the endpoint of the witness decisions after all laps through the lineup . Columns 4 - 5 break o ut the sequential subjects into those who did not request more than one viewing (SeqHold) and those who opted for more than one viewing (Seq+). Otherwise stated, each witness who saw a sequential lineup is represented in three columns on this table: R esp onse Double - blind sequential lineups 67 after one viewing (Seq1), final decision (SeqFinal), and response in either in the one - lap (SeqHold) or multiple - lap (Seq+) category . Table 7 includes all 372 witness participants. Table 8 details the subset of participants for whom the offender was a stranger (n=253), and Table 9 shows the remaining witnesses who report familiarity with the perpetrator from interactions prior to the study (n= 11 9). Sixty - eight percent of participant - witnesses stated that the perpetrator of the crime was a stranger t o them , perhaps coincidentally very close to the 66% H ennepin C ounty proportion. A fter one viewing of the lineup (standard procedure) , p articipant - witness choosing rates were at 35% overall , and 35% and 36% for simultaneous versus sequential lineups , re spectively. These are lower choosing rates than are typical for laboratory or field outcomes (see Table 3 ) , particularly given that 32% of the witnesses reported that the perpetrator was familiar to them from experiences prior to the lab study . I t appe ar s that either th is crime scenario was quite challenging for memory encoding, the lineup task particularly difficult , witnesses very cautious, or all of the above . I w ill return to this point below. Hypothesis 1 Repeated viewing of the sequential lineup The expectation for the first hypothesis was that performance of eyewitnesses on a repeated sequential lineup should approximate that of witnesses in the simultaneous lineup: higher choosing rates , increased correct identifications in the target - present c ondition, and increased false identifications in the target - absent condition. Fifty - three percent of witnesses elected to repeat the sequential lineup. However, only six witnesses opted for a Double - blind sequential lineups 68 third viewing (lap) ; therefore the a nalyses com bine second an d third laps , and simply refer to first versus final laps. Target present lineups . Table 7 details witness choosing rate

s for first and final viewings of the lineup, in the columns headed Seq 1 and Seq Final . The cho osing rate increase d from one lap to the final lap, from 4 6 % to 6 2 % , and the pattern of witness decisions was significantly different from first to final lap (Sign test produced N = 16, x = 1, p .000 1 ) . As predicted, correct identifications increase d , by 6% ; filler rates (errors) also gr e w 10 % . Another way of looking at this pattern is to consider the individual choices of subjects who elected to review the lineup a second time , specifically whether their decisions improved as they moved from the first to final lap. For 69 % of the se witness es , lineup responses remained the same from first to final lap . T he remaining 3 1 % changed their answer ; for 6 5 % of these subjects the change was not productive (a change from a “no choice” to a filler) and for 3 5 % the subject moved from an incorrect (fille r or “no choice”) to perpetrator selection. Otherwise stated, the second viewing of the lineup improved performance for 11 % of witnesses who went on to a second look (6 of 54) ; for almost twice as many ( 20 %), performance declined. Target absent lineups . The choosing rate in target absent sequential lineups increased from the first to the final repetition (2 6 % to 42 %) , and a significant difference in response pattern was evident (sign test produced N =16, x = 0, p .000 1 ) . F o r those subjects who opted fo r a repeat of the lineup, there was no gain in performance: 70 % maintained their initial answer; 3 0 % made a worse response (changing from correct rejection to a filler selection). Witnesses viewing a stranger crime were slightly more likely to take anothe r Double - blind sequential lineups 69 lineup lap when the target was absent versus present, although this difference (64% vs. 59%) was not statistically significant. A logical question is whether those witnesses who did not request a repetition of the lineup generated greater accuracy com pared to the entire group at the end of one lap . As can be seen on Table 7 in the Seq Hold versus Seq1 column s , this seems to be the case if the target is present; a higher choosing rate led to correct identifications at 40% and just 1% increase in filler s elections above the entire group ( Seq 1 ) . However, when the target is absent, the 10% higher choosing rate results in a 10% higher error (filler) selection rate. Thus, those witnesses who stop after one lap are not necessarily a more reliable group. A d iagnosticity ratio confirms this: At the end of one lap, the ratio for all witnesses is 6.50 (.26/.04); for those who elected just one lap, the diagnosticity ratio is 6.67 (.40/.06). The same pattern is true, if one calculates diagnostic ity ratios for the subset of stranger perpetrators (Table 8). In sum, w hen witnesses were allowed to repeat the sequential lineup, errors increased above the rate of sequential 1 - lap lineups. The negative effects of additional laps in the sequential lineup produced error s not equal to, but significantly greater than the simultaneous lineup, X 2 (2) = 7.98, p = .02, target - present, and X 2 (1) = 4.1 0, p = .04, target absent. Hypothesis 2 Attrition for familiarity The second hypothesis addressed the impact of downward l ineup size adjustments due to the witness’s familiarity with lineup members ( A ttrition f or Familiarity or AF ). Within the simultaneous lineup condition, AF w as expected to be associated with a higher level of choosing and therefore a greater number of fals e identifications in target - absent

Double - blind sequential lineups 70 lineups and correct identifications in target - present lineups. Within the sequential lineup, AF decreases in lineup size w ere not expected to affect choosing or accuracy rate s . Choosing and accuracy . Data for stra nger crimes ( n = 253) were used in this analysis as were witness responses at the end of the first sequential la p only . The question of interest is whether there w ere change s in choosing and accuracy from the full lineup to lineups sizes reduced by AF . Be cause of the very small numbers of participants who knew 3 - 6 lineup members, post - AF lineup size was collapsed into just two categories , 0 - 3 and 4 - 5 . These two post - AF categories are compared to a full lineup size of 6 in Table 10 . (Only one witness knew all six lineup members; the lineup was a target - absent simultaneous display and the witness correctly rejected the lineup.) The data first are collapsed across target present and absent conditions, to produce overall choosing and accuracy rate s for each l ineup format as post - AF size de creased . T he simultaneous lineup show ed a decrease in choosing (4 0 % vs. 30%) and increase in accuracy rates (32% vs. 45%) as the lineup dropped from a full 6 - person lineup to half that size or smaller (0 - 3) , z = .77, p = .22 , r = .09 for choosing and z = 1.0, p = .16 , r = .12 for accuracy , but neither statistically significant . The sequential lineups produced a significant (30%) drop in choosing as the lineup size diminished, 38% vs. 8%, z = 2.14, p = .02 , r = .24 . As with the simultaneous lineup , with in the sequential format a non - significant in crease in overall accuracy , 36% vs. 54% , was associated with a drop in lineup size . Thus, c ontrary to prediction, AF decreased choosing rate s for both lineup formats, and sequentia l lineups were significantly affected by AF . As can be seen in Table 10 , the impact of AF on s i multaneous lineup choosing rate showed up in a n on - significant 10% increase in hit rate when the target was present and Double - blind sequential lineups 71 increased correct rejections ( 12 %) when the target was absent. In sequential lineups, t here was a dramatic drop in choosing for both target present (34%) and target absent ( 28 %) lineups from the full to small er lineup . This reduction in choosing reduced accuracy to zero in the target - present small (0 - 3) lineu p — w here no lineup choices were mad e — and increased accuracy by 29 % for the target - absent small lineup , in which only one witness made a choice from the lineup . It should be noted that some sample sizes in these AF conditions were quite sma ll. T he 0 - 3 AF lineup size is a substantially truncated lineup , a point that would only be apparent to witnesses viewing a simultaneous lineup. Thirty percent of w itnesses chose from a simultaneous lineup of size 0 - 3 — whe ther the perpetrator was present or absent. Witnesses viewing a sequential lineup were not aware of lineup size ; f or most witnesses who could easily eliminate 3 - 6 of the lineup members due to familiarity , the lineup ended before they ma d e a choice. This differential choosing rate in sim ultaneous versus sequential lineups (30% vs. 8%, z = 1.47, p = .07 , r = .26) suggests that relative judgment may be guiding the former group and that absolute judgment is more strongly present in those witnesses faced with a sequential lineup. The impact of duds appears to be more dramatic in a simultaneous lineup, where the witness can see all remaining alternatives and in which the attrition of lineup size is very sal

ient. The sequential presentation, on the other hand, does not allow the witness to kno w the number of additional photos to be shown or to review those already passed, thus changing the impact of attrition. Confidence . Increased foil familiarity was expected to be associated with higher levels of confidence in both types of lineups followi ng the logic of the dud - alternative effect . Confidence was measured a fter completion of the entire lineup , rather than at the time of Double - blind sequential lineups 72 identification of a specific photo . For witnesses who were familiar with the perpetrator, the predicted relationship did exist: r (119) = - .31, p = .001, 2 - tailed; as the lineup became smaller, confidence grew. However, f or stranger identifications, there was no significant linear relationship between confidence (measured 1 - 6) and AF (0 - 6) for the overall data, r (253) = .02, p = .72 (2 - tailed) or any of the four conditions separately: S imultaneous TP, r (62) = .07 , p = .58 ; Simultaneous TA, r (56) = .19, p = .16; Sequential TP, r (66) = - .04, p = .74; and Sequential TA, r (69) = - .05, p = .71. Stranger and familiar pe rpetrators. The pattern of witness response to stranger versus familiar perpetrators in the lab was similar to that in HC . P erhaps just coincidentally but also conveniently, 2/3 of the witnesses in both samples reported that the perpetrator was a strange r. We must keep in mind that there was no variation in the perpetrator as stimulus in the lab (just one perpetrator for all witnesses), however in the field the perpetrators of course varied across crime scenes, as did viewing conditions for the witness . T he absolute numbers that detail witness response should not be expected to be similar across lab and field. However, the relative differences may be usefully examined. In both the lab and field, the choosing rate for familiar perpetrators was h igher than that for strangers , not an unexpected finding . In the laboratory target - present sequential lineup condition, choosing rate for final decisions was at 53 % for strangers and 76 % for familiar perpetrators; in the field the figures were 46% and 93% , respectively. As seen in Table 5 , suspect identifications in the field for familiar perpetrators w ere 55% above that of strangers; in the lab, this difference was 52 %. Comparison of sequential to simultaneous lineups Double - blind sequential lineups 73 Choosing and accuracy . Overa ll, the simultaneous condition produced 42% and 27% choosing rates in target - present and target - absent arrays, respectively. The sequential lineup produced choosing rates of 46% and 26%, target - present and absent, respectively. There was no significant di fference in choosing rate between lineup formats in either target - present ( z = . 57 , p = .2 8 ) or target - absent ( z = .1 7 , p = .4 3 ) conditions. The two lineup formats also generated similar accuracy at the end of one viewing, as can be seen by comparing the S im and S eq1 columns in Table 7 . Separate analyses of target - present and. target - absent conditions show no significant differences in witness decisions between the simultaneous and sequential formats, X 2 (2) = .375, p = .41 and X 2 (1) = .047, p = .41. Diagnosticity . The statistic of diagnosticity compares the identification rate of the perpetrator to that of an innocent person in the lineup (Wells & Turtle, 1986; Wells & Olson, 2002). For practical reasons, the diagnosticity of different lineup proc edures can be compared to determine which yields the greater probative value of its identi

fication evidence. Diagnosticity calculations require an estimate of the risk to an innocent suspect. In this case, there was no a - priori specification of an innoce nt suspect in the target - absent lineup; therefore, the average rate of identification per lineup member (percent of identifications divided by 6) was used to calculate the diagnostic value of the sequential vs. simultaneous display . Given the similarity o f sequential vs. simultaneous lineup results, it is not surprising to see similar diagnosticity figures for correct identifications: 6.05 for the sequential lineup and 5.56 for the simultaneous lineup for the overall group; for the subset of stranger ident ifications, 1. 51 for sequential and 1.6 7 for simultaneous lineups. Translated into practical terms, this means that identification was 6.05 times as likely to be the culprit Double - blind sequential lineups 74 rather than an innocent suspect when the sequential format was used, and 5.56 tim es as likely to be the culprit when the simultaneous lineup was used. . Confidence in sequential versus simultaneous lineups . Overall, simultaneous lineups produced a significantly higher average level of confidence, F (1, 368) = 3.83, p = .05, measured at the end of just one viewing of the lineup, ( m s = 4.62 vs. 4.39, simultaneous vs. sequential, respectively ) . This same effect held in the subset of w itnesses who were unfamiliar with the perpetrator and the lineup members . Those who viewed a simultaneous lineup were more confident than those who viewed a sequential lineup ( m s = 4.36 vs 3.86 , simultaneous vs. sequential, respectively , t (120) = 2.44, p = .016, 2 - tailed. For stranger lineups of size 0 - 3 (AF) , mean confidence was equal b etween simultaneous and sequential conditions (both m s = 4.00). Same versus cross - race identifications . Thirty - three cross - race stranger identifications were attempted, with accuracy rates not significantly different from same - race stranger identific ation attempts, 39% versus 37%, respectively. There was also no difference in accuracy between same and cross - race identification attempts within simultaneous versus sequential formats: 35% versus 38%, same - race simultaneous versus sequential lineups; 39% versus 40%, cross - race simultaneous versus sequential lineups. Discussion Two hypotheses were tested to determine the impact of revisions to recommended lineup protocol on eyewitness decisions . The first hypothesis was supported, in that repeated viewi ng of a sequential lineup reduced eye witness accuracy. This finding reinforces the Hennepin County field data and leads to a recommendation that repeated laps be allowed only with great caution. The second hypothesi s — a prediction for a dud - alternative e ffec t — Double - blind sequential lineups 75 was just partially supported , as eyewitness c onfidence was enhanced by lineup attrition only for witnesses familiar with the perpetrator . Rather than the increase in lineup selections predicted for the simultaneous format condition, choosing rates de creased as the lineup size diminished due to familiarity, for both simultaneous and sequential lineup formats . An unexpected effect was that the choosing rate dropped significantly only for sequential lineups , a dramatic reduction such that only one of 13 witnesses made a lineup selection when post - AF lineup size was three or fewer . On the other hand, 30% of witnesses viewing a simultaneous array made lineup selections even as lineup size decreased to three or fewer , and regardless of whether the target was present o

r absent . The se study outcomes have implications for field practice and for our understanding of witness cognitive processes, to be discussed in the final segment of this report. One third of witness - participants were familiar with the per petrator from encounters prior to the laboratory task , allow ing an unanticipated opportunity to compare laboratory and HC field data on the variable of offender familiarity. Lab and field data show parallel effects : i ncreased lineup selections from witnes ses who are familiar with the perpetrator compared to those for whom the culprit wa s a stranger, resulting in higher suspect identification rates in the field and perpetrator selections in the lab . While this is not an unexpected finding in itself , it rei nforces the soundness of extrapolat ion from lab to field settings for these new lineup procedures . As expected from past literature (Steblay, et al., 2001) simultaneous lineups produced significantly greater witness confidence than did sequential lineups. This finding is intriguing given the lack of performance differences between the two groups and the overall low level of perpetrator identifications. The lack of decision difference s between Double - blind sequential lineups 76 sequential and simultaneous lineup performance is unexpected. A number of reasons may be advanced and explored to explain this result , among them the following four considerations . (1) Lack of serious attention to the task by witness - participants. Participants appeared to seriously apply themselves to the lineup ta sk, presumably spurred by the context of the experiment and the message of its instructions. However, it is necessary to go beyond anecdot al impressions. If witness accuracy failed to rise above the level of chance, this raises the possibility of effects due to participant carelessness or guessing. A first consideration is the pattern of filler selections. C hoices of innocent lineup members in this lineup were not random ly distributed ; the rate of filler photo selections rang ed from 0% to 15% . This s ugges t s at least some intention among witness es to discern the perpetrator from among the lineup members . Beyond this, t he pattern of eyewitness response falls in line with that expected for a single - suspect lineup (Wells & Turtle, 1986) and indicates witness accuracy rates that are low but above chance. For example, the individual lineup member receiving the greate st proportion of responses should be the actual target if eyewitness performance is above chance level. Fifty - nine percent of the 8 2 subjects who chose from the target - present lineup identified the perpetrator , an appropriately greater proportion above chance (17%), z = 5.25, p .0001 . This pattern also was apparent in target - present stranger lineups, where 27% of choosers ide ntified the perp etra tor, although this was not significantly above chance level of 17% , z = 1.11, p = .13. In target - absent lineups, none - of - the - above responses should be more frequent than in target - present lineups , again what is expected if eyewitness performance is above chance levels . In this study, n one - of - the - above responses were indeed more frequent in the target - absent condition than the targ et - present condition, 74% vs. 57 % for all subjects; although for perpetrators who were strangers, this difference did not hold , Double - blind sequential lineups 77 66% vs. 68%. T he pattern of witness response to these lineups indicates a conscientious rather than cavalier approach to

what may have been a very difficult memory task. T he witnesses appear to have be en cautious, with approximately 2/3 making no choic e from the lineup in both target - present and target - absent condition s . (2) Limited viewing conditions . The study included a 10 - second view of the perpetrator . The film was shot from the victim’s perspective, allowing a very close range short confrontati on with the offender, with his face fully in view. However, perhaps this was too short a time for solid memory encoding to occur for most subjects . Sporer (1993) noted deterioration in accuracy to close to chance levels when view of the perpetrator wa s l imited to 5 seconds . Although the participants in this study did not demonstrate chance levels of accuracy, Sporer’s experience may speak to the level of task difficulty in this study . For the witnesses who reported familiarity with the perpetrator from previous encounters , 88% identified him when he was present in the lineup (2 witnesses made no choice). For those subjects for whom the perp etrator was a stranger, the identification rate was only 9%. (3) Lack of high similarity between lineup member s and offender . The data from Steblay and colleagues (2001) indicate that the sequential superiority effect was particularly strong when the innocent suspect in a target - absent lineup was highly similar to the culprit. Perhaps the protection of the sequ ential lineup against false identification is primarily a function of lineups in which the suspect is a close match to the perpetrator. In the current study, lineup members were rated as moderately similar to the perpetrator, no high - similarity “innocent suspect” was employed, and all lineup members (including the perpetrator) were rated as moderate matches to the description. Otherwise stated, this was a lineup that by Double - blind sequential lineups 78 structure should protect against bias , perhaps minimizing the likelihood of a visible sequential superiority effect . (4) Appearance change . An offender’s appearance may change across time, especially between an early mugshot and current identification task, or in the delay between crime event and apprehension. Even when the perpetrato r is arrested shortly after a crime, he or she may have intentionally and significantly changed aspects of his or her appearance. Just as the offender would hope, transformation of appearance does diminish the likelihood of recognition by eyewitnesses (Sh apiro & Penrod, 1986). Hair in particular is a primary descriptor reported by witnesses and used as a key marker for identification (Pozzulo & Warren, 2003). W itnesses are significantly less likely to identify the offender from a lineup when his hairst yle ha s changed (Pozzulo & Balfour, 2006). T he perpetrator’s hairstyle i n the current study changed from crime event to lineup , perhaps exacerbating identification challenges for witnesses who had only a 10 - second exposure to him . If appearance change hin dered recognition, it seemingly did so under both simultaneous and sequential formats. The NIJ Guide included a recommendation that all witnesses be reminded that the perpetrator’s appearance may change from the time of crime to the time of lineup photo: “Instruct the witness that individuals depicted in lineup photos may not appear exactly as they did on the date of the incident because features such as head and facial hair are subject to change.” (p. 32). The distinct change in the perpetrator’s hair style between crime event and lineup in this study made an Appearance - Change Instruction (ACI) appealing as a reasonable attempt to provide context for the witness ’s identification task . However, the s urprisingly low

rate of witness lineup selections can generate speculation about the impact Double - blind sequential lineups 79 of the culprit’s appearance change and the likelihood of the ACI’s remediation of any effect. Interestingly, the ACI was a recommendation to law enforcement in the NIJ Guide that was not backed by a foundation of emp irical research (Charman & Wells, 2007 ). More recently, t he effectiveness of the ACI has been empirically tested once with a simultaneous lineup format; use of the ACI produced no overall benefits to identification accuracy (Charman & Wells, 2007 ). The authors did find positive identification results from the ACI in one condition in which the perpetrator had changed appearance quite dramatically . Pe rhaps this is exactly where the ACI will be influential. The research literature also lacks a theoretic al foundation that might help explain ACI operation in th e current study . However, implicit in ACI use is the assumption that some correct identifications from culprit - present lineups will be lost when the offender ’s appearance in the lineup is not exact ly as it was during the crime event (Charman & Wells, 2007) . Also implicit is the assumption that the eyewitness expects a close match between the offender’s appearance in the witnessed event and the lineup photo. These reasonable conjectures lead to the prediction that an ACI will urge witnesses to more carefully search for any recognizable features of the lineup members, to see beyond t he appearance change, and ultimately to recognize the culprit at higher rates than will witnesses who are not provided an ACI. This prediction relies on the notion that the ACI will prompt a detailed scrutiny of lineup members that will enhance access to eyewitness memory; the witness will “dig more deeply” into his or her memory, with productive results. If this theory is accurate, the ACI may have helped to elevate recognition accuracy in the current study . Double - blind sequential lineups 80 This suggests that without the ACI, positive identifications of the perpetrator may have been even lower. Another scenario is possible however . The obtained low rates of witness lineup selections in our current lab study suggest that the ACI may be inhibiting witness choosing. It is possible that the ACI serves only to alert the witness to the difficulty of the identification task , perhaps rais ing the criterion th reshold for making a choice from the lineup or reducing pressure to choose . In this case, one might predict that the ACI will inhibit lineup selection s , particularly in the case where appearance change is significant as is likely in this study . This circ umstance suggests that the ACI does not improve access to memory, but only affects witness motivation to make a selection from the lineup. Choosing will be inhibited, with parallel effects in target - present and absent conditions. Finally, there is a thi rd possibility (Charman & Wells, 2007 ). Perhaps t he ACI only affects witnesses with poor memory, prompting them to choose. T his would generate more choosing as they allow substantial discrepancy between memory and suspect photo. In practical terms, this scenario would increase choosing rates and severely diminish witness performance . Testing of the impact of the ACI is currently underway in our lab , however the data are not yet available to help ascertain the exact reasons for witness performance in thi s study. Re gardless of the fact of and reason for a lack of expected sequential - simultaneous difference in witness performance, this study clearly accomplished its original objectiv

es. The data demonstrate the impact of a witness’s second viewing of the l ineup. The study also answered the question of how reduced lineup siz e — attrition due to familiarit y — can affect witness response. The study points out the potential danger to an innocent suspect Double - blind sequential lineups 81 when a witness repeats the sequential lineup or when lineup size is reduced with in a simultaneous format lineup. In summary, the laboratory portion of this work successfully lend s empirical information for effective field practice. G eneral Discussion The objective of this set of studies was to provide usefu l information -- practical, empirical, and theoretical – to bridge laboratory research with field practice. The successful implementation of double - blind sequential lineups in Hennepin County provides a strong model for jurisdictions across the country f or reform of lineup practice. As documented above, the Hennepin County experience generated a list of necessary considerations and realistic solutions to concerns that arise as the preferred protocol is translated to field practice. Equally useful is the quantitat ive product of the pilot program — a descriptive summary of witness decisions and the conditions under which they occurred. The HC pilot has provided the first available baseline summary of double - blind sequential lineup performance under controll ed conditions in the field. Also, a s other jurisdictions look to assess lineup reform, this pilot project offers recommendations for designing effective field protocol and appropriate interpretation of field data. A training DVD produced by the HC Attor ney’s Office is available for other jurisdictions, helping ease the changes and costs to these departments. These lessons are essential to inform practice and public policy . This project’s joining of field practice and laboratory testing generated new an d timely knowledge about lineup construction and protocol. T he rapid application of scientific research to practice has placed lineup researchers in the perhaps enviable position of being Double - blind sequential lineups 82 called upon to answer practical and important questions. For examp le, a recent writer has asked : What happens when the witness is permitted to view the sequential lineups a second time? (Diamond, in press). The practice of repeated witness viewing of sequential lineups was found to boost error rates in both lab and fie ld. Caution is strongly advised when this procedure is used in the field, particularly for witnesses who complete the first viewing of the lineup without voicing recognition of a ny lineup member. The importance of presenting a full 6 - person lineup also was underscored by results of the laboratory study. A phenomenon that may occur in practic e — a ttrition of lineup size due to the witness’s familiarity with lineup foil s — was found to boost errors for witnesses who viewed a simultaneous target - absent lineup. This pattern was not apparent for sequential lineups, suggesting that the sequential format may offer some protection against attrition in lineup size when it occurs under these circumstances. However, t his is not to suggest that use of sequential lin eup negates the need for a full lineup. An unanticipated extension to the original research plan also provides a model for future field tests. Mock witness procedures previously have been employed to examine the quality of lineup fillers for indivi dual lineups. This project was the first to develop and employ the mock witness test for the purpose of assessing overall lineup qua

lity in the jurisdiction of interest. At the same time, the mock witness procedure provided necessary information to more confidently interpret the Hennepin County field outcomes. Along with data on revisions to lineup structure and procedure, t he laboratory segment of this project provided the first available lab data on w itness response to a perpetrator famili ar to the witness from a previous context. As with the HC field data, Double - blind sequential lineups 83 reasonable patterns of witness response suggest that the sequential protocol is effective for accurate identification of both known and stranger offenders. This report would be remi ss if it did not also point to specific areas for future research collaboration. Most salient is the pending challenge and opportunity of computer technology for efficient and effective lineup delivery . Adaptation of lineup construction and procedure usi ng electronic photo repositories and laptop formats present s empirical questions that can be usefully subjected to laboratory evaluation. Similarly , creative innovations (e.g., the envelope procedure and the battleship laptop position) as well as currentl y accepted “fixes” such as the Appearance Change Instruction may benefit from further scrutiny and refinement. A somewhat more amorphous suggestion for future research is in the evaluation of witness comments . The range of witness lineup responses — from immediate absolute certainty to the qualified , slow , and tentative decision — presents field data that may inspire laboratory research. Are witness comments diagnostic of decision process or accuracy , indicative of real or false confidence, differentially p roduced by specific lineup procedures or added laps , or likely to affect an investigator ’s judgment of witness veracity ? These explorations demand a continuing dynamic dialogue between lab and field. Theoretical implications At the same time that prac tical experience should push the research agenda in new directions, the touchstone of science demands the grounding of new knowledge in empirical investigation and sound theory. These data offer information regarding the relative judgment principles that underlie our understanding of eyewitness decisions . Surprisingly, there was no difference in lineup accuracy between the simultaneous and sequential formats in this study. It is difficult to know the reasons for this unexpected Double - blind sequential lineups 84 outcome: A very short e ncoding opportunity, the perpetrator’s appearance change between event and lineup, and specific instructions for caution each may have contributed to the seemingly hesitant decisions of the witnesses. It may be that these conditions are those that minimi ze the likelihood of sequential superiority. However, it is fascinating that even under these apparently constrained conditions, witnesses who faced a simultaneous lineup were more likely than those with a sequential format to choose from the array when t hey saw duds in the lineup. The attrition of the lineup due to filler familiarity did not deter a lineup choice for these witnesses even in the face of a target - absent lineup. This choosing from a narrowed set of photos suggests relative judgment at work . Witnesses in the sequential lineup who could rule out several faces (one at a time, of course) appeared to rely more on absolute judgment, continuing to work through the lineup looking for a match to memory beyond the duds . An unexpected benef it of sequential lineup delivery can be appreciated when one

views the lap data. The traditional simultaneous lineup format does not allow us to know the level of comparison shopping (relative judgment) employed by the witness prior to the lineup decision . In contrast, the sequential lineup with a repetition option has revealed an objective indicator of eyewitness laps through the lineup. The effect of r elative judgment was evidenced in both field and laboratory as witnesses engaged in additional laps of the sequential lineup. The lab demonstrated that greater choosing in a second (or third) viewing helped to secure a handful of positive identifications but at the expense of four times as many identification errors . The diagnosticity of identification dr op ped from 6.05 to 4.57 after a repeated viewing of the lineup. The sequential lineup was mo re effective without repeated viewing . Double - blind sequential lineups 85 A concern of the police investigators, that jump - out identifications may be sacrificed with the sequential format , did no t appear to be the case. Jump - outs occurred at a relatively high level, particularly for instances of familiar perpetrators. Jump - outs may be the prototype of absolute recognition. If so, we would not expect a reduction in jump - outs from a sequential li neup , the sequential lineup being a better spur to absolute judgment than the traditional form . Jump - out identifications in the first viewing also suggested that absolute judgments were common in the field (55% of witnesses choosing). Implications for t he Future Th is convergent evidence from field experience and laboratory trials o ffers unprecedented opportunity for an increased understanding of eyewitness decision - making processes and the means to achieve more effective police lineup outcomes. Althoug h some of the lessons may apply more narrowly to H ennepin County’s unique local conditions and logistical circumstances, it is anticipated that the central research outcomes will be informative across jurisdictions. The laboratory research questions, whil e originating from local experience, were selected for exploration specifically because of their likely interest and relevance to the wider law enforcement and research communities A next generation of research must continue to effectively join theory and research with practice, particularly to address disruptions in recommended protocol that may (or may not) affect eyewitness decisions and accuracy. Diamond (in press) speaks of benefits to learning “when well - documented field investigations are combined wi th laboratory backup . ” She specifically cited the HC test o f filler quality – “applying laboratory procedures to test the hypothesis that the lineup construction rather than the memory of the eyewitness produced the identification.” (p. 13). Assistant Hen nepin County Attorney Scoggin has Double - blind sequential lineups 86 forcefully stated , “We in the law enforcement arena cannot help but look over our shoulders at a time when DNA exoneration cases are shaking confidence in the criminal justice system. We welcome objective research that su ggests improvements in the identification process. ” (2003, p. 1). Researchers must move forward swiftly to address this compelling need. Eyewitness research has been recognized as one of the most successful collaborations between psychology and the legal system (see, e.g., Doyle, 2005). M ore recently , Doyle and colleagues (2006) urge street investigators, laboratory scientists and courtroom litigators to take responsibility for integrating science into practice and to combine their perspectives into a p r

oductive working consensus for action informed by science. T he translation of knowledge from the laboratory to field must be informed by scientific expertise at the same time that it is enriched by those who have a textured appreciation for the circumsta nces in which lineups are performed. One of the most powerful aspects of this research project is in its collaboration among p rofessional communities , a model for the bridging of research and practice. Double - blind sequential lineups 87 Table 1 . Descriptive data of the Hennepin C ount y Pilot Project ________________________________________________________________________ Population 280 Lineups 117 Cases 206 Eyewitnesses Minneapolis 382,618 138 (49%) 38 89 Bloomington 85,172 86 (31%) 48 69 Minnetonka 51,301 30 (11%) 14 26 New Hop e 20,873 26 (9%) 17 22 Results of 280 Lineups Most Frequent Crimes (of 27 Categories): Murder Assault Theft Forgery/Check 30% 24% 12% 8% Suspect Race: African American Caucasian Hispanic Native American Other No Report 163 75 16 8 6 12 58% 27% 6% 3% 3% 4% Witness Race: African American Caucasian Native American Hispanic Other No Report Cross - race IDs Same - race IDs No information 77 63 8 5 5 122 38 113 129 28% 22% 3% 2% 2% 44% 14% 40% 46% Double - blind sequential lineups 88 Suspect gender Male Female 249 30 8 9% 11% Witness gender Male Female No report 181 96 3 65% 34% 1% Weapon presence Gun Knife Other No weapon No report 90 16 13 14 147 32% 6% 5% 5% 53% Time between event and lineup Lineup within 1 week: within 2 weeks: within 3 weeks: within 4 weeks: Cum % 50% 64% 75% 82% Double - blind sequential lineups 89 Table 2 Instructions to Investigators (from Scoggin 2003) _____________________________________________________________________  Use existing Minnesota Repository of Arrest Photos parameters. These defaults incl ude the use of photographs depicting suspects of similar age, skin color, complexion, hairstyle, and build. Consistency is also required as to backdrop, the use of color or black and white suspect photos, and distinguishing characteristics such as facial hair, scars, eyeglasses, and clothing.  Use no less than six photographs. The suspect’s photo should be randomly placed in the array.  Preserve a copy of the photos in the order in which they were displayed. One way is to preserve the traditional simultane ous six - photo display.  Assemble a different group of photos using new fillers for each suspect.  Interview witnesses in private, separate from other witnesses.  Do not tell the witness that the suspect is in a group of photos. Rather, the witness should be told the suspect “may or may not be” in the group of photos displayed.  Tell the witness that the displaying officer does not know whether the suspect is in the group of photos.  If a witness is able to recognize the suspect from the photos, a statement fro m that witness should include a description of how certain the witness is of the identification. Numerical certainty (percentages) should be avoided, but a description of why the photo resembles the suspect is encouraged. The witness should initial and d ate any photo identified.

Double - blind sequential lineups 90  The officer displaying the photographs should report on how the identification was made, including the speed of the identification, statements of certainty made during the process, and any comments about why the photos do or do no t look like the suspect. The officer should not encourage the witness to focus on any particular photo.  Photos should be shown one at a time. While one photograph is being displayed, the other photographs should be face down or otherwise hidden.  The witn esses may look through the photos more than once, but all the photos should be shown each time. The number of times the photos were shown should be reported. The witness may take as long as necessary to examine each photograph.  If a witness identifies a suspect before looking at all the photos, the rest of the display should be shown and the witness asked to identify or eliminate each photograph.  The officer showing the display should not know wh ich photo depicts the suspect.  The officer assembling the ph otos should not be in the witness’s view during the display.  A knowledgeable officer should be available to clarify questions that arise during the identification process and to provide support after the process is completed.  Exceptions:  Sequential display s should not be used with witnesses of twelve years of age or younger.  The blind examination requirement may be disregarded if necessary. Officers should document why an uninformed officer was not available Double - blind sequential lineups 91 ( e.g. , it is 3:00 a.m. and no uninformed officer is available). Reports should also include why sequential identifications are not possible. Double - blind sequential lineups 92 Table 3 . Eyewitness Performance _____________________________________________________________________ Hennepin County (HC) Results HC SIM (Field)** Sim (Fi eld)*** SIM(Field) **** Suspect ID 54% 50% 52% 4 0% Filler ID 8% 24% 15% 2 0% No Choice 38% 26% 33% 40% In the laboratory* Simultaneous Lineups Sequential Lineups Perpetrator Present Perpetrator absent Perpetrator present Perpetrator absent Perp et rator ID Filler ID No Choice 50% 24% 26% 51% 49% 35% 19% 46% 28% 72% Notes to table: * Steblay et al., (2001) ** Behrman, & Davey, (2001). *** Behrman & Richards (2005) **** Valentine, Picking, & Darling (2003) Double - blind sequential lineups 93 Table 4 . Witness Deci sions for repeated viewing of the Sequential Lineup _____________________________________________________________________ All witnesses [Reporting lineups = 128] Laps Lineups (n) Suspect ID Filler ID No Choice 1 68 66% 3% 31% 2 42 50% 10% 40% 3 14 50% 14% 36% 4, 5 or 6 4 25% 75% Stranger crimes only [Reporting lineups = 78] Laps Lineups (n) Suspect ID Filler ID No Choice 1 33 42% 3% 55% 2 31 32% 13% 55% 3 or more 14 43% 29% 29% Double - blind sequential lineups 94 Table 5 . Eyewitness Response: Stranger versu s Familiar Perpetrator ______________________________________________________

__________________ In the Field: Hennepin County Stranger Perpetrator (n = 178) Familiar Perpetrator (n = 93) Suspect ID 35 % 90 % Filler 11 % 3 % No Choice 53 % 6 % In t he Lab: Sequential format , target present , final decisions Stranger Perpetrator (n = 66 ) Familiar Perpetrator (n = 35 ) Accurate ID 14 % 66 % Filler 39 % 11 % No Choice 47 % 23 % Double - blind sequential lineups 95 Table 6 . Opportunity to view perpetrator __________________________ ______________________________________________ STRANGER IDENTIFICATIONS Very Brief Interaction (157 Lineups) Eyewitness Decisions : 50 Suspect 18 Filler 89 No Choice 32% 11% 57% Choosing Rate Jump - outs 43% 12% Reported qualifiers to the choice R eported qualifiers about other lineup members 29 of 68 (43%) 28 of 157 (18%) Eyewitness status : Observer Victim Other knowledge 68% 27% 4% Weapon involved (reported): 29% Type of crime : (categor�ies 8%) Assault Theft Murder Forgery Burglary 31% 17% 15% 10% 10% Longer Interaction (22 Lineups) Eyewitness Decisions : 13 Suspect 3 Filler 6 No Choice 59% 14% 27% Choosing Rate Jump - outs 73% 32% Reported qualifiers to the choice Reported qualifiers about other lineup members 6 of 16 (38%) 4 of 22 (18%) Double - blind sequential lineups 96 Eyewitness status : Observer Victim 67% 33% Weapon involved (reported): 9% Type of crime : (categor�ies 8%) Fraud Assault Forgery Credit Fraud Theft/swindle Robbery Theft 23% 14% 9% 9% 9% 9% 9% FAMILIAR PERPETRATOR IDENTIFICATION S “Regular Customer”/Multiple Views (45 Lineups) Eyewitness Decisions : 38 Suspect 3 Filler 4 No Choice 84% 7% 9% Choosing Rate Jump - outs 91% 60% Reported qualifiers to the choice Reported qualifiers about other lineup members 12 of 41 (29%) 4 of 45 (9%) Eyewitness status : Observer Victim Other 62% 33% 2% Weapon involved (reported): 60% Type of crime : (categor�ies 8%) Murder Assault Theft 42% 24% 9% Know Well (including gang associations) (48 Lineups) Eyewitness Decisions : 46 Suspect 2 No Choice 96% 4% Choosing Rate Jump - outs 96% 90% Reported qualifiers to the choice 1 of 46 (2%) Double - blind sequential lineups 97 Reported qualifiers about other lineup members 5 of 48 (10%) Eyewitness status : Observer Victim Other 44% 31% 25% Weapon involved (reported): 81% Type of crime : (categor�ies 8%) Murder Assault 71% 19% Double - blind sequential lineups 98 Table 7 Eyewitness decisions in the laboratory: all participant - witnesses _____________________________________________________________________ Overall N=372 TARG ET PRESENT Sim Seq 1 Seq Final Seq Hold Seq + Correct ID 25% 26% 32% 40% 24% 22 26 32 19 13 Filler ID (error) 17% 20% 30% 21% 3 7 % 15 20 30 10 20 No choice (error) 58% 55% 39% 38% 39% ` 52 55 39 18 21 _________________

Choosing rate 42% 46% 62% 61% 63% N 89 101 101 47 54 TARGET ABSENT Sim Seq 1 Seq Final Seq Hold Seq + No choice (correct) 73% 74% 58% 64% 54% 59 75 59 30 29 Filler ID (error) 27% 26% 42% 36% 46% 22 26 42 17 25 _________________ Choosing rate 27% 26% 42% 36% 46% N 81 101 101 47 54 Double - blind sequential lineups 99 Table 8 Eyewitness Decisions in the laboratory: stranger perp etrators only ________________________________________________________________________ N= 253 (68% of 372 witnesses) TARGET PRESENT Sim Seq 1 Seq Final Seq Hold Seq + Correct ID 10% 8% 14% 15% 13% 6 5 9 4 5 Filler ID ( error) 23% 2 6 % 39% 37% 41% 14 1 7 26 10 16 No choice (error) 68% 6 7 % 47% 48% 46% ` 42 4 4 31 1 3 18 _________________ Choosing rate 33% 3 4 % 53% 52% 5 4 % N 62 66 66 27 39 TARGET ABSENT Sim Seq 1 Seq Final Seq Hold Seq + No choice (correct) 64% 68% 45% 44% 45 % 36 47 31 11 2 0 Filler ID (error) 36% 32% 55% 56% 55 % 20 22 38 14 2 4 _________________ Choosing rate 36 % 32% 55% 56% 55 % N 56 69 69 25 44 Double - blind sequential lineups 100 Table 9 Eyewitness decisions in the laboratory: familiar perpetrators only ________________________________________________________________________ N= 119 TARGET PRESENT Sim Seq 1 Seq Final Correct ID 59 % 60 % 66% 16 21 23 Filler ID (error) 4 % 9 % 11% 1 3 4 No choice (error) 37 % 3 1 % 23% ` 10 1 1 8 _________________ Choosing rate 63 % 6 9 % 77% N 27 35 35 TARGET ABSENT Sim Seq 1 SeqFinal No choice (correct) 92 % 8 8 % 88% 23 28 28 Filler ID (error) 8 % 1 3 % 13% 2 4 4 _________________ Choosing rate 8 % 1 3 % 13% N 25 32 32 Double - blind sequential lineups 101 Table 10 Eyewitness decisions: Lineup size adjusted for attrition due to familiarity (AF) _________________________________________________________________________ ( Subset of witnesses for whom the p erpetrators w as a stranger) Simultaneous Sequential Lineup size Choosing Accuracy Choosing Accuracy Overall (n) (n) 6 (full) 40% (53) 32% 38% (69) 36% 4 - 5 29% (45) 36% 32% (53) 38% 0 - 3 30% (20) 45% 8% (13) 54% Target - present 6 (full) 38% (29) 10% 34% (32) 9% 4 - 5 26% (23) 4% 38% (29) 7% 0 - 3 30 % (10) 20% 0% (5) 0%

Target - absent 6 (full) 42% (24) 58% 41% (37) 59 % 4 - 5 32% (22) 68% 25% (24) 75% 0 - 3 30% (10) 70% 13% (8) 88% Double - blind sequential lineups 102 Footnotes 1. Structural elements of the process (sequential format, suspect position in t he lineup, blind administration) were reported conscientiously , while communication elements (cautionary instruction, blind instruction, and explanation of the sequential procedure for the witness) were less often documented. Witness responses beyond choi ce were less frequently cited (eyewitness confidence, qualifiers, number of repetitions of the lineup). The nuances of “no choice” responses (“not there” vs. “don’t know”) were not required or reported. 2. Hennepin County’s procedure has been wrongly portrayed in the recent Mecklenburg report of the Illinois pilot study . Contrary to Mecklenburg’s claim, there was not an instruction to the witness to hold his/her responses until the end of the array. The Mecklenburg report, titled “Report to the Legisla ture of the State of Illinois: The Illinois Pilot Program on Sequential Double - Blind Identification Procedures” is available at: http://www.chicagopolice.org/IL%20Pilot%20on %20Eyewitness%20ID.pdf . 3. Note that in this discussion “choosing” includes any pick from the lineup, suspect or filler. “No choice” indicates that the witness did not pick any photo from the lineup. 4. Alpha level for all statistical tests was set at .05. 5. Thanks are due to the many student research assistants involved with data collection in Studies 2 and 3; Special thanks to Hannah Dietrich, Miranda Nelson, and Shannon Ryan. Double - blind sequential lineups 103 References Behrman, B . W., & Davey, S.L. (2001). Eyewitness i dentif ication in a ctual c riminal c ases: An a rchival a nalysis . Law & Human Behavior , 25 , 475 - 491. Behrman, B. W., & Richards, R.E. (2005). Suspect/foil identification in actual crimes and in the laboratory: A reality monitoring analysis. Law and Hum an Behavior , 29, 2 79 - 301 . Brigham, J. C., Meissner, C.A., & Wasserman, A.W. (1999). Applied issues in the construction and expert assessment of photo lineups. Applied Cognitive Psychology , S73 - S92. Charman, S . D ., & Wells, G.L. (2007 ). Eyewi tness lineups: Is the appearance - change instruction a good idea? ” Law and Human Behavior , 31, 3 - 22 . Clark, S . E., & Davey, S.L. (2005). The t arget - to - f oils s hift in s imultaneous and s equential l ineups ,” Law and Human Behavior , 29 , 151 - 172. C utler, B. L., & Penrod, S.D. (1988). Improving the reliability of eyewitness identification: Lineup construction and presentation,” Journal of Applied Psychology , 73, 28 1 - 290 . Diamond, S. (in press). Psychological c ontributors to e valuating w itness t estimony . In E. Borgida & S. Fiske (Eds.), Psychological Science in Court . B lackwell . Douglass, A . (nee Bradfield), Smith, C., & Fraser - Thill, R. (in press). A p roblem w ith d ouble - b lind p hotospread p rocedures: Photospread a dministrators u s e o ne e yewitness’s confidence to influence the i dentification of a nother e yewitness ,” Law and Human Behavior . Doob, A. N., & Kirshenbaum, H.M. (1973). Bias in police lineups – Partial remembering,” Journal of Police Science and Admi nistration , 1 (3), 287 - 293. Douglass, A . (nee Bradfield) & Steblay, N. (in press). Memory d istortion in e yewitnesses.

Double - blind sequential lineups 104 A m eta - a nalysis of the p ost - i dentification feedback e ffect . Journal of Applied Cognitive Psychology . Doyle, J . (2005). True Witness: Cops, Courts, Science and the Battle Against Misidentification . , New York: Palgrave Macmillan . Doyle, J . M., P en rod, S . , Kovera, M . B . , & Dysart, J. (2006). The street, the lab, the courtroom, the meeting room. Unpublished manuscr ipt . Greene, E. (1988). Judge’s instructions on eyewitness testimony: Evaluation and revision . Journal of Applied Social Psychology , 18 , 252 - 276 . Harris, M. J., & Rosenthal , R. (1985). Mediation of i nterpersonal e xpectancy e ffects: 31 m et a - a nalyses . Psychological Bulletin , 97 , 363 - 386 H aw, R. & Fisher , R.P. (2004). Effects of a dministrator - w itness c ontact on e yewitness i dentification a ccuracy . Journal of Applied Psychology , 89, 1106 - 1112. Klobuchar, A. (2004). Year 2004 Goals. A Pub lication of the Hennepin County Attorney’s Office . Klobuchar, A ., & Knight , S. (2005) New l ineup p rocedures c an r educe e yewitness m istakes . Minneapolis Star Tribune , ( Jan 12 ): 11A. Klobuchar, A . , Steblay, N.K.M., & Caligiuri , H.L. (2006). I mproving eyewitness identifications: Hennepin County’s blind sequential lineup pilot project,” Cardozo Public Law, Policy, and Ethics Journal , 2, 381 - 414. Lindsay, R.C.L., Lea, J.A., Nosworthy, G.J., Fulford, J.A., Hector, J., LeVan, V., & Seabrook, C. (1991). Biased lineups: Sequential presentation reduces the problem. Journal of Applied Psychology, 76 (6) , 796 - 802. Lindsay, R.C.L., Smith, S.M., & Pryke, S. (1999). Measures of lineup fairness: Do they Double - blind sequential lineups 105 postdict identifi cation accuracy?” Applied Cognitive Psychology 13 (1999). Lindsay, R.C.L., & Wells, G.L. (1985). Improving eyewitness identifications from lineups: Simultaneous versus sequential presentation. Journal of Applied Psychology, 70 , 55 6 - 561 . Malp ass, R. S. (1981). Effective size and defendant bias in eyewitness identification lineups. Law and Human Behavior , 5 (4), 299 - 309. Malpass, R. S., & Lindsay, R.C.L. (1999). Measuring lineup fairness,” Applied Cognitive Psychology , 13, 1 - 7. Mas on v. Baithwaite 432 U.S. 98, 114 (1977) McQuiston - Surrett, D ., Malpass, R.S., & Tredoux, C.G. (in press). Sequential vs. s imultaneous lineups: A review of methods, data, and theory . Psychology, Public Policy, and Law . Mecklenburg , S. (2006). Report to the Legislature of the State of Illinois: the Illinois p ilot p rogram on s equential d ouble - b lind i dentification p rocedures . Retrieved August 2006 from www.chicagopolice.org. Meissner, C.A., & Brigham, J.C. (2001). Thirty y ears of investigating the own - race bias in memory for faces: A meta - analytic review,” Psychology, Public Policy, & Law 7, 3 - 35. Neil v. Biggers 409 U.S. 188, 199 (1972) . National Institute of Justice , October, NCJ 178240. North Carolina Actual In nocence Commission . (2003). Recommendations for Eyewitness Identification 1 - 6 . Northampton Police Department (2005). Administration & Operations Manual; Eyewitness Identification Procedure . ( V isited September 1, 2006) available at Double - blind sequential lineups 106

http://www.innocenceproject.org/docs/Northampton_MA_ID_Protocols.pdf . O’Toole, T. P. (2006). What’s the matter with Illinois ? How an opportunity was squandered to conduct an important study on eyewitness identification procedures. The Champion (August), 18 - 2 3 . Patenaude, K . (2006). Police i dentification p rocedures: A t ime f or c hange . Cardozo Public Law, Policy, and Ethics Journal , 2, 415 - 419. Penrod, S . (200 3 ). Eyewitne ss identification evidence: How w ell a re w itnesses and p olice p erforming? Criminal Justice Magazine, 54 , 36 - 47. Phillips, M . R., McAuliff, BD., Kovera, M.B. & Cutler , B.L. (1999). D ouble - b lind p h o toarray a dministration as a s afeguard a gainst i nvestigator b ias . Journal of Applied Psychology , 84 , 940 - 951. Police Chief’s Association of Santa Clara County. (2002). Lineup Protocol for Law Enforcement. Available at http://www.innocenceproject.org/docs/Santa_Clara_Lineup_Protocols.p df. Pozzulo, J . D., & Balfour, J. (2006). Children and a dults’ e yewitness i dentification a ccuracy w hen a c ulprit c hanges h is a ppearance: Comparing s imultaneous and e limination l ineup p rocedures. Legal and Criminological Psychology , 11 , 25 - 34 ; Pozzulo, J . D., & Warren, K.L. (2003). Descriptions and i dentifications of s trangers by y outh and ad ult e yewitnesses. Journal of Applied Psychology , 84, 167 - 176. Rattner, A . (1988). Convicted but innocent: Wrongful conviction and the criminal just ice system . Law and Human Behavior , 12 , 28 9 - 293. Report of the Governor’s Commission on Ca pital Punishment (2002), 31 - 40. Retrieved September 17, 2006 from Double - blind sequential lineups 107 http://www.idoc.state.il.us/ccp/ccp/reports/commission_report/index.html. Rosent hal, R . (2002). Covert c ommunication in c lassrooms, c linics, c ourtrooms, and c ubicles . American Psychologist , 57, 834 - 849. Rosenthal, R., & Rubin, D.B. (1978). Interpersonal expectancy effects: The first 345 studies. Behavioral and Brain Sci ences , 3, 377 - 386 Scoggin, P . (2003). P ilot Program for the Sequential Identification Process, ” 1 (October 27) Memorandum , Hennepin County Attorney’s Office , 1 - 3. Scoggin, P. (2004). Personal co mmunication, March 29. Scoggin, P. (2005). The us ual suspects: A pilot project to improve eyewitness identifications from lineups . Hennepin County report, 1 - 5 . Scoggin, P. (2005). Sequential identification, revised protocol. Hennepin County r eport (May 19), 1 - 4. Shapiro, P.N., & Penrod, S. (1986). Meta - a nalysis of f acial i dentification s tudies . Psychological Bulletin , 100 , 1 39 - 156. Simmons v. United States. 390 U.S. 377, 384 (1968). Sporer, S . (1993). Eyewitness i dentification a ccuracy, c onfidence, and d ecision t imes in s imult aneous and s equential l ineups . Journal of Applied Psychology , 78(1) , 22 - 33. Sporor, S . , Penrod, S., Read, J.D., & Cutler , B. (1995). Choosing, c onfidence, and a ccuracy: A m eta - a nalysis of the c onfidence - a ccuracy r elation in e yewitness i dentification s tudies . Psychological Bulletin , 118 , 315 - 327. State o f New Jersey .. (2002). Attorney General g uidelines for p reparing and c onducting p hoto and l ive l ineup i dentification p rocedures . State of Illinois, Report of the Governor’s Commission on Capital Punishment i (Apr 15) , 31 - 40. Double - blind sequentia

l lineups 108 State of Wisconsin Office of the Attorney General. (2005). Eyewitness i dentification p rocedure r ecomme ndations. Model Policy and Procedure for Eyewitness Identification . (April). State of Wisconsin Offi ce of the Attorney General. (2006, July 21). Response to Chicago Report on Eyewitness identification Procedures. Wis. Act 60, 17 (to be codified as Wis. Stat 175.50(2), 1 - 4. Steblay, N . (1997). Social i nfluence in e yewitness r ecall: A m eta - a n alytic r eview of l ineup i nstruction e ffects ,” Law and Human Behavior , 21 , 283 - 297. Steblay, N. (2006). Observations on the Illinois data. Available at http://www.augsburg.edu/p sychology/steblay.html . Steblay, N., Dysart, J., Fulero, S., & Lindsay, R.C.L.. (2001). Eyewitness a ccuracy ra tes in s equential and s imultaneous l ineup p resentations: A m eta - a nalytic c omparison. Law and Human Behavior , 25 , 459 - 473. Stova ll v. Denno, 388 U.S. 301 - 302 (1967). Technical Working Group for Eyewitness Accuracy . (1999). Eyewitness Evidence: A Guide for Law Enforcement , Research Report, Washington, DC: U.S. Department of Justice . The Innocence Project: Mistaken I.D., Retri eved August 31, 2006 from http://www.innocenceproject.org/causes/mistakenid.php The Innocence Project Retrieved July 2006 from www.innocenceproject.org . United States v. Ash 413 U.S. 300, 321 (1973). U.S. v. Wade 388 U.S. 224 - 225 (1967). V alentine, T ., & H eaton , P. (1999). An e valuation of the f airness of p olice l ineups and v ideo i dentifications . Applied Cognitive Psycholo gy , 13, S59 - S72 . Double - blind sequential lineups 109 Valentine, T., Pickering, A., & Darling, S. (2003). Characteristics of eyewitness identification that predict the outcome of real lineups. Applied Cognitive Psychology, 17 , 969 - 993. Virginia State Crime Commission . (2005). Report of the Virginia State Crime Commission on Mistaken Eyewitness Identification to the Governor and the General Assembly of Virginia, House Document No. 40 1, 14 - 15. Wells, G . L. (1984). The p sychology of lineup i dentifications . Journal of Applied Social Psychology , 36 , 1546 - 1557. Wells, G.L. (1988). Eyewitness Identification: A System Handbook . Toronto: Carswell Legal Publications. Wells, G. L. (1993). What d o w e know a bout e yewitness i dentification? American Psych ologist , 48 , 553 - 571. Wells, G. L. (2006 a ). Does the sequential lineup reduce accurate identifications in addition to reducing mistaken identifications? Document r etrieved from www.psychology.iastate.edu/faculty/gwells/SequentialNotesonlossofhits.htm . Wells, G . L. (2006b). An i mportant n ote on f ield s tudies of e yewitness i dentifications from l ineups: Filler i dentifications are “ c onditional p ro xy m easures”. Draft document available at http://www.psychology.iastate.edu/faculty/gwells Wells, G.L., & B radfield, A. (1998). ‘ Good, y ou i dentified the s uspect ’ : Feedback to e yewitnesses d istorts t heir r eports of the w itnessing e xperience . Journal of Applied Psychology, 83 , 360 - 376. Wells, G. L., & Bradfield, A.L. (1999). Measuring the goodness of lineups: Parameter estimation, question effects, and limits to the m ock witness paradigm,” Applied Double - blind sequential lineups 110 Cognitive Psychology , 13, S 27 - S 39. Wells, G. L., Leippe, M.R., & Ostrom, T.M. (1980). Guidelines

for empirically assessing the fairness of a lineup. Law and Human Behavior , 3 (4), 285 - 292. Wells, G. L., Luus, C.A .E., & Windschitl, P. (1994). Maximizing the utility of eyewitness identification evidence. Current Directions in Psychological Science , 3(6), 194 - 197. W ells, G.L. , Malpass, R . S., Lindsay, R.C.L., Fisher, R . P., Turtle, J . W., & Fulero, S. M. (2000). From the l ab to the p olice s tation: A s uccessful a pplication of e yewitness r esearch ,” American Psychologist , 55 , 581 - 598 . Wells, G. L., & Olson, E. (2002). Eyewitness identification: Information gain from incriminating and exonerating behaviors. Journal of Experimental Psychology: Applied, 8 (3) , 155 - 167. Wells, G . L., & Olson , E.A. (2003). Ey e witness t estimony , Annual Review of Psychology 54, 277 - 295 . Wells, G. L., & Seelau, E. P, (1995). Eyewitness identification: Psyc hological research and legal policy on lineups. Psychology, Public Policy and Law , 1, 776 - 778. Wells, G . L., Small , M., Penrod, S., Malpass, R.S., Fulero, S.M., & Brimacombe , C.A.E. (1998). Eyewitness i dentification p rocedures: Recommendati ons for l ineups and p hotospreads . Law and Human Behavior , 22 , 603 - 647. Wells, G . L., & Turtle , J. (1986). Eyewitness identification: The importance of lineup models. Psychological Bulletin , 99 , 320 - 329. Windschitl, P . D., & Chambers, J.R. (20 04). The dud - alternative effect in likelihood judgment . Journal of Experimental Psychology: Learning, Memory, and Cognition , 30 (1) , 19 8 - 215 . The author(s) shown below used Federal funds provided by the U.S. Department of Justice and prepared the following final report: Document Title: Double - Blind Sequential Police Lineup Procedures: Toward an Integrated Laboratory & Field Practice Perspective Author(s): Nancy K. Steblay Document No.: 246939 Date Received: May 2014 Award Number: 2004 - IJ - CX - 0044 This report has not been published by the U.S. Department of Justice. To provide better customer service, NCJRS has ma de this Federally - funded grant report available electronically. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 5 sequentially, and(4) to assess the impact on eyewitness decisions of a naturally-occurring reduction of lineup size due to a witness’s familiarity with lineup foils.The collaboration between the County Attorney’s Office and the research lab also facilitated the final project objective, (5) the integration of laboratory and field data to generate practical, empirical, and theoretical knowledge of effective lineup procedure, including identification of lessons as to how more effective lineup procedures can best be brought to practice. Brief rationale for the project Past scientific research has led to a cohesive lineup prototype that promises significant improvement in eyewitness accuracy (Well, Small, Penrod, Malpass, Fulero, & Brimacombe, 1998).Advancement has been achieved through creative application of insightful theory and rigorous laboratory testing. One compelling feature of the research endeavor is the relative speed at which lab results have been usefully applied to police practice (Wells, Malpass, Lindsay, Fisher, Turtle, & Fulero, 2000). The laboratory research has resulted in guidelines for law enforcement aimed at preventing witness errors in police lineup identifications (Technical Working Group for Eyewitness Accuracy, 1999, hereinafter The NIJ Guide).The recommendations include refinements in both lineup structu

re and procedure. More recently, scientists have advised police to also corporate double-blind administration and a sequential photo presentation format into their lineup procedures (Wells, et al., 2000). Versions of th newer protocol now are being introduced into practice in a number of jurisdictions; one such instance is a year-long lineup pilot project in Hennepin County, Minnesota. As scientific research is translated to public policy, as is the case in Hennepin County, it is desirable to maintain a timely exchange of information among scientists, law This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 4 Blind/Sequential Police Lineup Procedures: Toward an Integrated Laboratory & Field Practice Perspective The purpose of this project was to join data from past scientific eyewitness research, current field experience, and new laboratory investigation to advance nowledge of best police lineup practice for law enforcement and research communities. e project includedcomponents: (1) a pilot program of double-blind sequential lineups in Hennepin County, Minnesota, and (2) a laboratory investigation of eyewitness memory under revised lineup procedures. ese combined sources of data field and laboratory have generated valuable new information about successful field practice as well as scientifically sound answers to questions regarding the effects of specific changes prescribed lineup protocol. Such knowledge can contribute enormously to policy reform efforts. Objectives The data generated by the Hennepin County (HC) pilot lineup program addresstwo primary objectives: ) collection and analysis of descriptive data detailing field lineup administration, situational context, and witness response, to form the first available scientific baseline measure for double-blind sequential lineup practice; and (2) a cohesive and instructive summary of the HC lineup implementation process. Two questions prompted HC program and data analysis: Can double-blind sequential lineup procedures be effectively implemented in the field? Do the number and quality of identifications change with the blind-sequential procedure? The laboratory objectives involved controlled testing of two specific deviations from prescribed lineup protocol: (3) to etermine the impact on eyewitness lineup decisions of a witness’s second (or third) viewing of a lineup in which the photos were presented This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 3 familiarity with fillers(5) Integration of laboratory and field data to generate practical, empirical, and theoretical knowledge of effective lineup procedure; and (6) Practical and scholarly presentations and publications as appropriate to law enforcement professionals, psycho-legal research community, and the NIJ Data Resource Program. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. ��

Double-blind sequential lineups 2 Abstract The project purpose was to join behavioral data from scientific research, current field experience, and new laboratory investigation to advance knowledge of best police lineup practice for law enforcement and research communities. The project s a collaborative effort between the Hennepin County (Minnesota) Attorney’s Office and the PI’sresearch laboratory. Three data collection and analysis components were completed(1) Hennepin County’s pilot implementation of double-blind sequential lineup procedures, including field lineups; (2) boratory evaluation of the quality of the Hennepin County lineups; and (3) an experimental laboratory test of how revisions to prescribed lineup protocol affect eyewitness lineup decisions. The Hennepin County (HC) results indicate a successful application of double-blind sequential lineups to street investigations. Double-blind sequential lineups are now established county-wide, providing a standardized scientifically-based lineup procedure that has been demonstrated to be practicable in real cases. field data and laboratory test data converged to demonstrate increased misidentifications when a witness is allowed to view the lineup more than once. The study also revealed how reduced lineup sizeattrition due to the witness’s recognition of can negatively affect eyewitness identification accuracy. Completed grant objectives included: (1) Descriptive data providing the first available baseline measure for blind sequential field lineup practice; (2) Summary of the field lineup implementation process(3) A laboratory test of the impact on eyewitness decisions of an opportunity for repeated viewing of the sequential lineup (4) A laboratory test of the effect on eyewitness decisions of a reduction in lineup size through witness This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 1 Double-Blind Sequential Police Lineup Procedures: Toward an Integrated Laboratory & Field Practice Perspective Final Report Grant -0044 March 31, 2007 Nancy K. Steblay Augsburg College This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 110 Cognitive Psychology,S27-S39. Wells, G. L., Leippe, M.R., & Ostrom, T.M. (1980). Guidelines for empirically assessing the fairness of a lineup. Law and Human Behavior,3 (4), 285-292. Wells, G. L., Luus, C.A.E., & Windschitl, P. (1994). Maximizing the utility of eyewitness identification evidence. Current Directions in Psychological Science,3(6), 194-197. Wells, G.L., Malpass, R.S., Lindsay, R.C.L., Fisher, R. P., Turtle, J. W., & Fulero, S. (2000). From the lab to the police station: A successful application ofeyewitness researchAmerican Psychologist-598Wells, G. L., & Olson, E. (2002). Eyewitness identification: Information gain from incriminating and exonerating behaviors. Journal of Experimental Psychology: Applied, 8 (3)-167. Wells, G. L., & Olson, E.A. (2003). ewitness testimony,Annual Review of Psychology -295Wells, G. L., & Seelau, E. P, (1995). Eyewitness identification: Psychological research and legal policy on lineups. Psychology, Public Policy and Law-778. Wells, G. L., Small, M., Penrod, S., Malpass, R.S., Fulero, S.M., & Brimacombe,C.A.E. (1998).

Eyewitness identification procedures: Recommendations for lineups and photospreads. Law and Human Behavior,, 603-647.Wells, G. L., & Turtle, J. (1986). Eyewitness identification: The importance of lineup Psychological Bulletin, -329. Windschitl, P.D., & Chambers, J.R. (2004). The dud-alternative effect in likelihood judgment.Journal of Experimental Psychology: Learning, Memory, and Cognition,30 (1),-215. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 109 Valentine, T., Pickering, A., & Darling, S. (2003). Characteristics of eyewitness identification that predict the outcome of real lineups. Applied Cognitive Psychology, 17, 969-993. Virginia State Crime Commission. (2005). Report of the Virginia State Crime Commission on Mistaken Eyewitness Identification to the Governor and the General Assembly of Virginia, House Document No. 40 1, 14-15. Wells, G. L. (1984). The psychology of lineup identifications. Journal of Applied Social Psychology-1557.Wells, G.L. (1988). Eyewitness Identification: A System Handbook. Toronto: Carswell Legal Publications. Wells, G. L. (1993). What dknow about eyewitness identification? American Psychologist553-571.Wells, G. L. (2006a). Does the sequential lineup reduce accurate identifications in addition reducing mistaken identifications? Document retrieved from www.psychology.iastate.edu/faculty/gwells/SequentialNotesonlossofhits.htm . Wells, G. L. (2006b). An important note on field studies of eyewitness identifications from lineups: Filler identifications are “conditional pxy measures”.Draft document available at http://www.psychology.iastate.edu/faculty/gwells Wells, G.L., & Bradfield, A. (1998). ‘Good, you identified the suspect: Feedback to eyewitnesses distorts their reports of the witnessing experience. Journal of Applied Psychology, 83-376. Wells, G. L., & Bradfield, A.L. (1999). Measuring the goodness of lineups: Parameter estimation, question effects, and limits to the mock witness paradigm,” Applied This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 108 State of Wisconsin Office of the Attorney General. (2005). Eyewitness identification procedure recommendations. Model Policy and Procedure for Eyewitness Identification(April). State of Wisconsin Office of the Attorney General. (2006, July 21). Response to Chicago Report on Eyewitness identification Procedures. Wis. Act 60, 17 (to be codified as Wis. Stat 175.50(2), 1-4. Steblay, N. (1997). Social influence in eyewitness recall: A meta-analytic review of lineup instruction effectsLaw and Human Behavior,-297. Steblay, N. (2006). Observations on the Illinois data. Available at http://www.augsburg.edu/psychology/steblay.html . Steblay, N., Dysart, J., Fulero, S., & Lindsay, R.C.L.. (2001). Eyewitness accuracy tes in sequential and simultaneous lineup presentations: A meta-analytic comparison. Law and Human Behavior, 459-473.Stovall v. Denno, 388 U.S. 301-302 (1967). Technical Working Group for Eyewitness Accuracy. (1999). Eyewitness Evidence: A Guide for Law Enforcement, Research Report, Washington, DC: U.S. Department of Justice.The Innocence Project: Mistaken I.D., Retrieved August 31, 2006 from http://www.innocenceproject.

org/causes/mistakenid.php The Innocence Project Retrieved July 2006 from www.innocenceproject.org . United States v. Ash 413 U.S. 300, 321 (1973). U.S. v. Wade 388 U.S. 224-225(1967). Valentine, T.,Heaton, P. (1999). An evaluation of the fairness of olice lineups and video identifications. Applied Cognitive Psychology,S59-S72 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 107 http://www.idoc.state.il.us/ccp/ccp/reports/commission_report/index.html. Rosenthal, R. (2002). Covert communication in classrooms, clinics, courtrooms, and cubicles.American Psychologist-849.Rosenthal, R., & Rubin, D.B. (1978). Interpersonal expectancy effects: The first 345 studies. Behavioral and Brain Sciences, 3, 377-386 Scoggin, P. (2003). Pilot Program for the Sequential Identification Process,1 (October 27) Memorandum, Hennepin County Attorney’s Office-3. Scoggin, P. (2004). Personal communication, March 29. Scoggin, P. (2005). The usual suspects: A pilot project to improve eyewitness identifications from lineups.Hennepin County report, -5 Scoggin, P. (2005). Sequential identification, revised protocol. Hennepin County report (May 19), 1-4. Shapiro, P.N., & Penrod, S. (1986). Meta-analysis of facial identification studies. Psychological Bulletin,-156.Simmons v. United States. 390 U.S. 377, 384 (1968). Sporer, S. (1993). Eyewitness identification accuracy, confidence, and decision times in simultaneous and sequential lineups. Journal of Applied Psychology,78(1),22-33. Sporor, S., Penrod, S., Read, J.D., & Cutler, B. (1995). Choosing, confidence, and accuracy: meta-analysis of the confidence-accuracy relation in eyewitness identification studies. Psychological Bulletin,-327.State of New Jersey.. (2002). Attorney General guidelines for preparing and conducting photo and live lineup identification proceduresState of Illinois, Report of the Governor’s Commission on Capital Punishmenti (Apr 15), 31-40. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 106 http://www.innocenceproject.org/docs/Northampton_MA_ID_Protocols.pdf . O’Toole, T. P. (2006). What’s the matter with Illinois? How an opportunity was squandered to conduct an important study on eyewitness identification procedures. The Champion(August), 18-2Patenaude, K. (2006). Police identification procedures: A time for change. Cardozo Public Law, Policy, and Ethics Journal,-419. Penrod, S. (2003). Eyewitness identification evidence: How well are witnesses and police performing? Criminal Justice Magazine, 54, 36-47.Phillips, M. R., McAuliff, BD., Kovera, M.B. & Cutler, B.L. (1999). Double-blind toarray administration as a safeguard against investigator ias. Journal of Applied Psychology,-951. Police Chief’s Association of Santa Clara County. (2002). Lineup Protocol for Law Enforcement. Available at http://www.innocenceproject.org/docs/Santa_Clara_Lineup_Protocols.pdf. Pozzulo, J.D., & Balfour, J. (2006). Children and adults’ eyewitness identification accuracy when a culprit changes his appearance: Comparing simultaneous and elimination lineup procedures. Legal and Criminological Psychology,-34; Pozzulo, J.D., & Warren, K.L. (2003). Descriptions and identifications of strangers by youth and ult

eyewitnesses. Journal of Applied Psychology,167-176. Rattner, A. (1988). Convicted but innocent: Wrongful conviction and the criminal justice system.Law and Human Behavior,-293. Report of the Governor’s Commission on Capital Punishment (2002), 31-40. Retrieved September 17, 2006 from This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 105 postdict identification accuracy?” Applied Cognitive Psychology13 (1999). Lindsay, R.C.L., & Wells, G.L. (1985). Improving eyewitness identifications from lineups: Simultaneous versus sequential presentation. Journal of Applied Psychology, 70- Malpass, R. S. (1981). Effective size and defendant bias in eyewitness identification lineups. Law and Human Behavior,5 (4), 299-309. Malpass, R. S., & Lindsay, R.C.L. (1999). Measuring lineup fairness,” Applied Cognitive Psychology,-7. Mason v. Baithwaite 432 U.S. 98, (1977) McQuiston-Surrett, D., Malpass, R.S., Tredoux, C.G. (in press). Sequential vs. simultaneous lineups: A review of methods, data, and theory.Psychology, Public Policy, and Law. Mecklenburg, S. (2006). Report to the Legislature of the State of Illinois: the Illinois pilot program on sequential double-blind identification procedures. Retrieved August 2006 from www.chicagopolice.org. Meissner, C.A., & Brigham, J.C. (2001). Thirty years of investigating the own-race bias in memory for faces: A meta-analytic review,” Psychology, Public Policy, & Law-35.Neil v. Biggers 409 U.S. 188, 199 (1972). National Institute of Justice, October, NCJ 178240. North Carolina Actual Innocence Commission. (2003). Recommendations for Eyewitness Identification 1-6. Northampton Police Department (2005). Administration & Operations Manual; Eyewitness Identification Procedure. (Visited September 1, 2006) available at This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 104 meta-analysis of the post-identification feedback effect. Journal of Applied Cognitive Psychology Doyle, J. (2005). True Witness: Cops, Courts, Science and the Battle Against Misidentification.New York: Palgrave Macmillan.Doyle, J.rod, S., Kovera, M.B., & Dysart, J. (2006). The street, the lab, the courtroom, the meeting room. Unpublished manuscript. Greene, E. (1988). Judge’s instructions on eyewitness testimony: Evaluation and revision.Journal of Applied Social Psychology,-276Harris, M. J., & Rosenthal, R. (1985). Mediation of interpersonal expectancy effects: 31 a-analyses.Psychological Bulletin363-386 Haw, R. & Fisher, R.P. (2004).Effects of administrator-witness contact on eyewitness identification accuracy. Journal of Applied Psychology-1112. Klobuchar, A. (2004). Year 2004 Goals. A Publication of the Hennepin County Attorney’s OfficeKlobuchar, A., & Knight, S. (2005) New lineup procedures can reduce eyewitness mistakes.Minneapolis Star Tribune(Jan 1211A. Klobuchar, A., Steblay, N.K.M., & Caligiuri, H.L. (2006). Improving eyewitness identifications: Hennepin County’s blind sequential lineup pilot project,”Cardozo Public Law, Policy, and Ethics Journal,-414. Lindsay, R.C.L., Lea, J.A., Nosworthy, G.J., Fulford, J.A., Hector, J., LeVan, V., Seabrook, C. (1991). Biased lineups: Sequential presentation reduces the problem

. Journal of Applied Psychology, 76 (6)-802. Lindsay, R.C.L., Smith, S.M., & Pryke, S. (1999). Measures of lineup fairness: Do they This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 103 References Behrman, B. W., & Davey, S.L. (2001). Eyewitness identification in actual criminal cases: An archival analysis. Law & Human Behavior,-491. Behrman, B. W., & Richards, R.E. (2005). Suspect/foil identification in actual crimes and in the laboratory: A reality monitoring analysis. Law and Human Behavior,29, 279-301Brigham, J. C., Meissner, C.A., & Wasserman, A.W. (1999). Applied issues in the construction and expert assessment of photo lineups. Applied Cognitive Psychology, S73-S92. Charman, S. D., & Wells, G.L. (2007). Eyewitness lineups: Is the appearance -change instruction a good idea?Law and Human Behavior, 31, -22Clark, S.E., & Davey, S.L. (2005). The target--foils shift in simultaneous and sequential lineupsLaw and Human Behavior-172. Cutler, B. L., & Penrod, S.D. (1988). Improving the reliability of eyewitness identification: Lineup construction and presentation,” Journal of Applied Psychology, 73, 281-290Diamond, S. (in press). Psychological contributors to evaluating witness testimony. In E. Borgida & S. Fiske (Eds.),Psychological Science in Court. Blackwell. Douglass, A. (nee Bradfield), Smith, C., & Fraser-Thill, R. (in press). A roblem with double-blind photospread procedures: Photospread administrators uyewitness’s confidence to influence the identification of another eyewitnessLaw and Human Behavior. Doob, A. N., & Kirshenbaum, H.M. (1973). Bias in police lineups Partial remembering,” Journal of Police Science and Administration, 1 (3), 287-293. Douglass, A. (nee Bradfield) & Steblay, N. (in press). Memory distortion in eyewitnesses. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 102 Footnotes Structural elements of the process (sequential format, suspect position in the lineup, blind administration) were reported conscientiously, while communication elements (cautionary instruction, blind instruction, and explanation of the sequential procedure for the witness) were less often documented. Witness responses beyond choice were less frequently cited (eyewitness confidence, qualifiers, number of repetitions of the lineup). The nuances of “no choice” responses (“not there” vs. “don’t know”) were not required or reported. Hennepin County’s procedure has been wrongly portrayed in the recent Mecklenburg report of the Illinois pilot study.Contrary to Mecklenburg’s claim, there was not an instruction to the witness to hold his/her responses until the end of the array. The Mecklenburg report, titled “Report to the Legislature of the State of Illinois: The Illinois Pilot Program on Sequential Double-Blind Identification Procedures” is available at: http://www.chicagopolice.org/IL%20Pilot%20on%20Eyewitness%20ID.pdf Note that in this discussion “choosing” includes any pick from the lineup, suspect or filler. “No choice” indicates that the witness did not pick any photo from the lineup.Alpha level for all statistical tests was set at .05. 5. Thanks are due to the many student research assistants involved w

ith data collection in Studies 2 and 3; Special thanks to Hannah Dietrich, Miranda Nelson, and Shannon Ryan. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 101 Table Eyewitness decisions: Lineup size adjusted for attrition due to familiarity (AF) _________________________________________________________________________ (Subset of witnesses for whom the perpetrators was stranger) Simultaneous Sequential Lineup size Choosing Accuracy Choosing Accuracy Overall (n) (n) 6 (full) 40% (53) 32% 38% (69) 36% -5 29% (45) 36% 32% (53) 38% 0-3 30% (20) 45% 8% (13) 54% Target-present 6 (full) 38% (29) 10% 34% (32) 9% -5 26% (23) 4% 38% (29) 7% 0-3 30% (10) 20% 0% (5) 0% Target-absent 6 (full) 42% (24) 58% 41% (37) 59% -5 32% (22) 68% 25% (24) 75% 0-3 30% (10) 70% 13% (8) 88% This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 100 Table 9 Eyewitness decisions in the laboratory: familiar perpetrators only ________________________________________________________________________ N= 119 TARGET PRESENT Sim Seq1 SeqFinal Correct ID 59% 60% 66% 16 21 23Filler ID (error) 4% % 11% 1 No choice (error) 37% 3% 23% ` 10 1 Choosing rate 63% 6% 77% N 27 35 35 TARGET ABSENT Sim Seq1 SeqFinal No choice (correct) 92 8% 88% 23 28 28Filler ID (error) 8% 1% 13% 2 4 Choosing rate 8% 1% 13% N 25 32 32 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 99 Table 8 Eyewitness Decisions in the laboratory: stranger perpetrators only ________________________________________________________________________ N= 253 (68% of 372 witnesses) TARGET PRESENT Sim Seq1 SeqFinal SeqHold Seq+ Correct ID 10% 8% 14% 15% 13% 6 Filler ID (error) 23% 2% 39% 37% 41% 14 1 26 10 16 No choice (error) 68% 6% 47% 48% 46% ` 42 4 31 1 18Choosing rate 33% 3% 53% 52% 5% N 62 66 27 39 TARGET ABSENT Sim Seq1 SeqFinal SeqHold Seq+ No choice (correct) 64% 68% 45% 44% 45% 36 47 31 11 2 Filler ID (error) 36% 32% 55% 56% 55% 20 22 38 2 Choosing rate 36% 32% 55% 56% 55% N 56 69 69 25 44 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. ��

Double-blind sequential lineups 98 Table 7 Eyewitness decisions in the laboratory: all participant-witnesses _____________________________________________________________________ Overall N=372 TARGET PRESENT Sim Seq1 SeqFinal SeqHold Seq+ Correct ID 25% 26% 32% 40% 24% 22 26 32 19 13Filler ID (error) 17% 20% 30% 21% 3% 15 20 30 10 20 No choice (error) 58% 55% 39% 38% 39% ` 52 55 39 18 21Choosing rate 42% 46% 62% 61% 63% N 89 101 101 47 54 TARGET ABSENT Sim Seq1 SeqFinal SeqHold Seq+ No choice (correct) 73% 74% 58% 64% 54% 59 75 59 30 29 Filler ID (error) 27% 26% 42% 36% 46% 22 26 42 25 Choosing rate 27% 26% 42% 36% 46% N 81 101 101 47 54 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 97 Reported qualifiers about other lineup members 5 of 48 (10%) Eyewitness status : Observer Victim Other 44% 31% 25% Weapon involved (reported): 81% Type of crime : (categor�ies 8%) Murder Assault 71% 19% This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 96 Eyewitness status : Observer Victim 67% 33% Weapon involved (reported): 9% Type of crime : (categor�ies 8%) Fraud Assault Forgery Credit Fraud Theft/swindle Robbery Theft 23% 14% 9% 9% 9% 9% 9% FAMILIAR PERPETRATOR IDENTIFICATION S “Regular Customer”/Multiple Views (45 Lineups) Eyewitness Decisions : 38 Suspect 3 Filler 4 No Choice 84% 7% 9% Choosing Rate Jump - outs 91% 60% Reported qualifiers to the choice Reported qualifiers about other lineup members 12 of 41 (29%) 4 of 45 (9%) Eyewitness status : Observer Victim Other 62% 33% 2% Weapon involved (reported): 60% Type of crime : (categor�ies 8%) Murder Assault Theft 42% 24% 9% Know Well (including gang associations) (48 Lineups) Eyewitness Decisions : 46 Suspect 2 No Choice 96% 4% Choosing Rate Jump - outs 96% 90% Reported qualifiers to the choice 1 of 46 (2%) This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 95 Table 6. Opportunity to view perpetrator ________________________________________________________________________ STRANGER IDENTIFICATIONS Very Brief Interaction (157 Lineups) Eyewitness Decisions : 50 Suspect 18 Filler 89 No Choice 32% 11% 57% Choosing Rate Jump - outs 43% 12% Reported qualifiers to the choice R eported qualifiers about other line

up members 29 of 68 (43%) 28 of 157 (18%) Eyewitness status : Observer Victim Other knowledge 68% 27% 4% Weapon involved (reported): 29% Type of crime : (categor�ies 8%) Assault Theft Murder Forgery Burglary 31% 17% 15% 10% 10% Longer Interaction (22 Lineups) Eyewitness Decisions : 13 Suspect 3 Filler 6 No Choice 59% 14% 27% Choosing Rate Jump - outs 73% 32% Reported qualifiers to the choice Reported qualifiers about other lineup members 6 of 16 (38%) 4 of 22 (18%) This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 94 Table 5. Eyewitness Response: Stranger versus FamiliarPerpetrator ________________________________________________________________________ In the Field: Hennepin County Stranger Perpetrator (n = 178) Familiar Perpetrator (n = 93) Suspect ID 35 % 90 % Filler 11 % 3 % No Choice 53 % 6 % In the Lab: Sequential format, target present, final decisions Stranger Perpetrator (n = 66 ) Familiar Perpetrator (n = 35 ) Accurate ID 14 % 66 % Filler 39 % 11 % No Choice 47 % 23 % This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 93 Table 4. Witness Decisions for repeated viewing of the Sequential Lineup _____________________________________________________________________ All witnesses [Reporting lineups = 128] Laps Lineups (n) Suspect ID Filler ID No Choice 1 68 66% 3% 31% 2 42 50% 10% 40% 3 14 50% 14% 36% 4, 5 or 6 4 25% 75% Strangercrimes only [Reporting lineups = 78] Laps Lineups (n) Suspect ID Filler ID No Choice 1 33 42% 3% 55% 2 31 32% 13% 55% 3 or more 14 43% 29% 29% This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 92 Table 3. Eyewitness Performance _____________________________________________________________________ Hennepin County (HC) Results HC SIM (Field)** Sim (Fi eld)*** SIM(Field) **** Suspect ID 54% 50% 52% 4 0% Filler ID 8% 24% 15% 2 0% No Choice 38% 26% 33% 40% In the laboratory* Simultaneous Lineups Sequential Lineups Perpetrator Present Perpetrator absent Perpetrator present Perpetrator absent Perp et rator ID Filler ID No Choice 50% 24% 26% 51% 49% 35% 19% 46% 28% 72% Notes to table: Steblay et al., (2001) ** Behrman, & Davey, (2001). *** Behrman & Richards (2005) **** Valentine, Picking, & Darling (2003) This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are t

hose of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 91 e.g., it is 3:00 a.m. and no uninformed officer is available). Reports should also include why sequential identifications are not possible. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 90 The officer displaying the photographs should report on how the identification was made, including the speed of the identification, statements of certainty made during the process, and any comments about why the photos do or do not look like the suspect. The officer should not encourage the witness to focus on any particular photo. Photos should be shown one at a time. While one photograph is being displayed, the other photographs should be face down or otherwise hidden. The witnesses may look through the photos more than once, but all the photos should be shown each time. The number of times the photos were shown should be reported. The witness may take as long as necessary to examine each photograph. If a witness identifies a suspect before looking at all the photos, the rest of the display should be shown and the witness asked to identify or eliminate each photograph. The officer showing the display should not know which photo depicts the suspect. The officer assembling the photos should not be in the witness’s view during the display. A knowledgeable officer should be available to clarify questions that arise during the identification process and to provide support after the process is completed. Exceptions: Sequential displays should not be used with witnesses of twelve years of age or younger. The blind examination requirement may be disregarded if necessary. Officers should document why an uninformed officer was not available This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 89 Table 2 Instructions to Investigators (from Scoggin 2003) _____________________________________________________________________ Use existing Minnesota Repository of Arrest Photos parameters. These defaults ude the use of photographs depicting suspects of similar age, skin color, complexion, hairstyle, and build. Consistency is also required as to backdrop, the use of color or black and white suspect photos, and distinguishing characteristics such as facial hair, scars, eyeglasses, and clothing. Use no less than six photographs. The suspect’s photo should be randomly placed in the array. Preserve a copy of the photos in the order in which they were displayed. One way is to preserve the traditional simultaneous six-photo display. Assemble a different group of photos using new fillers for each suspect. Interview witnesses in private, separate from other witnesses. Do not tell the witness that the suspect is in a group of photos. Rather, the witness should be told the suspect “may or may not be” in the group of photos displayed. Tell the witness that the displaying officer does not know whether the suspect is in the group of photos. If a witness is able to recognize the suspect from the pho

tos, a statement from that witness should include a description of how certain the witness is of the identification. Numerical certainty (percentages) should be avoided, but a description of why the photo resembles the suspect is encouraged. The witness should initial and date any photo identified. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 88 Suspect gender Male Female 249 30 8 9% 11% Witness gender Male Female No report 181 96 3 65% 34% 1% Weapon presence Gun Knife Other No weapon No report 90 16 13 14 147 32% 6% 5% 5% 53% Time between event and lineup Lineup within 1 week: within 2 weeks: within 3 weeks: within 4 weeks: Cum % 50% 64% 75% 82% This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 87 Table 1. Descriptive data of the Hennepin County Pilot Project ________________________________________________________________________ Population 280 Lineups 117 Cases 206 Eyewitnesses Minneapolis 382,618 138 (49%) 38 89 Bloomington 85,172 86 (31%) 48 69 Minnetonka 51,301 30 (11%) 14 26 New Hop e 20,873 26 (9%) 17 22 Results of 280 Lineups Most Frequent Crimes (of 27 Categories): Murder Assault Theft Forgery/Check 30% 24% 12% 8% Suspect Race: African American Caucasian Hispanic Native American Other No Report 163 75 16 8 6 12 58% 27% 6% 3% 3% 4% Witness Race: African American Caucasian Native American Hispanic Other No Report Cross - race IDs Same - race IDs No information 77 63 8 5 5 122 38 113 129 28% 22% 3% 2% 2% 44% 14% 40% 46% This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 86 forcefully stated, “We in the law enforcement arena cannot help but look over our shoulders at a time when DNA exoneration cases are shaking confidence in the criminal justice system. We welcome objective research that suggests improvements in the identification process. (2003, p. 1).Researchers must move forward swiftly to address this compelling need. Eyewitness research has been recognized as one of the most successful collaborations between psychology and the legal system (see, e.g., Doyle, 2005). More recently,Doyle and colleagues (2006) urge street investigators, laboratory scientists and courtroom litigators to take responsibility for integrating science into practice and to combine their perspectives into a productive working consensus for action informed by science. The translation of knowledge from the laboratory to field must be informed by scientific expertise at the same time that it is enriched by those who have a

textured appreciation for the circumstances in which lineups are performed. One of the most powerful aspects of this research project is in its collaboration among professional communities, a model for the bridging of research and practice. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 85 concern of the police investigators, that jump-out identifications may be sacrificed with the sequential format, did not appear to be the case. Jump-outs occurred at a relatively high level, particularly for instances of familiar perpetrators. Jump-outs may be the prototype of absolute recognition. If so, we would not expect a reduction in jump-outs from a sequential lineup, the sequential lineup being a better spur toabsolute judgment than the traditional form. Jump-out identifications in the first viewing also suggested that absolute judgments were common in the field (55% of witnesses choosing). Implications for the convergent evidence from field experience and laboratory trials offers unprecedented opportunity for an increased understanding of eyewitness decision-making processes and the means to achieve more effective police lineup outcomes. Although some of the lessons may apply more narrowly to Hennepin County’sunique local conditions and logistical circumstances, it is anticipated that the central research outcomes will be informative across jurisdictions. The laboratory research questions, while originating from local experience, were selected for exploration specifically because of their likely interest and relevance to the wider law enforcement and research communities A next generation of research must continue to effectively join theory and research with practice, particularly to addressdisruptions in recommended protocol that may (or may not) affect eyewitness decisions and accuracy. Diamond (in press) speaks of benefits to learning “when well-documented field investigations are combined with laboratory backup.She specifically cited the HC test of filler quality“applying laboratory procedures to test the hypothesis that the lineup construction rather than the memory of the eyewitness produced the identification.”(p. 13). Assistant Hennepin County Attorney Scoggin has This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 84 outcome: A very short encoding opportunity, the perpetrator’s appearance change between event and lineup, and specific instructions for caution each may have contributed to the seemingly hesitant decisions of the witnesses. It may be that these conditions are those that minimize the likelihood of sequential superiority.However, it is fascinating that even under these apparently constrained conditions, witnesses who faced a simultaneous lineup were more likely than those with a sequential format to choose from the array when they saw in the lineup. The attrition of the lineup due to filler familiarity did not deter a lineup choice for these witnesses even in the face of a target-absent lineup. This choosing from a narrowed set of photos suggests relative judgment at work. Witnesses in the sequential lineup who could rule out several faces (one at a time, of course) appeared to rely more on absolute judgment, continuing to work through

the lineup looking for a match to memory beyond the duds. An unexpected benefof sequential lineup delivery can be appreciated when one views the data. The traditional simultaneous lineup format does not allow us to know the level of comparison shopping (relative judgment) employed by the witness prior to the lineup decision. In contrast, the sequential lineup with repetition option has revealed an objective indicator of eyewitness laps through the lineup. The effect of relative judgment was evidenced in both field and laboratory as witnesses engaged in additional laps of sequential lineup. The lab demonstrated that greater choosing in a second (or third) viewing helped to secure a handful of positive identifications but at the expense of four times as many identification errors. The diagnosticity of identification droppedfrom 6.05 to 4.57 after repeated viewing of the lineup. The sequential lineup was moeffective without repeated viewing. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 83 reasonable patterns of witness response suggest that the sequential protocol is effective for accurate identification of both known and stranger offenders. This report would be remiss if it did not also point to specific areas for future research collaboration. Most salient is the pending challenge and opportunity of computer technology for efficient and effective lineup delivery. Adaptation of lineup construction and procedure using electronic photo repositories and laptop formats presents empirical questions that can be usefully subjected to laboratory evaluation. Similarly, creative innovations (e.g., the envelope procedure and the battleship position) as well as currentlaccepted “fixes” such as the Appearance Change Instruction may benefit from further scrutiny and refinement. A somewhat more amorphous suggestion for future research is in the evaluation of witness comments. The range of witness lineup responsesfrom immediate absolute certainty to the qualified,slow, and tentative decisionpresents field data that may inspire laboratory research. Are witness comments diagnostic of decision process or accuracy, indicative of real or false confidence, differentially produced by specific lineup procedures or added laps, or likely to affect an investigatorjudgment of witnessveracityThese explorations demand a continuing dynamic dialogue between lab and field. Theoretical implications At the same time that practical experience should push the research agenda in new directions, the touchstone of science demands the grounding of new knowledge in empirical investigation and sound theory. These data offer information regarding the relative judgment principles that underlie our understanding of eyewitness decisions.Surprisingly, there was no difference in lineup accuracy between the simultaneous and sequential formats in this study. It is difficult to know the reasons for this unexpected This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 82 called upon to answer practical and important questions. For example, a recent writer has asked: What happens when the witness is permitted to view the sequential lineups a second time? (Diamond, in press).The practice of repeated witne

ss viewing of sequential lineups was found to boost error rates in both lab and field. Caution is strongly advised when this procedure is used in the field, particularly for witnesses who complete the first viewing of the lineup without voicing recognition of alineup member. The importance of presenting a full 6-person lineup also was underscored by results of the laboratory study. A phenomenon that may occur in practiceattrition of lineup size due to the witness’s familiarity with lineup foilwas found to boost errors for witnesses who viewed a simultaneous target-absent lineup.This pattern was not apparent for sequential lineups, suggesting that the sequential format may offer some protection against attrition in lineup size when it occurs under these circumstances. However, this is not to suggest that use of sequential lineup negates the need for a full lineup. An unanticipated extension to the original research plan also provides a model for future field tests. Mock witness procedures previously have been employed to examine the quality of lineup fillers for individual lineups. This project was the first to develop and employ the mock witness test for the purpose of assessing overall lineup quality in the jurisdiction of interest. At the same time, the mock witness procedure provided necessary information to more confidently interpret the Hennepin County field outcomes. Along with data on revisions to lineup structure and procedure, the laboratory segment of this project provided the first available lab data on witness response to a perpetrator familiar to the witness from a previous context. As with the HC field data, This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 81 when a witness repeats the sequential lineup or when lineup size is reduced within a simultaneous format lineup. In summary, the laboratory portion of this work successfully lends empirical information for effective field practice. General Discussion The objective of this set of studies was to provide useful information practical, empirical, and theoretical to bridge laboratory research with field practice. The successful implementation of double-blind sequential lineups in Hennepin County provides a strong model for jurisdictions across the country for reform of lineup practice. As documented above, the Hennepin County experience generated a list necessary considerations and realistic solutions to concerns that arise as the preferred protocol is translated to field practice. Equally useful is the quantitative product of the pilot programdescriptive summary of witness decisions and the conditions under which they occurred. The HC pilot has provided the first available baseline summary of double-blind sequential lineup performance under controlled conditions in the field. Also, as other jurisdictions look to assess lineup reform, this pilot project offers recommendations for designing effective field protocol and appropriate interpretation of field data. A training DVD produced by the HC Attorney’s Office is available for other jurisdictions, helping ease the changes and costs to these departments. These lessons are essential to inform practice and public policy. This project’s joining of field practice and laboratory testing generated new timely knowledge about lineup construction and protocol. The rapid application of scientific research to practice has placed lineup researchers in the perhaps enviable position of being This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are

those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 80 This suggests that without the ACI, positive identifications of the perpetrator may have been even lower. Another scenario is possible however. The obtained low rates of witness lineup selections in our current lab study suggest that the ACI may be inhibiting witness choosing. It is possible that the ACI serves only to alert the witness to the difficulty of the identification task, perhaps raising the criterion threshold for making a choice from the lineup or reducing pressure to choose. In this case, one might predict that the ACI will inhibit lineup selections, particularly in the case where appearance change is significant as is likely in this study. This circumstance suggests that the ACI does not improve access to memory, but only affects witness motivation to make a selection from the lineup. Choosing will be inhibited, with parallel effects in target-present and absent conditions. Finally, there is a third possibility (Charman & Wells, 2007Perhaps the ACI only affects witnesses with poor memory, prompting them to choose. This would generate more choosing as they allow substantial discrepancy between memory and suspect photo. In practical terms, this scenario would increase choosing rates and severely diminish witness performance. Testing of the impact of the ACI is currently underway in our lab, however the data are not yet available to help ascertain the exact reasons for witness performance in thistudy. Regardless of the fact of and reason for a lack of expected sequential-simultaneous difference in witness performance, this study clearly accomplished its original objectives. The data demonstrate the impact of a witness’s second viewing of the lineup. The study also answered the question of how reduced lineup sizeattrition due to familiaritycan affect witness response. The study points out the potential danger to an innocent suspect This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 79 of the culprit’s appearance change and the likelihood of the ACI’s remediation of any effect. Interestingly, the ACI was recommendation to law enforcement in the NIJ Guide that was not backed by a foundation of empirical research (Charman & Wells, 2007More recently, the effectiveness of the ACI has been empirically tested once with a simultaneous lineup format; use of the ACI produced no overall benefits to identification accuracy (Charman & Wells, 2007The authors did find positive identification results from the ACI in one condition in which the perpetrator had changed appearance quite dramatically.rhaps this is exactly where the ACI will be influential. The research literature also lacks a theoretical foundation that might help explain ACI operation in e current study.However, implicit in ACI use is the assumption that some correct identifications from culprit-present lineups will be lost when the offenderappearance in the lineup is not exactly as it was during the crime event (Charman & Wells, 2007). Also implicit is the assumption that the eyewitness expects a close match between offender’sappearance in the witnessed event and the lineup photo. These reasonable conjectures lead to the prediction that ACI will urge witnesses to more carefully search for any recognizable features of the lineup members, to see beyond the appearance change, and ultimately to recognize the culprit at high

er rates than will witnesses who are not provided an ACI. This prediction relies on the notion that the ACI will prompt a detailed scrutiny of lineup members that will enhance access to eyewitness memory; the witness will “dig more deeply” into his or her memory, with productive results. If this theory is accurate, the ACI may have helped to elevate recognition accuracy in the current study. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 78 structure should protect against bias, perhaps minimizing the likelihood of a visible sequential superiority effect. (4) Appearance changeAn offender’s appearance may change across time, especially between an early mugshot and current identification task, or in the delay between crime event and apprehension. Even when the perpetrator is arrestedshortly after a crime, he or she may have intentionally and significantly changed aspects of his or her appearance. Just as the offender would hope, transformation of appearance does diminish the likelihood of recognition by eyewitnesses (Shapiro & Penrod, 1986). Hair in particular is a primary descriptor reported by witnesses and used as a key marker for identification (Pozzulo & Warren, 2003).Witnesses are significantly less likely to identify the offender from a lineup when his hairstyle has changed (Pozzulo & Balfour, 2006). he perpetrator’s hairstyleincurrent study changed from crime event to lineup,perhaps exacerbating identification challenges for witnesses who had only a 10-second exposure to him. If appearance change hindered recognition, it seemingly did so under both simultaneous and sequential formats. The NIJ Guide included a recommendation that all witnesses be reminded that the perpetrator’s appearance may change from the time of crime to the time of lineup photo:“Instruct the witness that individuals depicted in lineup photos may not appear exactly as they did on the date of the incident because features such as head and facial hair are subject to change.” (p. 32). The distinct change in the perpetrator’s hairstyle between crime event and lineup in this study made an Appearance-Change Instruction(ACI) appealing as a reasonable attempt to provide context for the witness’s identification task. However, the surprisingly low rate of witness lineup selections can generate speculation about the impact This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 77 66% vs. 68%. The pattern of witness response to these lineups indicates a conscientious rather than cavalier approach to what may have been a very difficult memory task. The witnesses appear to have becautious, with approximately 2/3 making no choice from the lineup in both target-present and target-absent conditions. (2) Limited viewing conditionsThe study included a 10-second view of the perpetrator. The film was shot from the victim’s perspective, allowing a very close range short confrontation with the offender, with his face fully in view. However, perhaps this was too short a time for solid memory encoding to occur for most subjects. Sporer (1993) deterioration in accuracy to close to chance levels when view of the perpetrator s limited to 5 seconds. Although the participants in this study did not de

monstrate chance levels of accuracy, Sporer’s experience may speak to the level of task difficulty in this study. For the witnesses who reported familiarity with the perpetrator from previous encounters, 88% identified him when he was present in the lineup (2 witnesses made no choice). For those subjects for whom the perpetrator was a stranger, the identification rate was only 9%. (3) Lack of high similarity between lineup members and offenderThe data from Steblay and colleagues (2001) indicate that the sequential superiority effect was particularly strong when the innocent suspect in a target-absent lineup was highly similar to the culprit. Perhaps the protection of the sequential lineup against false identification is primarily a function of lineups in which the suspect is a close match to the perpetrator. In the current study, lineup members were rated as moderately similar to the perpetrator, no high-similarity “innocent suspect” was employed, and all lineup members (including the perpetrator) were rated as moderate matches to the description. Otherwise stated, this was a lineup that by This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 76 sequential and simultaneous lineup performance is unexpected. A number of reasons may be advanced and explored to explain this result, among them the following four considerations. (1) Lack of serious attention to the task by witness-participants.Participants appeared to seriously apply themselves to the lineup sk, presumably spurred by the context of the experiment and the message of its instructions. However, it is necessary to go beyond anecdotimpressions. If witness accuracy failed to rise above the level of chance, this raises the possibility of effects due to participant carelessness or guessing. A first consideration is the pattern of filler selections. Choices of innocent lineup members in this lineup were not randomly distributedthe rate of filler photo selections rangfrom 0% to 15%suggests at least some intention among witnessto discern the perpetrator from among the lineup members. Beyond this, the pattern of eyewitness response falls in line with that expected for a single-suspect lineup (Wells & Turtle, 1986) and indicates witness accuracy rates that are low but above chance. For example, the individual lineup member receiving the greateproportion of responses should be the actual target eyewitness performance is above chance level. Fifty-nine percent of the 8subjects who chose from the target-present lineup identified the perpetrator, an appropriately greater proportion above chance (17%), = .0001. This pattern also was apparent in target-present strangerlineups, where 27% of choosers identified the perpetrator, although this was not significantly above chance level of 17%, = 1.11, = .13. In target-absent lineups, none-of-the-aboveresponses should be more frequent than in target-present lineups,again what expected if eyewitness performance is above chance levels. In this study, none-of-the-aboveresponses were indeed more frequent in the target-absent condition than the targ-present condition, 74% vs. 57for all subjects; although for perpetrators who were strangers, this difference did not hold, This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. ��

Double-blind sequential lineups 75 was just partially supported, as eyewitness confidence was enhanced by lineup attrition only for witnesses familiar with the perpetrator. Rather than the increase in lineup selections predicted for the simultaneous format condition, choosing rates decreased as the lineup size diminished due to familiarity, for both simultaneous and sequential lineup formats. unexpected effect was that the choosing rate dropped significantly only for sequential lineups, a dramatic reduction such that only one of 13 witnesses made a lineup selection when post-AF lineup size was three or fewerOn the other hand, 30% of witnesses viewing a simultaneous array made lineup selections even lineup size decreased to three or fewer, and regardless of whether the target was present or absent. study outcomes have implications for field practice and for our understanding of witness cognitive processes, to be discussed in the final segment of this report. One third of witness-participants were familiar with the perpetratorfrom encounters prior to the laboratory task, allowing an unanticipated opportunity to compare laboratory and HC field data on the variable of offender familiarity. Lab and field data show parallel effectsincreased lineup selections from witnesses who are familiar with the perpetrator compared to those for whom the culprit s a stranger, resulting in higher suspect identification rates in the field and perpetrator selections in the lab. While this is not an unexpected finding in itself, it reinforces the soundness of extrapolation from lab to field settings for these new lineup procedures. As expected from past literature (Steblay, et al., 2001) simultaneous lineups produced significantly greater witness confidence than did sequential lineups. This finding is intriguing given the lack of performance differences between the two groups and the overall low level of perpetrator identifications. The lack of decision differences between This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 74 rather than an innocent suspect when the sequential format was used, and 5.56 times as likely to be the culprit when the simultaneous lineup was used. . Confidence in sequential versus simultaneous lineupsOverall, simultaneous lineups produced a significantly higher average level of confidence, (1, 368) = 3.83, .05, measured at the end of just one viewing of the lineup, (s = 4.62 vs. 4.39, simultaneous vs. sequential, respectively). This same effect held in the subset of witnesses who were unfamiliar with the perpetrator and the lineup members. Those who viewed a simultaneous lineup were more confident than those who viewed a sequential lineup s= 4.36 vs simultaneous vs. sequential, respectively, (120) = 2.44, = .016, 2-tailed. For stranger lineups of size 0-3(AF), mean confidence was equal betweensimultaneous and sequential conditions (both s = 4.00). Same versus cross-race identifications. Thirty-three cross-race stranger identifications were attempted, with accuracy rates not significantly different from same-race stranger identification attempts, 39% versus 37%, respectively. There was also no difference in accuracy between same and cross-race identification attempts within simultaneous versus sequential formats: 35% versus 38%, same-race simultaneous versus sequential lineups; 39% versus 40%, cross-race simultaneous versus sequential lineups. Discussion Two hypotheses were tested to determine the impact of revisions to recommended lineup protocol on eyewitness decisions. The first hypothesis was supported, in that repeated viewing of a sequential lin

eup reduced eyewitness accuracy. This finding reinforces the Hennepin County field data and leads to a recommendation that repeated laps be allowed only with great caution. The second hypothesisprediction for a -alternative effect This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 73 Choosing and accuracy. Overall, the simultaneous condition produced 42% and 27% choosing rates in target-present and target-absent arrays, respectively. The sequential lineup produced choosing rates of 46% and 26%, target-present and absent, respectively. There was no significant difference in choosing rate between lineup formats in either target-present (z = .) or target-absent (= .1 = .4) conditions. The two lineup formats also generated similar accuracy at the end of one viewing, as can be seen by comparing the and Seq1columns in Table . Separate analyses of target-present and. target-absent conditions show no significant differences in witness decisions between the simultaneous and sequential formats, (2) = .375, = .41 and (1) = .047, = .41. Diagnosticity. The statistic of diagnosticity compares the identification rate of the perpetrator to that of an innocent person in the lineup (Wells & Turtle, 1986; Wells & Olson, 2002). For practical reasons, the diagnosticity of different lineup procedures can be compared to determine which yields the greater probative value of its identification evidence. Diagnosticity calculations require an estimate of the risk to an innocent suspect. In this case, there was no a-priori specification of an innocent suspect in the target-absent lineup; therefore, the average rate of identification per lineup member (percent of identifications divided by 6) was used to calculate the diagnostic value of the sequential vs. simultaneous display. Given the similarity of sequential vs. simultaneous lineup results, it is not surprising to see similar diagnosticity figures for correct identifications: 6.05 for the sequential lineup and 5.56 for the simultaneous lineup for the overall group; for the subset of stranger identifications, 1.51for sequential and 1.67 for simultaneous lineups. Translated into practical terms, this means that identification was 6.05 times as likely to be the culprit This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 72 identification of a specific photo. For witnesses who were familiar with the perpetrator, the predicted relationship did exist: (119) = -.31, = .001, 2-tailed; as the lineup became smaller, confidence grew. However, for stranger identifications, there was no significant linear relationship between confidence (measured 1-6) and AF -6) for the overall data, (253) = .02, = .72 (2-tailed) or any of the four conditions separately: Simultaneous TP, (62) = .07, = .58; Simultaneous TA, (56) = .19, = .16; Sequential TP, (66) = -.04, .74; and Sequential TA, (69) = -.05, = .71. Stranger and familiar perpetrators. The pattern of witness response to stranger versus familiar perpetrators in the lab was similar to that in HC. Perhaps just coincidentally but also conveniently, 2/3 of the witnesses in both samples reported that the perpetrator was a stranger. We must keep in mind that there was no variation in the perpetrator as

stimulus in the lab (just one perpetrator for all witnesses), however in the field the perpetrators of course varied across crime scenes, as did viewing conditions for the witness.The absolute numbers that detail witness response should not be expected to be similar across lab and field. However, the relative differences may be usefully examined. In both the lab and field, the choosing rate for familiar perpetrators was higher than that for strangers, not an unexpected finding. In the laboratory target-present sequential lineup condition, choosing rate for final decisions was at 53% for strangers and 76% for familiar perpetrators; in the field the figures were 46% and 93%, respectively. As seen in Table 5, suspect identifications in the field for familiar perpetrators were 55% above that of strangers; in the lab, this difference was 52%. Comparison of sequential to simultaneous lineups This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 71 increased correct rejections (12whenthe target was absent. In sequential lineups, there was a dramatic drop in choosing for both target present (34%) and target absent (28lineups from the full to smalllineup. This reduction in choosing reduced accuracy to zero in the target-present small (0-3) lineupwhere no lineup choices were madeand increased accuracy by 29for the target-absent small lineup, in which only one witness made a choice from the lineup. It should be noted that some sample sizes in these AF conditions were quite small. The -3 AF lineup size is a substantially truncated lineup, a point that would only be apparent to witnesses viewing a simultaneous lineup. Thirty percent of witnesses chose from a simultaneous lineup of size 0-3whether the perpetrator was present or absent. Witnesses viewing a sequential lineup were not aware of lineup size; for most witnesses who could easily eliminate 3-6 of the lineup members due to familiarity, the lineup ended before they de a choice. This differential choosing rate in simultaneous versus sequential lineups (30% vs. 8%, = 1.47, = .26) suggests that relative judgment may be guiding the former group and that absolute judgment is more strongly present in those witnesses faced with a sequential lineup. The impact of duds appears to be more dramatic in a simultaneous lineup, where the witness can see all remaining alternatives and in which the attrition of lineup size is very salient. The sequential presentation, on the other hand, does not allow the witness to know the number of additional photos to be shown or to review those already passed, thus changing the impact of attrition. Confidence. Increased foil familiarity was expected to be associated with higher levels of confidence in both types of lineups following the logic of the dud-alternative effectConfidence was measured after completion of the entire lineup, rather than at the time of This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 70 lineups and correct identifications in target-present lineups. Within the sequential lineup, AF decreases in lineup size were not expected to affect choosing or accuracy rates. Choosing and accuracy. Data for stranger crimes (= 253) were used in this analysis as were witness respons

es at the end of the first sequential only. The question of interest is whether there were changes in choosing and accuracy from the full lineup to lineups sizes reduced cause of the very small numbers of participants who knew -6 lineup members, post-AF lineup size was collapsed into just twocategories, 0-3and 4-5These two post-AF categories are compared to a full lineup size of 6 in Table 10. (Only one witness knew all six lineup members; the lineup was a target-absent simultaneous display and the witness correctly rejected the lineup.) The data first are collapsed across target present and absent conditions, to produce overall choosing and accuracy rates for each lineup format as post-AF size decreased. The simultaneous lineup showa decrease in choosing 0% vs. 30%) andincrease in accuracy rates (32% vs. 45%) as the lineup dropped from a full 6-person lineup half that size smaller -3), = .77, = .09 for choosing and = 1.0, = = .12 for accuracy, but neither statistically significant. The sequential lineups produced a significant (30%) drop in choosing as the lineup size diminished, 38% vs. 8%, = 2.14, = .24. As with simultaneous lineup, within the sequential format a non-significant crease in overall accuracy, 36% vs. 54%, was associated with a drop in lineup size. Thus, contrary to prediction, AF decreased choosing rates for both lineup formats, and sequential lineups were significantly affected by As can be seen in Table 10, the impact of AF on simultaneous lineup choosing rate showed up -significant 10% increase in hit rate when the target was present and This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 69 lineup lap when the target was absent versus present, although this difference (64% vs. 59%) was not statistically significant. A logical question is whether those witnesses who did not request a repetition of the lineup generated greater accuracy compared to the entire group at the end of one lap. As can be seen on Table in the Seqversus Seq1 columns, this seems to be the case if the target is present; a higher choosing rate led to correct identifications at 40% and just 1% increase in filler selections above the entire group (Seq 1). However, when the target is absent, the 10% higher choosing rate results in a 10% higher error (filler) selection rate. Thus, those witnesses who stop after one lap are not necessarily a more reliable group. iagnosticity ratio confirms this: At the end of one lap, the ratio for all witnesses is 6.50 (.26/.04); for those who elected just one lap, the diagnosticity ratio is 6.67 (.40/.06). The same pattern is true, if one calculates diagnosticity ratios for the subset of strangerperpetrators (Table 8). In sum, when witnesses were allowed to repeat the sequential lineup, errors increased above the rate of sequential 1-lap lineups. The negative effects of additional laps in the sequential lineup produced errors not equal to, but significantly greater than the simultaneous lineup, (2) = 7.98, = .02, target-present, and (1) = 4.10, = .04, target absent. Hypothesis 2Attrition for familiarity The second hypothesis addressed the impact of downward lineup size adjustments due to the witness’s familiarity with lineup members (Attrition for Familiarity or AF). Within the simultaneous lineup condition, AF expected to be associated with a higher level of choosing and therefore a greater number of falsidentifications in target-absent This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily refle

ct the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 68 third viewing (lap); therefore the analyses combine second and third laps, and simply refer to first versus final laps. Target present lineupsTable 7details witness choosing rates for first and final viewings of the lineup, in the columns headed Seq1andSeqFinal. The choosing rate increasedfrom one lap to the final lap, from % %, and the pattern of witness decisions was significantly different from first to final lap (Sign test produced = 16, = 1, .000). As predicted, correct identifications increasedby 6%filler rates (errors) also ew %. Another way of looking at this pattern is to consider the individual choices of subjects who elected to review the lineup a second time, specifically whether their decisions improved as they moved from the first to final lap. For 69% of thewitness, lineup responses remained the same from first to final lap. The remaining 31changed their answer; for 6% of these subjects the change was not productive (a change from a “no choice” to a filler) and for % the subject moved from an incorrect (filler or “no choice”) to perpetrator selection. Otherwise stated, the second viewing of the lineup improved performance for 11% of witnesses who went on to a second look (6 of 54); for almost twice as many (20%), performance declined. Target absent lineupsThe choosing rate in target absent sequential lineups increased from the first to the final repetition (2% to 42, and a significant difference in response pattern was evident (sign test produced =16, = 0, 1). For those subjects who opted for a repeat of the lineup, there was no gainin performance: 70% maintained their initial answer; 3made a worse response (changing from correct rejection to a filler selection). Witnesses viewing a stranger crime were slightly more likely to take anothe This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 67 after one viewing (Seq1), final decision (SeqFinal), and response in either in the one-lap (SeqHold) or multiple-lap (Seq+) category. Table 7 includes all 372 witness participants. Table 8 details the subset of participants for whom the offender was a stranger (n=253), and Table 9 showsthe remaining witnesses who report familiarity with the perpetrator from interactions prior to the study (n=119). Sixty-eight percent of participant-witnesses stated that the perpetrator of the crime was a stranger to them, perhaps coincidentally very close to the 66% Hennepin County proportion. After one viewing of the lineup (standard procedure),articipant-witness choosing rates were at 35% overall,and35% and 36% for simultaneous versus sequential lineups, spectively. These are lower choosing rates than are typical for laboratory or field outcomes (see Table 3), particularly given that 32% of the witnesses reported that the perpetrator was familiar to themfrom experiences prior to the lab study. It appes that either is crime scenario was quite challenging for memory encoding, the lineup particularly difficult, witnesses very cautious, or all of the above.will return to this point below. Hypothesis 1 Repeated viewing of the sequential lineup The expectation for the first hypothesis was that performance of eyewitnesses on a repeated sequential lineup should approximate that of witnesses in the simultaneous lineup: higher choosing rates, increased correct identifications in the target-present condition, andincreased false identificati

ons in the target-absent condition. Fifty-three percent of witnesses elected to repeat the sequential lineup. However, only six witnesses opted for a This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 66 I am here to record any comments that you may have about the photos, but I cannot answer questions about the crime video or the lineup. The computer maintaina data file that captured aspects of the method and subject responses, including demographics (age, sex, race), target presence/absence, lineup format (simultaneous or sequential), position of the perpetrator (randomly determined, with the exception that the perpetrator never appear as the first photo), number of viewings of the sequential lineup, subject choice, and subject rating of confidence. The subjects were also prompted to report their familiarity with any of the crime scene and lineup participants,and these responses were recorded through the computer application. In addition, the experimenter kept a written record of any subject comments and questions, including verbatim comments about familiarity with any of the men in the photos,ualifiers offered, and expressions regarding confidence.Results Overall eyewitness responses Tables 7, 8, and 9 document witness performance in simultaneous lineups (column 1) and sequential lineups (columns 2-5). The direct comparison of simultaneous to sequential lineup performance is available from columns 1 and2; these figures, representing data from all witnesses, represent performance of the simultaneous array (Sim) and sequential procedure after one showing of the lineup (Seq1). Column three (SeqFinal) indicates the endpoint of the witness decisions after all laps through the lineup. Columns 4-5 break out the sequential subjects into those who did not request more than one viewing (SeqHold) and those who opted for more than one viewing (Seq+). Otherwise stated, each witness who saw a sequential lineup is represented in three columns on this table: Response This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 65 perpendicular to the subject, rather than in a direct sight line. A minor deception was employed to establish the blind status of the experimenter for witness-participants (see instruction below) by informing the witness that the specific perpetrator in the witness’s film was one of many employed in multiple versions of the film, thus not knowable to the experimenter. The computer promptthe subject to wait for specific instructions from the experimenter at one key point in the process. Subjects were informed that the study is funded by the National Institute of Justice and results were to be used to inform police practice nationally, therefore to “please pay serious and close attention to the procedures.” After a brief introduction, subjects worked through a computer application that requested demographic information and showed the crime scenario. The experimenter was out of the room during this time. At the conclusion of the crime video, the experimenter re-entered the room and provided additional instructions. The script for the experimenter emphasizedfive key points: The film segment you saw is one of a number we use. It was randomly generated by t

he computer, so that I don’t know who the perpetrator was in your display. In the lineup you’re about to see, we are interested in whether you recognize one of the lineup members to be the perpetrator of the crime in the film. However, the lineup may or may not include the perpetrator you saw in the film. Sometimes the appearance of the offender will change between the event and the lineup. Please keep this in mind. This procedure works best if you view all of the photos So the program will allow you time to examine each photo in the full lineup. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 64 perpetrator stand out as the best fit to his own description. Each photo was a full frontal head-and-shoulders view with a gray background. All lineup members including the offender wore their street clothes for the lineup, none matching the clothing worn by the perpetrator at the time of the crime. For each subject, the computer program generateda -person lineup sequential or simultaneous format, target-absent or target-present. The six fillers rotated in and out of the target-present lineup. The position of fillers in all lineups and the position of the perpetratorin the target-present array were balanced, with the exception that the offender never appeared in position 1. The stimulus crime video and subsequent lineup featuredcurrent or recently graduated Augsburg College students, a point central to this projectStimulus materials were prepared in April of 2003. The currency of the video allows the natural occurrence of subjects’ previousfamiliarity with lineup foils and the perpetrator, and subsequent evaluation of lineup performance for student subjects who recognizedlineup members from outside the lab. pilot study demonstratfamiliarity percentages of 45%, 15%, 20%, 15%, and 5% for subjects who recognize 0 3, and 4 foils, respectively. The current udy found familiarity percentages of 35%, 20%, 14%, 12%, 7%, 5%, and 7%, for subjects who recognized 0, 1, 2, 3, 4, 5, and all lineup members, respectively. Procedure Each subject s run individually with instructions provided through the computer program and by the experimenter during a 15-miniute laboratory interaction. The experimenter maintain a blind status by remaining across the room from the subject, with the laptop computer facing the subject. The experimenter also positioned herself This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 63 credit (for participant pool members) or a $5 payment. Studentactorsfeatured in the video and lineup segments of the laboratory materials provided written consent for their images to be used for laboratory purposes. Stimulus materials The experiment was delivered to subjects via laptop computer within an Authorware software program. A video of a short (30-second) purse-snatching incident servedas the stimulus incident. The event s filmed in color with audio, shot from the victim’s perspective, and the male Caucasian perpetrator, age 21, is visible for approximately 10seconds. -person lineup providedthe memory test for subject eyewitnesses. The lineup consistof Caucasian males of approximately 18-23 years of age and wa

s constructed to capture moderate physical similarity between lineup members and the perpetrator. In a prior pilot study, 17 participants watched the video and were asked to describe the perpetrator first via free recall, followed by prompts for clothing, race, hair/eye color, height and age. The modal description, excluding clothing,was used to find nine potential lineup members (a fit--description ). Next, 11 additional participants who had not seen the crime video were presented with a 10-photo perpetratorpresent lineup and asked to decide “Which one of these photos best meets the description?” The participants were allowed multiple, rankordered answers. The same 11 participants also were shown a side--side comparison of each filler photo with the perpetrator and asked to rate the similarity of each of the nine (one at a time) foils to the perpetrator, on a 1-10 scale. This produced a range of scores for the nine photos from 2.09-5.18 (low to moderate similarity). Finally, based on these measures, 6 fillers were selected. No one filler stood out as an obvious “innocent” suspect nor did the This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 62 The study design is a 2 X 2 factorial with independent variables of lineup presentation (simultaneous or sequential), and target presence in the lineup (present or absent). Number of lineup viewings and functional size of the lineup are ex post facto (non-manipulated) variables, varying naturally depending on each subject’s previous exposure to the foils and the subject’s decision as to whether to repeat the lineup viewing. The inclusion of these two variables in an ex post facto configuration is intentional; as a naturally occurring phenomenon, the behavior affords greater ecological validity to the laboratory results. Key dependent measures are witness choice, accuracy, and confidence. For target-present lineups, aggregate outcome measures are recorded as frequencies and percentages correct identifications, filler selections,and incorrect rejections of the lineup; for target-absent lineups, outcome measures are correct rejections and error identifications of any foil. Confidence is assessed through subject report on a 1-6 scale. Participants 3undergraduate students took part in the study, male and female, ages 18 to with a median age of 20 years. Eighty-seven percent of participants identified themselves as Caucasian. Participants were treated in accordance with American Psychological Association Ethical Standards for Treatment of Human Subjects. The experimental protocol was approved by Augsburg’s Institutional Review Board In order to achieve appropriate levels of statistical power (.70 to .80, depending on the specific analysis), the study requiredsubjects, with a greater proportion in the sequential condition. Participants were recruited through the existing college research participant pool, recruitment signs posted on campus, and announcements in classes. Participants were provided the choice of course This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 61 selected person matches memory for the culprit. The current study will examine both accuracy and confidence as a means to determine whether the du

d-alternative effect extends to familiar foils in lineups. It is predicted that within the simultaneous lineup condition,decreases in the number of real filler choices will be associated with a higher level of choosing and a greater number of both false identifications in target-absent lineups and correct identifications in target present lineups. Given the sequential lineup’s tendency to hold down choosing rates, it is expected that decreases in the number of real foil choices will not affect choosing or accuracy rates. Foil familiarity is expected to be associated with higher levels of confidence in both types of lineups. Study 3 was structured to explore these two deviations from prescribed lineup protocolrepeated viewing of the sequential lineup andreduction in lineup size due to familiaritywhile also testing two variables common to lineup lab investigation, simultaneous versus sequential presentation format and presence versus absence of the perpetrator in the lineup. e inclusion of standard conditions allows comparison of utcomes past research and integration of findings into the theoretical foundations of the research, as well as comparative evaluation of the sequential versus simultaneous lineup format. The procedures for both simultaneous and sequential lineup formats includedunbiased instructions, blind administration, and fair lineup construction. The sequential lineup differ from the simultaneous in its sequential positioning of the photos, in instructions specific to sequential lineup process, and in the option for additional reviews (laps). Method This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 60 ideal lineup with five unfamiliar foils is unrealized. The research question : How does familiarity with lineup foils affect the decision of the eyewitness? A lineup size of at least six is prescribed as a safeguard to protect the innocent suspect. Errors of memory should be distributed across fillers in a manner that reduces the likelihood of false identification. A reduction in number of fillers is worrisome, as there are fewer foils to absorb errors. Researchers refer to the number of photos in the lineup as its size. A lineup with fillers that are disparate from the perpetrator description or that can be easily dismissed by the witness for reasons other than a lack of match to memory is unfair. In such case the size of the lineup (real choices for the witness) becomes smaller nominal size. A reduction in number of fillers changes the chance distribution of choices, thus should logically exacerbate the effects of differential choosing that occur in simultaneous versus sequential lineups. If a highchoosing rate is maintained for simultaneous lineups despite reduced functional size, witness choices will be spread over a smaller number of photos. The result will be an elevation of correct identifications when the target is present but also an increase in identification errors for target-absent lineups compared to the sequential lineup. Recent research provides evidence for other concern, thatthe addition of very weak alternatives () to a judgment task increases the perceived likelihood of the focal outcome (Windschitl & Chambers, 2004). Although the investigators of this phenomenon did not use a lineup task in their research, they positthat confidence judgments of lineup witnesses are a real-world example for which the dud-alternative effectis relevant. More specifically, their work suggests that the inclusion of ill-fitting lineup distracters may boost the eyewitness’s confidence about whoever is chosen from the lineup and about how well the This document is a research

report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 59 Repeated laps of the sequential lineupHennepin County has structured a lineup procedure that allows a smooth logistical process and makes theoretical and practical sense. All key components of a blind sequential method are included except one: Instead of limiting the eyewitness to the recommended single viewing of the lineup, HC allowwitness to repeat the lineup. The HC field data revealed that repeated laps of the lineup were associated with increased numbers of decision errors as revealed by filler selections. In the laboratory this investigation can be extended to measure identification accuracy. The research question : What is the effect of a witness’s repeated viewing of the lineup on witness accuracy? e decision to allow a hesitant witness a second chance at the lineup was a revision aimed at securing greater number of true positive identifications. However, one could extrapolate from theory and research to surmise that this deviation w move the lineup format in the direction of a simultaneous presentation and give the witness greater opportunity to engage in relative judgment.Based on this logic, it is hypothesized that the performance of eyewitnesses on a repeated sequential lineup will not improve identification accuracy, but rather will approximate that of witnesses who view a simultaneous lineup: more choosing, increased correct identifications in the target-present condition, and increased false identifications in the target-absent condition. Impact of familiar foilsThe second research question relates to lineup structure. HC investigators recognize that in some instances lineup foils are familiar to the witness, a fact unknown to the lineup administrator until the lineup is underway. This circumstance is particularly likely to occur with gang-related crimes in minority populations, wherein photo pool -race, same-age persons available for lineup foils is constrained. The This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 58 procedures one a method to prevent suggestibility effects stemming from the investigator and the second an assessment of how well lineup construction has constrained suggestive presentation bring greater confidence in the accuracy of witness decisions. In the field we must depend on the design of the study to help rule out non-memory explanations owitness responses. An understanding of field lineup outcomes demands that those who design, undertake and analyze field research understand central principles about human memory, lineup research, sound scientific method, and common police practice. The usefulness of laboratory lineup research is to methodically identify and strip away factors that reduce witness accuracy and so to devise techniques that add precision our attempts to access witness memory. The new double-blind sequential lineup procedure successfully employed by Hennepin County was designed to limit the impact of non-memory factors that may occur in traditional lineups. The fruitful collaboration of HC and lab is extended now in Study 3, in a further investigation of best practices. Study 3: Laboratory tests of lineup protocol revisions Underlying recommended lineup protocol is a theoretical mode

l that explains how eyewitnesses process information, how memory can be effectively tapped or unfortunately contaminated, and the conditions under which better memory products may be expected. Empirical work has further demonstrated that relatively simple changes in lineup structure or procedure (Steblay, 1997)can dramatically influence witness accuracy. Therefore, even seemingly benign changes to lineup protocol may require empirical examination.Study 3 addresses two questions regarding adjustments to recommended lineup procedure. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 57 “…There was one other glaring defect in the report’s methodology: It equated a pick of the police suspect with a correct identification[italics in original]. To put fine point on it, this means that the Mecklenburg Report would have counted every single one of the DNA exonerations as “correct” identifications. Even worse, the benchmark picked by the Mecklenburg Report actually rewards suggestive police procedures by skewing the results in favor of any method that encouraged witnesses to make more “suspect” selections regardless of accuracy.”(p. -20 Employ procedures that rule out alternative (non-memory) interpretations of witnesses responses. The objective of a lineup is to accurately access the eyewitness’s memory of the perpetrator, circumventing non-memory factors that may otherwise influence the witness’s decision. Examples of non-memory factors include expectations and beliefs of the witness, inferences that the witness may draw from the lineup construction or presentation, the relative judgment process, and influences (imagined by the witness or real) from the lineup administrator. Challenges to eyewitness identification evidence typically claim at the witness’s decision can be attributed to non-memory factors. Assertions of sound eyewitness evidence must be able to demonstrate that the identification is of the highest quality that extraneous influences on memory have been avoided. To that end, field practice must rely on procedures that have been devised to secure memory accuracy, and evaluation of field data must recognize the extent to which the outcomes represent procedural safeguards against memory error. One requisite safeguard, as noted above,is double-blind administration of the lineup (Wells, 2006b); a second is reflected in the mock witness test of filler quality conducted by HC. Both This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 56 knowledge about the likelihood that the perpetrator and the suspect are one and the same. he suspect in the lineup represents the investigator’s belief about who the perpetrator isand the witness’s response may provide diagnostic information about whether the investigator has in fact placed the true perpetrator in the lineup (Wells & Olson, 2002). The lineup and its procedure must be set up in a manner that will challenge the witness not to simply confirm the investigator’s belief, but to t memory for the offender. In the field,Suspect IDs cannot be directly equated to Accurate IDs, because any false identification of an innocent suspect (dangerous error) is buried within Suspect s. A Suspect IDmay

be an accurate identification of the offender a false identification of an innocent suspect. A filler selectionis a known error but should not be confused with a dangerous false identification. There is no direct danger to the selected filler, but this choice by the witness is a signal to investigators that the witness has a poor memory of the offender, is guessing oruncooperative, or perhaps that the filler is a better match to the perpetrator than is the suspect.No choice responses will include those witnesses unable or ling to select from the lineup, or, particularly when a witness responds with “he’s not there” will indicate that the suspect is not the offender. Left with field measures that are inherently open to multiple interpretations, researchers must be very cautious as they determine the meaning of the data. The appropriate and effective evaluation of a study must occur at the nexus of witness response and the method used to generate e data. A powerful example of the confusion that ensues when this point is missed is provided by O’Toole (2006) in his critique of the Illinois pilot program. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 55 audience and policy-makers know the particulars of both the witnesses’ exposure to the offender and of the identification experience and how these affect the overall statistics. Be clear and realistic about the meaning of eyewitness performance measures. primary benefit of the laboratory test is that the data analyst can tally precise and meaningful percentages of correct identifications, filler selections, and false rejections target-present lineup. Knowledge of correct lineup rejections and identification errors in target-absent arrays likewise are quite straightforward because the lab researcher has controlled the offender’s presence or absence in the lab lineup. It is somewhat more challenging to simulate and evaluate the risk to an innocent suspect in laboratory studies in as much as there are many ways in which an innocent suspect may find his way into a lineup in the real world. However, it is likely that an innocent suspect has at least modest physical similarity to the true offender in most cases. In the lab, the researcher therefore typically plants a suspect who looks similar to the perpetrator in a target-absent lineup. This procedure simulates the worst possible scenario that may occur in a real police investigation i.e., a lo-alike to the offender has become a suspect. An alternative method is to calculate the risk to any one innocent suspect as 1/6 of the overall filler identification rate in a (6-person) target-absent lineup. Either way, the lab researcher is certain that the “filler” category has captured all mistaken identifications. Field lineup results can be much more challenging to evaluate. There is no parallel in the field to the laboratory’s “targetpresent” and “targetabsent” lineups, because the true status of the suspect as offender or innocent is unknown. While laboratory protocol is established to assess the witness’s memory, witness responses in the field have implications for both the witness the lineup. The purpose of the field lineup to gain additional This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. ��

Double-blind sequential lineups 54 is the use of repeated laps in the sequential procedure. In HC, a witness’s additional viewings of the sequential lineup generated more known errors (fillers). Would filler rates decline if witnesses were held to a single viewing of the lineup? Now we know: yes. Some variability in lineup protocol is not a problem, particularly as it may represent natural field practice and if it is well documented. The challenge is to evaluate lineup outcomes when procedural changes violate prescribed protocol or when they confound interpretation of the outcome. For example, placement of more than one suspect in the lineup runs counter to recommended protocol; blind sequential protocol is very clear about a-priori definition of one suspect and at least five fillers. While it may be reasonably argued that the sequential format may help to minimize effects from some aspects of faulty lineup construction (Lindsay, Lea, Nosworthy, Fulford, Hector, LeVan, & Seabrook, (1991), post-hoc estimation of such effects will be difficult. Finally, field data may be parsed many ways for analysis, with subsequent effects on the overall aggregate figures that become the simple summary of the more complex data. For example, HC chose to include every lineup conducted for felony crimes, including crimes in which the witness was familiar with the offender. Most HC lineups followed crimes of very short duration for which the witness did not know the perpetrator; not surprisingly, this subset resulted in lower suspect ID rates and somewhat higher filler selection rates compared to situations in which the witness had longer or even multiplexposures to the perpetrator. Of course, the aggregate statistics change if the mix of stranger versusfamiliar perpetrator percentages that make up the data set is adjusted. Clarity of results and subsequent comparisons and recommendations demand that the This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 53 a policy of using only single identification attempts. An effective scripting of the new protocol pulled together previous and new procedures into a coherent package. An example from other jurisdictions that shows a more problematic scenario is the evidentiary requirement in New York and Illinois that demands live lineup identification by the eyewitness. The practice that flows from this is that the eyewitness will have first successfully selectthe suspect from a photo lineup. This introduces the possibility that he witness’s recognition of the suspect in the live lineup stems from the previous procedure rather than the crime or that the witness can infer the suspect’s identity due to his presence as a common denominator across two lineup tasks. Furthermore, unsuccessful witnesses will have been screened out in earlier tasks, distorting overall identification rates. Filler selection rates can be expected to be tamped down and suspect identification rates elevated for reasons unrelated to memory accuracy Behrman and Davey’s2001 documentation of traditional simultaneous field lineups in California showed that the practice of using second identification task increased suspect ID rates 17% for the later identification. Interpretation of data from a recent pilot project in Illinois (Mecklenburg, was problematic due to this and other problems with a non-standardized protocol (see e.g., Diamond, 2006; Doyle, Penrod, Kovera, Bull, & Dysart, 2006; O’Toole, 2006; Steblay, 2006; Wells, 2006b). At the same time that clarity of initial protocol is essentialit is also critical to pay close attention to subsequent deviation from prescribed protocolBlind

-quential procedure is a package of components that together form best practice. It is important track how these components are implemented in the field, how revisions to the recommended protocol are employed, and their impact on the outcome data. One example This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 52 Lessons from the field: The interpretation of field studies As lineup reform continues, it is important that additional evaluative field data becollected and reviewed. Both scientists and non-scientists then will attempt to draw conclusions about the meaning of field data. The following points are offered to guide this potentially confusing endeavor. Interpretation of the HC data illustrates three important considerations regarding lineup field studies. First, it is essential to be clear about what is being measured; this is particularly true when simple aggregate figures will represent the study to policy-makers and laypersons. Second, field measures are imperfect. There are no established absolute levels of “goodnessand the measures by themselves tell us very little about eyewitness memory accuracy. Instead, lineup outcomes must be evaluated within the context of the study design and the type of data that contributed to the statistical outcome, as well as estimated gains or losses in accuracy that are likely from the procedures employed. Finally, good field practice and tests must employ procedures that rule out alternative (non-memory) fluences on eyewitness decisions. Be very clear about what is measured. Any hope of comprehending field study results will fade quickly if researcherfield administrators, and intended audience have not shared a specified and detailed protocol for lineup procedure. Prior to lineup reforms, investigators have long-seated habits for conducting lineups, sometimes following a written departmental procedure, often without. It is important to ascertain which conventions the status quo are intentionally or inadvertently transferred to the new lineup routine. HC had the benefit of established procedures that fed easily into lineup reform, e.g., one-suspect lineups of uniform size, lineup construction using a statewide photo repository, and This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 51 The core rationale for these information-gathering and documentation requirements is illustrated in the HC written protocol for investigators: “Sequential identification and the following documentation requirements do abetter job of conveying what a witness really means. In some ways this is Dr. Wells’ real point. The witness didn’t mean this is the guy –he meant this might be the guy, the photo looks a lot like him. We’re learning that eyewitness identification isn’t all or nothing. Identification takes place on a sliding scale. Some identifications “jump out” –the witness is absolutely certain …who committed the crime. Others are less sure or place conditions on the identification (“it’s him except he’s older” –thinner, or has different hair). The procedure…encourages the investigator to document what those judgments are. This gives the rest of us (prosecutors, judges, and jurors) a clearer view of th

e truth and much more confidence in the result.” (Scoggin, 2005, -5) (5) Caution with repeated viewings of the lineup. A second lap should occur only at the witness’s request and the witness should not be told beforehand that a second viewing is an option. The lineup should end with maximum of two repetitions and a record kept of the eyewitness’s response to each viewing of the lineup. Given the HC data, it must be presumed that an identification made on the second viewing, particularly with no recognition of a photo on the first lap, incorporates some increased level of relative judgment. (6Development of functional equivalents. For situations in which a blind administrator is unavailable, the practical need for emergency substitutions should be met with convenient but valid alternatives. It is desirable that researchers follow up in the lab establish the effectiveness of proposed equivalents. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 50 when a photo is chosen, or of non-identification at the end of the array. Witness commentshould be recorded as stated (“I’m positive”) rather than as the officerassessment of the ’s position(“He was positive”). Alternatively, the witness may write a statement in his or her own words. Clarification and precise recording of witness responses is critical. For example, the witness failed to identify the suspect”is ambiguous; did the witness identify a filler (which one?), indicate that the culprit is not in the lineup (“he’s not there”) express indecision (don’t know”)? Filler identificationshould be noted, including multiple identifications, as these may be diagnostic regarding the quality of the lineup and the identity of the suspect perpetratorJust as with a witness, feedback, forgetting and post-event information can intrude upon the administrator’s memory of the lineup interaction. The blind administrator should document the results before conversation or feedback from the investigator, and delayed recording of the lineup should be avoided. (4) Standardized training, procedure, and reporting.Until the procedure becomes habit for all officers, some mechanism for memory prompts is likely to be necessary. The latter could be achieved with a scripted reporting form and perhaps a simple laminated wallet card. Training should include a brief but clear explanation of the underlying rationale for the double-blind and sequential components. Understanding of basic principles of memory and the potential benefits to police investigations is likely to ease acceptance of the new protocol in the field and to spur the enthusiasm and creativity essential to generating solutions to the specific operational challenges of each local jurisdiction. In particular, investigators must be assured that the new procedure is not an assault on their professional integrity or a hindrance to securing timely and accurate identifications. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 49 process occur prior to the lineup proper, demanding early documentation of the witness’s description of the offender along with a detailed summary of lineup preparation and all interactions with the witness preceding the lineu

p. From that point, memorializing the lineup itself is essential, as recommended by the NIJ Guide. Particularly in jurisdictions where the witness may have attempted an earlier identification, recording of the witness’s history in the investigationis crucial including specific results from each identification task and the suspect’s history of appearance in each of the earlier identification arrays. The witness’s relationship with the offender stranger or familiar should be recorded as well. (2) Productive information gathering from the witness during the lineup procedureProcedure and instructions to the witness are tightly prescribed in the new protocol,and the blind lineup administrator is not to volunteer information beyond the protocol. However, the interaction between the blind administrator and the witness is not intended to be rigid or the conversation stifled. The lineup interview is an opportunity to learn from the witness, and a conversational flow focused on the witness’s experience and memoryis desirable. The blind-sequential interview may allow for but not demand comments or qualifiers from the witness for each photo.Yes,” “” “’m not suret pictureand such phrases are all reasonable responses from the witness as a means for him or her to advance the next photo. A forced “yes” or “no” may push the witness into a response that does not truly capture his or her position. (3Effective recording of witness response. The lineup administrator should keep a careful record of verbatim witness comments about all photos (preferably with videotape supplement), including qualifiers to the identification and comments about photos not chosen. The should be asked about his or her certainty of identification immediately This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 48 eyewitness response to strangerversus familiarperpetrators square well with what is known regarding eyewitness memory, with strangersuspect identifications at a lower level. The Hennepin County data also indicate that blind sequential lineups work well in situations of familiar perpetrators and confirmatory lineups. Given that even a confirmatory lineup is still a test of memory (and the investigating officer is unlikely to know just how familiar the perpetrator is to the witness), sequential lineups provide an appropriate protocol. The purpose of the project was to determine how recommended lineup procedures can best be brought into practice. The experience of the pilot project indicates that the lind sequential protocol is workable for police in both large and small departments without undercutting the ability to solve cases. At the same time, the protocol elicits valuable new information for the effective investigation and prosecution of criminal cases. Lessons from the field: Implications for policy and practice The Hennepin County experience offers specific suggestions for successful field use of the new lineup protocol beyond the recommended components of the double-blind sequential technique itself. Many of these recommendations have previously been put forth by other researchers and policy-makers, but deserve to be underscored as lineup reform efforts move ahead. (1) Background information: Transparent and close adherence to the underlying operating assumptions of double-blind sequential protocol. Operating principles for effective lineup protocol include: a single-suspect lineup with properly chosen fillersisolation of each witness from the potential influence of others; and an identification that is the witness’s first view of theapprehended suspect. These components of the lineup This

document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 47 validity of identification evidence. The County Attorney’s Office notes: “The procedures give us more information about what an eyewitness really means when an identification is made and, in the end, a clearer view of the truth”(Scoggin, 2005, p. 5). The Hennepin County pilot project also brought the first available information regarding the effect of repeated viewing of the lineup. One can extrapolate from theory and existing laboratory data to surmise that a repeated sequential lineup will ease the witness into relative judgment, with predictable reduction in performance accuracy. Now, Hennepin County has provided field data about the effects of lineup repetition: identifications are likely to be more reliable when the witness has made a decision after a single lap. While it is impossible to know the extent to which suspect identifications include false identifications of innocent persons, repeated lineup laps are shown to be associated with increased likelihood of error in the form of filler selections. Interestingly, witnesses who made an initial comment of recognition during the first lap were highly likely to identify the suspect in the second lap. Perhaps these are cases of witnesses who have reasonably strong memories of the perpetrator, but simply need to be sure with a second reinforcing look at the lineup. It remains important that lineup administrators record procedural details and witness comments carefully and methodically, particularly if a witness opts for a repeated viewing of the lineup. Laboratory studies of blind-sequential lineups typically deal with strangercrimes of short duration, which might be considered the most difficult test of memory. The subset of Hennepin County data that involved stranger crimes provides evidence that laboratory principles generalize well and productively to the field. In addition, the Hennepin County data examined eyewitness responses to lineups with familiar perpetrators. Patterns of This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 46 has established standardized scientifically-based practice that has been demonstrated to work in real street investigations. The key objective of the new blind sequential procedure is to secure better quality identifications based on what eyewitnesses actually remember. How do we know these HC lineup results are “better”? In the field, we cannot know identification accuracy in a hard factual manner. Instead we must look to procedures that have been demonstrated to protect witness memory and therein engender accuracy. What we know are the following: (1) Laboratory evidence provides a compelling foundation for belief in the superiority of sequential over simultaneous format as a means to acquire a more trustworthy rendition of eyewitness memory and thereby reduce the risk f misidentification. (2) The double-blind procedure facilitates the witness’s independent judgment and the investigator’s objective documentation of thjudgment. (3) Hennepin County investigators perceived no drop suspect identifications or in theability to do their jobs. The preservation of an effective suspect identification rate withthe safeguards o

f the blind-sequential procedure is a very positive development. And, (4) Hennepin County lineups providappropriate risk management for innocent suspects through effective blind sequential procedure. This is evidenced by the low filler rate, a conclusion reinforced by the results of the mock witness test of HC lineup fillers. The new lineup technique reduces concerns that administrator influence and relative judgment haaffected witness decisions. Because of these safeguards, Hennepin County prosecutors and investigators anticipate benefits in court. Most officers recognize the potential for harsh cross-examination should lineup evidence go to trial. Properly conducted, the blind sequential lineup can be expected to better withstand challenges to the This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 45 The lineup’s functional sizewas calculated as the reciprocal of the proportion of mock witnesses choosing the suspect, 4.54 (Wells, Leippe, & Ostrom, 1980).corresponds to the median functional size across all 37 lineups. Averageeffective sizeof the lineupEffective size is an index of the number of plausible lineup members, without specific reference to the suspect (Malpass, 1981). measure is useful as an indicator of the spread of choices across the lineup (operable alternatives). Median effective sizeese lineups was 3. In essence, the calculation reduces the nominal size of the lineup for each filler photo that fails to draw its proportionate level of identifications. The specific calculation is offered by Valentine and Heaton (1999). In summary, the fillers were found to be functioning well, drawing mock witness responses in appropriate proportions. Brigham, et al., (1999) contend that a lineup should contain a minimum of three viable alternatives to the suspect in a 6-person lineup, a criterion met by the statistics of this HC sample. Importantly, witnesses did not avoid the fillers, thus alternative interpretation for low filler rateeffective fillerscan be ruled out in the Hennepin County study. DiscussionHennepin County Pilot Program Overall Summary: What has been gained? The Hennepin County pilot project has been judged a successful application of double-blind sequential lineups in the fieldBlind sequential procedures are now county-wide, and adjacent Ramsey County has rolled out the new protocol. The new procedure This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 44 effective sizeof the lineup (see Brigham, Meissner, Wasserman, 1999for a useful discussion of the strengths and weaknesses of each index of lineup fairness). Averageproportionof mock witnesses who choose the suspect. Lineup bias is most simply calculated as the proportion of mock witnesses who choose the suspect (Doob & Kirshenbaum, 1973). This proportions technique as been noted to be the most useful estimate of lineup bias (Brigham, et al., 1999). Also, the proportion of witnesses who choose the suspect is a significant predictor of the proportion of witnesses who identify the lineup member in a perpetrator-absent lineup(Lindsay, Smith, & Pryke, 1999). The median proportion of mock witnesses who chose the subject was 22%, a non-significant difference from the rate of chance 1

7%. A compliment to this figure is more relevant to the current question of whether fillers were effective in drawing mock witness responses: An average 78% of mock witnesses chose a lineup member other thanthe suspect. photos drew choices (median across the 37 lineups = 6). Averagefunctional size of the lineup.Because lineup size directly contributes to fairness, a supplement to this proportion indexof lineup fairness is a measure of nctional size. The number of members appearing in the lineup regardless of their physical appearance or fit to witness description is referred to as nominal sizeFunctional sizethe other hand, is the number of lineup members that are physically similar to the witness’s description of the perpetrator (Wells & Turtle, 1986). Wells and Bradfield (1999) explain that the lineup task must control for simple recall of the description by ensuring that all lineup members fit the eyewitness’s verbal description equally well Functional size represents the number of plausible choices for the witness the suspect and some number of viable alternatives. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 43 The testing was done by two experimenters. The first was blind to which lineup member was the suspect; the second experimenter (the PI) was not blind, but left the subject’s cubicle during the rating of the photos. Neither experimentinteractwith the subject during the rating task. Presentation order of the 37lineups was counterbalanced across subjects. The participants were instructed as follows: In this packet is a series of real police lineups. Connected to each one is a description of the perpetrator that was provided by the real witness to the crime. Some of these descriptions are very brief, others provide more detailed information. The entire description provided by the witness is given, even if some features cannot be seen in the photo. For example, an attribute of the perpetrator’s body, not visible in the picture, may be part of the witness’s description. Also, just as would a real witness, you can assume that clothing and sometimes appearance will have changed between the witnessed event and the time of the lineup. Your task for each lineup is to read the description, view the lineup, and make a choice, as best you can, as to which of the lineup membersyou think is the accused. That is, who do you think the suspect is? Results and Discussion To evaluate the quality of HC lineups, three statistics were calculated: the proportion of mock witnesses who choose the suspect, functional sizeof the lineup, and This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 42 would limit the damage of such a witness in a real identification task, in that all of the lineup members would fit the witness’s description to the same extent; therefore, the suspect would be at no greater risk of identification than any one filler. Method college students each evaluated a sample of 37 lineups.The study was approved by Augsburg’s Institutional Review Board and by the Hennepin County Court. Participants were recruited with posters and through the campus research participant pool and compensated with class credit or cash. Participants were req

uired to sign a consent form and treated in accordance with APA ethical principles. Lineup bias is not a property of a lineup per se, but rather a result of the interaction between the lineup and the verbal description provided by a specific witness (Wells & Bradfield, 1999). In order to use the mock witness protocol, it was necessary to find lineups in which the police report included the witness’s prior verbal description of the perpetrator. Only one lineup per case was allowed for this analysis, reducing the available 280 HC blind-quential lineups to 117. A sample of 50 lineups was requested and 52 provided by HC. Of these, 37 were appropriate for this research task. Lineups were eliminated from the sample of 52 for the following reasons: 1) the police report failed to indicate which lineup member was the suspect; 2) the perpetrator was familiar (known) to the witness, so a physical description was not provided; 3) a non-witness source implicated the suspect (e.g., license plate, weapon, citizen tip); or 4) it was not clear as to which eyewitness provided the description used to construct the lineup. Procedure This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 41 whether lineup fillers provide effective alternatives for the unreliable witness (size). This assessment is accomplished using the mock witness method.The mock witness paradigm at first appears somewhat counterintuitive: Persons who did not witness the crime in fact have no knowledge of it view the lineup and are asked to posit who the suspect is in the array (“Which person is the accused?”). The mock witness is provided one clue: the real witness’s pre-lineup verbal description of the perpetrator. Armed with this information, the mock witness is hazarding a best guess based not on memory, but on inference derived from the physical description of the perpetrator. As recommended by Wells and Bradfield (1999), use of the specific form of the question “Which person is the accused?” instead of “Which person best fits the description?” also serves to capture the mock witness’s use of any other non-memory information also available to the eyewitness (e.g., subtle aspects of lineup display that make the suspect stand out.) The reasoning from this point is straightforward in regard to determination of bias. The real witness’s identification is interpreted in the context of mock witnesses’ ability to identify the suspect in the lineup. That is, the mock witnesses serve as a type of control condition against which to compare the actual witness’s lineup choice (Wells, Luus, & Windschitl, 1994). If mock witnesses show preference for the suspect despite their absence of any memory of the crime, the real witness’s selection of the suspect may also be due to non-memory factors rather than true recognition; any conclusion about accuracy of the real witness is confounded. As Valentine and Heaton (1999) explain,“A mock witness simulates the worst possible scenario of a witness who has no memory of the culprit beyond that which they gave in a verbal description…”(p. 61). A maximally fair lineup This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 40 procedure. A faulty lineup structure prior

to the blind-sequential procedure may jeopardize the usefulness of this new protocol. A blind-sequential protocol should lower the filler selection rate by inhibiting witness guessing, thereby reducing eyewitness error and danger for the innocent suspect.The HC data suggest that this objective was achieved. However, when the HC results were released, critics reasonably argued that the low filler selection rate (8%) could be due to poor lineup construction, i.e., perhaps the results are due not exclusively to the merits of the double-blind sequential procedure, but to poor construction of the lineups prior to the identification process. A low filler selection rate could occur if the lineup is too “easy”: fillers are not plausible options, so eyewitnesses can arrive at the suspect through inference rather than memory. Taking this concern very seriously, we developed the means to analyze quality of the lineup structure through a laboratory mock witness procedure (Doob Kirshenbaum, 1973) applied to a sample of the HC lineups. Study 2: Evaluation of Hennepin County lineup fillerquality Assessing lineup fairness: The mock witness method The evaluation of lineup fairness involves two related qualities: lineup bias and lineup size (Malpass & Lindsay, 1999). First, the lineup must not be biased against the suspect, i.e., the suspect should not be identifiable on the basis of position in the lineup, photo quality, or other non-memory cue. Second, all members of the lineup must be reasonably plausible suspects based on the description provided by the witness. Lineup evaluation is accomplished, then, by determining whether the suspect stands out in the lineup in a manner not attributable to the witness’s memory of the event (bias),and This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 39 involved perpetrators who were well known to the witness. The remaining data leave a subset of 109 lineups. In this group, cross-race identification decisions resulted in 53% suspect choices, and 11% filler identifications (a 64% choosing rate); same-race decisions generated a similar level of suspect choices, 55%, and 4% filler identifications (59% choosing rate). The 7% difference in known errors (fillers) is not significantly different, .08. ummary of data analysisHennepin County double-blind sequential field tests produced acceptable suspect identification rates relatively comparable to those in prior laboratory and field tests. Repeated viewing of the lineup was associated with increased filler selections (known errors). The new procedure donot appear to have sacrificed jump-out identifications. Patterns of eyewitness response to strangerand familiarperpetrators were reasonable, with strangersuspect identifications at a lower level. Confidence and suspect identifications were significantly related, particularly for jump-out identifications. For other categories of expressed confidence (even ), confidence and decision outcome were not significantly related. A positive outcome of the project was the low filler identification rate, which suggests increased protection for innocent suspects. However, additional attention to the fillers is necessary. Could there be an alternative explanation for the HC results: poor fillers? Functional size of the lineupthe number of lineup members who effectively match the description of the perpetratoris an important consideration for all lineups. Lineup fairness depends on adequate lineup construction as well as appropriate administration procedures, and lineups are vulnerable to bias in both structure and This document is a research report submitted to the U.S. Departme

nt of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 38 A similar relationship between confidence and decision outcome exists when only stranger identifications are examined: (9) = 31.08, = .0001 and s = .63,when jump- outs are included ( = 50); s = -.07 when jump-outs are eliminated from the analysis ( 24). Again, filler rates are higher than suspect ID rates in the three remaining confidence categories: 75%, 60%, and 63%, respectively. Confidence and witness decisions are significantly related in this sample when jump-out identifications are included in the analysis. Beyond jump-out decisions, however, even witnesses who express high confidence in their decision choose a filler more often than the suspect. Three caveats to these results are important. First, the data do not directly address the confidence-accuracy relationship, as ground truth (actual guilt or innocence) is impossible to ascertain in this archival field data. Second, only a very small number of lineup reports included a clear statement of witness confidence, therefore this sample may not well represent the broader data set. Finally, it is difficult in some cases (particularly the “jumpouts”) to determine whether the confidence rating is directly that of the witness or is the officer’s assessment of witness confidence. Given these difficulties, no further data regarding between confidence and other witness behaviors are reported. Cross-race and same-race identifications. The research literature has established a reliable same-race identification advantage such that faces of one’s own race are better remembered when compared with faces another, less familiar race (Meissner & Brigham, 2001). The field data here do not offer clear information, particularly because accuracy of identification is impossible to ascertain in the field. In addition, while 96% of lineup reports ascribed the race of the offender, only 44% of reports included mention of the witness’s race. Finally, it is appropriate to eliminate from analysis the lineups that This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 37 greater levels of qualifiers to the choice occur with stranger crimes than with familiar perpetrators. Witness confidence and decision outcomes. Lineup administrators were asked to record verbatim any eyewitness comments regarding confidence. However, this requirement led to specific comments in only 15% of lineup reports (n = 42). Assuming that jump-out identifications also indicate a meaningful level of certainty and can therefore be added to the analysis, a total of 125 lineups afford information regarding witness confidence (45% of the overall lineups, and 71% of lineups in which the witness made a lineup selection). Witnesses expressed confidence in a decision to choose from the lineup in only two of the 125 cases (1%). For analysis, the memorialized comments were sorted into four categories of decreasing certainty: (1) -outs; (2) high confidencethose not included as jump-outs but involving statements of 80%+ certainty or phrasing such as “quite certain,” “sure,” and “positive”; (3) moderate confidence, involving “pretty sure,” “fairly sure” and estimates between 50% and 80%; and (4) low confidence, with such phrasing as “not sure,” “not very,â

€ or “not too” and “low.” Within this subset of lineup data, witness confidence and decision outcomes were significantly related, with greater confidence associated with higher levels of suspect identifications,2 (9) = 83.73= .0001and = .74. This significant statistical relationship was largely due to the impact of jump-out decisions, 99% of which were suspect identifications. If the jump-out category is removed from analysis (leaving ), a significant relationship no longer appears. In each of the remaining three confidence categories (high, moderate, and low), witnesses selected fillers at a slightly higher rate than suspects (filler identification rates of 58%, 67%, and 63%, respectively), = .08 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 36 longer time (more than ten minutes) generated a 59% suspect identification and 14% filler choice rate. We do not know the accuracy of suspect identifications, of course. These rates suggest that witnesses are more willing to choose from the lineup when they have had longer exposure to the culprit. Lower suspect identification rates occur in situations where one might expect weak memory. An interesting ancillary finding is that 43% of the witnesses who made a lineup selection after viewing a very brief strangercrime were reported to have expressed some qualifications to their identification (see table 6A similar breakdown of the familiar (known) perpetratorcategory also shows an outcome which is at least intuitively coherent. Witnesses who reported some familiarity with the perpetrator (e.g., a familiar face seen on multiple prior occasions) chose from the lineup at a very high rate (91%), selecting the suspect in 84% of the lineups recorded, fillers in only 7%, and making no choice in 9% of the lineups. And, those witnesses who knew the perpetrator, often by a street name,made suspect identifications at 96% with 4% making no lineup choice. It should be noted that, in this latter group, persons making lineup choices included not only observers and victims of crim25% of the witnesses were those involved through indirect knowledge of the crime. Thus, this category included “confirmatory” lineups that are not typically part of laboratory studies.Given the many factors that co-occur in each of the four viewing categories, it is risky to infer cause and effect between any set of them. Table 6describes the correlates of these eyewitness experiences. The patterns of eyewitness response make sense, given the context for the identification. For example, brief interaction with a stranger produces the lowest choosing rates, jump-outs, and suspect identifications of the four groups. Also, This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 35 selected a photo from the lineup), ninety-six (55%) were jump-outs. The resulting choices produced 99% suspect identifications and only one filler choice. The “immediate choice” criterionis similar to that used by Behrman and Richards (2005) in their field research as a signal of automatic processing versus eliminative memory strategies of witnesses. Six percent of their witnesses were reported by police to use such an automatic memory response. In this HC data, we use a broa

der criterion,including not just officer interpretation but also strong positive witness responses,and 34% of witnesses fall into this “jumpout” category. Stranger vs. familiar perpetratorsThe Hennepin County program required blind sequential lineups for all felony cases, regardless of familiarity between eyewitness and perpetrator. In 66% of the lineups, witnesses reported that the offender was a stranger. The following analyses explore eyewitness response as a function of familiarity and of the witness’s opportunity to view the culprit. First is a comparison between situations in which the perpetrator was a stranger to the witness, based on the best knowledge of the case investigator, and those in which the culprit was at least familiar to the witness and sometimes known quite well. Not surprisingly, suspect identification rates were significantly lower for strangercrimes (35% as opposed to 90% for known perpetrators), (2) = 74.68 .001, with lower choosing rates (47% versus 94% for known perpetrators). Filler rates were low in both categories (see top frame of Table 5). The lower suspect identification rate in the strangerset of lineups may elicit concern from investigators. A finer distinction is perhaps useful (Table 6). Crimes of brief duration committed by strangers (estimated as only a few minutes) produced 32% suspect identifications and 11% filler choices. Crimes in which the witness viewed a stranger for a This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 34 Returning to the subset of 128 lineups, over half of the witnesses (53%) viewed the lineup just once. Another way to look at the impact of repeated viewing is to examine witness decisions in the remaining 47% of lineups, 60 lineups in which witnesses requested additional laps. In 15 of 36eups in which complete information was available, witnesses indicated recognition of a photo during the first showing of the lineup, with nine jump-outs (, the witness made an immediate choice or made comments such as “that’s the guy,”) and six tentative identifications. The tentative identifications included comments such as “hold that one” and “that looks like him.” The second lap (for two witnesses, a third lap) was the point at which the witness confirmed his or her choice; 100%, all 15selected the suspect. Although the additional 21witnesses selected a photo from the lineup, they did not indicate recognition until at least after the second showing of the lineup. In these cases, climbing error levels are apparent. Thirteen identifications made after two lineup showings produced 62% suspect and 38% filler choices; eight witnesses made a decision after three or more showings and generated more filler identifications (50%) than suspect identifications (38%). In summary, the evidence shows that a witness who has not made even a tentative identification at the end of the first lineup viewing presents a subsequent risk for misidentification. Jump-out” identificationsA concern of investigators was that sequential lineups might diminish the likelihood of “jumpout” identifications, inhibiting desirable witness expressions of absolute certainty. For this analysis, jump-outs were taken to be those lineups for which the investigator recorded an “immediate choice”and/or witness comments such as “that’s the guy!”Of 175 choosers in the data set (those who actually This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position

or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 33 laboratory sequential rates, and the filler selection rate (8%) much lower than other procedures. These rates change somewhat as the context of the crime shifts, particularly the relationship between witness and perpetrator, as discussed below. Witness performance on sequential repetitions (“laps”). Ideal blind sequential protocol calls for only one viewing of the lineup per witness.Researchers reason that a second lap through the sequential lineup may produce a de factosimultaneous array. That is, subjects may begin to mentally compare photos and lapse into relative judgment. However, HC brought a practical worry to the considerations some number of good identifications would be lost due to overly cautious eyewitnesses. A revision to the procedure was determined: witnesses in the pilot project were allowed multiple repetitions, or laps, through the lineup, and lineup administrators were to record the details process. The details were not as thoroughly recorded as hoped, with 46% of lineup reports specifically indicating the number of repetitions. The results in Table 4are based on these 128 lineups, summarizing witness decisions for those who viewed a lineup just oncetwicethree times. Repeated viewing of the lineup was associated with significantly increased likelihood of filler choices (known errors), (10) = 31.23 .001= .13 For crimes involving perpetrators familiar the witness, this is somewhat less evident. With familiar perpetrators, the suspect identification rate was 92% for witnesses who took either one or two laps through the lineup, and 50% for the few witnesses who requested a third viewing, (4) = 7.34= .06. For lineups in which the perpetrator was a stranger to the eyewitness, a risk of additional laps was most apparent in the filler selection rate, which increased from 3% to 29%, (10) = 22.00= .007 This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 32 identification procedures involving the same suspect). Fifty-two percent identified the spect; 15% identified a foil. Three teams of researchers from England (Valentine, Pickering Darling, 2003) corded the responses of over 300 eyewitnesses to simultaneous live lineups, including situations of suspects known and unknown to witnesses. As summarized by Valentine et al., results of these studies are quite consistent: approximately 40% of witnesses identified the suspect; approximately 20% identified a filler, thereby making a known mistaken identification; and approximately 40% made no identification. Comparative laboratory data from simultaneous and sequential lineups. Steblay and colleagues (2001) summarized thirty laboratory comparisons of simultaneous and sequential lineup performance. This review, representing 4145 test witnesses, demonstrated that the sequential procedure reduced eyewitness choosing rates, with significant positive effects on accuracy (see Table 3). Regardless of whether the criminal was in the lineup (perpetrator-presentcondition) or not (perpetrator-absent), the simultaneous format produced an almost even split between filler choices and no choiceresponses. There were significantly more errors in simultaneous versus sequential lineups when the perpetrator was not in the lineup, 51% versus 28% filler selection rate, respectively. Witness decisionsFor comparison purposes, the Hennepin County data (HC) is presented in the upper section of Table 3alongside simultaneous

(SIM) field data sets.Hennepin County generated a suspect identification rate of 54%, filler selections 8%, and 38% of witnesses made no choice from the lineup. The 54% suspect identification rate comparable to that achieved with simultaneous lineups in the field and higher than This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 31 riskof false identification (not dangerous false identification itself) is present in this Hennepin County pilot: Administrator masking allows filler identifications to work as a conditional proxy measure.Assessment of specific changein lineup outcomes resulting from use of the new lineup procedure would require suitable data from concurrent blind simultaneous field lineups against which to compare blind sequential lineup performance. There are no suchcomparative baseline data for blind simultaneous lineups in Hennepin County or elsewhere. However, existing sources of relevant information from other venues can be examined provide a picture, albeit limited, of (presumably non-blind) simultaneous lineup field performance.The reader should keep in mind that the HC field data include some witnesses who elected multiple laps through the lineup and some lineups in which the perpetrator was familiar to the witness due to interactions prior to the crime. Also, in any field test, suspect identifications may include some unknown proportion of false identifications. Comparative field data from simultaneous lineupsBehrman and Davey (2001),a California field study,reported eyewitness decisions for fifty-eight simultaneouslive lineups, with a suspect identification rate of 50%, filler choice of 24%, and no choice at 26%. In 284 photo arrays, 48% of the identifications were of the suspect; foil identifications were not recorded. Behrman and Richards 005) later reportarchival police data (a portion of the cases formed part of the database for the earlier 2001 study) for 424 photographic lineups and 37 physical lineups. These were single-suspect simultaneous lineups and all were first identification attempts by the witnesses (none were preceded by This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 30 the police believe is the offender; a filler(known error) selection; or a rejection of the lineup (no choice from the lineup). Witness performance in the field is measured, then, through percentage rates of suspect identifications, filler selections, and no choiceresponses. In the field, however, we cannot assess ground truthmemory accuracy or errorsuspect identifications and filler selections become proxy measures.Unfortunately, these measures alone cannot tell us much about memory accuracy. At the simplest level, a new lineup procedure that produces a plummeting drop in suspect identifications, a perilous rise in filler selections,or overwhelming confusion in witnesses,will and should raise eyebrows and concerns. Such outcomes signal that a procedure may not be conducive to tapping witness memory or perhaps that earlier rates of field eyewitness performance were dramatically misleading. Similarly, a sharp rise in suspect identification rates should give pause; while this outcomemay indicate a better lineup procedure, it may alternatively indicate lineups with very

poor fillers. Ambiguity inherent in field measures, so an absolute level of goodness is not established. In a subsequent section of this report, I will return to the issue of how field data might best be interpreted. Of importance at present is cautious and nuanced understanding of filler identifications. Filler selections in the field, asin the lab, are counted as eyewitness mistakes. Filler selections are not dangerous errors (they typically do not lead to a prosecution of the filler), but the rate of filler selections, under the right conditions, can provide information about the risk for an innocent suspect in the lineup. Wells (2006b) has termed filler identifications “conditional proxy measures”that can work well under some conditions but not others. The key condition for use of filler selections as an indication of This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 29 produced an effective DVD for training and review. A similar positive experience has been reported by Lt. Kenneth Patenaude (2006) of the Northampton Police Department (MA), where his department ultimately came to strongly prefer the new procedure. In 2003, the Northampton Police Department made an “additional modification to the policy, by making the blind administration of photo arrays the mandatory, rather than preferred, method of presentation.”(p. 418). Research Question 2Do the Number and Quality of Identifications Change with the Double-Blind Sequential Procedure? The sequential format is expected to lower eyewitness choosing rates.Researchers believe this is due to the witness’s movement from relative to absolute judgment, a process that also may involve an upward criterion shift (Wells & Olson, 2003). The witness knows that there could be another person, a better match to memory, coming later in the sequence, and at any given time the witness does not know how many photos remain in the array (Lindsay & Wells, 1985). Thus, he or she is forced to dig a bit more deeply into memory. If a match between memory and photo is compelling, a choice from the lineup is likely; otherwise, the witness should reject the lineup. Guessinglucky or otherwiseshould be reduced. The sequential procedure is seemingly a more conservative test of memory, and, if this technique is working well, a low rate of filler selections likely. In the best of situations, the blind sequential lineup would maintain a reasonable level of accurate identifications, but an appreciable drop in false identifications as guessing is reduced. One of three outcomes will occur as the endpoint of a witness’s decision for a police lineup in which a single suspect is present: a suspect identificationof the person who This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 28 involved minimal cost. From an administrative perspective, the police chiefs had wondered if the need for blind administrators would significantly increase man-hours. However, as Minnetonka Police Chief Joy Rikala noted, “There [are] no cost implications of this. It’s negligible (p. 409). The double-blind sequential lineup procedure is well within the reach of police agencies. Yet-unmeasured practical advantages of the new procedure are likely in the

future. Better lineup screening devices allow police to move more quickly in their work, to find the right perpetrator without wasting time on false leads, and to remove perpetrators from the streets before additional offenses are committed. he potential for damaging cross-examination at trial regarding cues consciously or inadvertently sent to witnesses is effectively removed by the use of a blind administrator. Chief Rikala’s report of numbers of filler picks deserves comment. A filler selection rate means that HC prosecutors and investigators can spend less time tracking down and clearing the filler to avoid defense challenges at trial. It is also important to recognize that a witness who identifies a filler (perhaps when the true offender is not in the lineup) is a “burned” witness, one who will not be seen as credible for the continuing investigation or for court testimony. Investigators do not want to spend their witnesses with lineup procedures that may prompt filler selections. Finally, filler identifications may imperil convictions even when the actual perpetrator is ultimately identified, as the mistaken identification will undercut the correct one. The biggest hurdle in implementation seemed to be a general resistance to change. even fewer problems are expected the longer the protocol is used. New investigators will be trained in the reformed procedure, “It was simple to pick up.”(p. 410). HC has This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 27 NIJ guidelines were positioned as a “framework for innovation”(NIJ Guide, p. 9) anticipation of technological developments such as computer-based imaging to add effectiveness and efficiency to lineup procedures. Sequential lineup presentation as well as double-blind procedure can be facilitated through laptop delivery of the lineup, as researchers do in the lab. Although computer-based lineups were not a formal part of the HC pilot, the lab technology may become the solution to some implementation issues. Investigators sometimes followther jurisdictions the creation of functional equivalents to blind lineup administration for situations in which a true blind administrator is unavailable. The most common substitute method is the fold(or envelope) technique. lineup photos are randomly placed into six numbered folders by a person uninvolved in the case, shuffled, and then provided to the witness by the investigating officer. The witness is asked to respond to each photo without displaying the photo to the officer. helps to maintain a blind status of the officer. In Hennepin County’s neighboring Ramsey County,a wood-crafted 6-door “lineup box” provides a similar functional equivalent when a second officer slides the randomly positioned 6-pack photos into the box and the lineup administrator is positioned in the room so as not to view the photos. The witness opens, examines, and closes each of the six doors, one at a time. Overall, police chiefs and investigators found the pilot project to be much easier to implement than anticipated. In smaller jurisdictions, investigators had the whole project underway in less than two weeks. In the larger jurisdictions, the process took less than a month. Initial skepticism and unease faded. “By the end of the project,” Minneapolis Police Chief William McManus reported, “the burden on investigators was far less than my department had anticipated”(Klobuchar, et al., 2006, p. 409). The pilot project also This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view express

ed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 26 and sergeants. New Hope, the smallest department, reported no problems with double-blind procedure despite employing only two investigators. Without greatly hindering collaboration, one larger department used property crime investigators as blind administrators for investigations dealing with crimes against persons and vice versa. The introduction of a second officer to a witness for lineup administration was not found to hinder investigations in any significant way. No witness refused to view the lineup with the new unknown officer. On the contrary, some witnesses reported appreciation of the procedure and an understanding of the reasoning behind it. Investigators reportno drop in effective suspect identifications or in their ability to “get the job done” as they compared their perception of blind sequential lineups to past simultaneous lineup outcomes (Klobuchar, et al, 2006, p. 408).Agencies also reported positive effects on lineup results due to the new procedures. Investigators noticed witnesses were now less likely to make a misidentification of a filler. “We’re not having a lot of people pick fillers,” Chief Rikala observed(p. 410). To address concerns about repeat offenders and multiple witnesses, as well as to explore options for investigators to deliver lineups for their own cases, the Minneapolis Police Department has been working with the Hennepin County Attorney’s Office to develop laptop computer lineup administration. Laptops can randomly order six photos for viewing by a witness and be positioned across the table from the administering officer (dubbed the “Battleship” position, ireference to the well-known table game). In cases with multiple witnesses, the laptop will randomly shuffle the six pictures for each administration. This new procedure will allow investigators to conduct a lineup without the aid of a second officer, dispelling lingering concerns about the double-blind procedure. As noted earlier, the This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 25 significant crime just after it happens. This “allhands” policy constrains the pool of uninformed officers available to meet the requirement of a blind administration. A related circumstance occurs with investigations of great urgency (e.g., an Amber Alert) or very high profile crimes. A concern specific to Minneapolis police was the problem of chronic offenders, whose mere presence in a lineup would reveal the identity of the suspect blind administrator (Scoggin, 2005). In Hennepin County, 34% of the pilot cases involved multiple witnesses. In these situations, one witness’s identification of a lineup member can compromise the blind status of the administrator for the next witness. A separate blind administrator for each witness may be prohibitively cumbersome and expensive. Finally, since a blind administrator must sometimes travel with the main case investigator to meet witnesses, the administrator loses time that would otherwise be spent working in a larger role on other cases. Investigators also worried that the rapport and trust that develop between an investigator and witness may be jeopardized when an officer unknown to the witness is introduced for the viewing of the lineup. The lack of prior relationship between investigator and witness also was anticipated to be troubleso

me when the case involved an uncooperative witness. Crime often takes place between not--innocent bystanders, thus witness motives may compromise any lineup procedure, sequential or simultaneous. The “no choice” category (and possibly filler choices of witness responses) in this study may hide some of these uncooperative witnesses. problems with the blind administrator requirement were less serious than anticipated. To solve problems related to limited availability of investigators to serve as blind administrators, smaller departments turned to staff, such as patrol officers, captains, This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 24 potential for misidentifications when eyewitnesses engage in relative judgment, investigators say they better understand and appreciate the new protocol. One investigator explained, “I like the format better. I like that the person is studying one picture. It’s larger, and you can see them concentrating on a single picture and reflecting back to whatever event they had. From that alone, I think it’s a success.”(p. 406). The larger size of photos in the sequential display was considered a collateral benefit of the sequential format. Ultimately, the change to a sequential lineup format caused few problems, none of them serious or enduring. Although New Hope investigators were initially reluctant to implement the sequential presentation requirement, Chief Gary Link remarked, “it took maybe two or three lineups before they realized that it wasn’t that big of a deal.” (p. 406). Double-Blind AdministrationPolice chiefs and investigators were significantly more concerned about blind administration of the lineups. With a traditional lineup, an investigator conductthe array as part of the case investigation with no additional staff required. number of operational challenges were anticipated with the new blind requirement and some problems emerged. In smaller departments with few investigators, all officers may be focused on a single case, so a blind administrator be difficult to find. Witnesses located at odd hours (middle of the night) or those in transient populations can make the coordination of a second investigator at the scene inconvenient. Most municipal departments operate independently and although municipalities have mutual aid packs for patrol response, cooperative arrangements are rare for follow-up investigations. Even in larger departments, difficulties surfaced. Some departments bring a near full force of officers to a This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 23 these meetings and interviews are summarized below. e introduction of double-blind administration and sequential presentation to lineup procedures initially posed problems, but proved to be less troublesome than originally presumed. The final action taken after the pilotto roll out lineup reform county-widis the strongest endorsement of the new procedures (Scoggin, 2005). Sequential Presentation. The problems experienced with implementation of sequential lineups were primarily in the simple details of orchestrating the new physical format of the photos. Prior to the pilot project, lineups consisted of “sixpacks,” six photographs presen

ted on a single sheet of paper. Filler pictures were selected from online photograph repositories of arrested persons by using culprit description parameters. It took some experimentation and innovation to determine how to best turn the downloaded single sheet of paper into a viewable sequence of six individual photos. vestigators had separate the six pictures and enlarge them, while maintaining consistency of background and color. lot of people spent a lot of time at the Xerox machine,” remarked Bloomington’s police chief Laux(Klobuchar, et al., 2006, p. 406). Within a week, however, one department created a new photo template to remedy the problem. In an attempt to make the process more efficient, ongoing efforts are being made in the county to create software and adapt the online photo repositories to the requirements of the new protocol. The introduction of the sequential lineup presentation made apparent for many investigators the strong desire of eyewitnesses to aid their lineup decisions by comparing otos side by side. Even after being instructed about the new lineup format, witnesses still would ask to see two photographs simultaneously.Now realizing the increased This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 22 occurred only at the witness’s request. If a review of the photos was requested by the witness, all photos were shown in the same order for the second “lap.” Results Research question 1: Can the laboratory procedures be effectively implemented in the field? Hennepin Countysituation was different from many jurisdictions that have implemented lineup reform.The Cromedy case in New Jersey prompted reform, the Avery case in Wisconsin, the Ryan Commission in Illinois. In Hennepin County, however, police chiefs could claim that existing procedures were working well. With the exception of one misidentification caught before trial,there had been no apparent problems with lineups (Klobuchar & Knight, 2005). Nothing seemed to be broken. County Attorney Klobuchar reports iscussions and training sessions about the new protocol nevertheless convincedfour department chiefs to move forward with a pilot test as a means to reduce the risk of false identification and to increase confidence in eyewitness evidence (Klobuchar, Steblay, & Caligiuri, 2006). The question of implementation effectiveness is largely informed by qualitative and anecdotal information. Midway through the pilot, a group of thirteen investigators conveyed their concerns and impressions at a meeting with the researcher and an attorney from the Hennepin County Attorney’s Office. Additional feedback from police agencies was fielded by the County Attorney’s Office. Interviews with police chiefs and law enforcement personnel were conducted by HC attorneys and are reported more fully recent article by Klobuchar, Steblay, and Caligiuri (2006). Issues that surfaced through This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 21 Sequential presentation. Lineup photos were presented one at a time. The witness was not allowed to compare photos side--side at any time. The full sequence was completed even if an early identification was made, and the witness was informed that this co

mpletion is required by the procedure. Each witness was instructed that he or she would view an entire photo series. The objective of the procedure was to neither inhibit nor demand a witness response after each photo. Therefore, the protocol did not require a specific “yes” or “no” answer to each individual photo. The witness generated the pace through the photos, commenting or signaling readiness to view the next photo. Under this procedure it is entirely appropriate if a witness makes an identification before seeing all the photos; such a witness has made a positive identification without needing to see the entire array. However, the protocol also requires that the witness views all the photos even if a choice is made before the lineup display is completed. When a witness identified a photo during the presentation of the array, the officer documented this, and then reminded the witness that the procedure required all the photos to be shown, and the remaining photos were shown. If the witness had not made any identification by the end of the photo array, the officer then solicited a response, asking the witness if any of the lineup members had been familiar. 2 practice allowed the benefits of a full lineup display and the blind recording of any witness responses to each lineup member. One benefit of this may be seen, for example, if identification of a filler occurred early in the lineup; the investigating detective will certainly want to know how the witness reacted to the suspect’s photo later in the lineup.Hennepin County adjusted the recommended protocol soto allow the witness view the entire sequential lineup display as many times as desired. Review of the lineup This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 20 to the lineup task. Incomplete reports were supplemented with data from the complete police file. Most investigators continued their prior reporting procedure, wherein each wrote a narrative summary of the lineup process. The Minnetonka Police Department also developed a supplementary form for the witness that provided instructions,including the cautionary instruction that the perpetrator may or may not be in the collection of photos to be displayed. Minnetonka witnesses were requested to check one of two options“I am unable to select any photo as the suspect in this case” or “I have selected photograph # ___ from the group”—and were offered space to write comments. Procedure HC employed single-suspect lineups,and all lineups involved witnessidentification attempt. The following five basic principles for blind sequential lineups were employed, the first three part of police procedures prior to the pilot program. (Additional detail is provided in Table 2.) Effective lineup construction. A six-member lineup included one suspect and five fillers, with fillers chosen to match the witness’s description of the perpetrator. Cautionary (unbiased) instruction. The witness was instructed that the perpetrator may or may not be in the lineup. Confidence statement. A statement of witness confidence, in the witness’s own words, was recorded at the time of the identification and before any feedback. Blind Administration. The lineup administrator did not know who the suspect was, and the witness was instructed that the administrator did not know which lineup member was the suspect. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Departme

nt of Justice. �� Double-blind sequential lineups 19 was to see if “the techniques suggested by lab investigation really work in real police investigations(Scoggin, 2004). Two questions framed the project: Can the procedures be effectively implemented in the field? Do the number and quality of identifications change with the blind-sequential procedure? (Scoggin, 2005). Method Sample The Hennepin County pilot program focused on felony cases in four municipal police departments, including both strangerand familiar perpetratorlineups. The cities representfour levels of population and included both urban and suburban locales. In Minneapolis, the largest of the four cities, the protocol was used exclusively by Central Investigations, which handles violent crimes. The project involved 280 lineups from 117 cases, representing 206 eyewitnesses over a twelve month period ending in November 2004. (See Table 1) The Augsburg College Institutional Review Board approved the research protocol. Hennepin County Judge Kevin Burke signed the appropriate documentation on behalf of Hennepin County to allow PI access to the confidential case filesJudge Burke prescribed that the data be extracted from these files without personal identifiers and that it be reported only in aggregate form. The case files were gathered by the Hennepin County Attorney’s Office and made available to the for the collection and analysis of descriptive data. Data were drawn directly from investigators’ lineup reportswhichprovided information regarding lineup structure and administration, lineup context (e.g., crime type), and eyewitness responses This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 18 the identification. (Witnesses indicated how certain they had been at the time of their identification.) Perhaps more alarming is that an extensive range of variables was inflated in conjunction with retrospective certainty, including witness reports of the quality of their view of the perpetrator, how much attention was paid, ease of the identification, and basis for the identification. Participant witnesses who received confirming feedback were also more willing to testify about their identification and reported a greater ability to remember strangers. These outcomes demonstrate the reliability and robustness of distortions produced by post-identification feedback and reinforce the need for double-blind lineup administration and prompt full recording of eyewitness certainty comments. The Hennepin County pilot program incorporated both double-blind and sequential components, offering the first objective baseline measure of eyewitness responses under these conditions in the field. Study 1 Hennepin County Field Pilot Program In November 2003, the Hennepin County Attorney’s Officeembarked on a year-long pilot study during which investigators used a revised lineup procedure in the city of Minneapolis and three suburban communities. The goal of the Hennepin County pilot to “implement an eyewitness identification pilot program with Hennepin County law enforcement agencies to reduce the possibility of incorrect identifications from photo lineups(Klobuchar, 2004). The strength of scientific laboratory results underlying the new lineup protocol hadconvinced HC leaders that the new methodsuperior. Thus, the purpose of the pilot s not a comparative evaluation between new and old techniques. Instead, the program was an attempt to identify and remedy implementation issues, to determine how recomm

ended lineup procedures can best be brought into practice. The task This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 17 The double-blind procedure counters the above problems by suring equal attention to all witness comments and -selective interpretation of witness responses. The witness must depend on memory and the administrator records an objective appraisal of the lineup procedure and results, thereby protecting the tegrity of the lineup. Blind administration is essential. A lack of administrator masking confounds interpretation of lineup outcomes because such a flaw changes the investigative question from “Is this suspect the perpetrator?” Did the witness pick the person police thought was the suspect?When double-blind administration of the lineup is not used, there is potential not only for erroneous identification, but for the investigatorbehavior to influence eyewitness’s subsequentconfidence in her or her decision. Research demonstrates a modecorrelation between witness confidence and accuracy (Sporer, Penrod, Read, & Cutler, 1995) andthis relationship is easily corrupted. Factors that produce witness accuracy (e.g., good viewing conditions, lengthy exposure, distinct perpetratorfeatures) are not the same as those that may affect confidence (e.g., confirmatory feedback from police, post-event information, or influence from other witnesses). Adangerous erosion of the confidence-accuracy relationship occurs when an eyewitness is exposed to factors that inflate confidence but have little relationship to accuracy (Wells & Bradfield, 1998). A recent meta-analytic review underscores the need for lineup administrators to assess eyewitness confidence before any feedback is provided (Douglass & Steblay, in press). Analysis of 20 available laboratory tests demonstrated that confirming post-identification feedback received by the witness immediately after the identification (i.e., “Good. You identified the actual suspect.”) inflated significantly the participant-witnesses’ retrospective confidence reports when compared with a control group told nothing about the accuracy of This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 16 Two additional conduits for influence are the lineup administrator’sselective attention to witness comments which can significantly shape the witness’s decision and selective recording of fillers especially in case of low-confidence filler selections.Examples may help to portray these threats. nsider the lineup scenario,using either simultaneous or sequential format,in which the witness pauses at Photo #3, the suspect’s photo. Note even a seemingly conscientious response from an attentive but non-blind lineup administrator may push the witness toward the suspect Witness“Hmmm…” (long pause) “Maybe…hat’s the look …” (shrug). Non-blind Administrator: “Tell me more…What’s familiar about #3?” Witness: “The eyes arefamiliar. The hair was different”… (more conversation ensues) Non-blind Administrator“Okay, should I indicate you chose #3?” The conversation may be quite different if the lineup administrator knows photo #3 to be a filler.The investigator may unintentionally push the witness aw

ay from the filler, a photo presumed to be not directly helpful to the investigation. Witness: “Hmmm..” (long pause) “Maybe… That’s the look …” (shrug). Non-blind Administrator: “Okay, not exactly? Look carefully, you don’t have choose unless you’re certain.” … Should we go to the next photo?” Consider a final scenario, again with the witness slowing down to examine Photo scenario reveals the easy error of selective recording. Witness: “Hmmm..” (longpause) “Could be… I’m just not entirely sure…” Non-blind Administrator: (If #3 is the suspect) records tentative ID of suspect (If #3 is a filler) records failure to ID the suspect. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 15 nonverbal cues. A supplement to this procedurenotification to the witness that the officer does not know which is the suspectfords the additional advantage that the witness is less likely to seek or infer cues from the officer’s behavior.Specifically, then, the blind lineup procedure includes the following key features: The officer displaying the photos does not know which photo depicts the suspect (“blind administration”).The witness is informed that the lineup administrator doesn’t know whichphoto, if any, is the suspect (“blind instruction”)An assessment of witness confidence is taken at the time of the identification and before feedback from police or others (Wells, et al., 1998). The NIJ Guide identified blind procedure as a direction for future exploration and field testing. The expressed concern was that this desirable feature was likely to be impractical in police field use. Focused research conducted since publication of the NIJ Guide indicates that the double-blind procedure essential to scientific integrity more broadly is also critical to the task of securing accurate eyewitness accounts (Douglass, et al., in press).Any operational difficulties in the blind procedure must be seriously weighed against the inherent defects of the non-blind lineup. In the lineup situation, there are multiple avenues for administrator influence (Wells & Seelau, 1994) st obvious are the unknowing leaks and “tells” that cue the witnessas to the suspect’s position in the lineup. These include verbal comment, facial expression, posture change, hesitancy, enthusiasm, a different handling of the photo, and any number of such unintentional signals. Theris no need to assume intentionality of influence on the part of the lineup administrator for these influences to be dangerous to an innocent suspect. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 14 Instruction to the witness that the perpetrator may or may not be in the collection of photos to be displayed (an “unbiasedor “cautionaryinstruction). Instruction to the witness that the sequence must be completed even if/after an identification is made, and the procedure conducted this manner. The witness unaware of how many photos are in the sequence. Photos presented one at a time, with a decision made before examining the next. The witness not allowed to “go back” over the sequence.The witness not allowed to place photos next to one another. Blind lineup administratio

n. Double-blind experimental procedure, in which neither experimenter nor subject know the subject’s treatment conditionan essential element of sound scientific method usedto prevent inadvertent contamination of research results. Interpersonal expectancy effects occur across a broad set of human interactions, necessitating a double-blind method for data accuracy (Harris & Rosenthal, 1985; Rosenthal, 1988; Rosenthal & Rubin, 2002). First noted as essential for lineup administration by Wells in 1988 and reinforced by a broader group of scientists in recommendations for lineups in 1998 (Wells, et al, 1998) there is wide agreement among eyewitness scientists that the double-blind lineup procedure is crucial in eyewitness procedures (see e.g, Wells, 2006b; McQuiston-Surrett, Malpass, & Tredoux, 2006; Haw & Fisher, 2004). A double-blind lineup protocol (often simply referred to as “blind”) helps to manage the inherently suggestive nature of the situation (Douglass, Smith, & Fraser-Thill, in press; Phillips, McAuliff, Kovera, & Cutler, 1999). A lineup administrator who does not know the identity of the suspect is unlikely to lead the witness to the suspect through verbal or This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 13 the simultaneous format. Why? In simultaneous lineups, witnesses with weak memories can pick the lineup member who looks most like the culprit shallow and relative judgment. With a sequential lineup, the witness with a weak memory is inhibited by the greater difficulty of the task and cannot simply compare photos to arrive at a best guess (Wells, 2006). It is also important to note that the sequential lineup is more diagnostic of guilt when the witness does make a choice, yielding odds of guilt almost twice that of the simultaneous lineup. For police, the critical question is: Is the identification a good predictor of guilt? The blind-sequential lineup procedure improves the odds that a suspect, if identified, is the actual culprit (Wells, 2006aResearchers use the phrase equential lineup” shorthand for what is actually collection of rules that represent best practice for conducting eyewitness identifications. For example, the sequential procedure assumes a single-suspect model (only one suspect in the array) and that the lineup task is the first identification attempt by the witness. Furthermore, an effective sequential procedure cludes the following (see e.g., Lindsay & Wells, 1985Cutler & Penrod, 1988)Effective lineup construction, designed to detect the unreliable witness and to protect an innocent suspect by drawing guesses to filler photos.lineup of at least six members, five of whom are foils unknown to the eyewitness. Each lineup member chosen to match the witness’s description of the perpetrator.The suspect’s position in the lineup determined in an(approximately) random manner. This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 12 perpetrator before moving to the next photo. The witness does not know how many photos will be shown, thereby decreasing the pressure to choose any one photo in the sequence. This one--a-time procedure is intended to discourage the eyewitness from simply deciding who most resembles the perpetrator, thus

forcing a more absolute cision criterion. The NIJ Guide only suggestedthe sequential method for practice, presumably because the available research in 1999 did not yet confirm the sequential lineup’sbenefit beyond the traditional simultaneous format. Since that time, a meta-alytic review has demonstrated reliable positive laboratory outcomes from use of a sequential procedure (Steblay, Dysart, Fulero, & Lindsay, 2001). Simultaneous versus sequential formats produce dramatically different choice and accuracy outcomes.Witnesses who view a simultaneous lineup array are more likely to choose a photo from the lineup. When the perpetrator is present, this higher choosing ratemay boost correct identifications, likely aided by relative judgment. In a target-absent simultaneous isplay, the increased tendency to choose translates into greater risk of false identification. The reverse is true for the sequential format. Particularly when the perpetrator is absent from the lineup, the sequential format is preferable,reducing false alarms by 23% and identification of a designatedinnocent suspect by 2/3 (27% to 9%). Some investigators may be concerned about the reduction of correct identifications for sequential lineups compared to simultaneous outcomes (3% and 15% average decreases, for the subgroup of “choosers” and for all witnesses, respectively). It is not known if this drop represents a loss of accurate eyewitness identifications, a reduction olucky guesses, or both. Penrod (2004) has presented a compelling argument that guessing is a significant component of eyewitness decisions. This is likely to be particularly so with This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 11 of which were instances in which the actual offender was not in the lineup illustrate exactly this roblem: witness ability to correctly reject a target-absentlineup (The Innocence Project, 2006). The results of controlled experiments predict negative outcome when police unknowingly place an innocent suspect in a lineup. The impact of relative judgment when the offender is absent from the lineup was demonstrated convincingly by Wells (19). rticipant-witnesses to a staged crime were shown one of two versions of a lineup. When the perpetrator was present in a six-person lineup, 54% of the witnesses selected him. All witnesses had been given an unbiased instruction (“the perpetrator may or may not be in the lineup”) and 21% opted not to choose from the lineup. Now, the key question: What will happen to the second group of witnesses who view the same lineup minus the perpetrator? If 54% of witnesses truly recognize the offender when he is present, is 54% who would have identified the offender had he been in the lineup should join the 21% who reject the lineup, producing a 75% “nochoice” response. What happenis quite different: Only 32% of the witnesses correctly rejected this target-absent lineup, only a small gain from the 21% expected in this response category. Sixty-eight percent of the witnesses chose from the lineup, most of the filler identifications falling on the photo that was the next-best match to the offender, placing this innocent suspect in jeopardy. This result has been dubbed the “target-foils (Clark & Davey, 2005). Even in a perpetrator-absent lineup, it is likely that one lineup member will provide a better relative match to memory than the others,thereby drawing the attention of the eyewitness andincreasing the risk of false identification. To address this problem, the sequential procedure presents the eyewitness with one lineup member at a time and requires the witness to decide whether o

r not that person is the This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 10 collection of sound scientific laboratory research.In the majority of -exoneration cases, mistaken eyewitnesswere the primary evidence leading to conviction (Wells, et al., 2000). The sychological research has shown that eyewitness reports are often unreliable and that unintentional police influence can exacerbate witness tendencies toward inaccurate lineup choices. The NIJ Guide clear about remediation of this problem: The eyewitness should be given unbiased lineup instruction (“The perpetrator may or may not be in this lineup”), lineups should be constructed fairly (e.g., foils matched to perpetrator description and the perpetrator not standing out in the lineup), and officers should record results in a prescribed manner. Also included in the NIJ Guide a recognition that “advances in social science and technology will, over time, affect procedures used to gather and preserve eyewitness evidence”(p. 8). Three examples were specified as areas for future exploration: the sequential method of lineup presentation; blind lineup administration procedure; and computer-based imaging systems facilitate the use of improved procedures. Sequential method of lineup presentation. A sequential lineup presentation attempts to remedy a troublesome aspect of the eyewitness decision-making process, relativeersus absolutejudgment(Wells, 1984).Standard police lineups present the eyewitness with all lineup members (e.g., six persons) at one time. Under thisimultaneous format, eyewitnesses tend to compare lineup members to each other to determine which most closely resembles the offender in memory, a process of relative judgmentIf the witness was able to encode a vivid memory of the perpetrator and this person is in the lineup (a target-presentarray), the likelihood of a positive and correct identification is increased. The concern, however, is whether the witness will recognize the absence of the offender when in fact the suspect is not the perpetrator. e DNA-exoneration cases the majority This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 9 researchers (Doyle, 2005). Farmer approved new lineup procedures with safeguards exceeding those recommended by the National Institute of Justice.Using the unique authority granted the Attorney General in that state, Farmer implemented mandatory statewide guidelines, making New Jersey the first state to uniformly adopt double-blind sequential lineup procedures (State of New Jersey, 2002). Other initiatives have been undertaken as well. In 2002,Governor Ryan’s Commission on Capital Punishment, charged with ensuring the accuracy and justness of capital punishment in Illinois, recommended the implementation of eyewitness identification reforms (Report of the Governor’s Commission on Capitol Punishment).The North Carolina Actual Innocence Commission created a series of recommendations in 2003 for state law enforcement officers, including a comprehensive lineup protocol. In early 2005, the Avery Task Force wrote recommendations for the Wisconsin criminal justice system, recently affirmed by the State of Wisconsin Office of the Attorney General (2006). The Virginia Gen

eral Assembly also instructed the Virginia State Crime Commission (2005) to create guidelines for improving lineup procedures in the commonwealth. Individual jurisdictions in a handful of states also have implemented improved procedures, Massachusetts, Virginia, Washington, Minnesota and California among them (see e.g, Northhampton Police Department, 2005; Police Chief’s Association of Santa Clara County, 2002). Scientific rationale reviewThe U.S. Department of Justice published Eyewitness Evidence: A Guide for Law Enforcementin 1999. This document followed from concerns about unjust convictions revealed through -exoneration cases,s recommendations based on the growing This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 8 was reliable. To varying degrees, these cases sought to remedy the effects of suggestive lineups, but did little to discourage use of prejudicial procedures. The subsequent introduction of DNA testing allowed the exonerations of many wrongfully convicted individuals,numbering 183(The Innocence Project), and drew substantial attention to eyewitness error as the major contributor to these unjust convictions (Wells et al., 1988). Investigators, attorneys, and testifying witnesses who have helped to prosecute a later-exonerated individual realize with extreme regret that even well-intentioned “by the book” procedures can end very badly. Along with the horrific effects on the es of the violated innocent person and his or her loved ones, a wrongful conviction leaves the true perpetrator on the streets to commit additional offenses. The reality of wrongful conviction also hasthe potential to substantially erode public confidence in the justice system and citizens’ sense of security. As the Court considered eyewitness evidence problems, scientists over four decades developed and reported findings shed light on the reasons for memory errors in the eyewitness experience (Wells, et al., 1998). Principles of human perception, memory, and social influence illuminated not only the causes of faulty memory but preventive measures to preclude eyewitness failure. Scientific recommendations about improved lineup structure and procedure provide one means to make it less likely that innocent suspects are prosecuted and more likely that true perpetrators are held responsible for thecrimes. DNA exonerations became the catalyst for lineup reform in some jurisdictions. For example, in the wake of the New Jersey Cromedy decisionan eyewitness evidence case in which a DNA test of biological evidence collected from the victim exonerated the defendantAttorney General John Farmer turned to the lineup reforms recommended by This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 7 constitutes a due process violation if it could lead to an irreparably mistaken identification. Therefore a defendant could move to suppress prejudicial identification testimony depending on the “totality of the circumstances” surrounding the testimony(p.302). The next year, in Simmons v. United States,ruled that each potential due process violation during a lineup must be examined on the facts of the individual case. Lineups would be excluded from trial if the “procedure was so impermissibl

y suggestive as to give rise to a very substantial likelihood of irreparable misidentification”(p. 384). In the 1970s, however, the Court began retreating from the broader safeguards guaranteed in WadeStovalland . In United States v. Ash(1973) the Court refused to extend the protection of a Sixth Amendment right to counsel to photographic lineups, reasoning thatphoto display did not involve such risk so as to require a safeguard.The Court also found that even extremely biased lineups were not per seexclusionary. Instead, it was necessary to determine whether an admittedly suggestive lineup was nonetheless reliable. In Neil v. Biggers(1972) the court considered five factors for determining the dependability of an eyewitness identification: the witness’s opportunity to view the perpetrator during the crime, the witness’s attentionto the perpetrator at time of the crime, the accuracy of the witness’s initial description of the perpetrator, the witness’s certainty at the lineup, and the length of time between the crime and the identification. In Manson v. Brathwaite(1977) the Court concluded, “reliability is the linchpin in determining the admissibility of identification testimony” (p. 114). This even more firmly emphasized that the important question was not whether the identification procedure was prejudicial to the criminal defendant, but whether the identification itself This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department of Justice. �� Double-blind sequential lineups 6 enforcement professionals, and legal practitioners. Law enforcement officials are in a good position to identify gaps in laboratory-generated knowledge, and laboratory researchers can use field research to inform their efforts to achieve desirable levels of thenticity and ecological validity in the lab (see e.g., Diamond, in press). At the same time, field practice prompts questions that compel subsequent laboratory examination.Questions of interest usually center on practical issues, but hold implications for broader theoretical issues as well. Of particular relevance is the matter of how adjustments in police lineup procedure that meet convenience or practical needs of the local jurisdiction might compromise or enhance witness accuracy. With its randomized controlled trials, the laboratory offers the most objective means to address such questions. Legal rationale and review Eyewitness identification is persuasive evidence of criminal wrongdoing. On the witness stand a confident eyewitness is likely to deliver compelling testimony (Greene 1988; Rattner 1988).However memory is fallible,and even a well-intentioned and confident eyewitness may bring flawed recall to a police lineup and falseincriminating evidence to court. Recognizing this problem,the United States Supreme Court in the 60s began to put into place safeguards to protect criminal defendants from wrongful convictions through misidentification. For example, in United States v. Wade(1967) the Court held that the Sixth Amendment right to counsel applies to critical stages of pretrial proceedings including the physical lineup procedure. The Court recognized the “vagaries of eyewitness identification” and the “innumerable dangers and variable factors which might seriously, even crucially, derogate from a fair trial”(p. 218). The United States Supreme Court ruled in Stovall v. Denno(1967) that an unduly suggestive lineup This document is a research report submitted to the U.S. Department of Justice. This report has not been published by the Department. Opinions or points of view expressed are those of the author(s) and do not necessarily reflect the official position or policies of the U.S. Department