Unfulfilled Promise: - PDF document

Unfulfilled Promise: Philadelphia’s Dropout Crisis, 2000–2005Ruth Curran Neild, Ph.D. e are deeply indebted to thevision and support of many citizens and civic leaders infutures of out-of-school youth led them to initiate this research. Our work was supported by the gen-erosity of the William Penn Foundation.Dr. Candace Bell, Education ProgramOfficer at the William Penn Foundation,encouraged us to creatively use the city’sdata resources to map the landscape ofhigh school dropout in Philadelphia.Through out this project, School District of Phila delphia CEO Paul Vallas was genuinely supportive of our efforts tolearn more about the landscape ofdropout in Philadelphia.At the University of Pennsylvania, Dr.Dennis Culhane and Dr. John Fantuzzo,who developed and currently direct the KIDS data infrastructure, providedinvaluable guidance in negotiating andunderstanding the data sets and the cityagencies that provided the data. The staff at Penn’s Cartographic ModelingLaboratory—Carnell Baugh, JeanettaChurchill, Dr. Tara Jackson, Nora Hunt-Johnson, Vicky Tam, and Diane-Louise(D.L.) Wormley—were collaborative andsympathetic colleagues. We especiallythank Jeanetta Churchill and Carnell Baughfor their unrelenting work in obtaining andorganizing data sets, and Vicky Tam for cre-ating the map that appears in this report.We express sincere thanks to researchassistants Christopher Boccanfuso andGreg Katz at Johns Hopkins and Ching-fuLan at the University of Pennsylvania.We are grateful for the enthusiasm andsage advice of staff at the PhiladelphiaYouth Network—among them JennyHamilton, Melissa Orner, Laura Shubilla,and Don Spangler—who have been apleasure to work with.Finally, we thank the members of thePhiladelphia Youth Transitions Collab Weinberger, Center for Literacy; SteveHoneyman, Carmen LeBron, and DoloresShaw, Eastern Pennsylvania OrganizingProject; James Randolph and Ellen Walker,Philadelphia Department of HumanJuvenile Probation Department; ColleenMcCauley Brown and Shelly Yanoff,Philadelphia Citizens for Children andYouth; Carol Fixman and Allie Mulvihill,Philadelphia Education Fund; Paul Socolar,Candace Putter, Reintegration Initiative;Courtney Collins-Shapiro and HeatherFrattone, School District of Philadelphia;and David Fair, United Way of SoutheasternPA). We also thank the members of theKIDS Advisory Board (the City of Philadelphia’sDepartment of Behavioral Health andMental Retardation Services, Departmentof Human Services, Department of PublicHealth, Division of Social Services, Officeof Emergency Shelter Services, andSolicitor’s Office; Philadelphia Safe andSound; the School District of Philadelphia;the University of Pennsylvania; and theWilliam Penn Foundation) who reviewedand provided feedback on this report. 2 young person in the UnitedStates who embarks on adult-hood without a high schooldiploma faces a grim economic future: an annual income that is likely to beinsufficient to support a family, a greaterlikelihood of long stretches of unemploy-ment, and restricted opportunities foroccupational advancement. Cities withlarge percentages of youth who lack highschool diplomas suffer as well: they cantake advantage of fewer economic devel-revenue, and experience higher socialservice costs, more crime, less civic participation, and high levels of concen-trated and inter-generational poverty. Acity of the 21st century cannot prosperwhen large numbers of its young peoplelack this basic academic credential.Despite the serious individual and collec-tive costs that result when youth fail tocomplete high school, until now we havenot had a clear picture of how many stu-dents in the Philadelphia public schoolsearn their high school diplomas and howmany drop out of school. Data are criticalfor assessing the numbers of dropoutsfor determining whether we are succeed-ing in our efforts to retain students inschool and to reconnect dropouts witheducational opportunities. This Study This report uses a unique set of dataobtained from the Kids Integrated DataSystem (KIDS), which is housed at theUniversity of Pennsylvania’s CartographicModeling Laboratory. The KIDS systemmerges individual-level data on youngpeople from the School District ofPhiladelphia and the city’s social serviceagencies, including the Department ofPublic Health, the Department of HumanServices, and the Office of EmergencyShelter and Services. The resulting de-identified data allow us to follow cohortsof students over multiple years, examin-ing their educational outcomes as well asthe predictors of graduation and dropout. This report addresses three central sets of questions:  How many students in grades 6 through12 drop out of Philadelphia’s publicschools in a single year? What are the key characteristics of these students,including their age, grade, race/ethnicity,gender, type of school attended, andneighborhood of residence?  What percentage of 9th graders gradu-ates within four years, five years, or sixyears of starting high school? What hasbeen the trend in these cohort gradua-tion rates over the past 5 years? What arethe trends in cohort graduation rates formales and females and for students ofdifferent racial/ethnic backgrounds?  Which student characteristics, knowableor potentially knowable by school personnel and agency staff, can identifystudents as being at high risk of droppingout of high school? 3 EXECUTIVE SUMMARY Dropout During a Single School Year: During the 2003–2004 school year,approximately 6% of the students ingrades 6–12 in the city’s public schools(including charter schools) dropped out ofschool. An additional 4% of students in grades 6–12 were technically enrolledbut were absent from school more thanhalf the time; we call these students the “near-dropouts.” In all, over 13,000students became dropouts or near-dropouts during 2003–2004.  Almost two-thirds of the students whodropped out of school in 2003–2004 werein grade 10 or lower; about one-third werein grade 9 or lower. However, there is nograde at which high school students areimmune to dropping out: over one-thirdof the students who dropped out were in11th or 12th grade. Despite being consid-erably younger than the legal school-leav-ing age, more than 500 students in grades 6–8 were officially listed as having dropped out of school.  During 2003–2004, 20% of the Latino students at the city’s publicly supportedhigh schools were either dropouts ornear-dropouts, as were 18% of AfricanAmerican students, 15% of White stu-dents, and 12% of Asian students. Maleswere more likely to be dropouts or near-dropouts than females. Despite differ-ences in severity, high school dropout inPhiladelphia is a serious problem in eachof the above racial/ethnic groups, and itis a problem for both males and females. Trends in Cohort Graduation Rates For cohorts of first-time freshmen whoform the Classes of 2000 through 2005,the four-year (“on-time”) graduationrates range from 45% to 52%. For thefour cohorts for which we have six-yeargraduation data, the percentage of students earning a high school diplomaranges from 54% to 58%. If we include allof the dropouts from the Classes of 2000 through 2005, about 30,000 students who began 9th grade in Phila delphia’s public high schools left without  In the six cohorts for which we have data,not a single racial or ethnic group had anon-time graduation rate greater than 71%.Consistent with the annual dropout ratefor 2003–2004, Asian students were mostlikely to graduate on-time, followed by  For the Classes of 2000 through 2003,only about 40% of Latino males earned a high school diploma within six years;only about half of African American andWhite males finished high school; andabout 65% of Asian males graduated.Among females, just over half of Latinofemales graduated, about 65% of AfricanAmericans and Whites graduated, and75% of Asians earned a diploma. Predictors of Dropping Out Two 8th grade factors gave students atleast a 75% probability of dropping out ofschool: 1) attending school less than 80%of the time in 8th grade (that is, missingat least 5 weeks of school), and 2) receiving a failing final grade in mathe-matics and/or English during 8th grade.Of those 8th graders who attendedschool less than 80% of the time, 78%became high school dropouts. Of those8th graders who failed mathematicsand/or English, 77% dropped out of highschool. Importantly, gender, race, age,and test scores did not have the strongpredictive power of attendance andcourse failure.  A second group of dropouts, who werenot classified as at-risk in 8th gradeaccording to our definition, were 9th gradersless than 70% of the time during 9thgrade, and/or 2) earned fewer than 2credits during 9th grade, and/or 3) werenot promoted to 10th grade on time. A ninth grader with just one of thesecharacteristics (who was not at-risk in 8th grade) had at least a 75% probabilityof dropping out of school.  About half of the dropouts in the city’spublic schools can be identified in 8thgrade, prior to their entrance to highschool. Eighty percent of the studentswho dropped out of school were eitherat-risk 8th graders or at-risk 9th graders. 4 7 Finding answers to the questions of howmany students graduate, how many dropout of school, and why they drop out iscritical to shaping a policy response. Aswe emphasize throughout this report,dropouts come in many shapes and sizes,figuratively speaking. Until we are clearabout the many pathways to dropout—for example, how many students drop out shortly before graduation, how manyleave school after having earned very few credits, how many have struggledacademically for years, and how manyhave good grades and high test scoresbut were thrown off-track by an unfore-seen life event, such as pregnancy—wewill have difficulty crafting a set of inter-ventions that meet the various needs ofout-of-school youth in Philadelphia.Data are critical for assessing the numbersof dropouts and their characteristics, andultimately for determining whether we aresucceeding in our efforts to retain studentsin school and to reconnect dropouts witheducational opportunities. This reportdraws on KIDS (Kids Integrated DataSystem), a database infrastructure housedat the University of Pennsyl enabled us to analyze an exceptional set of merged data files, including data fromthe School District of Philadelphia, theDepartment of Public Health, and theDepartment of Human Services. These data enable us to follow students overtime as they move through Philadelphia’spublic schools—or drop out of school. The data permit us to focus on studentswith particular characteristics, for example,students who drop out but eventuallyreturn to a public school, students who are served by Philadelphia’s social serviceagencies, female students who have chil-dren, students who drop out in the 9thgrade or before, or students who make it almost all the way through to the 12thgrade but leave school before obtaining a diploma. We are also able to consider asingle year in detail to examine who dropsout and from which types of schools. Inshort, the KIDS data set provides a windowonto the dropout crisis in Philadelphia withsufficient detail so that informed publicpolicy can result.This report has three chapters. The firstchapter examines a basic question—what are the high school graduation and dropout rates in Philadelphia’s publicfor which good data, includingsubgroups, have been sorely lacking. This chapter shows how these rates havechanged over time and how they vary byage, race/ethnicity, and gender, as well as by high school type and poverty level.We look in depth at a single school year(2003–2004) and also follow multiplecohorts of students as they progressthrough high school from the mid 1990sto the spring of 2005. The second chapterexplores the characteristics of the stu-dents who drop out. Specifically, it exam-ines pre-high school characteristics ofdropping out and assesses which factorsare most predictive. It looks at how stu-dents who drop out in the early grades of high school (9th and 10th grade) differfrom students who drop out in the laterhigh school grades (11th and 12th grade).Further, it shows the relationship betweendropping out of high school and socialservice involvement (for example, fostercare or juvenile justice placements), andfor females, the relationship between having a child and leaving school withouta diploma. The concluding chapter syn-thesizes the key findings and highlightsimplications for policy and practice. 8 n this chapter, we use recent data from the School District of Philadelphia to provide basicinformation on high school completion:how many students graduate and drop out, demographic characteristics of gradu-ates and dropouts, and graduation anddropout trends over time. We present several analyses, each of which provides a complementary picture about studentprogress through high school. When wecombine these pictures, we develop amuch more sophisticated image of gradua-tion and dropout in Philadelphia. Before wepresent this analysis, however, we addressis the best way to calculate graduationrates? And which students should be classified as high school dropouts? Which Is the Best Way to Determine the Graduation and Dropout Rates? The “best way” to determine rates of highschool completion and non-completioninvolves two things: 1) high-quality dataand 2) a method of calculating graduationand dropout that is appropriate to thequestion being asked. The “gold standard”for graduation and dropout calculationsuses data about individual students thatallow their progress through high school to be followed over time. These are thekind of data that are typically available toschool districts and, increasingly, to states.Beginning in the 2006–2007 school year, for example, each student attending aunique, anonymous identification numberthat will allow the state to keep more accu-rate records of graduation and dropout.The National Governors Association hasissued a call for states to upgrade theirdata collection systems so that they cantrack individual students over time.Graduation rate calculation methods gen-erally fall into one of two types: (sometimes called the “event rate”) pro-vides information on the number of stu-dents who graduate or drop out of schoolin a single year. When Pennsylvania reportsa statewide dropout rate of 1.9% for stu-dents in grades 7 through 12, as it did for the 2003–2004 school year,it is using an annual rate. In contrast, the provides information about the graduationand/or dropout rate of a single cohort ofstudents, for example, a group of studentswho all started 9th grade in a given year. Both methods have their advantages anddisadvantages. The annual method pro-vides a window on the magnitude of thedropout challenge that a district or statefaces in any given year. It can provide information, for example, on how many students might need a dropout recoveryprogram or intervention. At the same time,the annual method has some drawbacks. It only provides information on the numberof dropouts in a given year, and some ofthose dropouts may return to school thenext year. In theory, if all of the dropoutswho left a district in a given year were toreturn to school the next year and stay untilthey graduate, the district would have ahigh annual dropout rate even though100% of its students ultimately earn a high school diploma. In practice, however,one of the drawbacks of the annual highschool graduation rate is that it tends tomake things seem better than they are.Assuming that many dropouts do notreturn to school, a district’s consistentannual dropout rate of 10% means thateach year the district loses 10% of its highschool students. As a single cohort of fresh-men passes through high school, it mightlose 10% in Year 1, 10% in Year 2, and so on, until 40% of the cohort has droppedout by the end of four years. CHAPTER 1: High School Graduation and Dropout Trends in Philadelphia 10 Students who were incarcerated in ajuvenile justice facility not under the juris-diction of the public schools.Incarcerated students are perhaps themost difficult to assess. Pennsylvania’sreporting guidelines for school districtscall for students who are in detentionsecondary educationalprograms to be classified as dropouts,while those in facilities with educationalprograms are not to be coded asdropouts even though they have left However, the data weuse for this report do not provide infor-mation on whether the facility has an educational program; for example, we have no information on whether the student was being held in an adult or juvenile facility. Further, because many students who are incarceratednever return to the public schools, it isreasonable to assume that a substantialpercentage have not earned a highWe classified as “incapacitated” any stu-dent who was deceased or was withdrawnfrom school because of mental or physicalillness. In addition, for any student who wascoded in district data as having transferredto a private school or to another publicschool district, we accepted the district’sdesignation and classified them as “trans-fers” in our analyses. In order to get thebest picture of the graduation and dropoutrates among students who were a) withoutphysical or mental impediments to obtain-ing schooling and b) not enrolled in anoth-er high school diploma-granting institution,we often exclude the “incapacitated” and“transfer” students from our subsequentanalyses. Each analysis indicates which students are included. What Happened to PhiladelphiaStudents in 2003–2004?Dropout Rates Using the In this section, we examine data for theapproximately 130,000 students who were enrolled in grades 6 through 12in Philadelphia public schools, , at any point during the2003–2004 school year. From September 2003 through June 2004, 26,224 students left the rolls ofPhiladelphia’s public schools.Of thosewho left, 41% (10,653 students) were graduating seniors. An additional 27%transferred to another school or schoolTransferring to another educa-tional institution was most common in the middle grades and in 9th grade; 70% of the transfers were in grades 6through 9 when they left. Less than 15% of the students who were coded as trans-ferring were in 11th or 12th grade. Aboutone-half of one percent of the studentswere removed from the rolls for involun-tary reasons such as illness.The rest of the school leavers—more than8,000 students or about 30% of all studentsin grade 6 through 12 who left during theyear—exited the district without earning adiploma or giving any indication that theywere transferring to a private school oranother school district. Some of these students re-enrolled in the district in a subsequent school year, but most did not.There is a great deal to learn aboutdropouts simply by looking at descriptivedata from school district records. Oneinstructive type of information is the partic-ular explanation (“code”) that the schoolprovides about why the student is beingremoved from the rolls. Of the studentsclassified as dropouts, less than 5% hadwithdrawal codes indicating that the student had formally withdrawn (e.g., “voluntary withdrawal” or “Job Corps”).Instead, two-thirds of the dropouts had acode indicating that they were over thecompulsory school age and were beingdropped from the rolls because of non-attendance. Twenty-two percent of thedropouts had a code of “whereaboutsunknown,” indicating that they were lessthan 17 years old but were not attendingschool and could not be located. While it is possible that the schools tended tounder-use the “voluntary withdrawal”code, assigning instead the code indicat-ing non-attendance, it is hard to imaginewhy that would be the case. We suggestthat a more logical explanation is thatmost dropouts do not announce that they are leaving school. They simply most of Philadel phia’s dropouts have earned few creditstoward graduation.If the “ungraded” students are removed from the analysis,almost two-thirds of the students whodropped out were in grade 10 or lower;about one-third were in grade 9 or lower(Table 1). It is also worth noting that morethan 500 students in grades 6 through 8were officially listed as having droppedout of school, despite being considerablyyounger than the legal school-leaving age. 12 In sum, more than 13,000 students ingrades 6–12 became out of school youth—that is, either dropouts or near-dropouts—during the 2003–2004 school year.Table 3shows the distribution of enrollment status-es for students in 9th through 12th grades(who comprise the majority of dropoutsand near-dropouts), as well as for the highAmong the2003–2004 9th graders, for example, 81%were enrolled and attending school at leasthalf of the time; an additional 10% wereenrolled but attending less than half thetime; and 8% dropped out of school duringthe year. Of all students who were enrolledin grades 9 through 12 during the year, 16%were either dropouts or near-dropouts, with10% having dropped out and 6% beingnear-dropouts. An out-of-school populationthis large (about 11,000 students in grades9–12), produced during just one schoolyear, would fill at least seven medium-sizedhigh schools. This fact points to the scaleand seriousness of the high school dropout crisis in Philadelphia.Among 9th and 10th graders, almost 20%of the students were dropouts or near-dropouts. While the percentages ofdropouts or near-dropouts are smaller in11th and 12th grade, even in these gradesmore than 10% of the students could be so designated. Variation in Annual Dropout Rates byRace/Ethnicity and Gender Table 4 shows the percentage of studentsin four major racial/ethnic groups whobecame dropouts or near-dropouts dur-ing the 2003–2004 school year. Withineach group, the data are also presentedby gender. There are three key points to note in thistable. First, some racial/ethnic groups areat greater risk of leaving high school with-Consistent with nationaldata (Fry, 2003; Laird et al, 2006), and con-sistent with the cohort rates that we showin a subsequent section of this report,Latino students and African American students were more likely than Asian orWhite students to drop out of school.These students were also more likely to be near-dropouts. Overall, almost 20% ofthe city’s Latino youth who were enrolled in public high schools at the beginning ofthe school year fell into one of these twodropout categories, as did about 18% ofAfrican American high school students.Because these two groups represent overthree-quarters of the students in the publichigh schools, their elevated dropout ratesmean that the sheer size of the out-of-school youth population in Philadelphia isquite large. The figures for Whites andAsians were approximately 15% and 12%,respectively—somewhat lower than those of Latinos and African Americans, but stillquite high by almost any standard. It isimportant to remember that these areannual dropout rates; as we show later inthis chapter, the cohort dropout ratesthat is, the dropout rates for students whostarted high school at the same time—aremuch higher.Within each racial and ethnic group, are considerably more likely than femalesto drop out of school but only somewhatmore likely to be near-dropouts.greater tendency of males to drop out ofschool has been documented for decadesin the United States and is apparent inrecent national statistics (Rumberger, 1983;Greene and Winters, 2006). The patterndescribed above, with Latinos having thehighest probability of dropout, followed byrepeated within in each gender category.Although there are differences in highschool dropout between racial and ethnicgroups, and between males and females,the data also show clearly that high schooldropout in Philadelphia is a serious prob-lem in each of the racial and ethnic groupswe identified, and it is a problem for bothmales and females.At the end of theschool year, no racial or ethnic group couldclaim that more than 90 percent of the stu-dents who started the year were a) stillenrolled in school and b) had attendancegreater than the very low standard of 50%. Distribution of Enrollment/Attendance Status and Dropout Status, by Grade Level,2003–2004 School Year in grades 9–129th10th11th12thEnrolled, attendance at least 50%83.8%81.2%81.1%85.4%89.7%Enrolled, near-dropouts6.2%10.4%6.4%3.5%2.2%Dropped out10.0%8.4%12.5%11.1%8.2%Total100.0%100.0%100.0%100.0%100.0%n68,73122,09818,68513,66814,280 Table 3 14 School Type and Dropout Among students in grades 9 through 12,dropouts and near-dropouts are muchmore likely to be found in the city’s neigh-borhood high schools and disciplinaryschools than in special admissions (“mag-net”) schools or vocational schools. Table 5 shows the breakdown by school type forstudents who were in grades 9 through 12 (or who were classified as “ungraded”)during the 2003–2004 school year and whoattended neighborhood, vocational, specialadmissions, or disciplinary schools. Table 5, in and of itself, is evidence neitherfor censure nor compliment for any partic-ular school type. It certainly may be thecase that some schools or types of schools,because of their size, mission, or dysfunc-tional climate, are extremely good at pro-ducing high school dropouts. At the sametime, some types of schools are able toavoid dropout-prone students by screen-ing applicants carefully before offeringadmission or by “returning” students totheir neighborhood high schools whenthey under-perform; this is certainly true ofthe special admissions high schools suchas Central or Girls. Further, the neighbor-hood high schools that serve students not admitted to the special admissions or vocational schools may simply be over-whelmed by the magnitude of the aca-demic and personal challenges that thesestudents bring with them. A study of theeffectiveness of certain schools or schooltypes at promoting graduation and dis-couraging dropout would require a sophis-ticated analysis with careful controls that isbeyond the scope of this report.Of all of the school types, special admis-sions schools have the lowest percentageof students who leave high school withoutgraduating. Vocational schools also have arelatively low percentage of students whoare dropouts or near dropouts. On aver-age, neighborhood high schools—that is, large comprehensive high schools that serve primarily students from their surrounding geographic areas—had about 13% of their students drop out in2003–2004. An additional 8% of students at neighborhood high schools were near-dropouts. In total, then, about one-fifth of the students at neighborhood highschools who were enrolled at any point in the 2003–2004 school year did notattend school on a regular basis. Disciplinary schools—that is, schools thatserve students who have been involvedwith the justice system or who need spe-cial assistance to work on their behavior—had the highest annual dropout rate of any of the school types. Given the seriouschallenges that students at disciplinaryschools face, it is not surprising that thedropout rates at these schools are higherthan at the other school types. But becausethe mission of these schools is to educateeffectively some of the school district’smost challenging students, the extremelyhigh dropout rates in the disciplinaryschools is also cause for concern.Thirty-six percent of the students at disciplinaryschools became dropouts during the year,and an additional 9% were near-dropouts.As we show in Chapter 2, the high annualdropout rate for these students—70% of whom are male, and 90% of whom areminority—contributes to a cohort dropoutrate in disciplinary schools that is veryclose to 100%. Annual Dropout Rates, by School Type, for Students in Grades 9–12 (and Ungraded),2003–2004 School Year High School Type (“Magnet”)VocationalNeighborhoodDisciplinary% who became dropouts0.7%2.9%12.8%36.1%% who were near-dropouts0.4%2.0%8.3%9.1%Total % dropouts or near-dropouts1.1%4.8%21.1%45.2%Total n (all students)6,5734,83643,8104,143 Table 5 16 Figure 1Percent of Students in Grades 9–12 (and Ungraded) Who Were Dropouts or Near Dropouts by Neighborhood, 2003–2004 EastwickMarconi PlazaRichmondWynnefieldElmwoodTaconyTiogaFrankfordOverbrookPennsportKingsessingGirard EstatesFishtownHunting ParkFairhillEast FallsBelmontBridesburgGrays FerryCobbsCreekByberrySomertonRoxboroughBustletonHolmesburgWest TorresdaleLawncrestPennypackOlneyChestnut HillMayfairRhawnhurstFox ChaseOxford CircleLoganPoplarWissahickonParkPennypack ParkJuniataParkOgontzW. Mount AiryMill CreekWest Oak LaneGermantownStrawberryMansionAlleghenyWestEastGermantownUniversityCityWhartonFairmountRiverfrontE. Mount AiryKensingtonNorthCentralEast Oak LaneHaddingtonHartranftManayunkCenterCity EastCedarbrookPointBreezeSchuylkillCenterCity WestPoweltonCedarParkHarrowgateWestKensingtonBrewerytownSouthPhila. Legend N/A Miles 18 schools, perhaps as a result of the increasein the number of charter high schoolThe highest on-time graduationrate is for the Class of 2005. At 46%, it isabout 3 percentage points greater than the average for the preceding five years. Figure 3 shows the on-time, five-year, andsix-year graduation rates for students did not transfer to other districts or to pri-vate schools or who were not removedfrom the system due to death or serious. For the four cohorts for which wehave six-year graduation data, the percent-age of students earning a high schooldiploma ranges from 54% to 58%. The percentage of students who had earneddiplomas by the six-year mark is higherthan the four-year percentage by about 8percentage points to 10 percentage points.Generally, the increase from the four-yearto the five-year graduation rate is greaterthan that from the five-year to the six-yearrate. After six years in high school, whilesome students continue to earn diplomas,the probability of graduating is very low.The Class of 2005—the most recent cohortfor which we have data—is the only first-time freshman cohort in our analysis in whichat least 50% of the students graduated infour years. Their on-time graduation rate(52.4%) is about four percentage pointshigher than the average for the precedingfour cohorts. As Figure 4 shows, this gain ingraduation rates occurred across magnet,vocational, and neighborhood high schools.Vocational high schools saw the greatestupswing, and magnet high schools crossedthe 90% threshold for the first time in theyears for which we have data. Notably, evenwith a small gain, the on-time graduationrate in Philadelphia’s neighborhood highschools remained below 50%. Among recent cohorts, the Class of 2005 istied for the highest percentage of studentslisted as transferring out of the district toother schools, and it also has the feweststudents still enrolled in school at the endof four years. Thus, it is possible that thegains in the four-year graduation rate couldreflect a higher percentage of studentsgraduating on time, rather than improve-ments in the total number of students whowill graduate within six years; it will beimportant to see whether the gains in theon-time graduation rate carry over to itsfive- and six-year graduation rates. Further,graduation rates generally fluctuate some-what from year to year, so it will be impor-tant to see whether the Class of 2006 isable to continue the higher graduation rate of the previous cohort.It is important to celebrate small success-es, such as the higher on-time graduationrate for the Class of 2005—after all, it isvery difficult to change large institutionslike public school districts and to createthe social conditions in families and com-munities that encourage students to stayin school. At the same time, it is also nec-essary to face just how far Philadelphia hasto go. A six-year graduation rate of 58%(as was the case for the Class of 2003)means that 42% of the class had not yetgraduated and probably would not gradu-ate from Philadelphia’s public schools. Put differently, about 5,000 Philadelphiastudents from the Class of 2003 aloneembarked on adulthood without the mini-mal academic credential of a high schoolIf we include all of the studentsfrom the Classes of 2000 through 2005,about 30,000 students who began 9thgrade in Philadelphia’s public high schoolsdropped out without earning a diploma. Percentage of Students Graduating in Four, Five, or Six Years for Six Cohortsof First-Time Ninth Graders Class of 2000Class of 2001Class of 2002Class of 2003Class of 2004Class of 2005 Figure 3* *Graduation rates for the Class of 2004 are estimated. See Footnote 18. 20 In all cases, the graduation rates for thefreshmen and transfer cohorts are three tofour percentage points than thosefor the first-time freshman cohorts. A fullanalysis of what drives the differences inthese rates is beyond the scope of thisreport. However, some part of the explana-tion may lie in the fact that the freshmenand transfer cohorts are numerically domi-nated by students who transfer into thedistrict’s neighborhood high schools. Manyof these students who transfer into the district are 9th graders, but substantialnumbers of the transfers are in the uppergrades. We do not have access to data onthe academic histories of these studentsprior to their entry into the Philadelphiapublic schools, however, we can observethat they have lower graduation rates thanstudents at neighborhood high schoolswho did transfer into the district after8th grade. It is possible that many of the“transfer-in” students experienced aca-demic difficulty at their prior high schoolsand transferred to Philadelphia publicschools for a second chance. Most importantly, what we learn from acomparison of these different ways ofdefining cohorts is that the graduationrates are not radically different.are within a few percentage points ofeach other. Therefore, in the followingsections of this chapter, we will continueto base our analyses on the freshman cohortsfor consistency. Further,our analysis of the predictors of droppingout, presented in the next chapter,requires that we have data on studentsprior to entering high school. Table 7* Cohort Graduation Rates Calculated in Two Ways Freshmen and Transfer CohortsFirst-Time Freshman CohortsOn-time5-year6-yearTotal nOn-time5-year6-yeargraduationgraduationgraduationforgraduationgraduationgraduation Class ofraterateratecohortrateraterate200150.7%58.7%61.0%20,70647.9%55.8%57.8%200249.3%59.1%61.4%20,98644.2%53.6%55.9%200353.8%62.1%63.0%21,02948.3%55.7%57.9%200448.0%56.7%n/a22,38242.9%50.9%n/a 200554.0%n/an/a22,06852.3%n/an/a*Graduation rates for the Class of 2004 are estimated. See Footnote 18. 22 Race and ethnicity variation In the six first-time freshmanwhich we have data, not a single racial orethnic group had an on-time graduationrate greater than 71%. Consistent with the annual dropout rate for 2003–2004,Asian students were most likely to graduateon-time, followed by Whites, AfricanAmericans, and Latinos. Figure 6 breaksdown the on-time graduation rates by race or ethnicity, for six cohorts of first-timefreshmen. The figure also includes the linear trend for each group.A shorthand description of the on-timegraduation rates prior to the Class of 2005 isthat Asian graduation rates tended to be in the 60-percent range; White rateswere in the 50-percent range; AfricanAmerican rates were in the 40-percentrange; and Latino rates were in the high 30-percent and low 40-percent range. linear trend for the six cohorts is modestlyupward for Asian, African American, andLatino students, driven primarily by thesharp increase in on-time graduation ratesfor the Class of 2005. African Americans in the Class of 2005, in particular, experi-enced a substantial increase in on-timegraduation rates, breaking the 50% mark forthe first time in the six cohorts for which wehave data. The trend for Latino students hasbeen very modestly upward. While gradua-tion rates for Whites show some year-to-year variation, the on-time graduation trend for this group is approximately flat.Driven by the increase in African Americanand Latino graduation rates, the on-timegraduation gap between White studentsand these two groups was narrower for themore recent cohorts than for the earliercohorts. In fact, on-time graduation ratesfor Whites and African Americans in theClass of 2005 were within two percentagepoints of each other. However, the sharpincrease in the percentage of Asian students graduating on-time in the Class ofand other racial or ethnic groups was widerat the end of the time period under consid-eration than at the beginning. As Figure 6shows, on-time graduation rates bouncearound from year to year, and it may be the case that the gap has narrowed againfor the Class of 2006. On-Time Graduation Rates for Males in Six Cohorts: Percentages by Race/Ethnicity Class of 2001Class of 2002Class of 2003Class of 2004Class of 2005 African AmericanLatino Figure 6* *Graduation rates for the Class of 2004 are estimated. See Footnote 18. 24 The intersection of race/ethnicity In the six first-time freshmanwhich we have on-time graduation data,the linear trend in on-time graduation hasbeen modestly upward for males in each of the four racial/ethnic groups we examineeven though the on-time graduation ratefor Latino males declined somewhat from2004 to 2005. Figure 7 shows that on-timegraduation rates for Asian males movedfrom 56% for the Class of 2000 to 62% forthe Class of 2005; for African Americanmales, the comparable figures are 38% and 47%; and for Latino males, 31% and37%. The trend for White males, althoughslightly positive, is flatter than for othergroups, with 51% graduating on-time in2000 and 52% in 2005.For females, the graduation trends are flatter than those for males, with the excep-tion of Asian females, who widened theiralready considerable advantage during the period from 2000 to 2005 (Figure 8). On-Time Graduation Rates for Males in Six Cohorts: Percentages by Race/Ethnicity Class of 2000Class of 2001Class of 2002Class of 2003Class of 2004Class of 2005 Asian African AmericanLatino Figure 7* *Graduation rates for the Class of 2004 are estimated. See Footnote 18. On-Time Graduation Rates for Females in Six Cohorts: Percentages by Race/Ethnicity Class of 2000Class of 2001Class of 2002Class of 2003Class of 2004Class of 2005 Asian African AmericanLatina Figure 8* *Graduation rates for the Class of 2004 are estimated. See Footnote 18. 26 ver the past several decades,scholars have developed a largebody of literature identifying predictors of dropping out of high school.Many of the predictors have been demon-strated so often, in so many differentstudies, that they are widely considered to be settled findings. For example, it is well-known that, in the United States,males, lower-income students, membersof racial or ethnic minority groups, thosewith lower academic achievement, andstudents who are older than the typicalstudent in their grade (usually the result of being held back in elementary school)are more likely than students withoutthese characteristics to leave high schoolwithout graduating (Rumberger, 2004).There is also a general consensus amongscholars that dropping out of school is theculmination of a process of disengagingfrom the academic or social aspects ofschool, or both; students who ultimatelydrop out tend to give “warning signals,”such as attending school less frequently or letting their grades slip (Finn, 1989;Newmann et al., 1992; Wehlage et al.,1989). Finally, there is growing evidencethat many dropouts can be identified priorto entering high school (Alexander et al.,1997) and that the rocky transition to 9thgrade often aggravates academic prob-lems that students have been accumulat-ing over their school years (Roderick andCamburn, 1999).In this report, we take a slightly differentapproach, examining factors that are know-able about students by school personnel orby staff at social service agencies. Thesefactors could be used to identify studentswho are at greatest risk of dropping outbased on what happened to similar stu-dents in earlier cohorts. The variables thatwe examine are gleaned from studentrecords kept by the public schools and thesocial service agencies with which some ofthe students are involved. The Hazard of Dropping Out WithinEight Years of Starting High School To understand who does not graduate, weemploy an analytical method known as ard analysis. The idea behind this type ofanalysis is simple: given that a student hasreached a certain point in his or her educa-tion, we ask what the probability (“hazard”)is of not graduating. Specifically, we ask What is the probability of not graduating,  A student is a first-time freshman?  A student has reached 10th grade?  A student has reached 11th grade?  A student has reached 12th grade?In this case, it is important to note thatwhen we say “10th grade,” we do notmean simply “the second year in which a student is enrolled in high school.”Rather, we refer to those students whohave earned enough credits to be classified by their school as 10th graders. This type of analysis is important for a bet-ter understanding of who drops out andwhy they drop out because although themajority of students who drop out of highschool do so when they are still in 9th or10th grade, there is a substantial subgroupof students who leave school in 11th or12th grade, when graduation would seemto be around the corner. Our motivation in examining students who drop out at dif-ferent points in their high school careers is to see whether there are distinct differ-ences between “early-grade leavers” and“late-grade leavers.” From a policy andintervention point-of-view, such informationis critical to tailoring programs that willwho leaves school at age 17 with almost no credits toward graduation will need adifferent sort of program that will enablehim or her to earn a high school diplomafrom a student who leaves at the same age lacking just a few credits.For simplicity’s sake, we use only the time freshman Class of 2000in this analysis.Further, we use an eight-year dropout raterather than a six-year rate; by the end ofeight years after starting high school,almost every student has exited the districtin one way or another. We have no reasonto believe that the general patterns andrelationships observed in the Class of 2000data would be appreciably different for anyof the other cohorts for which we have data(even though the overall levels of graduat-ing or dropping out may be different, as CHAPTER 2: Who Does Not Graduate from High School? The First-Time Freshman Class of 2000 28 of Dropping Out Forty-five percent of the first-time 9thgraders dropped out of school but, ofcourse, 55% did not. Which factors height-en the probability that a first-time 9th grad-er ultimately will drop out of high school? In this analysis, we examine the predictivepower of factors that are known—or couldbe known—by school and agency person-nel, as well as by parents. We begin byexamining information that is knowableabout students when they are in 8th grade:school attendance, report card grades, testscores, age, and demographic factors suchas gender and race/ethnicity. We continueby assessing which factors among 9thgraders and upperclassmen are most pre-dictive of dropout. We do not argue thatthese particular variables are the “rootcauses” of dropout, which involve complexissues related to the student’s academichistory, family and school environment,community factors, and individual personal-ity. For example, “poor grades” is a factorthat may be directly linked to dropping out of school; researchers sometimes callthese kinds of factors proximal variablesBut the “root cause,” which produces the low grades, may be more example, an undiagnosed learning disabili-ty or a family situation that makes it difficultfor a child to concentrate on schoolwork.Proximal factors are relevant to our analysis,however, because they serve as signals thata child has a heightened probability ofleaving high school without a diploma. Which 8th grade factors are strong predictors of dropout? For this analysis, we examine 8th gradedata for our entire first-time freshmanthat made up the Class of 2000; byPhiladelphia public schools during the1995–1996 school year. Eighth graders inthis cohort who had lower attendance,weaker test scores, who failed core aca-demic courses, were overage for theirgrade, and/or who were male were morelikely to drop out of school. Each of thesefactors exerted a statistically independenteffect on the odds of dropping out. How ever, while each of these factors contri to dropping out, we were most interestedin factors that were strongly predictive ofdropping out. We identified two factors from 8th gradethat gave students at least a 75% proba-bility of dropping out of school: 1) attend-ing school less than 80% of the time in8th grade (that is, missing at least 5 weeksof school), and 2) receiving a failing finalgrade in mathematics and/or English dur-ing 8th grade. Of those 8th graders whoattended school less than 80% of thetime, 78% became high school dropouts.Of those 8th graders who failed mathe-matics and/or English, 77% dropped outof high school. Importantly, gender, raceor ethnicity, age, and test scores did not have the strong predictive power of attendance and course failure. Clearly, there are numerous factors thatcontribute to the risk of dropping out. But in this analysis, we define “at-risk 8thas those who attended less than80% of the time and/or who failed mathe-matics and/or English in 8th grade. four percent of the dropouts in the Classof 2000 were at-risk 8th graders accordingto this definition, even though they madeup only 34% of the entire cohort. Thesedata indicate that about half of thedropouts in the city’s public schools canbe identified in 8th grade, prior to theirentrance to high school. In fact, a separate analysis of a cohort ofmiddle grades students in the Philadelphiapublic schools during the 1996–1997 schoolyear shows that many of the students whobecame dropouts could be identified asearly as 6th grade using similar data on Table 11 Percentage of Students At-Risk in 8th Grade and Percentage of At-Risk 8th GradersWho Dropped Out, by Key Categories, Class of 2000 % of students% of at-riskin this category 8th graders inwho were at-riskthis category who 8th gradersdropped outAfrican American35.1%74.1%Asian15.8%83.8%Latino40.8%77.3%White27.6%78.4% Females31.4%70.3%Males35.6%79.8% 13 and under27.7%70.4%14 and over53.0%84% 30 of Latinos, 50% of African Americans, 40%of Whites, and 30% of Asians were at-risk.More than half of the males, and close tohalf of the females, were at-risk. And 70%of those who were at least 15 years or olderat the start of the school year had one ormore of the risk factors we have identified. Eighty percent of the students whodropped out of school were either at-risk8th graders or at-risk 9th graders, in theway we have defined them. Given this statistic, it is not surprising that when webreak down the data by school (that is, bythe high school attended during the 9thgrade) we find that the percentage of thecohort who dropped out closely tracks thepercentage of students in the cohort whowere either at-risk in 8th grade or 9thgrade. Figure 10 presents in a graph theclose correspondence between these twofigures. Each school is represented by avertical line. In most cases, the percentageof students at-risk in 8th or 9th grade inthe Class of 2000 was greater than thepercentage of students dropping out, areminder that some of the at-risk studentscompleted high school despite the odds.The data in Figure10, in and of themselves,are not evidence that particular highschools are especially good at producingdropouts—or that they are powerless toreduce the number of dropouts because of the academic weaknesses that enteringstudents bring with them. How studentsperform in high school, and ultimatelywhether they drop out, is the result of theinteraction of factors operating before highschool and students’ experiences duringhigh school. aresaying is that,given our description of the risk factors in recent cohorts, the vast majority ofpotential dropouts can be identified—and perhaps targeted for intervention—at the start of high school. In fact, many at-risk 9th graders who arerunning into trouble can be identified asearly as the end of the first marking periodof the freshman year. For example, of theat-risk 9th graders who had an overallattendance rate for the year of less than70%, almost half (49%) attended school lessthan 70% of the time during the first mark-ing period. And 80% had attendance ratesof less than 70% during the first and/or sec-with similar recent findings from Chicago(Allensworth and Easton, 2005), suggestthat high schools need not wait until theend of the year to identify a large percent-age of their dropouts—data from 8th gradeplus the information from the first and/or second marking period will suffice. By High School: Percentage of Dropouts and Percentage At-Risk in 8th or 9th Grade, Class of 2000 100%80%0% % at-risk, 8th or 9th% dropping out Schools Figure 10 Table 13 Percentage of Students At-Risk in 9th Grade and Combined Percentage of At-Risk in 8th or 9th Grade, by Key Categories % of students% of studentsin this categoryin this categorywho were at-riskwho were at-risk9th graders8th graders at-risk 9th gradersAfrican American16.6%51.7%Asian14.5%30.3%Latino21.3%62.1%White14.6%42.2% Females15.3%46.7%Males18.1%53.8% 14 and under at beginning of 9th grade16.5%44.2%15 and over at beginning of 9th grade17.4%70.4% 32 In each year, the number of credits earnedis a good predictor of whether the studentwill drop out of school. Among on-time10th and 11th graders, students whoearned fewer than 5 credits during the yearhad a higher probability of dropping outthan graduating within eight years of start-ing high school. Among 12th graders, it isonly those who earn fewer than 3 creditsduring the year who have more than a 50%probability of dropping out; seniors maynot need to earn 5 credits in order to grad-uate and, if they do, they may be more like-ly to return for another year to acquire theclasses they need to graduate.We find two non-academic factors thatalso give students more than a 50% prob-ability of dropping out of high school. Ofthe 104 young women in the Class of 2000who reached 10th grade on time but whogive birth to a baby during the school year(or following summer), 55% left highschool without graduating. However, the141 on-time 11th graders and the 135 12thgraders who gave birth during the year orwho had ever had a baby by that pointhad relatively high probabilities of gradu-ating (70% or above). Finally, we also findthat students who had a scrape with thewere assigned by the courts to anout-of-home juvenile facility dropped outof high school at very high rates. In thiscase, of the 411 on-time 10th graders whohad one of these placements during 10thgrade (or during the previous summer),93% did not earn a high school diplomafrom the School District of Philadelphia. Non-Academic Predictors of Dropout The vast majority of the Class of 2000who dropped out of school struggledacademically and/or attended infrequent-ly, sometimes prior to entering highschool, sometimes after entering highschool, and sometimes both. Some ofthese students also had contact with the city’s social service agencies and/orgave birth to at least one child. Table 15shows the percentage of students whohad a substantiated case of abuse orafter starting high school, a foster care placement,in a juvenile justice facility,or who gavebirth in Philadelphia within four years ofstarting high school. The percentages for dropouts are presented, along withpercentages for graduates and dropouts,for sake of comparison.Students who dropped out of school weremore likely to have given birth to a childand/or to have had contact with social serv-ice agencies. Even among the dropouts,however, relatively few—less than 3%—hada substantiated case of abuse or neglectduring their high school years. Less than10% of the dropouts had a foster careplacement. However, close to one-quarterof the males who dropped out had beenplaced in a juvenile justice facility for someperiod of time after starting high school.One-third of the young women whodropped out of school had a baby withinfour years of starting high school, and 40%had a child within five years. Table 15 Social Service Agency Contact After Starting High School, for Students Who Dropped Out of School Percent of allPercent of all Percent of all dropoutsgraduatesstudents*Substantiated case of abuse or neglect2.8%.89%1.8%Foster care placement7.4%2.0%4.5%Juvenile justice placement (all students)14.4%1.3%7.2%Juvenile justice placement (males only)22.6%2.2%12.8%Gave birth within 4 years of starting high school (females)32.8%9.7%18.7%Gave birth within 5 years of starting high school (females)41.4%15.2%25.5% Number of students (male and female)6,0537,29613,393*including those still enrolled in school Other cities—Portland, Boston, and NewYork City among them—currently have abroad array of these types of options forstudents wishing to return to school. Inaddition, the School District of Philadelphiahas opened a number of small schools foryouth over age 17 who have few highschool credits. These schools are a strate-gic part of the overall high school reformagenda. Slots in these schools are beingexpanded annually. Many of the dropouts with few credits haveneeded to succeed in high school. Themost recent standardized test data that wehave for these students is from their 8thgrade year,and students may have experi-enced some academic advances duringtheir time in high school, making thesedata an underestimate of their academicskill when they dropped out. Nevertheless,it is instructive to see that many studentswho dropped out as 9th or 10th gradershad a grade equivalent of 5th grade orbelow on the SAT-9 reading and/or mathe-matics tests when they were in 8th gradeand that the vast majority scored belowgrade level (Table 17). In order to enablethese students to produce high school-level work to earn a diploma, high schoolcompletion programs will need to help alarge proportion of dropouts develop thereading comprehension skills and middlegrade mathematical knowledge assumedby high school level work. It is important to remember, however, that a substantial subgroup of studentsdropped out when they were not far fromhigh school graduation. These studentsalso scored higher on the standardizedmath and reading tests in 8th grade thanthe 9th and 10th grade dropouts did, andthey are more likely to be candidates forpost-secondary education. These students,too, need a program tailored to theirneeds. An example of a program thatworks with such students is the Gateway to College Program, which allows studentswho 1) are within 10 credits of high schoolgraduation and 2) score at the 8th gradelevel or above on an adult education testto earn high school and community collegecredits simultaneously. This model is basedon one at Portland Community Collegeand has been replicated in several cities,including Philadelphia. Dual enrollmentlegislation at the state level in Pennsylvaniawas also specifically designed to include this population of youth. 34 Table 17 Grade Equivalents on 8th Grade SAT-9 Reading and Math Tests, for Three Groups of Students, Class of 2000 Dropped out in 9th gradeDropped out in 10th gradeAll 8th graders% at this % at this % at this % at this % at this % at this Reading level, Math level,Reading level,Math level, Reading level, Math level, 8th grade8th grade8th grade8th grade8th grade8th grade5th grade or below57.9%48.5%49.5%43.1%34.4%30.6%6th–7th grade20.4%35.1%22.7%36.6%22.4%33.9% 8th grade or above21.7%16.4%27.8%20.3%43.2%35.6%n1,3211,0561,1991,07811,98711,164 Highest Grade of Students WhoDropped Out, Class of 2000 12th Grade23% 11th Grade 10th Grade 9th Grade Figure 11 36 School type and dropout During the years examined in this report,nearly all of Philadelphia’s students grad-uated or dropped out from three types of schools. Philadelphia’s academicallyselective public high schools educateabout 10% of the high school students in the non-charter public schools. Collect ively, from 2000 to 2005, these schoolshad six-year cohort graduation rates near80%. The school district’s vocational highschools, which educate another 8% of the enrolled high school students, hadcollective six-year graduation rates in the 60-percent range. During this timeperiod, nearly three-quarters of the dis-trict’s high school students attendedneighborhood high schools, where thecollective on-time graduation rate from2000 to 2005 was in the upper 40-percentrange and the six-year graduation rate inthe 50-percent range. This average rateobscures considerable variation amongthe neighborhood high schools. Dropout and graduation rates are highlycorrelated with a high school’s povertylevel. Twenty-nine thousand students inPhiladelphia attend 24 high schools inwhich 75% or more of the students are eligible for free or reduced price lunch.These high-poverty high schools have an annual total dropout rate (formaldropouts and near dropouts combined) of 25%. Dropout rates by gender and Depending on the year and the specificcomparison being made, there is a 10 to 15percentage point graduation gap betweenmales and females in Philadelphia. This gap holds true across all racial and ethnicgroupings, and while the female advantagereflects national and historical trends, thegender gap in Philadelphia is considerablygreater than the national average. The gen-der gap for the four-year, on-time gradua-tion rate appears to have slightly narrowedfrom 2000 to 2005, with males obtainingtheir highest graduation rate in 2005. Buteven with this gain, males in Philadelphiahad a 47% on-time graduation rate com-pared to 58% for females. Asian students have the highest graduationrates and lowest dropout rates in the schooldistrict. Whites and African Americans havesimilar rates. Latinos, who according to therecent U.S. census data are the fastestgrowing population in the Philadelphiaregion, have the lowest graduation ratesand high dropout rates. Graduation rates,however, are much too low across all racialand ethnic groups. For the cohorts forwhich we have six-year graduation data,more than half of the Latino students, 40%of White and African American students,and about 30% of Asians did not earn ahigh school diploma in six years. Early identification of dropouts Over half the eventual dropouts fromPhiladelphia’s public schools can be identi-fied prior to the start of high school. Themajority of the students who become dropouts failed their English or mathematicscourses or attended school less than 80%of the time when they were in the middlegrades. Another 15% do not show this levelof academic difficulty or disengagementfrom school during the middle grades buthave a rocky transition to high school andearn poor grades and/or attend schoolinfrequently in the 9th grade. This meansthat by the first year of high school, 80% ofthe students who eventually drop out havesignaled clearly that they have fallen off thepath to graduation. Students who attendthe 9th grade less than 70% of the time orearn fewer than two credits, for example,have dropout rates of over 75%. Sixty per-cent of Latino and half of African Americanhigh school students signal either at thestart of high school or by the end of theirfirst year in high school that they are on the way to dropping out. About 20% of eventual dropouts cannot be readily identified by the first year in highschool. These are the students who make it to 10th, 11th, or 12th, often on time,before they dropout. Once a student hasadvanced to the upper grades of highschool, it becomes more difficult to identifywho ultimately will drop out or graduate.However, one constant remains: studentswho do not earn sufficient credits in a givengrade to be promoted to the next grade ontime are at increased risk of dropping out. 40 The reason a school district gives for why astudent leaves can make all the differencein the dropout rate. Should a district inten-tionally try to be deceptive, as was allegedabout some districts in Texas (Dobbs,2003), there is plenty of opportunity fordents have not formally withdrawn from thedistrict. Even without the intent to deceive,dropout codes can be assigned in differentways in different schools, using differentlevels of evidence for a student’s where-abouts (Hammack, 1986). For example, districts contend with the question ofwhether the statement from a student’sfriends that he or she moved from the district is sufficient evidence to have himcount as a transfer (National Forum onEducation Statistics, 2006). A recent reporton dropout in the Pittsburgh Public Schoolsystem by RAND tries to correct for theslipperiness of district data on dropout byassessing the probability that a studentwho is listed as transferring is actually adropout (Engberg and Gill, 2006). Ourunderstanding is that, like many districts in the United States, the guidance given to Philadelphia schools about how toassign codes is relatively weak.We cannot independently verify whetherthe codes that have been assigned to stu-dents are the best descriptors of why theyleft the district. We do note that, in the2003–2004 cohort, students in grades 9–12who were assigned a code indicatingtransfer to another school tended to beyounger (that is, not older than 15), andalmost half were listed as 9th graders.Further, those who transferred were dis-proportionately White: 27% of the trans-fers were White, while 16% of all highschool students were White. BecauseWhite students in the district are less likelyto be low-income, and thus more likely toattend private schools, this statistic maybe an indication of some level of veracityin the data on student transfers.In the end, there are likely two kinds offorces at work in the coding: students whoactually dropped out of school but weremistakenly coded as transferring to anotherschool (which would bias the graduationrates upward), and students who trans-ferred to another school but were given acode indicating dropout or whose informa-tion was never entered into the computerand so have been counted as dropouts inthis analysis (which would bias the gradua-tion rates downward). We suggest thatthese two forces cancel each other to somedegree, although the extent to which theycancel each other cannot be determineddefinitively from these data. 42 These estimates, however, face two poten-tially confounding data problems. Thenumber of students enrolled in a particulargrade in a given year includes those whoare in that grade for the first time andthose who are repeating the grade (forexample, the 9th grade numbers wouldinclude first-time 9th graders and 9th graderepeaters). Further, the number of studentsearning a diploma in a given year includesstudents who are on-time graduates andthose who took an extra high school yearor two or even more to obtain their diplo-ma. In addition, students transfer in andout of high school between the initial yearthe enrollments were calculated and theyear of the diploma count. If the number of 9th grade repeaters does not “balance”the number of extra-time graduates, or the number of “transfer ins” does not “bal-ance” the number of “transfer outs,” thengraduation rate estimates comparing thenumber of students enrolled in one year to the number of graduates in another willeither over- or under-estimate the actualgraduation rate. Because these variablescan fluctuate from year to year, graduationrate estimates calculated using enrollmentand diploma data can be fairly accurateone year, an under-estimate the next year,and an over-estimate the following year. The researchers who make graduation rateestimates are aware of these data issuesand make different choices on how to cor-rect or control for them. As seen in the fol-lowing charts and tables, the end result isthat some graduation rate estimates endup being better estimates of Philadelphia’ssix-year or total cohort graduation rate andothers better estimates of Philadelphia’sfour-year or on-time graduation rate. How Do Graduation Rate EstimatesCompare to Our Longitudinal CohortGraduation Rates? Figure A compares graduation rates calculated in five ways:1) The six-year graduation rate thatwe calculated in Chapter 1 of thisreport for the first-time freshmenThe six-year graduation rate thatwe calculated for the freshmen andtransfer cohorts;3) The Average Freshman Graduationrate, used by the U.S. Departmentof Education National Center onEducational Statistics; 4) The Greene Method; and 5) The method that compares 8thgrade enrollment to diplomas. Average Freshman Graduation Ratemethod, used by the U.S. Department ofEducation, divides the number of diplo-mas issued in a Year Y by the average ofthe number of 8th, 9th, and 10th gradersenrolled in Year Y-4, Year Y-3, and Year Y-2,respectively. The Greene Method,used inreports issued by the Manhattan Institute,a population change correction. Thegrade enrollment to diplomas”divides the number of diplomas issued in Year Y by the number of 8th graders in Year Y-5. Comparing Graduation Rate Estimates to Philadelphia’s Six-Year Graduation Rate Class of 2001Class of 2002Class of 2003 Philadelphia 6 Year Longitudinal Cohort (Without Transfers In) NCES Average Freshman Graduation Rate 8th Grade Enrollment to Diplomas (5 Years Later) Greene Method (Avera e Freshmen Graduation Rate Plus Population Chan e Ad ustment) Philadelphia 6 Year Longitudinal Cohort (IncludingTransfers In) Figure A 44 Both of these methods produced esti-mates that are typically within 5 percent-age points of our calculated four-year longitudinal graduation rate. The CPI and the 9th grade-to-diploma measure,however, consistently produce substantialunder-estimates of Philadel phia’s cohort graduation rate. In this case itshould be noted that the CPI was designedto estimate four-year graduation rates. The method of calculating the graduationrate that appears to be the most inaccurateis the graduation rate currently used by theCommonwealth of Pennsylvania under NoChild Left Behind accountability.Figure C, the Pennsylvania NCLB rate sub-stantially overestimates the four-year gradu-ation rate for students in both our first-timefreshman cohorts and our freshmen andtransfer cohorts. This overestimation occursbecause the Commonwealth’s methodcompares the total number of graduates ina given year to the number of 12th gradedropouts from that year, the number of11th grade dropouts from the year before,and so on. But as we have shown, the totalnumber of graduates in a given yearincludes significant numbers of studentswho take five or six years to graduate. As a result, the method really compares thetotal number of graduates in a cohort tojust four years of dropouts, rather than six.This is significant because the stated inten-tion of the No Child Left Behind legislationis to measure the percent of students grad-uating with a regular diploma in the stan-dard number of years—that is, within fouryears of starting high school. Comparing State NCLB Graduation Rates to Philadelphia’s 4-Year (On-Time)Graduation Rate 4-Year Graduation Rate Philadelphia 4-Year Longitudinal Cohort (IncludingTransfers In)Philadelphia 4-Year Longitudinal Cohort (Without Transfers In) Figure C 46 Alexander, Karl, Doris Entwisle, and Carrie Horsey. (1997). "From first grade forward: Earlyfoundations of high school dropout." Sociology of educationAllensworth, Elaine. (2005). Graduation and dropout trends in Chicago: A look at cohortsof students from 1991 through 2004.Chicago: Consortium on Chicago School Research.Allensworth, Elaine, and John Easton. (2005). The on-track indicator as a predictor ofhigh school graduation.Chicago: Consortium on Chicago School Research.students on track to graduation.http://www.philaedfund.org/powerpoint/dropoutresearch_4.06.pptBoesel, David, Nabeel Absalam, and Thomas M. Smith. (1998). Education and labor market outcomes of GED certification. Washington, DC: National Library of Education,Office of Educational Research and Improvement, U.S. Department of Education.Brookings Institution Center on Urban and Metropolitan Policy. (2003). focus: A profile from the Census 2000.Washington, DC: Brookings.Cameron, Steven V., and James J. Heckman. (1993). “The nonequivalence of high schoolJournal of labor economicsDobbs, Michael. (November 8, 2003). “Education ‘miracle’ has a math problem.” The Washington Post.Page A01.Engberg, John, and Brian Gill. (2006). Estimating graduation and dropout rates with longitudinal data: A case study in the Pittsburgh Public Schools.RAND Working Finn, Jeremy D. (1989). "Withdrawing from school." Review of educational researchFry, Richard. (2003). Hispanic youth dropping out of U.S. schools: Measuring the chal-Washington, DC: The Pew Hispanic Center. REFERENCES 47 Greene, Jay, and Marcus Winters. (2006). Leaving boys behind: Public high schoolgraduation rates.New York: Center for Civic Innovation, The Manhattan Institute. Hammack, Floyd. (1986). "Large school systems' dropout reports: An analysis of definitions, procedures, and findings." Teachers College RecordJunn, Jane. (2005). “The political costs of unequal education.” Paper presented at the Symposium on Social Costs of Inadequate Education, Teachers’ College,Columbia University.Laird, Jennifer, Stephen Lew, Matthew DeBell, and Chris Chapman. (2006). Dropoutrates in the United States: 2002–2003. U.S. Department of Education NCES ReportLegters, Nettie, Robert Balfanz, Will Jordan, and James McPartland. (2002).Comprehensive reform for urban high schools: A talent development approach.New York: Teachers College Press.Moretti, Enrico. (2005). “Does education reduce participation in criminal activities?”Paper presented at the Symposium on Social Costs of Inadequate Education,Teachers’ College, Columbia University.Murnane, Richard J., John B. Willett, and John H. Tyler. (2000). “Who benefits fromobtaining a GED? Evidence from high school and beyond.” Review of economics National Forum on Education Statistics. (2006). A taxonomy for standard student exit codes (NFES 2006-804).of Education. Washington, DC: National Center for Education Statistics.National Governors Association. (2005). Graduation counts: A report of the NationalGovernors Association Task Force on State High School Graduation Data.Washington, DC: National Governors Association. Newmann, Fred M., Gary G. Wehlage, and Susie D. Lamborn. (1992). "The signifi-cance and sources of student engagement." Pp. 11–39 in Student engagement andachievement in American secondary schools, edited by Fred Newmann. New York:Teachers College Press.Pennsylvania State Department of Education. (2005). Electronic dropout/graduatereport (EDGR): Dropout instructions for school year 2004–05.Pennsylvania State Department of Education. (2005). Public secondary schooldropouts in Pennsylvania: 2003–04Roderick, Melissa, and Eric Camburn. (1999). "Risk and recovery from course failure inthe early years of high school." American Educational Research JournalRouse, Cecelia Elena. (2005). “The labor market consequences of an inadequate education.” Paper presented at the Symposium on Social Costs of InadequateEducation, Teachers’ College, Columbia University.Rumberger, Russell. (1983). “Dropping out of high school: The influence of race, sex,and family background.” American educational research journalRumberger, Russell. (2004). “Why students drop out of school.” Pp. 131–155 inDropouts in America: Confronting the graduation rate crisis,edited by Gary Orfield.Cambridge, MA: Harvard Education Press.Waldfogel, Jane, Irwin Garfinkel, and Brendan Kelly. (2005). “Public assistance programs: How much could be saved with improved education?” Paper present-ed at the Symposium on Social Costs of Inadequate Education, Teachers’College, Columbia University.Wehlage, Gary, Robert Rutter, Gregory Smith, Nancy Lesko, and RicardoReducing the risk: Schools as communities of support.Philadelphia: Falmer Press. For more information about the Philadelphia YouthTransitions Collaborative and Project U-Turn, please contact:Philadelphia Youth Transitions Collaborative c/o Philadelphia Youth Network714 Market St., Suite 304Philadelphia, PA 19106www.projectUturn.net Additional funding for Project U-Turn has been provided by: