HowEffectiveIsDrugAbuseResistanceEducationAMetaAnalysisofProjectDAREO - PDF document

1 HowEffectiveIsDrugAbuseResistanceEducati
HowEffectiveIsDrugAbuseResistanceEducation?AMeta-AnalysisofProjectDAREOutcomeEvaluationsSusanT.Ennet,PhD,NancyS.Tobler,MS,PhD,ChristopherL.Ringwalt,DrPH,andRobertL.Flewelling,PhDIntroductionSchool-baseddrugusepreventionprogramshavebeenanintegralpartoftheUSantidrugcampaignforthepasttwodecades."2Althoughprogramshaveproliferated,noneismoreprevalentthanProjectDARE(DrugAbuseResistanceEducation).3Createdin1983bytheLosAngelesPoliceDepartmentandtheLosAngelesUnifiedSchoolDistrict,DAREusesspeciallytrainedlawenforcementofficerstoteachadrugusepreventioncurriculuminelementaryschools4and,morerecently,injuniorandseniorhighschools.Sinceitsinception,DAREhasbeenadoptedbyapproximately50%oflocalschooldistrictsnationwide,anditcontinuestospreadrapidly.3DAREistheonlydrugusepreventionprogramspecifi-callynamedinthe1986Drug-FreeSchoolsandCommunitiesAct.Some10%oftheDrug-FreeSchoolsandCommunitiesActgovernors'funds,whichare30%ofthefundsavailableeachfiscalyearforstateandlocalprograms,aresetasideforprograms"suchasProjectDrugAbuseResistanceEducation,"5amountingtomuchoftheprogram'spublicfunding.Givenitswidespreaduseandtheconsiderableinvestmentofgovernmentdollars,schooltime,andlawenforcementeffort,itisimportanttoknowwhetherDAREisaneffectivedrugusepreventionprogram.Thatis,towhatextentdoesDAREmeetitscurriculumobjectives,mostprominently"tokeepkidsoffdrugs"?DARE'scorecurriculum,offeredtopupilsinthelastgradesofelementaryschool,istheheartofDARE'sprogramandthefocusofthisstudy.Weevaluateherethecorecurriculum'sshort-termeffectivenessbyusingmeta-analytictech-niquestointegratetheevaluationfindingsofseveralstudies.6'7WesearchedforallDAREevaluations,bothpublishedandunpublished,conductedoverthepast10yearsandselectedforfurtherreviewthosestudiesthatmetspecifiedmethod-ologicalcriteria.Wecalculatedeffectsizesasamethodforestablishingacomparableeffectivenessmeasureacrossstudies.7-9Inaddition,toputDAREinthecontextofotherschool-baseddrugusepreventionprograms,wecomparedtheaveragemagnitudeoftheDAREeffectsizeswiththoseofotherprogramsthattargetyoungpeopleofasimilarage.DARE'sCoreCuriculumTheDAREcorecurriculum's17lessons,usuallyofferedonceaweekfor45to60minutes,focusonteachingpupilstheskillsneededtorecognizeandresistsocialpressurestousedrugs.4Inaddition,lessonsfocusonprovidinginformationaboutdrugs,teachingdecision-makingskills,buildingself-esteem,andchoosinghealthyalternativestodruguse.4DAREofficersuseteachingstrategies,suchaslectures,groupdiscussions,question-and-SusanT.Ennett,ChristopherL.Ringwalt,andRobertL.FlewellingarewiththeResearchTriangleInstitute,ResearchTrianglePark,NC.NancyS.TobleriswiththeStateUniver-sityofNewYork,atAlbany,NY.RequestsforreprintsshouldbesenttoSusanT.Ennett,PhD,CenterforSocialResearchandPolicyAnalysis,ResearchTri-angleInstitute,POBox12194,ResearchTrianglePark,NC27709-2194.ThispaperwasacceptedFebruary15,1994.Note.Theviewsexpressedherearetheauthors'anddonotnecessarilyrepresenttheofficialpositionoftheUSDepartmentofJusticeortheUSDepartmentofHealthandHumanServices.September1994,Vol.84,No.9 DAREMeta-Analysisanswersessions,audiovisualmaterial,workbookexercises,androle-playing.4ThetrainingthatDAREofficersreceiveissubstantial.Theyarerequiredtoundergo80traininghoursinclassroommanagement,teachingstrategies,commu-nicationskills,adolescentdevelopment,druginformation,andcurriculuminstruc-tion.4Inaddition,DAREofficerswithclassroomexperiencecanundergofurthertrainingtoqualifyasinstructors/men-tors.4Theseofficersmonitortheprogramdelivery'sintegrityandconsistencythroughperiodicclassroomvisits.MethodsIdentificationofEvaluationsWeattemptedtolocateallquantita-tiveevaluationsofDARE'scorecurricu-lumthroughasurveyofDARE'sfiveRegionalTrainingCenters,computerizedsearchesofthepublishedandunpub-lishedliterature,andtelephoneinter-viewswithindividualsknowntobein-volvedwithDARE.Eighteenevaluationsin12statesandoneprovinceinCanadawereidentified.Severalevaluationswerereportedinmultiplereportsorpapers.(SeeAppendixAforabibliographyofthestudiesconsidered.)EvaluationSelectionCritenaTobeselectedforthismeta-analysis,anevaluationmusthavemetthefollowingcriteria:(1)useofacontrolorcomparisongroup;(2)pretest-posttestdesignorpost-testonlywithrandomassignment;and(3)useofreliablyoperationalizedquantita-tiveoutcomemeasures.Quasi-experimen-talstudieswereexcludediftheydidnotcontrolforpreexistingdifferencesonmeasuredoutcomeswitheitherchangescoresorcovariance-adjustedmeans.10Inaddition,toensurecomparability,wefocusedonresultsbasedon

2 lyonimmedi-ateposttest.Becauseonlyfourev
lyonimmedi-ateposttest.Becauseonlyfourevaluationstudieswerelongterm(twoofwhichwerecompromisedbyseverecontrolgroupattritionorcontamination),wewereun-abletoadequatelyassesslonger-termDAREeffects.Weexaminedseveralothermethod-ologicalfeatures,suchasthecorrespon-dencebetweentheunitofassignmentandanalysis,theuseofapaneldesign,matchingofschoolsintheinterventionandcontrolconditions,andattritionrates.Althoughthesefactorswereconsideredinassessingthestudies'overallmethodologi-calrigor,wedidnoteliminateevaluationsonthebasisofthesecriteria.DataAnalysisForeachstudy,wecalculatedaneffectsizetoquantifythemagnitudeofDARE'seffectivenesswithrespecttoeachofsixoutcomesthatreflecttheDAREcurriculum'saims.Aneffectsizeisdefinedasthedifferencebetweentheinterventionandthecontrolgroupmeansforeachoutcomemeasure,standardizedbydividingbythepooledstandardde-viation[effectsize=mean,-meanc/SD].7-9Ifmeansandstandarddeviationswerenotavailable,wecalculatedeffectsizesusingformulasdevelopedtoconvertotherteststatisticsandpercentagestoeffectsizes.9Inallcases,weusedstatisticsreflectingcovariance-adjustedmeans,withpretestvaluesascovariatesratherthanunadjustedmeanssothatanydifferencesbetweenthecomparisongroupsbeforetheinterventionwouldnotbereflectedintheeffectsizes.10Thesixoutcomemeasureclassesincludeknowledgeaboutdrugs,attitudesaboutdruguse,socialskills,self-esteem,attitudetowardpolice,anddruguse.Somestudiesdidnotincludeallsix,andsomeoutcomesweremeasuredbymorethanoneindicator.Whenmultipleindica-torswereused(e.g.,twomeasuresofsocialskills),wecalculatedseparateeffectsizesandthenaveragedthem.6"0Thisprocedureyieldedoneeffectsizeperstudyforeachmeasuredoutcometype.Intheonestudythatreportedonlythatameasuredoutcomewasnotstatisticallysignificant(anddidnotprovideanyfurtherstatistics),weassignedazerovaluetothateffectsize.10Tocalculateeffectsizesfordruguse,weconsideredonlyalcohol,tobacco,andmarijuanause;weaveragedeffectsizesacrossthesesubstances.Inasupplementaryanalysis,weconsidereduseofthesesubstancesseparately.Theprevalenceofotherdrugs,suchascocaine,wastoosmalltoproducemeaningfuleffects.Inadditiontocalculatingoneeffectsizeperoutcomeperstudy,wecalculatedtheweightedmeaneffectsizeand95%confidenceinterval(CI)foreachoutcometypeacrossprograms.Theweightedmeaniscomputedbyweightingeacheffectsizebytheinverseofitsvariance,whichisareflectionofthesamplesize.8'9Theeffectsizeestimatesfromlargerstudiesaregenerallymoreprecisethanthosefromsmallerstudies.8Hence,theweightedmeanprovidesalessbiasedestimatethanthesimple,unweightedmeanbecauseestimatesfromlargersamplesaregivenmoreweight.The95%CIindicatestheestimatedeffectsize'saccuracyorreliabil-ityandiscalculatedbyaddingtoorsubtractingfromthemean1.96multipliedbythesquarerootof1dividedbythesumofthestudyweights.8ComparisonofDAREwithOtherDnrgUsePreventionProgramsForcomparisonwithDARE,weusedtheeffectsizesreportedinTobler'smeta-analysisofschool-baseddrugusepreventionprograms.10ToallowthemostappropriatecomparisonswithDAREeffectsizes,weobtainedTobler'sresultsforonlythoseprograms(excludingDARE)aimedatupperelementaryschoolpupils.Theseprogramsareasubsetof25fromthe114programsinTobler'smeta-analysis,whosestudiesarereferencedinAppendixB.Weselectedthismeta-analysisforcomparisonbecauseofitsgreatersimilar-itytooursthanothermeta-analysesofdrugusepreventionprograms."1-'4To-bler'sstudiesmetthesamemethodologi-calstandardsthatweusedfortheDAREstudies.TheonlydifferenceswerethatToblerexcludedstudiesthatdidnotmeasuredruguseandconsideredresultsfromlaterposttests,whereasweconsid-eredonlyimmediateposttestresults.Nei-therofthesedifferences,however,shouldseriouslycompromisethecomparison.TheevaluationstudiesincludedinTobler'smeta-analysisareclassifiedintoAmericanJournalofPublicHealth1395TABLE1DAREEvaluationStudiesSelectedforReviewLocationReferencesaBritishColumbiaWalker1990(BC)Hawaii(HI)Manos,Kameoka,andTanji1986Illinois(IL)Ennettetal.1994(inpress)Kentucky-AClaytonetal.(KY-A)1991a,1991bKentucky-BFaineand(KY-B)Bohlander1988,1989MinnesotaMcCormickand(MN)McCormick1992NorthCarolinaRingwalt,Ennett,(NC)andHolt1991SouthCarolinaHarmon1993(SC)aSeeAppendixAforfullreferences.September1994,Vol.84,No.9 EnnettetaLtwobroadcategoriesbasedonthepro-grams'contentandprocess.Processde-scribestheteachingapproach(howthecontentisdelivered).ProgramsclassifiedbyTobleras"noninteractive"emphasizeintrapersonalfactors,suchasknowledgegainandaffectivegrowth,andareprimar-ilydeliveredbyanexpert."Interactive"programsemphasizeinterperso

3 nalfactorsbyfocusingonsocialskillsandgen
nalfactorsbyfocusingonsocialskillsandgeneralsocialcompetenciesandbyusinginterac-tiveteachingstrategies,particularlypeertopeer.Consistentwithothermeta-analysesshowingthatprogramsemphasiz-ingsocialskillstendtobethemostsuccessful,'"'13,5interactiveprogramspro-ducedlargereffectsizesthannoninterac-tiveprograms.WecomparedDAREwithbothcategoriesofprograms.ResultsCharacteristicsofEvaluationsOftheoriginal18studies,8metthecriteriaforinclusion.Oneadditionalstudymetthemethodologicalcriteriabutdidnotadministerthefirstposttestuntil1yearafterDAREimplementation;therefore,itcouldnotbeincludedinouranalysisofimmediateeffects.'6"17ThelocationandprimaryreferenceforeachevaluationareshowninTable1,andstudycharacteristicsaresummarizedinTable2.Eachevaluationrepresentsastateorlocaleffort.Thenumberofstudentsubjectsinallstudieswaslarge,eachstudycomprisingatleast10schoolswithapproximately500to2000students.Al-thoughdemographicinformationwasnotgivenforthreestudies,theremainingfivestudiesinthesampleprimarilyconsistedofWhitesubjects.AssignmentofDAREtointerven-tionandcontrolgroupswasbyschoolforalleightstudies.Inonestudy,DAREalsowasassignedbyclassroomincertainschools.18Becauseofpotentialcontamina-tioninthisstudyofthecontrolgroupclassroomsbytheircloseproximitytoDAREclasses,weeliminatedthesecon-trolclassrooms;onlycontrolschoolswithnoDAREclasseswereincluded.TwostudiesusedatrueexperimentaldesigninwhichschoolswererandomlyassignedtoDAREandcontrolconditions;athirdstudyusedrandomassignmentfortwothirdsoftheschools.Theremainingfiveevaluationsusedanonequivalentcontrolgroupquasi-experimentaldesign.Becausetherewererelativelyfewsamplingunitsacrossstudies-rangingfrom11to63schools,withallexceptonestudyinvolvingfewerthan40schools-itisunlikelythatequivalencebetween1396AmericanJournalofPublicHealthTABLE2-SampleandMethodologicalCharacteristicsoftheDAREEvaluations(n=8)Schools,Subjects,UnitofPretestScaleStudynnResearchDesignMatchingAnalysisEquivalency8ReliabilitiesAttritionBC11D=287Quasi,cross-sectionalYesIndividualYesNoNotapplicableC=175Hi26D=1574Quasi,panelNoIndividualNoNoNoC=435IL36D=715Experimental/quasi,YesSchoolbasedYesYesYesbC=608panelKY-A31D=1438Experimental,panelNoIndividualYesYesYesbC=487KY-B16D=451Quasi,panelYesIndividualYesYesNoC=332MN63D=453Quasi,panelNoIndividualYesYesYescC=490NC20D=685Experimental,panelNoSchoolbasedYesYesYesbC=585SC11D=295Quasi,panelYesIndividualYesYesYescC=307Note.SeeTable1forinformationonstudylocationsandreferences.D=DARE;C=comparison.aPretestequivalencyondemographicvariablesassessedandcontrolledifnecessary.bAttritionratesreportedanddifferentialattritionacrossexperimentalconditionsanalyzed.cAttritionratesreportedonly.TABLE3-UnweightedEffectSizesAssociatedwithEightDAREEvaluationsAttitudeAttitudesSocialSelf-towardDrugStudyKnowledgeaboutDrugsSkillsEsteemPoliceUseaBC.68.00..........02Hi....07.34......IL....03.15.15.12.05KY-A....11.10.07....00KY-B.58.19.30.14.27...MN.19.06.08-.03.05...NC....19.17.00....11SC....32.19.06.08.10Note.SeeTable1forinformationonstudylocationsandreferences.aLUmitedtoalcohol,tobacco,andmarijuana.September1994,Vol.84,No.9 DAREMeta-Analysisgroupswasobtainedwithoutpriormatch-ingorblockingofschools,evenwithrandomization.Onlyhalfthestudiesmatchedcomparisonschoolsonselecteddemographiccharacteristics.Moststudies(75%),however,assessedtheequivalencyofthecomparisongroupsatpretestandmadeadjustmentsforpretestdifferencesondemographiccharacteristics.Allstud-iesadjustedforpretestdifferencesonoutcomemeasures.Allbutonestudyusedapaneldesignthatmatchedsubjectsfrompretesttoposttestwithauniqueidentificationcode.OutcomemeasuresusedintheDAREevaluationswerebasedonre-sponsestoself-administeredquestion-naires.Sevenstudiesusedstandardizedscalesorrevisedexistingmeasures;sixstudiesreportedgenerallyhighscalereliabilities(usuallyCronbach'salpha).Validityinformation,however,wasrarelyreported,andnostudyusedeitherabiochemicalindicatoror"boguspipeline"techniquetovalidatedruguseself-reports.'9Moststudies(75%)didnotuseadataanalysisstrategyappropriatetotheunitofassignment.Becauseschools,notstudents,wereassignedtoDAREandcontrolconditions,itwouldhavebeenappropriatetoanalyzethedatabyschoolswithsubjects'dataaggregatedwithineachschoolortouseahierarchicalanalysisstrategyinwhichsubjectsarenestedwithinschools.20321Sixstudiesignoredschoolsaltogetherandanalyzedindi-vidualsubjects'data,therebyviolatingthestatisticalassumptionofindependenceofobservations.Ignoringschoolsasaunitofanalysisresultsinapos

4 itivebiastowardfindingstatisticallysigni
itivebiastowardfindingstatisticallysignificantprogrameffects.21ThisbiasmaybereflectedinCIsreportedforeachoutcome'sweightedmeaneffectsize.Fivestudiesreportedgenerallysmallattritionrates.Noneofthethreestudiesthatanalyzedattritionfoundthatratesdifferedsignificantlyacrossexperimentalandcontrolconditions.Inaddition,sub-jectsabsentfromtheposttestwerenotmorelikelytobedrugusersoratriskfordruguse.Althoughattritionusuallyisgreateramongdrugusers,22giventhesample'syoungage(whenschooldropoutisunlikelyanddruguseprevalenceislow),theseresultsarenotsurprising.DAREEffectSizesStudyeffectsizesareshowninTable3.Ingeneral,thelargesteffectsizesareforknowledgeandsocialskills;thesmall-estarefordruguse.MeanEffectSize.9.8....8....................................6......................................................................4..4................................3.....................................19........1..111.0KnowledgeAttitudesSocialSkillsDruguseincludesalcohol,tobacco,andmarijuana.Figure1showsthemeanweightedeffectsizeand95%CIforeachoutcomebasedontheeightstudiescombined.Thelargestmeaneffectsizeisforknowledge(.42),followedbysocialskills(.19),atti-tudetowardthepolice(.13),attitudesaboutdruguse(.11),self-esteem(.06),anddrugbehavior(.06).Theeffectsizesforknowledge,socialskills,attitudeto-wardthepolice,attitudesaboutdruguse,andself-esteemarestatisticallysignifi-cant.TheCIforthemeandruguseeffectsizeoverlapswithzero(i.e.,itisnotsignificantlydifferentfromzero).Becauseaveragingalcohol,tobacco,andmarijuanauseforthedruguseeffectsizecouldobscuresubstantialdifferencesamongthesubstances,wecalculatedDARE'smeanweightedeffectsizessepa-ratelyforthesesubstances.Theweightedmeaneffectsizeforalcoholuseis.06(95%CI=.00,.12);fortobaccouse,.08(95%CI=.02,.14);andformarijuanause,-.01(95%CI=-.09,.07).Onlythemeanfortobaccouseisstatisticallysignificant.MeanEffectSizesforDAREvsOtherDnrgUsePreventionProgramsWecomparedbytypeofoutcomethemeanweightedDAREeffectsizewiththeSelf-PoliceDrugEsteemUse'meanweightedeffectsizefornoninterac-tive(n=9)andinteractive(n=16)pro-grams;effectsizesforthecomparisonprogramsarederivedfromTobler.'0ThecomparisonprogramstargetyouthofthesamegraderangetargetedbyDARE.TheoutcomesassessedbybothDAREandthecomparisonprogramsareknowl-edge,attitudes,socialskills,anddrugusebehavior.Acrossthefouroutcomedomains,DARE'seffectsizesaresmallerthanthoseforinteractiveprograms(Figure2).MostnotableareDARE'seffectsizesfordruguseandsocialskills;neithereffectsize(.06and.19,respectively)ismorethanathirdofthecomparableeffectsizesforinteractiveprograms(.18and.75,respectively).DARE'seffectsizefordruguseisonlyslightlysmallerthanthenoninteractiveprograms'effectsize.DARE'seffectsizesforknowledge,atti-tudes,andsocialskills,however,arelargerthanthosefornoninteractivepro-grams.Comparisonofeffectsizesseparatelyforalcohol,tobacco,andmarijuanauseshowsthatDARE'seffectsizesaresmallerthanthoseforinteractivepro-grams(Figure3).Exceptfortobaccouse,AmericanJounalofPublicHealth1397FIGURE1-MagnitudeofDARE'sweightedmeaneffectsize(and95%CI),byoutcomemeasure.September1994,Vol.84,No.9 Ennettetal.theyalsoaresmallerthanthosefornoninteractiveprograms.DiscussionTheresultsofthismeta-analysissuggestthatDARE'scorecurriculumeffectondruguserelativetowhateverdrugeducation(ifany)wasofferedinthecontrolschoolsisslightand,exceptfortobaccouse,isnotstatisticallysignificant.Acrossthestudies,noneoftheaveragedruguseeffectsizesexceeded.11.Reviewofseveralmeta-analysesofadolescentdrugusepreventionprogramssuggeststhateffectsizesofthismagnitudearesmall.1014ThesmallmagnitudeofDARE'seffectivenessondrugusebehaviormaypartiallyreflecttherelativelylowfre-quencyofdrugusebytheelementaryschoolpupilstargetedbyDARE'scorecurriculum.However,comparisonoftheDAREeffectsizeswiththoseofotherschool-baseddrugusepreventionpro-gramsforsame-ageadolescentssuggeststhatgreatereffectivenessispossiblewithearlyadolescents.ComparedwiththeprogramsclassifiedbyToblerasinterac-tive,DARE'seffectsizesforalcohol,tobacco,andmarijuanause,bothcollec-tivelyandindividually,aresubstantiallyless.10Exceptfortobaccouse,theyalsoarelessthanthedruguseeffectsizesformoretraditional,noninteractivepro-grams.IthasbeensuggestedthatDAREmayhavedelayedeffectsondrugusebehavioroncepupilsreachhighergrades.23'24Longer-termfollow-upstudiesareneededtotestthispossibility.Onlyfourreviewedstudiesadministeredmul-tipleposttests,andfortwoofthesetheresultsfromsomelaterposttestsareuninterpretable.However,basedontw

5 oexperimentalstudiesforwhichreliableinfo
oexperimentalstudiesforwhichreliableinformation1and2yearsafterimplemen-tationisavailable,thereisnoevidencethatDARE'seffectsareactivatedwhensubjectsareolder.25'26Mostlong-termevaluationsofdrugusepreventionpro-gramshaveshownthatcurriculumeffectsdecayratherthanappearorincreasewithtime.27,28DARE'simmediateeffectsonout-comesotherthandruguseweresome-whatlarger(especiallyforknowledge)andwerestatisticallysignificant.Theseeffectsizes,however,alsowerelessthanthecomparableeffectsizesforsame-ageinteractiveprograms.ThatDARE'seffectsizesforknowledge,attitudes,andskillsweregreaterinmagnitudethanthoseofnoninteractiveprogramsmaynotbepar-ticularlymeaningfulbecausemanyofthesetypesofprograms,suchasprogramsusing"scaretactics"oremphasizingfac-tualknowledgeaboutdruguse,havebeendiscreditedasunsuccessful.293'0ComparisonofDARE'scorecurricu-lumcontentwiththeinteractiveandnoninteractiveprograms'curriculamaypartiallyexplaintherelativedifferencesineffectsizesamongtheseprograms.Inter-activeprogramstendtoemphasizedevel-opingdrug-specificsocialskillsandmoregeneralsocialcompetencies,whereasnon-interactiveprogramsfocuslargelyonintrapersonalfactors.BecauseDAREhasfeaturesofbothinteractiveandnoninter-activeprograms,itisperhapsnotsurpris-ingthattheeffectsizeswereportedshouldfallsomewhereinbetween.Per-hapsgreateremphasisintheDAREcorecurriculumonsocialcompetenciesandlessemphasisonaffectivefactorsmightresultineffectsizesnearertothosereportedforinteractiveprograms.How-ever,itisdifficulttospeculateontheeffectofaddingorsubtractingparticularlessonstoorfromDARE'scurriculum.Mostschool-basedpreventionprogramevaluationshaveassessedtheeffective-nessofanoverallprogramratherthanvariousprogramcomponentsorcombina-tionsofcomponents.WhoteachesDAREandhowitistaughtmayprovideotherpossibleexplana-tionsforDARE'slimitedeffectiveness.DespitetheextensiveDAREtrainingreceivedbylawenforcementofficers,theymaynotbeaswellequippedtoleadthecurriculumasteachers.NostudieshavebeenreportedinwhichtheDAREcur-riculumwasofferedbyanyoneotherthanapoliceofficer;resultsfromsuchastudymightsuggestwhetherteachersproducebetter(orworse)outcomesamongpupils.Regardlessofcurriculumleader,however,thegenerallymoretraditionalteachingstyleusedbyDAREhasnotbeenshowntobeaseffectiveasaninteractiveteachingmode.1014Althoughsomeactivitiesencouragepupilinterac-tion,thecurriculumreliesheavilyontheofficerasexpertandmakesfrequentuseoflecturesandquestion-and-answerses-sionsbetweentheofficerandpupils.Infact,itisinteachingstyle,notcurriculumcontent,thatDAREmostdiffersfromtheinteractiveprogramsexaminedbyTobler.1398AmericanJournalofPublicHealthMeanEffectSize.9.80.76.7.60.53.50.42.40.33-3.20.160.90.180.11.1~~~~~~~~~~0.080.60.08.0KnowledgeAttitudesSocialSkillsDrugUseIEDAREONoninteractiveKInteractiveNote.ComparisonprogramsselectedfromTobler.101Druguseincludesalcohol,tobacco,andmanjuana.FIGURE2-Weightedmeaneffectsize,byoutcome,forDAREandotherdrugusepreventionprograms.September1994,Vol.84,No.9 TheDAREcorecurriculumrecentlywasmodifiedtointroducemoreparticipatoryactivities,whichmayleadtogreaterprogrameffectiveness.Severallimitationsshouldbeconsid-eredinevaluatingourfindings.Thenumberofevaluationsreviewed(eight)isnotlargewhencomparedwiththevastnumberofsiteswhereDAREhasbeenimplemented.Theconsistencyofresultsacrossstudies,however,suggeststhattheresultsarelikelytoberepresentativeofDARE'scorecurriculum.Evenso,wewouldhavepreferredafullsetofeighteffectsizesforeachoutcome.ItispossiblethattheeffectsizesfortheDAREstudiesmayhavebeenattenu-atedcomparedwiththedrugusepreven-tionprogramsreviewedbyToblerbe-causethecontrolgroupswerenotpure"notreatment"groups.AsdocumentedbyTobler,effectsizesarelowerwhenthecontrolgroupreceivessomesortofdrugeducation.10'14TheDAREevaluationsgenerallylackedinformationonalterna-tivetreatmentsreceivedbythecontrolgroups,butitislikelythatmostcontrolgroupsreceivedsomedrugeducationbecausethestudiesoccurredafterthe1986Drug-FreeSchoolsandCommuni-tiesAct.However,approximatelyhalf(54%)oftheprogramsreviewedbyTobleralsowereconductedbetween1986and1990,suggestingthattheymaysufferfromthesameeffect.10MostofthedrugusepreventionprogramsevaluatedbyToblerwereuniver-sityresearch-basedevaluationstudies,whereasDAREisacommerciallyavail-ablecurriculum.AlthoughthemagnitudeoftheresourcesinvestedinDAREisconsiderable,theintensityofeffortde-votedtosmaller-scaleprogramsmaybegreater.Somediminishedeffectivenessisperhapsinevitableonceprogramsarewidelymarketed.Althoughwefoundlimitedimm

6 edi-atecorecurriculumeffects,somefeature
edi-atecorecurriculumeffects,somefeaturesofDAREmaybemoreeffective,suchasthemiddleschoolcurriculum.Inaddition,DARE'scumulativeeffectsmaybegreaterinschooldistrictswhereallDAREcur-riculaforyoungerandolderstudentsareinplace.OtherDAREoutcomes,suchasitsimpactoncommunitylawenforcementrelations,alsomayyieldimportantben-efits.However,duetotheabsenceofevaluationstudies,considerationofthesefeaturesisbeyondthisstudy'sscope.DARE'slimitedinfluenceonadoles-centdrugusebehaviorcontrastswiththeprogram'spopularityandprevalence.AnSeptember1994,Vol.84,No.9ijAnmivieta-AnatysisMeanEffectSize.85.6~~~~~~~~~~~~~~~~~0*.5..4.'O..0-.2A_S!AlcTobaccoMarijuanaUeAREONoninteractiveUIweractiveNote.CompaisonpgssidfmTobler.10FIGURE3-Weightedmeaneffectsize,bydrug,forDAREandotherdrugusepreventionprograms.importantimplicationisthatDAREcouldbetakingtheplaceofother,morebeneficialdrugusecurriculathatadoles-centscouldbereceiving.Atthesametime,expectationsconcerningtheeffec-tivenessofanyschool-basedcurriculum,includingDARE,inchangingadolescentdrugusebehaviorshouldnotbeover-stated.3"0AcknowledgmentsThisresearchwassupportedbyawardsfromtheNationalInstituteofJustice,OfficeofJusticePrograms,USDepartmentofJustice(91-DD-CX-K053)andtheNationalInstituteonDrugAbuse,USDepartmentofHealthandHumanServices(5R01DA07037).WethankSusanL.Bailey,KarlE.Bauman,GeorgeH.Dunteman,RobertL.Hubbard,RonaldW.Johnson,andJ.ValleyRachalfortheirthoughtfulcommentsonanearlierdraftofthispaper.Inaddition,weacknowledgethesupportandassistanceofJodyM.GreeneonthisprojectandtheeditorialassistanceofBerylC.Pittman,LindaB.Barker,andRichardS.Straw.References1.GlynnTJ.Essentialelementsofschool-basedsmokingpreventionprogram.JSchHealth.1989;59:181-188.2.HealthyPeople2000.Washington,DC:USDeptofHealthandHumanServices;1988.DHHSpublicationno.PHS91-50212.3.RingwaltCL,GreeneJM.Resultsofschooldistricts'drugpreventioncoor-dinator'ssurvey.PresentedattheAlcohol,Tobacco,andOtherDrugsConferenceonEvaluatingSchool-LinkedPreventionStrat-egies;March1993;SanDiego,Calif.4.ImplementingProjectDARE:DrugAbuseResistanceEducation.Washington,DC:BureauofJusticeAssistance;1988.5.Drug-FreeSchoolsandCommunityAct.PubLNo.101-647,§5122(e)(1).6.Bangert-DrownsRL.Reviewofdevelop-mentsinmeta-analyticmethod.PsycholBull.1986;99:388-399.7.RosenthalR.Meta-AnalyticProceduresforSocialResearch.AppliedSocialResearchMethodsSeries.Vol6.NewburyPark,Calif:SagePublications;1991.8.HedgesLV,OlkinI.StatisticalMethodsforMeta-Analysis.NewYork,NY:AcademicPress;1985.9.PerryPD,ToblerNS.Meta-analysisofadolescentdrugpreventionprograms:finalreport.In:ManualforEffectSizeCalcula-tion:FormulaUsedintheMeta-AnalysisofAdolescentDrugPreventionPrograms.Rock-ville,Md:NationalInstituteonDrugAbuse;1992:appendix3.10.ToblerNS.Meta-AnalysisofAdolescentDrugPreventionPrograms:FinalReport.AmericanJournalofPublicHealth1399 Ennettetal.Rockville,Md:NationalInstituteonDrugAbuse;1992.11.Bangert-DrownsRL.Theeffectsofschool-basedsubstanceabuseeducation:ameta-analysis.JDrugEduc.1988;18:243-264.12.BruvoldWH.Ameta-analysisofadoles-centsmokingpreventionprograms.AmJPublicHealth.1993;83:872-880.13.BruvoldWH,RundallTG.Ameta-analysisandtheoreticalreviewofschool-basedtobaccoandalcoholinterventionprograms.PsycholHealth.1988;2:53-78.14.ToblerNS.Meta-analysisof143adoles-centdrugpreventionprograms:quantita-tiveoutcomeresultsofprogrampartici-pantscomparedtoacontrolorcomparisongroup.JDrugIssues.1986;16:537-567.15.HansenWB.School-basedsubstanceabuseprevention:areviewofthestateoftheartincurriculum,1980-1990.HealthEducRes.1992;7:403-430.16.NyreGF,RoseC.DrugAbuseResistanceEducation(DARE)LongitudinalEvalua-tionAnnualReport,January1987.LosAngeles,Calif:EvaluationandTrainingInstitute;1987.Unpublishedreport.17.NyreGF,RoseC,BolusRE.DrugAbuseResistanceEducation(DARE)LongitudinalEvaluationAnnualReport,August1987.LosAngeles,Calif:EvaluationandTrainingInstitute;1987.Unpublishedreport.18.ManosMJ,KameokaKY,TanjiJH.EvaluationofHonoluluPoliceDepartnent'sDnrgAbuseResistanceEducationProgram.Honolulu,Hawaii:UniversityofHawaii-Manoa,SchoolofSocialWork,YouthDevelopmentandResearchCenter;1986.Unpublishedreport.19.BaumanKE,DentCW.Influenceofanobjectivemeasureontheself-reportsofbehavior.JApplPsychol.1982;67:623-628.20.MoskowitzJM.Whyreportsofoutcomeevaluationsareoftenbiasedoruninterpret-able.EvalProgrPlan.1993;16:1-9.21.MurrayDM,HannanPJ.Planningfortheappropriateanalysisinschool-baseddrugusepreventionstudies.JConsultClinPsychol.1990;58:458-468.22.BiglanA,AryDV.

7 Methodologicalissuesinresearchonsmokingp
Methodologicalissuesinresearchonsmokingprevention.In:BellCS,BattjesR,eds.PreventionResearch:DeterringDrugAbuseAmongChildrenandAdolescents.Rockville,Md:NationalInsti-tuteonDrugAbuse;1985:170-195.23.ClaytonRR,CattarelloA,DayLE,Wal-denKP.Persuasivecommunicationanddrugabuseprevention:anevaluationoftheDAREprogram.In:DonohewL,SypherH,BukoskiW,eds.PersuasiveCommunicationandDrugAbusePrevention.Hillsdale,NJ:LawrenceErlbaumAssoci-ates,Inc;1991:295-313.24.RingwaltCL,EnnettST,HoltKD.AnoutcomeevaluationofProjectDARE(DrugAbuseResistanceEducation).HealthEducRes.1991;6:327-337.25.ClaytonRR,CattarelloA,WaldenKP.Sensationseekingasapotentialmediatingvariableforschool-basedpreventioninter-vention:atwo-yearfollow-upofDARE.HealthCommun.1991;3:229-239.26.EnnettST,RosenbaumDP,FlewellingRL,BielerGS,RingwaltCL,BaileySL.Long-termevaluationofDrugAbuseResis-tanceEducation.AddictBehav.1994;19:113-125.27.EllicksonPL,BellRM,McGuiganK.Preventingadolescentdruguse:long-termresultsofajuniorhighprogram.AmJPublicHealth.1993;83:856-861.28.MurrayDM,PirieP,LuepkerRV,Pal-lonenU.Five-andsix-yearfollow-upresultsfromfourseventh-gradesmokingpreventionstrategies.JBehavMed.1989;12:207-218.29.BotvinGJ.Substanceabuseprevention:theory,practice,andeffectiveness.In:TonryM,WilsonJQ,eds.DrugsandCrime.Chicago,Ill:UniversityofChicagoPress;1990:461-519.30.MoskowitzJ.Preventingadolescentsub-stanceabusethrougheducation.In:GlynnT,LeukefeldCG,LudfordJP,eds.Prevent-ingAdolescentDrugAbuse:InterventionStrategies.Rockville,Md:NationalInsti-tuteonDrugAbuse;1983:233-249.31.DryfooJG.Preventingsubstanceabuse:rethinkingstrategies.AmJPublicHealth.1993;83:793-795.APPENDIXA-BibliographyofDAREEvaluationsAniskiewiczRE,WysongEE.ProjectDAREEvaluationReport:EvaluationandTrainingInstitute.DAREEvaluationReportforKokomoSchoolsSpring,1987.Kokomo,Ind:IndianaUniversity-1985-1989.LosAngeles,Calif:EvaluationandTrainingInstitute;Kokomo;1987.Unpublishedreport.1990.Unpublishedreport.AniskiewiczR,WysongE.EvaluatingDARE:drugeducationandtheFaineJR,BohlanderE.DrugAbuseResistanceEducation:Anmultiplemeaningsofsuccess.PolicyStudRev.1990;9:727-747.Assessmentofthe1987-88KentuckyStatePoliceDAREProgram.DAREEvaluation,Sprin.1987.Pittsburgh,Pa;BowlingGreen,Ky:WesternKentuckyUniversity,SocialResearchAnonymous.ProjectDAREEvalLaboratory1988Unpublishedreport.1987.Unpublishedpaper.'FamneJR,BohlanderE.DAREinKentuckSchools1988-89.BowlingBeckerHR,AgopianMW,YehS.ImpactevaluationofDrugAbuseGreen,Ky:lWesernKentuckyUrShooilResearchBoratnResistanceEducation(DARE).JDrugEduc.1990;22:283-291.1989.Unpublishedreport.lClaytonRR,CattarelloA,DayLE,WaldenKP.Persuasivecommuni-HarmonMA.Reducingtheriskofdruginvolvementamongearlycationanddrugabuseprevention:anevaluationoftheDAREadolescents:anevaluationofDrugAbuseResistanceEducationprogram.In:DonohewL,SypherH,BukoskiW,eds.Persuasive(DARE)EvalRev1993;17221-2CommunicationandDrugAbusePrevention.Hillsdale,NJ:Lawrence(ErlbaumAssociates;1991a:295-313.KethineniS,LeamyDA,GuyonL.EvaluationoftheDrugAbuseResistanceEducationPrograminIllinois:PreliminaryRep!ort.Normal,Ill:ClaytonRR,CattarelloA,WaldenKP.SensationseekingasaInisStatce.unDeparmentofCreiminalJusic.NScIecspotentialmediatingvariableforschool-basedpreventionintervention:1991.Unublihdreport.atwo-yearfollow-upofDARE.HealthCommun.1991b;3:229-239..npub1SherepoMcCormick,FC,McCormickER.AnEvaluationoftheThirdYearDrugDeJongW.Ashort-termevaluationofProjectDARE(DrugAbuseAbuseResistanceEducation(DARE)PrograminSaintPaul.SaintPaul,ResistanceEducation):preliminaryindicatorsofeffectiveness.JDrugMinn:educanalOperationsConepts,Sc;1992UnPulEduc1987-17:279-294.Minn:EducationalOperationsConcepts,Inc;1992.UnpublishedEduc.198,;17279-294..report.DukesR,MatthewsS.EvaluationoftheColoradoSpringsCitywideMcDonaldRM,TowbermanDB,HagueJL.VolumeII:1989ImpactDAREProgram.ColoradoSprings,Co':UniversityofColorado,AssessmentofDrugAbuseResistanceEducationintheCommonwealthofCenterforSocialScienceResearch;1991.Unpublishedreport.Virginia.Richmond,Va:VirginiaCommonwealthUniversity,InstituteEarleRB,GarnerJ,PhillipsN.EvaluationoftheIllinoisStatePoliceforResearchinJusticeandRiskAdministration;1990.UnpublishedPilotDAREProgram.Springfield,Ill:A.H.TrainingandDevelopmentreport.Systems,Inc;1987.Unpublishedreport.ManosMJ,KameokaKY,TanjiJH.EvaluationofHonoluluPoliceEnnettST,RosenbaumDP,FlewellingRL,BielerGS,RingwaltCR,Department'sDrugAbuseResistanceEducationProgram.Honolulu,BaileySL.Long-termevaluationofDrugAbuseResistanceEduc

8 ation.Hawaii:UniversityofHawaii-Manoa,Sc
ation.Hawaii:UniversityofHawaii-Manoa,SchoolofSocialWork,YouthAddictBehav.1994;19:113-125.DevelopmentandResearchCenter;1986.Unpublishedreport.1400AmericanJournalofPublicHealthSeptember1994,Vol.84,No.9 DAREMeta-AnalysisAPPENDIXA-ContinuedNyreGF,RoseC.DnugAbuseResistanceEducation(DARE)LongitudinalEvaluationAnnualReport,January1987.LosAngeles,Calif:EvaluationandTrainingInstitute;1987.Unpublishedreport.NyreGF,RoseC,BolusRE.DrugAbuseResistanceEducation(DARE)LongitudinalEvaluationAnnualReport,August1987.LosAngeles,Calif:EvaluationandTrainingInstitute;1987.Unpublishedreport.RingwaltC,EnnettST,HoltKD.AnoutcomeevaluationofProjectDARE(DrugAbuseResistanceEducation).HealthEducRes.1991;6:327-337.WalkerS.TheVictoriaPoliceDepartmentDnigAbuseResistanceEducationProgramme(DA.RE.)ProgrammeEvaluationReport#2(1990).Victoria,BritishColumbia:TheMinistryoftheSolicitor-General,FederalGovernmentofCanada;1990.Unpublishedreport.APPENDIXB-BibliographyofComparisonProgramEvaluationsAllisonK,SilverG.DignamC.EffectsonstudentsofteachertrainingMcAlisterA.Approachestoprimaryprevention.Presentedattheinuseofadrugeducationcurriculum.JDrugEduc.1990;20:31-46.NationalAcademyofScienceNationalResearchCouncilConferenceonAlcoholandPublicPolicy;May1983;Washington,DC.DielmanT,ShopeJ,ButchartA,CampanelliP.Preventionofadolescentalcoholmisuse:anelementaryschcolprogram.IPediatrMoskowitzJ,MalvinJ,SchaefferG,SchapsE.EvaluationofanPsychol.1986;11:259-281.affectivedevelopmentteachertrainingprogram.JPrimaryPrev.1984;4:150-162.DielmanT,ShopeJ,CampanelliP,ButchartA.ElementarySarvelaP.EarlyadolescentsubstanceuseinruralnorthernMichiganschool-basedpreventionofadolescentalcoholmisuse.Pediatrician:IntandnortheasternWisconsin.In:DissertationAbstractsIntemational.JChildAdolescHealth.1987;14:70-76.1984:4601,2000A.UniversityMicrofilmsNo.84-22,327.DielmanT,ShopeJ,LeechS,ButchartA.DifferentialeffectivenessofSarvelaP,McClendonE.Animpactevaluationofaruralyouthdruganelementaryschool-basedalcoholmisusepreventionprogram.JScheducationprogram.JDrugEduc.1987;17:213-231.Health.1989;59:255-262.SchaefferG,MoskowitzJ,MalvinJ,SchapsE.TheEffectsofaDuboisR,HostelterM,Tosti-VaseyJ,SwisherJ.ProgramReportandClassroomManagementTeacherTrainingProgramonFirst-GradeEvaluation:1988-1989SchoolYear.Philadelphia,Pa:CorporateStudents:OneYearFollow-Up.Napa,Calif:TheNapaProject,PacificAllianceforDrugEducation;1989.Unpublishedreport.InstituteforResearch;1981.Unpublishedreport.FlayB,KoepkeD,Thomnson5,Santi5,BestJ,Brown5.Six-yearSchapsE,MoskowitzJ,CondonJ,MalvinJ.Aprocessandoutcomefollow-upofthefirstWaterlooschcolsmokingpreventiontrial.AmJevaluationofanaffectiveteachertrainingprimarypreventionprogram.PublicHealth.1989;79:1371-1376.JAlcoholDrugEduc.1984;29:35-64.SchinkeS,BebelM,OrlandiM,BotvinG.PreventionstrategiesforFlayB,RyanK,BestJ,etal.Cigarettesmoking:whyyoungpeopledoitvulnerablepupils:schoolsocialworkpracticestopreventsubstanceandwaysofpreventingit.In:McGrathP,FirestoneP,eds.Pediatricabuse.UrbanEduc.1988;22:510-519.andAdolescentBehavioralMedicine.NewYork,NY:Springer-Verlag,SchinkeSP,BlytheBJ.Cognitive-behavioralpreventionofchildren's1983:132-183.smoking.ChildBehavTher.1981;3:25-42.FlayB,RyanK,BestJ,etal.AresocialpsychologicalsmokingSchinkeS,GilchristL.Primarypreventionoftobaccosmoking.JSchpreventionprogramseffective?TheWaterloostudy.JBehavMed.Health.1983;53:416-419.1985;8:37-59.SchinkeS,GilchristL,SchillingR,SnowW,BoboJ.SkillsmethodstoGersickK,GradyK,SnowD.Social-cognitivedevelopmentwithsixthpreventsmoking.HealthEducQ.1986;13:23-27.gradersanditsinitialimpactonsubstanceuse.JDrugEduc.Schinke5,GilchristL,SnowW.Skillsinterventionstoprevent1988;18:55-70.cigarettesmokingamongadolescents.AmJPublicHealth.1985;75:665-GilchristL,SchinkeS,TrimbleJ,CvetkovichG.Skillsenhancementto667-preventsubstanceabuseamongAmericanIndianadolescents.IntJSchinkeS,GilchristL,SnowW,SchillingR.Skills-buildingmethodstoAddict.1987;22:869-879.preventsmokingbyadolescents.J.AdolescHealthCare.1985;6:439-444.JohnsonC,GrahamJ,HansenW,FlayB,McGuiganK,GeeM.ProjectSmartAfterThreeYears:AnAssessmentofSixth-GradeandMultiple-ShopeJ,DielmanT,LeechS.Anelementaryschool-basedalcoholGradeImplementations.Pasadena,Calif:UniversityofSouthernmisusepreventionprogram:twoyearfollow-upevaluation.PresentedCalifornia,InstituteforPreventionResearch;1987.UnpublishedattheAmericanPublicHealthAssociationAnnualMeeting;Novem-report.ber1988;Boston,Mass.Sentember1994.Vol.84.No.9AmericanJournalofPublicHealth140