InPartIIweestimateseveralmodelsthataccountforvariationindeviant behaviormeasuredintheninthgradeintermsofconstructsmeasuredinthesev enthandeighthgrades. InChapter3weestimate,forallsubjectswhoprovideddata duringtheseventh,eighth,andninthgradetestings,aseriesoffourincreasingly elaboratedmodelsinwhichanewvariableisaddedtotheimmediatelypreceding modelasplayingcommonantecedentand/orinterveningexplanatoryroles. The vii viii Preface mostinclusivemodelhasbeenpublishedpreviouslyinitspresentform(Kaplan & Johnson,1991);thethreelessinclusivemodelshaveappearedinsomewhatdiffer entform(Johnson &Kaplan,1987;Kaplan,Johnson, &Bailey,1986,1987;Kaplan, Martin, &Johnson,1986). Thepresentversionsdifferfromtheearlierversionsin termsofexogenousvariables,measurementvariables,samplesize,andinclusionof correlateddisturbancesinthemodels. Therepetitionofthemostinclusivemodel andthereestimationoftheotherthreemodelspermitustodescribeandeasilycom municatetheeffectsofeachsuccessivetheoreticalelaborationandtodiscussthe resultsingreaterdetailthanpublicationintheperiodicalliteraturegenerallyper mits. Thereproductionofthemostinclusivemodelalsoprovidesabaseagainst whichwecancomparethemodelsestimatedinthenexttwochapters. InChapters 4and5,respectively,weestimatethismodelforsubgroupsinordertotestassump tionsabouttheequivalenceofmeasurementmodelsandstructuralinvariance betweenthedifferentsocialgroupings. InChapter4weestimatethemodelfor malesandfemales. InChapter5weestimatethemostinclusiveexplanatorymodel forsubgroupsdifferentiatedaccordingtorace/ethnicity. InPartIIIwereportanalysesthataddressthevalidityofthemodelasan explanationofdeviantbehaviorthatismeasuredandexpressedinyoungadult hood. InChapter6theelaboratedmodelasanexplanationofdeviantbehaviorin youngadulthoodisestimatedforthemostinclusivegroupingavailable,white malesandfemalesconsideredtogether. Whenwewereattemptingtoconductsub groupanalysisforgroupsdifferentiatedaccordingtorace/ethnicity,itbecame apparentthat-unlikethesituationinpredictingdeviantbehaviorinearlyadoles cence-themeasuresofdeviantbehaviorusedfortheyoungadultanalyses formedavalidconstructonlyforthewhitesubjects. Themeasuresoftheconstruct werenotsufficientlyvalidorreliabletopermitestimationinasampleofyoung blackadults. Further,forHispanicsubjectsthesamplesizesimplywastoosmall toprovidestableestimatesrelativetothenumberofparametersinthemostcom prehensivemodel. Hence,inordertoconsiderthedifferentialexplanatorypower oftheelaboratedmodelinaccountingfordeviantbehaviorinyoungadulthood andearlyadolescence,respectively,weusedthemodelsforwhitesubjectsas pointsofreference. InChapter7weapplytheelaboratedmodeltotheexplana tionofdevianceinyoungadulthoodformalesandfemalesconsideredseparately. InPartIVwesummarizetheanalysesandofferconclusionsaboutthe significanceofthiswork. Chapter8considerstheusefulnessofthecomprehen sivetheoryinexplainingsomeofthevarianceindeviantbehavioratdifferent pointsinthelifecycleandspecifyingthenatureoftherelationshipsamong explanatoryvariables. Inclosing,weofferillustrativesubsequentstudiesthat buildontheearliertheoryandanalyticstrategies. Acknowledgments ThisworkwassupportedinpartbyresearchgrantsROIDA02497andROI DA04310andbyaResearchScientistAward(K05DAOOI36)tothefIrst-named author,allfromtheNationalInstituteonDrugAbuse. Wearepleasedtorecognizetheunwaveringdedicationof"Sam"McLean, HollyGroves,andJeffersonRogerstothetechnicalproductionofthemanuscript. Wherewerecognizetheoriginsofourideas,weacknowledgethesesources byappropriatetextualcitations. However,manyofourideas-particularlyas thesearesynthesizedinourcomprehensivetheoryofdeviantbehavior-arethe productsoflifetimesofscholarlyactivityandtheprecisesourcesororiginalityof theseideascannolongerbedetermined. Thus,oftenwemustleaveittoothersto makejudgmentsaboutthehistoricalroots,originality,orindependentcreationof thetheoreticalstatementsinthisvolume. ix Contents PART I. THEORETICAL AND METHODOLOGICAL APPROACHES TO THE STUDY OF DEVIANT BEHAVIOR 1. Toward a G
