dispp43.com

1PATTERNSOFDEVIANCEUNDERLYINGTHEAGE-CRIMECURVE:THELONGTERMEVIDENCESusanMcVieCentreforLawandSocietyUniversityofEdinburghAbstractThehighprevalenceofdelinquentbehaviourintheteenageyearsiswelldocumented.
Thephenomenonthatistheage-crimecurve,whichtendstopeakinthemidtolateteens,iswidelyagreedtocrossbothjurisdictionalandtemporalboundaries.
How-ever,analysisatsuchanaggregatelevelconcealsimportantunderlyingdifferencesbetweenindividualsandwithindifferentoffencetypes.
Furthermore,shiftsinpreva-lenceratesarenotnecessarilymirroredbysuchconsistentchangesinincidencerates(seeFarrington,1986).
Theplethoraofcross-sectionalstudiescarriedouthavebeenunabletoshedlightonthenuancesofindividualoffendingcareers.
UsinglongitudinaldatacollectedfromthefirstfivesweepsoftheEdinburghStudyofYouthTransitionsandCrime,thispapershallexplorepatternsandtrendsindelin-quentbehaviourofasingleagecohortfrom12to16yearsofage.
Trendsinbothprevalenceandincidenceshallbeexploredinanattempttoexploretherelationshipbetweenthesetwofundamentalaspectsofoffendingbehaviour.
Andsomeexplora-toryworkwillbedonetoidentifydistinctgroupsofoffenderbasedontheirinvolve-mentindelinquencyoverthecourseoftheirearlyteens.
IntroductionTheexistenceoftheage-crimecurvehasbecomeoneoftheleastcontendedissueswithincriminology.
HirshiandGottfredson(1983;andGottfredsonandHirshi,1990)havearguedthatduringalongperiodofchangeanddevelopment,theage-crimecurveremainedconstant–followingacloselysimilarpatternacrossdifferenttimeperiods,amongstdifferentpopulationsandevenbetweenthesexes.
Certainly,manysourcesofofficialcriminaldata–predominantlycautionsandconvictions–haveconsistentlyshowntheretobeasharpinclineinoffendingbehaviourduringearlyadolescence(fromaroundtheageofcriminalresponsibility),peakingduringthemid-lateteenageyearsandthendeclining,steeplyatfirst(tothemid20s)and,thereafter,moresteadily(Farrington,1986;192-5).
Therealityisnotquitesoclearcut.
Despitebroadsimilaritiesinappearance,theubiquitouscurvedoesinfactvaryquitewidelyinspecificterms:thepeakmaybemoreorlesssharp,thepeakagemaybehigherorlower,theskewmaybemoretotheleftorright,andsoon.
Allofthesedifferenceshavedefinitesignificanceintermsofinterpretationandshouldnotbeignored.
Toillustratetheexistenceandformoftheage-crimecurve,Figure1revealsthemostrecentlypublisheddataonthenumberofcautionsissuedinEnglandandWalesdur-ing2002byageband(HomeOffice,2003).
Thefirstpointtonoteisthatthebanding2ismuchnarrowerfortheyoungeragegroupsthantheolder,buttheoverallpatternisunambiguous.
Thecharacteristicpeakemergesbetweenage15and17;however,thereisanothermini-peakaroundage30to39.
Thismaybeanartefactcausedbyaggregationofthedataintoagebands,oritmayreflectasurgeinlateonset.
What-evertheprecisereason,itistruetosaythatsuchodditiestendtogounexplainedbydataanalystssolongasthedominantteenagepeakisevident.
Figure1:NumberofoffenderscautionedinEnglishcourtsin2002,byage0510152025303510to1112to1415-1718-2021-2425-2930-3940-4950-5960+AgewhencautionedNumberin1000'sAtapracticallevel,gettingaccesstothedatanecessarytoanalysesuchtrendsishighlyproblematic.
Publisheddatafromofficialsourcestendtovaryintermsofavailabilityandage-banding.
Forexample,dataoncriminalconvictionspublishedinthesameHomeOfficereportaggregatedeveryoneaged21oroverintoonegroup,makingitimpossibletoidentifytrendsinadulthood.
Onceagain,itseems,solongasthepeakintheteensisevidentthepatternswithintheremainingagegroupsareir-relevant.
Thereareotherfrailtiesintheofficialsourcesofdataweusetoevidencethisrelation-ship.
Twentyyearsofnationalcrimesurveyshaveroutinelydemonstratedthehighlevelofcrimeabsentfromthepolicerecordedcrimestatistics(formostrecentreportsseeSimmonsandDodd,2003andMVA,2002;alsoseeAnderson,1999).
Levelsofattritioninthejusticeprocessmeanthattheproportionofcrimesrecordedforwhomsomeoneiseventuallyconvictedisfractional.
Andconvictionscanrepresentseveralcrimesandoftenoccurlongafterthesewerecommitted.
Thereforerelianceoneitherarrest,cautionorconvictiondataappears,atseverallevels,tobefundamentallyflawed.
Self-reportstudiesarenowwidelyregardedasamoreaccuratesourceofinformationonoffendingpatterns(seeJunger-Tasetal,1994).
Theyprovideapotentiallymuchmorerichsourceofinformationabouttrendsinoffending,yettheyarefarandfewbetween,particularlyintheUK,andanalysisoftheresultshastendedtobesimplisticanddescriptive.
Cross-sectionaldatafromboththe1992youthcrimesurveyreportedbyGrahamandBowling(1995)andthe1998/99YouthLifestylesSurvey(Flood-Pageetal,2000)presentboththeprevalenceandfrequencyofoffendingwithindif-ferentoffencetypes,butfailtoexploretheintricaciesoftherelationshipbetweenthetwo.
AnexceptionistheCambridgeStudyofDelinquentDevelopment,whichhasprovidedafarmoresophisticatedanalysisofthetrendsinoffending(Farrington,31990)andcontinuestodoso;however,thisstudyisratherlimitedinitscoverageandexcludesfemaleoffendingcompletely.
Explainingtheage-crimecurveisdifficultbecausedifferencesreflectbothdevelop-mentalchangewithinindividualsandhistoricalshiftsovertime.
Farrington(1986,1990)distinguishesbetween3effects:'ageeffects'whichoccurwithmovementthroughthelife-cyclee.
g.
maturity,puberty;'periodeffects'whicharehistoricchangesaffectingallindividuals,regardlessofagee.
g.
availabilityofdrugs;and'co-horteffects'thataffectallindividualsofthesameagewhoshareacommonexperi-ence.
Disentanglingtheseeffectsisdifficultsincecross-sectionalstudiesholdperiodconstantbutcannottrackagingorcohorteffects.
Longitudinalstudiesholdthecohortconstantandcantracktheprocessofaging,butcannotexamineperiodeffects(al-thoughthiscouldtoanextentbeachievedthroughacceleratedcohorts).
Inadditiontotheproblemofattemptingtodisentangletheseeffects,wemustrecog-nisethattheage-crimecurveisacumulativeproductofmanydifferentoffencecate-gories,whichinthemselvesdisplaygreatvariationsintheagecrimerelationship.
Soothilletal(2004),forexample,foundthepeakageofconvictionforsomecrimetypes,suchasburglary,tobefarlowerthanothersataround16orless.
Motoringanddrugoffencespeakedbetween21-25beforedeclining;whereasconvictionsforfraudandforgeryalsopeakedbetween21-25,butremainedhighuntil30.
Thoughthepeaksalloccurredwithinyoungadulthood,thepatternsandtrendswereverydiffer-ent.
Therearealsoopponentstothefundamentalprinciplesunderpinningthesocialcon-structionofthecrimerates.
TremblayandNagin(2004)intheirstudyofyouthfulaggression,forexample,believethatifearlymisbehaviourandconductproblemsweretobetakenintoaccount,theage-crimecurvewouldstartfarearlierandpeakfarlesssharplyinadolescence.
Theyassertthatphysicalaggressionpeaksaroundtheageof2,andsubsequentlyshowsdistinctindividualtrajectories,butageneraldown-wardtrend.
Thisissueisnotdevelopedfurtherinthispaper,butitisofmajorimpor-tanceasthistypeofworkrepresentsashiftinemphasisofthedevelopmentalunder-standingofviolence(Fonagy,2003).
Theaimofthispaperistoreviewtwoofthekeydebatessurroundingtheage-crimecurve.
Firstly,thequestionofwhetherthecurveisexplainedbyincreasesinpreva-lence,incidenceorbothAnd,secondly,doestheage-crimecurverequiretobedis-aggregatedusingoffencetypologyoroffendertaxonomy,orisageneraltheoryofcrimesufficientThereafter,usingdatafromtheEdinburghStudyofYouthTransi-tionsandCrime,Iwanttoexploresomeoftheintricaciesoftheage-crimecurvebycomparingprevalenceandfrequencyofoffendingandexaminesomedifferencesinoffenceandoffendertypes.
Keydebatesontheage-crimerelationship:PrevalenceversusincidenceTheage-crimecurveistypicallypresentedasacountofthetotalnumberofcrimescommittedwithinaspecifictimeperiod.
Itmayalsobepresentedintermsofpreva-lenceofoffending,i.
e.
asacountofthenumberofpeoplewithinthepopulationwho4haveoffended(orbeencautionedorconvicted).
Theshapeofthesetwocurvestendstobebroadlysimilar.
Adebateisongoingaboutwhetheritistheprevalenceofoffendersortheincidenceofindividualoffendingthatmostreadilyexplainstheage-crimecurve(seeSmith,2002).
Thedebateisessentiallythis:istheage-crimecurveforthetotalcountofoffencescommitted(ortotalnumberofconvictionsmade)accountedforbytheknownin-creaseinprevalenceofoffendersduringthesametimeperiod,oristhereagreatercausalinfluenceofanincreasedfrequencyofoffendingamongstoffenders.
Sincetheearly80's,GottfredsonandHirshi(1986)havestatedthatthereissimplynopointinstudyingtheindividual'sfrequencyofoffendingbecausecrimeasafunctionofagefollowsa'unimodal'curve,theformofwhichisinvariantovertimeandplace.
However,Blumsteinetal(1988),intheirreviewofdatafromtheGlueck'slongitudi-nalBostoncohort,foundthatthenumberofarrestsperpersonremainedconstantovertime,whereasthetotalnumberofarrestsfell.
Theyconcludedthatthefallinthenumberofarrestsafterthepeakageofoffendingwasduetoadecreaseinprevalencebuttherewasnoreductioninfrequencyamongsttheactiveoffenders.
Farringtonalsofoundthatthepeaknumberofconvictionsamongsthislongitudinalcohortwasprimarilyaffectedbyprevalenceratherthanfrequency,butproblemswithsamplesizeandstatisticalpowermeanttheresultswerenotconclusive.
Nevertheless,hecontendedthattheagecrimecurvereflectedvariationinprevalenceratherthanincidence,astheage-crimecurvesforindividualsdidnotresembletheaggregatecurve,andincidencedidnotchangeconsistentlyoverthecourseofacriminalcareer(seeFarrington,1986).
ReturningtoFarrington'sdata10yearson,NaginandLand(1993)usednewstatisti-calmodellingmethodstounderstandpatternsofconvictions.
Dividingthesampleinto4groups(thoseneverconvictedandadolescentlimited,lowratechronicandhighratechronicoffenders)theyfoundthatthefrequencyofoffendingroseandfellgentlyamongthechronicgroupsbutsharplyamongsttheadolescentlimitedgroupfromage10to30.
Thisledthemtoconcludethatbothprevalenceandoffendingfrequencycontributedtotheoverallpatternofconvictions.
Thebalanceofexistingevidence,then,suggeststhattheage-crimecurvelargelyre-flectschangesinprevalenceofoffendingratherthanfrequency,thereforethesharpriseincrimeratesintheteenageyearsisduetoanincreaseinnewrecruitsratherthananincreasingproliferationofoffendingamongstexistingoffenders.
However,NaginandLandshowedthatthiswastrueforsomeoffendersmorethanothers.
AndBlum-stein(1994)emphasisedthat"itisimportanttorecognisethattheaggregatecrimerateistheresultofthecombinedeffectofparticipationandfrequency".
Unfortunately,thebulkofthedataunderpinningthisdebatearefromofficialsourcesratherthanself-reportedoffending.
AndSampson&Laub(2003)arguethatinci-denceofoffendingvariesgreatlyforsometypesofcrimeandsometypesofoffenderandthatthisischaracteristicoftheenormousheterogeneityinindividualcriminalbe-haviouroverthelife-span.
Thedebatecontinues,butitisclearthatmuchmoreevi-denceneedstobeamassedfromself-reportstudiesontheincidenceofoffendinganditsrelativeimportanceinunderstandingcrimerates.
5OffenceandoffendertypologiesversusageneraltheoryofcrimeTreatingcrimeandcriminalsashomogeneousgroupsmakesanassumptionthat,de-spitedifferencesinnature,therearebasicsimilaritieswhichconnectbehaviourandage.
Atatheoreticallevelthereareargumentsforthisposition(seeSampsonandLaub,2004),yetintuitivelyitishardtoconnectvandalismandpaedophilia,orcom-puterfraudanddomesticviolence,otherthansimplyasbreachesofthecriminallaw.
Francisetal(2004)arguethatthefocusontotalityorfrequencyofoffendingismis-leadingasitencouragestheviewthatpatternsofcrime(orconvictions)arestaticandunchanging.
However,thereareenormousdifferencesinoffencetypesandpatternsofoffending,soatypologyapproachisfarmoreusefulforunderstandingcrimeandit'scauses.
Classificationofcrimeandcriminalsisnotnew–ithasbeencommonsincethe1950's,butstartedevenearlierwithpeoplelikeCesareLombroso.
Thisearlyworkwasconcernedprincipallywithcomparingformsofcriminalbehaviour,usingbothsimpleempiricaldataandafixedtheoreticalapproach.
Morerecentworkhastakenbothadefinitemethodologicalshifttowardstheuseofquantitativedataandadvancedstatisticalmethods,andatheoreticalandsemanticshiftawayfromrigid'typologies'infavourofamoreflexiblenotionofcriminal'pathways'and'trajectories'(SampsonandLaub,1993).
Nevertheless,theresearchresultsrevealsomecommonalities.
Intheearly1970sWolfgangetal(1972)showedthatmostyouthoffenderswereversatile,withonlysomespecialisation.
Inthe1980s,Farringtonandothers(1988)developedamoreanalyticalapproach–usingthe'forwardspecialisationcoeffecient'–toproducere-sultsthatbroadlyagreedwithWolfgang'sfindings.
Butheidentifiedconsiderablevariationbyoffencetype,withgreaterspecialisationinoffencessuchasburglaryandvehicletheft,andlessforvandalism,weaponsandtrespass.
Researchhasgenerallyshowngreaterversatilitythanspecialisationamongstallagegroups,althoughspe-cialisationinsomecrimetypesdoesincreasewithage.
Intermsofoffendertaxonomy,oneofthemostinfluentialauthorsofrecentyearsisTerrieMoffitt(1993),whoclassified2distinctcategoriesofoffender:thelife-coursepersistentoffenderandtheadolescentlimitedoffender.
AccordingtoMoffitt,offend-ingamongstthelifecoursepersistentgroupstartsearlyinlifeandcontinuesthrough-outthelifecourse,butitsformsofexpression,thesocialreactionsitreceivesandthewayitisdealtwith,varyatdifferentstagesofthelifecycle.
Whileadolescentlimitedbehaviourincreasesrapidlyinadolescence,peaksandthendeclines.
Moffittbelievestheage-crimecurveistheresultofsuper-imposingtherateofoffendingovertimeamongstonegroupovertherateofoffendingovertimeamongsttheother,withtheadolescentlimitedgroupbeingalmostentirelyresponsiblefortheteenagepeakHowever,Moffitt'sapproachhasbeencriticisedforbeingtoosimplisticandnotdif-ferentiatingsufficientlybetweenoffendinggroups.
NaginandLand'sanalysisofthe4groups,mentionedearlier,alsofoundthatwhiletheadolescentlimitedgroupdif-feredonkeyaspectsfromlowandhighratechronicoffenders,theyalsosharedkeycharacteristicswiththemanddifferedsignificantlyfromtheneverconvictedgroup.
Moffitt(1997)respondedtothisbyarguingthatthedifferencesbetweenadolescentlimitedandlifecoursepersistentoffenderswerepredominantlyintermsofpsychopa-6thology.
However,Fergussonetal(1996)conceptualisedoffendingbehaviourasacontinuumwithadolescentlimitedatoneextremeandlifecoursepersistentattheother,andconsiderableblurringinthemiddle.
Recentstatisticaladvanceshaveseentheapplicationofasemi-parametricgroupbasedapproachtounderstandingtheunderlyingtrajectoriesofcrime.
Francisetal(2004)appliedtheclusteringcapabilityoflatentclassanalysistotheHomeOfficeOffender'sIndextodescribeoffendingpatternsandinvestigatewithin-individualchangesovertime,allowingmoreadvancedinvestigationofcriminalpathwaysandcrimespecialisation.
Studyingconvictionhistoriesofthe1953birthcohortupto1993(age40),within5yearagebands,Francisetaldemonstratedthatpatternsofoffendingvariedmarkedlybothintermsofdiversityandspecialisation.
Nineoffendingclassesemergedformales,withquitedistinctageprofilesfordifferenttypesofoffending,andspecialisa-tionofoffendingincreasingwithage.
Onlythreeoffendingclassesemergedforfe-males,withnodistinctageprofilesorchangesovertime.
Theyclaimthatthefactthatclustersemergeddemonstratesthatoffendingispatterned–sincetheclustersshowfarstrongerpatternsthanindividualoffencetypes.
Francisetal'sworkdoesprovideevidenceinsupportofanoffencetypologyap-proach.
However,SampsonandLaub(1993,2004)haverejectedthistypeofworkontwoaccounts.
First,theyrejectitonconceptualandpragmaticgrounds.
Theyassertthattheideathatoffencesandoffenderscanbedissectedandpackagedintoneatgroupsisflawedand,inpractice,isimpossibletodo.
Theybelievethefundamentalcausesofcrimearethesameforeveryone,eventhoughtheoffendingpathways(sin-gleormultiple)betweenindividualsaredifferent,thereforeageneraltheoryofcrimeismostappropriate.
Thisisbasedontheassumptionthat,whilethemanifestationsofcrimemaydiffer,thesamegeneralprocessesunderpinthem.
Theyrejecttheasser-tionthatdistinctoffendergroupshavedistinctcausalmechanismsandarguethatmanyofthesetypologiesareidentifiedretrospectively,basedonoutcome,ratherthanprospectively.
Theirsecondobjectioniswiththestatisticalmethodologyitself.
Theyassertthattherobustnessoflatentclassmodellinghasnotbeensufficientlyvalidatedand,indeed,foundconsiderablevariationsintrajectorieswithintheirowndatade-pendingonhowtheydealtwithcertaincharacteristicsofthecohort(Egglestonetal,2004).
Clearlytherearemanyissuesthatremainunresolvedintermsoftheexactreasonsun-derpinningthisubiquitouscurve.
Yet,itisroutinelytakenforgrantedasacrimino-logicalgiven,andcommonlyusedasapoliticaltoolforjustifyingthedevelopmentofcrimecontrolpoliciesandinitiativestargetedattheyoung.
Itishardtoseehowbroadpoliciesandspecificinitiativescanbeeffectiveiftheydonotunderstandsuchsimpleissuesaswhethertheage-crimecurveisexplainedbyprevalenceorincidence.
Thispaperarguesthatitisnotvalidtoexaminethecurveatacumulativelevel,andthatitmustitbedisaggregatedtoitsconstituentpartstofullyunderstandthemulti-fariousnatureofjuvenileoffending,andthusbegintodosomethingaboutit.
7FindingsfromtheEdinburghStudyTheanalysisforthispaperusesdatafromtheEdinburghStudyofYouthTransitionsandCrime,acontemporarylongitudinalstudyoftheoffendingcareersofover4000youngpeopleintheScottishcapital(forafulldescriptionoftheaims,studydesignandmethodsofthisstudy,seeSmithandMcVie,2003).
Theapproachoftheanalysisisexploratoryandsimplistic,incomparisontothemanycomplexstatisticalmodellingapproachesprevalentincriminalcareersresearch.
However,itisusefulindemon-stratingsomeoftheproblemswiththecurrentconceptualisationofcrimetrendsatanaggregatelevel.
Furtherstatisticalmodellingiscurrentlybeingdeveloped.
SowhatmighttheEdinburghStudycontributetothedebateThestudyhascollectedextensivedataonalargecohortofyoungpeople,pro-vidingdetailedanalysis(onprevalenceandfrequencyofoffending)atmanylevels.
Datacollectedincludesabroadrangeofindividual,familial,social,peer-relatedandneighbourhood-relatedcharacteristics.
Theresearchcanprovidecomparisonofbothofficialandself-reportdatatoil-lustratepointsofsimilarityanddifferencebetweenthetwo.
Theresearchhasaprospectivedesignwithlongtermpredictivecapability.
ThedataIwillbepresentingshowstheoffendingpatternsofthecohortacross5ofthe6sweepsconductedtodate,withanalysisrestrictedtothoseindividualswhopar-ticipatedinall5sweeps(n=3523).
Analysisisfocusedontheprevalenceandinci-denceofinvolvementintwelveparticulartypesofoffendingoranti-socialbehaviourwhichrespondentswereaskedaboutatall5sweeps.
1Acountwasmadeofthetotalnumberofincidentsofdelinquencyreportedbycohortmembersateachsweep.
Formethodologicalreasons,aclosedquestionwasaskedtodeterminethenumberoftimesindividualshadoffendedineachway.
Thisgroupedtogetherthosewhohadoffendedbetween6and10times(assumingaconservativeminimumof6),andcappedthemaximumnumberreportedat11forthosewhosaidtheyhaddonethings'morethan10times'.
Thedisadvantageofthisapproachisthatanaccuratecountofthetotalnumberofcrimescommittedbythecohortisnotpossi-ble.
Inanycase,theaccuracyofindividual'sresponsesafteracertainlevelisques-tionable,andanadvantageofourapproachisthatitpreventsspuriousoutliersinthedatafromskewingmeansanddistributions.
I,therefore,presenta'conservativeminimum'totalvolumeofincidentsofoffending,whichismorethansufficienttomeetthecriteriaof'persistence'usedinmostotherstudiesorbyyouthjusticeagen-cies.
Figures2aand2bshowaveryfamiliarpattern:theminimumnumberofincidentsofoffendingreportedbycohortmembersateachofthefirstfivesweepsrisessteadilyfromage12(thefirstsweep)toage13,thensteeplytoage14,thendeclinestoage16.
Figures2aand2bdifferinoneimportantrespect:thefirstdisplaysthemeanvol-umeofoffendingreportedbyindividualoffenders;whilethesecondshowstheper-1Thesewerefire-setting;drivingstolenvehicles;theftfromvehicles;physicalviolence;theftusingviolence;shoplifting;carryingaweapon;breachofthepeace;non-farepayment;housebreaking;van-dalismandgraffiti(thesearedescribedmorefullyinAppendixAofSmithandMcVie,2003).
8centageofcohortmembersreportinginvolvementinatleastoneofthetwelveoffencetypesincludedintheanalysis.
Interestingly,inouranalysis,themeanvolumeofof-fendingappearstoexplainfarmoreofthevarianceintheoverallcountofincidentsreportedthantheproportionofpeopleinvolvedinoffending.
Infact,prevalenceoverthefirstthreesweepsisincrediblystableandthevarianceinthetotalcountappearstobeexplainedsolelybyoffendingfrequency.
Inthelatertwosweeps,prevalenceandincidenceappeartohavemoreofacombinedeffectonthetotalcount,sinceneitherlinematchesthedescentofthetotalcountabsolutely.
Figure2a:Minimumno.
ofincidentsFigure2b:Minimumno.
ofincidentsandmeanvolumeofoffendingandprevalenceofoffendingOurmeasureofprevalencedoesnotshowthetypicalagecrimecurveassociatedwithoffendingprevalence.
Thelevelofkurtosis(orpeak)isfarmoreshallow,suggestingamuchmorestablepictureofparticipationintheearlyteenageyears.
Thisispartlyanartefactofthereferenceperiodforthefirstsweep,whichwasmuchbroaderthanthesubsequentsweeps('ever'ratherthan'thelastyear'),therefore,thenumbersatthispointarenodoubtinflatedtoagreaterorlesserextent(offence-specificanalysispresentedlatershowstheeffectisgreaterforsomecrimetypesthanothers).
Ourpeakageforbothprevalenceandfrequencyofoffendingalsoappearstobesomewhatlowerthansomeotherstudiesofthistype,althoughtheYouthLifestylesSurveydidshowthatoverallprevalenceofoffendingamongstbothboysandgirlspeakedatage14-15(andpeakedagainforboysatage18-21)(Flood-Pageetal,2000).
Lookingatthepictureoverall,thecharacteristicsteepriseinthecrimecountcurveamongsttheEdinburghstudycohortappearstobemorereadilyexplainedbythesharppeakinoffendingfrequencyratherthanprevalenceofoffending,whereas,thedeclineappearstobebasedonacombinationofthetwofactors.
However,thisover-archingprofileconcealsimportantdifferencesinbothprevalenceandfrequencyatthelevelofspecificoffencetypes.
ThereisinsufficientspaceheretolookateverytypeofdelinquentbehaviourcoveredbytheEdinburghStudy,sofocusshallberestricted010203040Uptoage12Age13Age14Age15Age160481216MinimumcountofoffendingMeanvolofoffending(exclnon-offenders)NumberinthousandsMeanvolume010203040Uptoage12Age13Age14Age15Age16020406080MinimumcountofoffendingPrevalenceofoffendingNumberinthousands%ofcohort9tofourwhichdemonstratethepoint.
Theselectedoffencespresentedinfigures3ato3d,below,demonstratefourdifferentgeneraltrendswhichweremirroredbyotheritemsincludedinthesurvey,althoughthisisnotpresentedasanoffencetypology.
Figure3ashowstherelationshipbetweenprevalenceandincidenceofoffendinginself-reportsoffaredodging(definedastravellingonabusortrainwithoutpayingorpayingtoolittle).
Clearly,therehasbeenasteadyincreaseinprevalenceofthisquitetrivial,yetnotuncommon,formofdelinquency,whichhasnotyetpeakedatage16.
Atthesametime,therehasbeenaparallelincreaseinfrequencyoverthesameperiod,whichalmostexactlymatchesthetrendinprevalence.
Inotherwords,notonlyaremoreindividualsinvolvedinthisformofbehaviour,butthosewhoarereportdoingsoincreasinglymoreoftenwithage(possiblyreflectiveoftheirgrowingsocialmobility).
Thepatternforfighting(definedashitting,kicking,punchingorattackingsomeonewiththeintentionofreallyhurtingthem)isquitedifferent.
Figure3bdemonstratesthatboththeprevalenceandfrequencyoffighting,usingthisbroaddefinition,de-clinedquitemarkedlyduringtheearlyteens.
Theapparentdeclinebetweenage12and13islikelytobeanartefactofthelongerreferenceperiodatage12,however,itseemsclearthatthisformofbehaviourplateauedaroundage13to14,thendeclinedsharply.
Asimilarpatternemergesforfrequencyofoffending,althoughthe'peak'ismoreclearlydemarcatedatage14(thesameproblemofreferenceperiodappliestovolumeofoffending),andthedeclineinfrequencyofoffendingisnotnearlysosteepasthedeclineinprevalence.
Thissuggeststhatthosewhocontinuedtobeinvolvedinfightinghadareducedfrequencyofviolentbehaviouroverall.
Someotherformsofoffendingshowedadistinctlydifferentpatterninprevalenceandfrequencyofoffending.
Figure3cshowstheexampleoffire-raising,whichshowedastaggeringincreaseinprevalencebetweenage12and13(despitetheextendedrefer-enceperiodatage12),followedbyafairlysteepdeclineafterage14.
Thetrendinfrequencyshowedamuchmoreconsistentpattern,however.
Therewasanoverallpeakatage14infrequency,buttheevidencesuggeststhatthevolumeofoffendingamongstthosejoiningandleavingthe'poolofoffenders'didnotdifferdramaticallyfromthosewhostayed.
Finally,Iexploretheexampleofweaponpossession(definedascarryingaknifeorotherweaponwithyouforprotectionorincaseitwasneededinafight).
Inthiscase,offendingprevalenceshowedafairlysteepinclinefromage12toage14,whereitpeaked,andthendeclinedrathermoresteadily.
Meanvolumeofoffendingalsoshowedanincreasefromage12to14,butthistrend-linecontinueduptoage16.
Inotherwords,despiteadropinthenumberofpeopleinvolvedinoffending,thenumberofweaponoffencescommittedcontinuedtoincreaseshowingadramaticescalationinindividualbehaviour.
Alloftheotherformsofdelinquencyanalysedforthispaperfellintooneofthesefourbroadcategories;however,therewerediscretedifferencesbetweeneachoffencetype.
So,attheaggregatelevel,prevalencewasmorestablethanvolumeofoffend-ing,leadingtotheconclusionthatincidencecontributesmorehighlytotheoveralltotalityofthese12formsofdelinquencythanprevalenceintheearlyteens.
However,itisclearthattheindividualpiecesofthejigsawarecomplexandthisneatoverall10picturemasksthesedifferentunderlyingtrends.
Forsomecrimetypes,volumeofof-fendingactuallyshowedmorestabilitythanprevalence,whileforothers,therewereremarkablydivergentpatternswhereprevalenceandincidencemovedinopposingdirections.
Figure3a:PrevalenceandmeanFigure3b:Prevalenceandmeanvolumeoffare-dodgingvolumeoffightingFigure3c:PrevalenceandmeanFigure3d:Prevalenceandmeanvolumeoffire-raisingvolumeofcarryingaweapon010203040Uptoage12Age13Age14Age15Age16012345PrevalenceofoffendingMeanvolumeofoffending%ofcohortMeanvolumeofoffending0102030405060Uptoage12Age13Age14Age15Age16012345PrevalenceofoffendingMeanvolumeofoffending%ofcohortmeanvolumeofoffending01020Uptoage12Age13Age14Age15Age16012345PrevalenceofoffendingMeanvolumeofoffending%ofcohortMeanvolumeofoffending0510152025Uptoage12Age13Age14Age15Age160246PrevalenceofoffendingMeanvolumeofoffending%ofcohortMeanvolumeofoffending11SohowconstantisthepopulationofoffendersovertimeIsitthesamepeopleof-fendingconstantlyovertime,whilethenon-offendersremainthesameThinkingaboutMoffit'staxonomy,Iderivedapictureofprevalencewhichdistinguishesbe-tweenoffendersintermsofthenumberofsweepsatwhichtheysaidtheyhadof-fended.
Figure4showsthat1in10cohortmembersdidnotreportoffendingatanysweepofthestudy.
Attheotherendofthespectrumaboutathirdofthecohortre-portedoffendingatall5sweeps.
Asignificantminorityateachsweepsaidtheyhadnotoffended,butwerenotedasbeingoffendersatothersweepsofthestudy.
Afairlysizeableproportioninthemiddleofthechart(variousshadesoforange)reportedof-fendingatafewsweepsofthesurvey–thesemightlooselybedescribedasthe'ado-lescentlimitedoffenders'.
Figure4:Prevalenceofoffendingateachsweep,bynumberofsweepsatwhichoffendingwasreported0%20%40%60%80%100%Uptoage12Age13Age14Age15Age16%ofcohortmemberswhooffendedNeveroffendedNotoffendedthissweepOffended1or2sweepsOffended3sweepsOffended4sweepsOffendedallsweepsOffendedallsweeps,serious&persistentthissweepLookinginalittlemoredetailatthegroupwhohadoffendedacrossallsweeps,wecanidentifyasmallergroupwhoreportedpersistentoffending(inthiscasedefinedasmorethanthemeannumberofincidentsreportedeachyear)includingatleastoneseriousoffence.
Thesemightbedefinedas'highlevelchronicoffenders',andtheycertainlyappeartoshowamazingstabilityintermsofnumbers.
However,thenum-berofthisgroupwhoreportedsuchbehaviourateverysweepfelltojust1.
1%ofthecohort–justaninthofthenumberwhohadbeenpartofthisgroupatanytime(9.
9%ofthecohort).
ConclusionMuchemphasisisplacedontheaggregatepatternsofcrimeprevalenceorfrequencyproducedregularlybyresearchonoffendingbehaviour,totheextentthattheage-crimecurvehasbecomeareassuringiconofcriminologicalresearch.
However,whatremainsrelativelyunder-researchedistheextenttowhichthisisexplainedbychangesintheprevalenceofoffendersorchangesinindividualfrequencyofoffendingisun-clear.
Self-reportdatafromtheEdinburghStudysuggeststhatchangesinoffendingfrequencyhavemostimpactoncrimeratesattheaggregatelevel,sinceprevalenceofoffendingstaysquitestaticovertheearlyteenageyears.
However,disaggregateddata12presentedbycrimetypeshowsenormousdivergenceintermsofbothprevalenceandfrequencyofoffending,provingthatitistoosimplistictoidentifyoneortheotherastheprimarycontributor.
Thecomplexnatureofthepatternsinprevalenceandfre-quencyofoffendingfordifferentoffencetypesimpliesthata'general'theoryofof-fendingisunwise,asittakesnoaccountofoffenceclassificationorattemptstoun-derstandthedifferentialgroupingsofoffenceoroffender.
ContraSampsonandLaub,itappearsessentialthatatypologyisdeveloped,atleasttounderstandjuvenileof-fendingwhichisdiverseandmultifariousinnature.
Tobemosteffective,crimecon-trolpoliciesmusttakeintoaccountthedynamicandmulti-facetednatureofyouthof-fendingdemonstratedhere.
LookingsimplisticallyattheextentofcohortmemberinvolvementinoffendingacrossthefirstfivesweepsoftheEdinburghStudy,Ihaveidentifieddifferentgroupsofindividualwhoclearlydifferintermsoftheircriminaltrajectories,whichstronglysuggeststheneedforsomeformofoffendertaxonomyapproach.
Ofparticularinter-est,inthiseraofpreoccupationwithpersistentoffenders,arethethirdofindividualswhoadmittedoffendingatall5sweeps.
Nevertheless,evenwithinthisgroup,therewasconsiderabledriftintermsofpersistenceandseriousnessofoffending,withonly1%ofthecohortoverallmeetingourdefinitionofseriousandpersistentoffendersacrossall5sweeps.
Inaddition,theoffendersfromourstudywhomightbedescribedas'adolescentlimited'formafarmorecomplexgroupthanMoffittcontendsandre-quiremuchmorescrutiny.
Theyspantherangeofone-offoffenderthroughtospo-radicbutseriousoffender.
Thislendssupporttothepropositionthatthereisacontin-uumofoffendingacrosswhichboundariesareblurred.
13ReferencesAnderson,S(1999)'CrimeStatisticsandthe'problemofcrime'inScotland'inDuff,P.
andHutton,N.
(Eds)CriminalJusticeinScotlandAldershot:Ashgate.
Blumstein,A.
,Cohen,J.
andFarrington,D.
P.
(1988)'Criminalcareerresearch:It'svalueforcriminology',Criminology,No.
26,pp1-35.
Blumstein,A(1994)'Nextstepsincriminalcareerresearch'inWeitekamp,E.
G.
M.
andKerner,H.
-J.
(Eds)Cross-NationallongitudinalresearchonhumandevelopmentandcriminalbehaviourNATOASISeriesD:BehaviouralandSocialSciences–Vol76.
Eggleston,E.
P.
,Laub,J.
H.
andSampson,R.
J.
(2004)'Methodologicalsensitivitiestolatentclassanalysisoflong-termcriminaltrajectories'inJournalofQuantitativeCriminology,Vol.
20,No.
1,pp1-26.
Farrington,D.
P(1986)'Ageandcrime'inTonry,M.
andMorris,N.
(Eds)Crimeandjustice:AnannualreviewofresearchVol7,pp189-250Farrington,D.
P.
,Snyder,H.
N.
andFinnegan,T.
A.
(1988)'Specialisationinjuvenilecourtcareers',Criminology,Vol.
26,No.
3,pp461-487.
Farrington,D.
P.
(1990)'Age,Period,CohortandOffending'inGottfredson,D.
MandClarke,R.
V.
(Eds)PolicyandTheoryinCriminalJustice,51-75,Aldershot:Avebury.
Fergusson,D.
M.
,Horwood,L.
J.
andLynskey,M.
T.
(1993)'Theeffectsofconductdisorderandattentiondeficitinmiddlechildhoodonoffendingandscholasticabilityatage13',Journalofchildpsychologyandpsychiatry,Vol.
34,No.
6;pp899-916.
Flood-Page,C.
Campbell,S.
,Harrington,V.
andMiller,J(2000)YouthCrime:Find-ingsfromthe1998/99YouthLifestylesSurveyLondon:HomeOfficeResearchStudy209Fonagy,P.
(2003)'Towardsadevelopmentalunderstandingofviolence'inBritishJournalofPsychiatry,No.
183,pp190-192Francis,B.
,Soothill,K.
andFligelstone,R.
(2004)'Identifyingpatternsandpathwaysofoffendingbehaviour:Anewapproachtotypologiesofcrime'inEuropeanJournalofCriminologyVol.
1,No.
1pp47-87.
Gottfredson,M.
andHirschi,T.
(1986)'Thetruevalueoflambdawouldappeartobezero:Anessayoncareercriminals,criminalcareers,selectiveincapacitation,cohortstudiesandrelatedtopics'inCriminology,No.
24,pp213-33.
Gottfredson,M.
andHirschi,T.
(1990)Ageneraltheoryofcrime,Stanford,Califor-nia:StanfordUniversityPress14Graham,JandBowling,B.
(1995)YoungpeopleandcrimeLondon:HomeOfficeResearchStudy145Hirschi,T.
andGottfredson,M.
(1983)'AgeandtheExplanationofcrime',inAmericanJournalofSociology,No.
89,pp552-84.
HomeOffice(2003)CriminalStatistics,EnglandandWales,2002TheStationeryOffice(seehttp://www.
homeoffice.
gov.
uk/rds/crimstats02.
html)Junger-Tas,J.
,Terlouw,J.
G.
andKlein,M.
(1994)Delinquentbehaviouramongyoungpeopleinthewesternworld:firstresultsoftheInternationalSelf-ReportDe-linquencyStudyAmsterdam:KuglerMoffitt,T.
E.
(1993)'"Life-coursepersistent"and"adolescent-limited"anti-socialbe-haviour:Adevelopmentaltaxonomy'inPsychologicalReview,No100,pp674-701Moffitt,T.
E.
(1997)'Adolescence-limitedandlife-coursepersistentoffending:Acomplementarypairofdevelopmentaltheories',inThornberry,T.
P.
(Ed)Advancesincriminologicaltheoryvolume7:Developmentaltheoriesofcrimeanddelinquency,pp11-54,NewBrunswickandLondon:Transaction.
MVA(2002)The2000ScottishCrimeSurvey:OverviewReportTheScottishEx-ecutiveCentralResearchUnit:Edinburgh(seehttp://www.
scotland.
gov.
uk/cru/kd01/green/csor-00.
asp)Nagin,D.
S.
andLand,K.
C.
(1993)'Age,criminalcareersandpopulationheterogene-ity:specificationandestimationofanon-parametric,mixedpoissonmodel'inCrimi-nology,Vol.
31,No.
3,pp327-362.
Sampson,R.
J.
andLaub,J.
H.
(2004)'Ageneralage-gradedtheoryofcrime:Lessonslearnedandthefutureoflife-coursecriminology'inFarrington,D.
P.
(Ed)Advancesincriminologicaltheory:Testingintegrateddevelopmental/life-coursetheoriesofof-fendingVol13Sampson,R.
J.
andLaub,J.
H.
(2003)'Desistancefromcrimeoverthelifecourse'inMortimer,J.
T.
andShanahan,M.
J.
(Eds)Handbookofthelifecourse.
NewYork:Plenum/Kluwer.
Sampson,R.
J.
andLaub,J.
H.
(1993)Crimeinthemaking:pathwaysandturningpointsthroughlifeCambridge,MA:HarvardUniversityPressSimmons,J.
andDodd,T.
(2003)CrimeinEnglandandWales2002/2003Research,DevelopmentandStatisticsDirectorate:HomeOffice(seehttp://www.
homeoffice.
gov.
uk/rds/crimeew0203.
html)Smith,D.
J.
(2002)'Crimeandthelifecourse'inMaguire,M.
,Morgan,R.
andReiner,R.
(Eds)TheOxfordHandbookofCriminology3rdedition,Oxford:OxfordUniversityPress15Smith,D.
J.
andMcVie,S.
(2003)'TheoryandmethodintheEdinburghStudyofYouthTransitionsandCrime'inBritishJournalofCriminology,No.
43,pp169-195.
Soothill,K.
,Ackerley,E.
andFrancis,B.
(2004)'ProfilesofCrimeRecruitment'inBritishJournalofCriminologyNo.
44,401-418Tremblay,R.
andNagin,D(2004)'Physicalaggressionduringearlychildhood:tra-jectoriesandpredictors'inPediatrics,Vol.
114,No.
1,pp43-50Wolfgang,M.
E.
,Figlio,R.
M.
andSellin,T.
(1972),DelinquencyinaBirthCohort,Chicago,ILL:UniversityofChicagoPress.