NBER-能否以低成本遏制槍支暴力-來自紐約市一項(xiàng)隨機(jī)實(shí)驗(yàn)的證據(jù)_第1頁
NBER-能否以低成本遏制槍支暴力-來自紐約市一項(xiàng)隨機(jī)實(shí)驗(yàn)的證據(jù)_第2頁
NBER-能否以低成本遏制槍支暴力-來自紐約市一項(xiàng)隨機(jī)實(shí)驗(yàn)的證據(jù)_第3頁
NBER-能否以低成本遏制槍支暴力-來自紐約市一項(xiàng)隨機(jī)實(shí)驗(yàn)的證據(jù)_第4頁
NBER-能否以低成本遏制槍支暴力-來自紐約市一項(xiàng)隨機(jī)實(shí)驗(yàn)的證據(jù)_第5頁
已閱讀5頁,還剩52頁未讀 繼續(xù)免費(fèi)閱讀

下載本文檔

版權(quán)說明:本文檔由用戶提供并上傳,收益歸屬內(nèi)容提供方,若內(nèi)容存在侵權(quán),請進(jìn)行舉報或認(rèn)領(lǐng)

文檔簡介

NBERWORKINGPAPERSERIES

CANGUNVIOLENCEBEDETERREDATLOWCOST?EVIDENCEFROMARANDOMIZEDEXPERIMENTINNEWYORKCITY

OludamilareAboaba

AaronChalfin

MichaelLaForestTucker

LucieParker

PatrickSharkey

WorkingPaper32455

/papers/w32455

NATIONALBUREAUOFECONOMICRESEARCH

1050MassachusettsAvenue

Cambridge,MA02138

May2024

TheauthorsgreatlyappreciatethesupportandpartnershipoftheNewYorkStateDivisionofCriminalJusticeServices(DCJS),NewYorkStateDepartmentofCorrectionsandCommunitySupervision(DOCCS),theManhattanDistrictAttorney’sOffice,andtheUniversityofChicagoCrimeLabincompletingthisstudy.DatausedinthisstudyisprovidedbyDCJSandDOCCS.Theopinions,findings,andconclusionsexpressedinthispublicationarethoseoftheauthorsandnotthoseofDCJSorDOCCS.NeitherNewYorkStatenorDCJSorDOCCSassumesliabilityforitscontentsorusethereof.TheauthorswouldalsoliketothankMeredithStricker,SibellaMatthews,ZacharyHonoroff,JuliaImperatore,andHaysGoldenfortheirinvaluableassistanceandAnthonyBraga,BenjaminHansen,andSaraHellerforhelpfulcommentsonearlierdraftsofthepaper.TheviewsexpressedhereinarethoseoftheauthorsanddonotnecessarilyreflecttheviewsoftheNationalBureauofEconomicResearch.

Atleastoneco-authorhasdisclosedadditionalrelationshipsofpotentialrelevanceforthisresearch.Furtherinformationisavailableonlineat

/papers/w32455

NBERworkingpapersarecirculatedfordiscussionandcommentpurposes.Theyhavenotbeenpeer-reviewedorbeensubjecttothereviewbytheNBERBoardofDirectorsthataccompaniesofficialNBERpublications.

?2024byOludamilareAboaba,AaronChalfin,MichaelLaForestTucker,LucieParker,andPatrickSharkey.Allrightsreserved.Shortsectionsoftext,nottoexceedtwoparagraphs,maybequotedwithoutexplicitpermissionprovidedthatfullcredit,including?notice,isgiventothesource.

CanGunViolencebeDeterredatLowCost?EvidencefromaRandomizedExperimentinNewYorkCity

OludamilareAboaba,AaronChalfin,MichaelLaForestTucker,LucieParker,andPatrickSharkey

NBERWorkingPaperNo.32455

May2024

JELNo.K49

ABSTRACT

Cangunviolencebedeterredatlowcost?Wereporttheresultsofarandomizedexperimentofamessaginginterventionwhichwasdesignedtoreducegunviolenceamongindividualsunderparolesupervisionwithapriorviolentfelonyconvictionorfirearmarrest.Theinterventionconsistedofagroupmeetinginwhichhigh-riskparoledindividualswerenotifiedofthesanctiontheywouldfaceuponreoffendingwhilebeingofferedcommunityresourcestosupportre-integrationintothecommunity.Theprogramdidnotleadtoareductioningunviolenceorcreatecommunityspillovereffectsbutdidreduceparoleviolationsby15%.Potentialmechanismsandimplicationsforsimilarprogramsarediscussed.

OludamilareAboaba

UniversityofChicagoCrimeLab

291Broadway

NewYork,NY10007

daboaba@

AaronChalfin

DepartmentofCriminology

UniversityofPennsylvania

558McNeilBuilding

3718LocustWalk

Philadelphia,PA19104

andNBER

achalfin@

MichaelLaForestTucker

404OswaldTower

UniversityPark,PA16802

mlaforest@

LucieParker

UniversityofChicagoCrimeLab

291Broadway

NewYork,NY10007

laparker@

PatrickSharkey

PrincetonUniversity

patrick.t.sharkey@

1

1.Introduction

IntheUnitedStates,71%ofindividualsleavingprisonarere-arrestedwithinfiveyearsofrelease(Durose&Antenangeli,2021).Despitemorethanfivedecadesofpolicyinnovationandalargeevaluationliteraturefocusedonthispopulation–includingsomeofthefoundationalworkforcetrainingfieldtrialsinlaboreconomics(Lalonde,1986)–today’srecidivismrateremainslargelyunchangedfromthatofthe1970s.Thesocialscienceliterature–includingalargeandgrowingliteratureineconomics–broadlysupportsthedeterrencevalueofinvestmentsinlawenforcement(Nagin,2013;Chalfin&McCrary,2017;Bragaetal.,2019)aswellastheimpactofsocialservicesupportssuchasmentalhealthtreatment(Helleretal.,2017;Jacome,2021),summerjobs(Heller,2016;Davis&Heller,2020;Gelberetal.,2016)andcommunitynonprofitsmoregenerally(Sharkeyetal.,2014).However,fewapproaches,includingtheprovisionofholistic“wraparound”services,haveademonstratedrecordofsuccessindeterringthehighest-riskoffenders–especiallythoseleavingprison–whodriveanoutsizeshareofthesocialcostsof

violence(Doleac,2021;Doleac&LaForest,2022).

Effortstocontrolcrimemustaccommodatethedualempiricalregularitiesthatcrimeisintensivelyconcentratedamongasmallnumberofpeople(MacDonald,2023)andplaces(Weisburd,2015).Theintensityofthatconcentrationsuggeststhatfocusedeffortstodeterorincapacitatethehighest-riskoffendershavethepotentialtogenerateoutsizeandpotentiallyscalabledividends.Withrespecttocrimecontrol,recentevidencehasfoundthattargetedenforcementefforts,whichfocusonparticularhigh-riskindividualsandgroups,canbeeffectiveinreducingviolentcrime(Groffetal.,2015;Bragaetal.,2019;Chalfinetal.,2021;Dominguez,2021).However,whilelaunchinglong-terminvestigationstoarrestthedriversofviolencecanleadtosizableimpacts,theapproachisdifficultandexpensiveandrequiresagreatdealofexpertiseandinvestigativeknow-howtopursueeffectively.Manyofthecitiesthathavelaunchedtaskforcesorsimilarinitiativestoinvestigatehigh-prioritylawenforcementtargetshaveexperiencedfewpublicsafetybenefitsaspoliticalinfightingandalackofresourcesandexpertise

frequentlycannibalizesuchefforts(Kennedy,2019).

Adifferentapproach–onethatisfarlessexpensiveandeasiertoimplement–istousethebullypulpitthatlawenforcementhasinsupervisingindividualsonparoletomaximizethesalienceoflawenforcementinputs(i.e.,deterrence)amonghigh-riskoffenders.Relativetoincapacitation,

deterrenceischeapascrimevictimsavoidthecostsofdeterredcrimesandsocialplannersdonot

2

havetofinancethecostofcourtinvestigationsandincarceration(ChalfinandMcCrary,2017).Evenmoreimportantly,inexpensivedeterrenceinterventionsofferthepromiseofdeliveringimpactsatscale,abarthataconsiderablemajorityofpublicsafetyinterventions,eventhosewhichareeffectiveamongasubsetofthepopulation,areunabletomeetduetohighadministrativecostsandlimitationsinherentinthedifficultyofscalingupcomplexandcontext-specificinterventions

(Davisetal.,2017).

Thisstudyteststhedeterrencevalueofmessaging–arelativelyinexpensiveapproachwhichhasbeenusedtoreducecostsandincreasecompliancewiththecriminaljusticesysteminavarietyofothercontexts(Cumberbatch&Barnes,2018;Fishbaneetal.,2020;Shah&LaForest,2022;Owens&Sloan,2023)–appliedtoahigh-riskgroupofoffendersinNewYorkCity.Theintervention,whichwascarriedoutfrom2013to2014,wasinspiredbytheprinciplesof“focuseddeterrence,”aviolencereductionstrategyimplementedthroughthecollaborativeeffortsoflawenforcementalongwithcommunity-basedsocialserviceorganizations(McDevittetal.,2006;Deckeretal.,2007;Papachristosetal.,2007;Bragaetal.,2014;Bragaetal.,2019).Theapproachbeginswiththeengagementoflawenforcement.High-riskindividualsareidentifiedthroughintelligenceorproactiveinvestigationsandarenotifiedthattheiractivitieshavecausedthemtobecomeapotentiallawenforcementtarget.Individualsarewarnedthatlawenforcementwillbesurveillingthem,withanintensivefocusoncriminalactivitiesthathaveanexustoseriousviolence.Individualsarealsomadeawareofcommunityorganizationsthatcanofferthemsupport

indisengagingfromahigh-risklifestyle.

Focuseddeterrenceisinspiredbyseveralempiricalregularitiesincrimeandviolenceresearch.First,violenceisconcentratedtoaconsiderabledegreeamongasmallnumberofindividualsandgroups–whoaregenerallyknowntolawenforcement–operatinginasmallnumberofcommunitieswithinacity(Weisburd,2015;MacDonald,2023).Theseindividualsincludebothfocalmembersofgangs–whohaveasizableimpactongangviolence–andrecentlyreleasedprisoners–whoaccountforasmanyas20%ofallarrestsforviolentcrimes(Rosenfeldetal.,2005).Second,mostviolenceisinterpersonal–andstemsfromdisputesandarguments–whichcaneasilyspilloverintocommunityviolenceascommunitymembersstepintodefendoneanother.Focuseddeterrencemaythushavetheabilitytoreducethemultipliereffectthatgangs

andcommunitygroupshaveonthesocialcostsofaninterpersonaldispute.

3

Finally,whilemostresearchsuggeststhattheprospectofalongprisonsentencegeneratesonlylimiteddeterrence(Nagin,2013;Chalfin&McCrary,2017),avarietyofevidencefromeconomics,criminologyandothersocialsciencedisciplinessuggeststhatcrime-whichisdisproportionatelydrivenbypresent-orientedoffenders,doestendtoberesponsivetothecertaintyofapprehension(Nagin,2013;Naginetal.,2015).Byengagingdirectlywiththedriversofviolence,successfullyconveyingthemessagethattheyareontheradaroflawenforcement,andsendingthemessagethatlawenforcementisknowledgeableabouttheiractivities,lawenforcementcanpotentiallydoubledownonthedeterrenceimpactsthattheyareabletogeneratewithafixedlevelofresources.Inotherwords,messaging,ifeffective,canincreasetheamountoflaw

enforcementcapital,makingeachunitoflabormoreproductiveintheproductionofpublicsafety.

Quasi-experimentalevaluationsoffocuseddeterrenceprograms,largelyleveragingmatchedcomparisongroups,havegenerallyfoundpositiveimpactsontheoutcomesofindividualswhotakepartintheprogramsaswellaspositivespillovereffectsfornearbyindividualsinthecommunitywhodonottakepart(Bragaetal.,2019;Mearesetal.,2009;McDevittetal.,2006).However,thereisstillsubstantialuncertaintyintheliteratureaboutthepreciseeffectsoftheseprogramsduetothequasi-experimentalnatureofpriorevaluationsandtheprospectfortheretobesubstantialselectionwithrespecttowhichgroupsweredesignatedtoreceivetheintervention

(Bragaetal.,2019).

Todate,therehavebeentworandomizedevaluationsoffocuseddeterrenceinterventionsdeliveredtoadultoffenders:astudyinSacramento,CAbyArieletal.(2019)andanevaluationofaprograminSt.LouisbyHamiltonetal.(2018).TheresearchbyArieletal.(2019)foundthatanintensiveeffortinwhichpoliceofficersvisitedthehomesofpreviously-arrestedindividualstoconveyamessageoffuturesurveillancereducedre-arrests.2Theotherstudy-thatofHamiltonetal.(2018)-waslimitedbyanextremelysmallsamplesizeandwasthushighlyunderpowered

todetecttreatmenteffectsatconventionallevelsofsignificance.3

Thisstudyreportstheresultsofarandomizedexperimentoftheefficacyofdeliveringlow-

2AthirdrandomizedevaluationofmessagingbyDavisetal.(2023)studiedasetoffourforumsdeliveredovera

four-dayperiodtoincarceratedjuvenilesinCookCounty,IL.Thisresearchalsofoundsizableeffectsthoughthereisaccumulatingevidencethatitiseasiertomodifyoffendingamongyouthwhohavenotyetacquiredagreatdealofcriminalcapital,orde-valuedtheirhumancapital,thanadultoffenderswhohavealifetimeofdeficitstoovercome

(Doleac,2021).

3ThesamplesizeinHamiltonetal.(2018)wasverysmall(112)andhadalowtake-uprate(63%),thusmakingitdifficulttoreachquantitativeconclusions.Thestudyfoundnulleffects,albeitwithlargestandarderrors.

4

costandeasy-to-implementdeterrence-basedmessagingtohigh-riskadultoffenders.Critically,ourstudyinvestigatestheeffectsofaparticularlyinexpensiveandlighttouchintervention,whichisrelativelyeasytoreplicatewithfidelityandhasthepotentialtoscalewellacrossjurisdictions,ifeffective.Theinterventionwasintendedtoincreasethesalienceoffuturepenaltiesaswellasemphasizetheavailabilityofsocialsupportsthroughasinglemeetingthatparoleeswererequiredtoattend.Westudythedirecteffectofthemessagingonindividualsand,recognizingthepotentialsalienceofspilloversinplace-basedrandomizedexperiments(MiguelandKremer2004),oncommunity-basedspilloversusingarandomizedexperimentexplicitlydesignedforthispurpose.Byidentifyingspillovers,weareabletoestimatewhethertheinterventionhasanimpactatscaleforthecommunityatlarge,thecriticalestimandforasocialplannerandanestimandthatis

ordinarilyconspicuouslyabsentfromprogrammaticevaluations.

Incontrasttomuchofthepriorliterature,wefindthattheinterventionhadnosignificantimpactonfuturearrestsamongthosereceivingtheintervention,nosignificantspillovereffectstonearbyparoleeswhowerenotrequiredtoattendaforum,andnosignificantimpactonneighborhoodcrime.However,wedofindevidencethattheforumsledtoreductionsinfutureparoleviolationsamongparticipants.Thesereductionswereconcentratedwithinthefirstsixmonthsafterforumattendance,andrepresentareductionofapproximately15%intotalviolationsand25%inabscondingviolationsamongprogramattendees.Thefindingsthussuggestthatwhiletheinterventiondidnothavetheintendedeffectsongunviolenceorothertypesofseriousoffending,themessagingwassuccessfulinmodifyingatleastsomebehaviorsamonghigh-risk

offenders.

2.Background

Study

TheNewYorkStateGunViolenceReductionProject(GVRP)wasimplementedacrossfivesitesinNewYorkStatein2013and2014.Eachmonth,ineachstudysite,recently-releasedparoleeswererequiredtoattendasingle“notificationforum”asaconditionoftheirsupervision.Eachforumwasonehourlongandincluded15-20paroledindividuals.Ateachforum,attendeesheardshortpresentationsdeliveredbyaseriesofspeakersrepresentingvariousarmsoflaw

enforcement,communitymembers,andsocialserviceprovidersfromwithintheattendees’

5

communities.Representativesoflawenforcementexplainedthatthegoaloftheprogramwastocreatesafercommunitiesbyreducingfirearmviolence,sharedrecentincidentsofviolenceinthecommunityandtheconsequencesofthatviolence,andremindedparoleesoftheharshpenaltiestheywouldface,personally,shouldtheyengageinguncrimesandgangviolenceintheircommunity.Thismessageofdeterrencewascomplementedbymotivationalstoriesfromformerlyincarceratedindividualsandencouragementfromcommunitymembersandsocialserviceproviderstomakeon-the-spotconnectionstocareercounseling,drugtreatment,andother

communitysupportservices.

TheNewYorkStateGVRPwasimplementedinfivesitesacrossthestateofNewYork.TwositeswereinAlbanyandSchenectady,mid-sizedcitiesseveralhoursnorthofNewYorkCity.TheremainingthreesiteswereinNewYorkCity,oneeachintheBronx,Manhattan,andBrooklyn.4Individualslivinginthesecommunitieswereeligiblefortheforumsifthey(1)wereonparole,(2)hadbeenconvictedofaviolentfelonyorhadapriorfirearmarrest,(3)didnothaveamentalhealthdiagnosisorapriorsexualassaultcharge,(4)hadlivedwithinthetargetedareaforatleast30consecutivedays,and(5)hadbeenreleasedfromprisonwithinthepriortwoyears.ForumstookplacebetweenAugust2012andJuly2014ineachofthefivesites,withrandomized

assignmentofattendeestakingplacebetweenFebruary2013andJuly2014.

TreatmentassignmentinvolvedtwostagesofrandomizationineachofthethreeNYCstudysites.Inafirststageofrandomization,censustractswithineachareawererandomlyassignedtoa“neighborhoodtreatmentgroup”anda“neighborhoodcontrolgroup.”Individualswholivedwithinthetreatedcommunitieswereeligibletobecalledintotheforums.Individualsinthecontrolcommunitieswerenoteligibletobecalledin.IntheAlbanyandSchenectadysites,noneighborhoodlevelrandomizationoccurredduetothesubstantiallysmallereligiblepopulationsinthesecities.Inthesetwosites,alleligibleindividualswereclassifiedasbelongingtothe

neighborhoodtreatmentgroup.

Inallfivestudysites,eligibleindividualswerethenrandomlyassignedtoan“individualtreatmentgroup”andan“individualcontrolgroup.”Thus,intheNYCstudysites,eligibleindividualsbelongedtooneoffourstudygroupsbasedontheircombinationofneighborhoodtreatmentgroupandindividualtreatmentgroup,asshowninFigure1.Onlyindividualsinboth

theneighborhoodandindividualtreatmentgroup(Group1)wereeligibletobecalledintothe

4SeeAppendixAforfurtherdetailsonthefivestudysites.

6

forums.IntheAlbanyandSchenectadysites,eligibleindividualsbelongedtooneoftwostudygroups–anindividualtreatmentgroupandanindividualcontrolgroup.Inthesesites,all

individualsintheindividualtreatmentgroupwereeligibletobecalledin.

3.Data

Inordertoevaluatetheeffectsofthenotificationforums,wereceiveddeidentifieddataonallstudyparticipantsfromtheNewYorkStateDivisionofCriminalJusticeServices(DCJS)andNewYorkStateDepartmentofCorrectionsandCommunitySupervision(DOCCS)(DivisionofCriminalJusticeServicesandDepartmentofCorrectionsandCommunitySupervision,2020).DCJSprovidedcriminalhistoryforallindividualsinthestudyfromJanuary1990throughJune2019.5Additionally,DCJSprovideddemographicinformationforallindividualsinthestudysuchasbirthdate,sex,race,andethnicity.DOCCSprovidedparoleviolationdataforthestudypopulationfromAugust2012throughJune2019,includingtype(e.g.,anewarrestoratechnicalviolation)anddateofviolation.DOCCSalsoprovidedinformationonwhether,andwhen,each

individualattendedaparoleforum.

Weconsideredtwocategoriesofindividual-leveloutcomesintheevaluation.Thefirstwasarrests,includingboth“allarrests”and“violentfelonyarrests.”6Thesecondsetofoutcomeswasdocumentedparoleviolationsthatleadtoparolerevocationandreincarceration.Thesemeasuresinclude“allparoleviolations”,aswellasindividualsubsetsofviolationspertainingto“violationsduetoanewarrest”,“violationsduetoabsconding”,or“othertechnicalparoleviolations.”7InordertostudytheeffectsoftheforumsonneighborhoodcrimeintheNYCsites,weusedpublicly-availableNewYorkCityPoliceDepartment(NYPD)crimedatafromtheNYCOpenDataportal

(NewYorkCityPoliceDepartment,2019).Thesedataincludedthedateandlocationofallcrime

5Thedatacontainedevent-levelcriminalhistoriesofadultfingerprintablearrests(i.e.,felonyandmisdemeanorcharges)andincludedinformationonthearrestdate,arrestcharges,arraignment-relatedprosecutorialandjudicialactions,andrelevantsentencingdecisions.Thedataincludedbothsealedandunsealedarrestrecords.

6Althoughtheprogramwasestablishedtotargetfirearmviolencespecifically,thereweretoofewweapons-relatedoffensestousethismeasureasanoutcome.

7“Abscondingparoleviolations”occurwhenparoleesdonotreporttotheirparoleofficers,changetheiraddresseswithoutapproval,ormovesuchthattheirwhereaboutsareunknowntotheirparoleofficers.“Othertechnicalviolations”occurwhenparoleesbreakotherindividual-specificconditionsofparolesuchascurfew,restrictionsondrugandalcoholconsumption,orrestrictionsonsocialcontacts,suchthatDOCCSrevokestheirparole.

7

complaints,arrests,andshootingsreportedtotheNYPDbetweenJuly2010andJune2019.8

Complaintandarrestdataincludeallvalidfelony,misdemeanor,andviolationcrimesandarrests.

Thetwostagesofrandomizationprovideassurancethatindividualsinthetreatmentgroupandcontrolgroupshouldbebalancedonbothobservableandunobservablecharacteristics.Table1presentsthemeancharacteristicsofsamplemembersforthefullsamplesusedfortheanalysisbrokendownbytreatmentgroup,aswellastheresultsofbalancetests.Acrossallcharacteristics,includingage,race/ethnicity,gender,numberofarrestspriortoentranceintothesample,numberofpriorarrestswithaweaponscharge,andnumberofpriorarrestsforviolentfelonies,thereisnoevidencetosuggestthatrandomizationresultedinanysignificantdifferencesbetweentheindividual-leveltreatmentandcontrolgroups.9Withrespecttosampledemographics,theaverageageofparoleeswas37andnearlyallweremale.Two-thirdswereBlack,20%wereHispanic,and10%werenon-Hispanicwhite.Despitepreviouslyspendingtimeinprison,theaverageparolee

inthesamplehadbeenpreviouslyarrested11times,ofwhichonaveragetwowereviolentfelonies.

3.ResearchDesign

Leveragingthetwostagesofrandomization,weconductfouranalyses.Thefirstanalysisisusedtoestimatetheeffectsofparticipationintheprogramonindividual-leveloutcomes.Forthisanalysis,theoutcomesofindividualsinGroup1(individualslivingintreatmentneighborhoodswhowererandomlyassignedtotheindividualtreatmentgroup)arecomparedwithindividualsinGroup2(individualslivingintreatmentneighborhoodswhowererandomlyassignedtotheindividualcontrolgroup)acrossallfivestudysites.Ourestimatefortheeffectof

forumattendanceonindividual-leveloutcomeyiisgivenbyYinEquation1:

yi=a+βxi+Y(GTOUP1)i+θb+εi,(1)

?i∈GTOUP1,GTOUP2.

8TheNYPDdatasetsincludedtheXYcoordinateoftheincident,whichwasthenmappedtocensustractinordertodeterminewhethertheincidentoccurredinatreatmentgrouporcontrolgroupneighborhood.0.3%ofcomplaintsaremissinglocationdata,whilenodocumentedarrestsorshootingsaremissinglocationdata.

9Whilethisimpliesthatthereisnoneedtocontrolforthesepotentialdifferencesinourmodels,weultimatelyincludedthesedemographicandcriminalcharacteristicsascontrolsintheregressionstomaximizetheprecisionofourestimatesandadjustforslightimbalancesthatwerepresentacrossthetreatmentandcontrolgroups.

8

In(1),yiistheoutcomeofinterestforindividual,,anindicatorofwhethertheindividualwasarrestedorcommittedaparoleviolationduringthepost-interventionobservationperiod.10xiisavectorofparticipantcharacteristics,includingrace/ethnicity,ageandsquaredage,gender,numberofarrestspriortoentranceintothesample(andsquarednumberofpriorarrests),andthemonthinwhichtheindividualenteredthesample.θbisasetoffixedeffectsfortheblockinwhicheachneighborhoodwasclassified.11Theprimaryvariableofinterestistheindicatorfortheindividual’streatmentstatus,labeled“GTOUP1,”whichtakesavalueofoneiftheindividualisamemberofGroup1andzerootherwise.Controlvariablesinthemodeladjustforanyincidentaldifferencesincharacteristicsofthesamplepopulationthatmightbeassociatedwithoutcomesandareincluded

toimprovetheprecisionoftheestimates.

Ourprimaryresultsestimatethe“intent-to-treat”(ITT)effectofbeinginvitedtoattendaforumasopposedtotheeffectofactuallyattendingaforum.Thismethodprovidesthemostrealisticestimateofhowtheprogramaffectsthosetargetedfortheprogram,inpractice,asnotallthosewhoareinvitedtoaforumwillactuallyattend.FortheNewYorkGVRP,81%ofindividualsassignedtothetreatmentgroupultimatelyattended,while14%ofindividualsassignedtothecontrolgroupattended.Inordertoadjustfortreatmentnon-compliance,wealsocompute“treatmentonthetreated”(TOT)estimateswheretheITTestimateisscaledupbythecomplianceratewiththeintervention.AsshowninAppendixTableA2,TOTestimatesleadtothesame

substantiveconclusionsabouttheeffectsoftheprogramastheITTestimates.

Thisinitialanalysisleveragesthefullsamplepopulationacrossallfiveprogramsitesandprovidesthemostbasicassessmentofwhetherbeingcalledintoattendaforumaffectsanindividual’soutcomes.However,theanalysisreliesontheassumptionthattheforumshavenoeffectonindividualswholiveintreatmentneighborhoodswhoarenotcalledintoattendtheforums—inotherwords,thatthereisnogeographicorsocialspillovereffectthatspreadsbeyondthosewhopersonallyattend.Wethusreporttheresultsofasecondanalysiswhichestimatestheeffectoftheforumsthatislessvulnerabletobiasfrompotentialspillovereffects.Forthisanalysis

theoutcomesofindividualsinGroup1(individualslivingintreatmentneighborhoodswhowere

10Forindividualsinthetreatmentgroup,theoutcomeperiodstartsatthedateofexpectedforumattendance.Forindividualsinthecontrolgroup,itstartsatthedateoftheforumtheindividualwouldhavebeenexpectedtoattendhadtheybeenassignedtothetreatmentgroup.

11Blocksaregroupsofneighborhoodsthatwerecreatedwithinthefiveprogramsites,priortorandomization,andwhichhadroughlysimilarnumbersofeligibleindividuals.

9

randomlyassignedtotheindividualtreatmentgroup)arecomparedwithindividualsinGroups3and4(individualslivinginthecontrolneighborhoods),amongtheNYCstudysites.TheequationforthisanalysisisthesameasEquation1exceptthatthecontrolgroupconsistsofindividualsin

Groups3and4insteadofGroup2:

yi=a+βxi+Y(GToUP1)i+θb+εi,(2)

?i∈(GToUP1,GToUP3,GToUP4)∩(BTonx,BTooklyn,Man?attan).

Becauseindividualsintreatmentgroupneighborhoodswhowerenotcalledintotheforumsareexcludedfromthisanalysis,thereislessofathreatthatadiffusionoftheforum’smessageacross

thecommunitycouldcontaminatetheestimatedprogramimpact.

Thethirdanalysisexplicitlytestsfortheexistenceofcommunityspilloverscausedbytheforums.Thisanalysiscomparesre-offendingamongindividualsinGroup2(individualslivingin

溫馨提示

  • 1. 本站所有資源如無特殊說明,都需要本地電腦安裝OFFICE2007和PDF閱讀器。圖紙軟件為CAD,CAXA,PROE,UG,SolidWorks等.壓縮文件請下載最新的WinRAR軟件解壓。
  • 2. 本站的文檔不包含任何第三方提供的附件圖紙等,如果需要附件,請聯(lián)系上傳者。文件的所有權(quán)益歸上傳用戶所有。
  • 3. 本站RAR壓縮包中若帶圖紙,網(wǎng)頁內(nèi)容里面會有圖紙預(yù)覽,若沒有圖紙預(yù)覽就沒有圖紙。
  • 4. 未經(jīng)權(quán)益所有人同意不得將文件中的內(nèi)容挪作商業(yè)或盈利用途。
  • 5. 人人文庫網(wǎng)僅提供信息存儲空間,僅對用戶上傳內(nèi)容的表現(xiàn)方式做保護(hù)處理,對用戶上傳分享的文檔內(nèi)容本身不做任何修改或編輯,并不能對任何下載內(nèi)容負(fù)責(zé)。
  • 6. 下載文件中如有侵權(quán)或不適當(dāng)內(nèi)容,請與我們聯(lián)系,我們立即糾正。
  • 7. 本站不保證下載資源的準(zhǔn)確性、安全性和完整性, 同時也不承擔(dān)用戶因使用這些下載資源對自己和他人造成任何形式的傷害或損失。

評論

0/150

提交評論