ORIGINALARTICLE IntimatePartnerViolenceandSubsequentDepression: ExaminingtheRolesofNeighborhoodSupportiveMechanisms EmilyM.Wright 1 • GillianM.Pinchevsky2 • MichaelL.Benson3 • DanaL.Radatz4 Publishedonline:21September2015 SocietyforCommunityResearchandAction2015 Abstract Thisstudyexaminesthedirecteffectsof neighborhoodsupportivemechanisms(e.g.,collective efﬁcacy,socialcohesion,socialnetworks)ondepressive symptomsamongfemalesaswellastheirmoderating effectsontheimpactofIPVonsubsequentdepressive symptoms.Amultilevel,multivariateRaschmodelwas usedwithdatafromtheProjectonHumanDevelopmentin ChicagoNeighborhoodstoassesstheexistenceofIPVand latersusceptibilityofdepressivesymptomsamong2959 adultfemalesin80neighborhoods.Resultsindicatethat neighborhoodcollectiveefﬁcacy,socialcohesion,social interactions,andthenumberoffriendsandfamilyinthe neighborhoodreducethelikelihoodthatfemalesexperi- encedepressivesymptoms.However,livinginareaswith highproportionsoffriendsandrelativesexacerbatesthe impactofIPVonfemales’subsequentdepressivesymp- toms.Theﬁndingsindicatethatneighborhoodsupportive mechanismsimpactinterpersonaloutcomesinbothdirect andmoderatingways,althoughdirecteffectsweremore pronouncedfordepressionthanmoderatingeffects.Future researchshouldcontinuetoexaminethepositiveand potentiallymitigatinginﬂuencesofneighborhoodsinorder tobetterunderstandforwhomandunderwhichcircum- stancesviolentrelationshipsandmentalhealthareinﬂu- encedbycontextualfactors. Keywords Intimatepartnerviolence Depression Neighborhoods Protectivefactors Collectiveefﬁcacy Socialties Introduction Intimatepartnerviolence(IPV)anddepressionaremajor publichealthconcernsthatarebothimpactedbyneigh- borhoodcontext(Bensonetal. 2003;Kim 2008;Mairetal. 2008;Miles-Doan 1998;PinchevskyandWright 2012; Ross 2000).Themajorityofextantresearch,however,has tendedtofocusonthedirectandnegativeimpactof neighborhoodsontheseoutcomes,whilelittleresearchhas exploredthepotentialprotectiveeffectsof positive neigh- borhoodconditions.Forinstance,thereissomeresearch whichsuggestssupportivemechanismssuchassocial cohesion,socialties,orcollectiveefﬁcacymayprotect individualsfromIPVordepressionandothermentalhealth issues(AneshenselandSucoff 1996;Emeryetal. 2011; GeisandRoss 1998;Kirstetal. 2015;Pinchevskyand Wright 2012;RossandJang 2000;Rossetal. 2000;Wright andBenson 2010, 2011),butmostofthesestudiesassess onlyneighborhooddirecteffectsandtreattheseoutcomes & EmilyM.Wright [email protected] GillianM.Pinchevsky [email protected] MichaelL.Benson [email protected] DanaL.Radatz [email protected] 1 SchoolofCriminologyandCriminalJustice,University ofNebraska,6001DodgeStreet,218CPACS,Omaha, NE68182-0149,USA 2 DepartmentofCriminalJustice,UniversityofNevada, LasVegas,4505SouthMarylandParkway, Box455009,LasVegas,NV89154,USA 3 SchoolofCriminalJustice,UniversityofCincinnati, 665HDyerHall,Cincinnati,OH45221,USA 4 DepartmentofCriminology&CriminalJustice,Niagara University,P.O.Box1941,NewYork,NY14109-1941,USA 123 AmJCommunityPsychol(2015)56:342–356 DOI10.1007/s10464-015-9753-8separately.Mostnotablyforthecurrentstudy,verylittle researchhasinvestigatedthepotentialforneighborhood supportivefeaturestomoderateimportantindividual-level relationships,suchastheimpactofIPVonsubsequent depression. Thus,whileweknowthatneighborhoodscandetri- mentallyimpactIPVanddepression(LatkinandCurry 2003;Mairetal. 2008;PinchevskyandWright 2012),we knowlessaboutthepositiveimpactthatneighborhood supportivefeaturesmighthaveontheseproblems.Addi- tionally,wedonotknowwhether,orhow,neighborhood supportivemechanismsmoderatetheimpactofIPVon subsequentdepression.Weattempttoaddtothesparse researchinthisareabyexaminingtherelationshipbetween intimatepartnerviolenceandsubsequentdepressionin neighborhoodcontext.Weusedatafrom2959females livingin80neighborhoodsinChicagotoinvestigatethe directeffectsofneighborhoodsupportivemechanisms, suchascollectiveefﬁcacy,socialcohesion,andinterac- tionswithothers,onfemales’depressivesymptoms,as wellastoexplorewhetherneighborhoodcontextalleviates thelong-termimpactofexperiencingIPVonlater depressionamongtheseadults. IntimatePartnerViolenceVictimization andDepression Scholarshavesuggestedthatvictimsofpartnerviolence maybeparticularlysusceptibletoexperiencingmental healthproblemsintheaftermathofabuseinpartbecause theirvictimization(andtheassociatedtrauma)isinﬂicted bypeoplewhomthevictimtrusts,loves,orconsiderstobe ‘‘safe’’(DeMarisandKaukinen 2005).Alongwithpost- traumaticstressdisorder,depressionisamongthemost prevalentmentalhealthconsequencesofIPV(Caldwell etal. 2012;Campbell 2002),withover45%ofvictims experiencingit(Golding 1999).Althoughdepression amongvictimsmaydeclineifandwhenthevictimization does,long-termeffectsdolinger,butmoreresearchinthis areaisneeded,especiallyamongadultpopulations (Bonomietal. 2006;Campbell 2002;Cokeretal. 2002a; Fletcher 2010;Johnsonetal. 2014). Violencefrompartnerscanbelinkedtodepressionvia mechanismssuchasstress,powerlessness,hopelessness, isolation,lowself-esteem,andphysicalpain(Campbell etal. 1996;Goodmanetal. 2009).Asasourceofstress, abusecancontributetodepressionbydisruptingdaily routines,increasingotherstressfuleventsinone’slife, loweringthevictim’sfeelingsofsecurityandsenseofself- esteem,orincreasingtheirfeelingsofpowerlessnessto controlthesituation(Campbell 2002;Campbelletal. 1996; Goodmanetal. 2009).Forinstance,becauseofthephysical painandinjurythatisoftenaconsequenceofIPV,victims maydevelopchronicpainorsufferconsequencesofthe abuse(e.g.,episodesoffaintingcausedbyextensivehead trauma;sexuallytransmitteddiseases,etc.,seeCampbell 2002),whichmayleadtodepressionovertime.Inaddition, socialisolationfromothers,whichisoftenpurposefully achievedbyabusersinviolentrelationships(e.g.,Campbell etal. 1996),canalsoleadtodepressionbecauseisolated victimsmayfeelthattheyhavenoonetoturntoforhelpor support.WhiletherearemanyavenuesbywhichIPVmay leadtodepressionandothernegativementalhealthdis- positions,fewstudieshaveexaminedsuchrelationships withlongitudinaldataamongadults,andmoreassessment isneededinthisarea(Campbell 2002;Fletcher 2010). Moreover,researchhasneglectedtoexaminethisrela- tionshipfromacontextualperspective. TheImportanceofNeighborhoodContext Neighborhoodcontextisrelatedtodepressionandother problematicmentalhealthoutcomesamongbothadults (e.g.,Kim 2008;Mairetal. 2008)andchildren(e.g.,Xue etal. 2005).Theliteratureregardingtheneighborhood predictorsofdepressiontendstocenteronsocioeconomic, poverty-related,orsocialdisordervariables,andgenerally demonstratesthatlowersocioeconomicstatus(SES)and moreneighborhooddisorderdetrimentallyimpactdepres- sionandothernegativementalhealthoutcomes.Mairetal. (2008)foundthat,of45studiesonneighborhoodsand depression,37demonstratedsigniﬁcantneighborhood effects,evenaftercontrollingforindividual-levelfactors suchasage,race,gender,andindicatorsofsocioeconomic status;inanotherreview,Kim(2008)foundthat11of22 studiesexaminingneighborhoodsocioeconomicstatus showedthatneighborhoodconditionshadsigniﬁcant effectsondepression.Mechanismsdirectlylinkingneigh- borhoodfactorstomentalhealthordepressionoften revolvearoundincreasedstresslevels,limitedaccesstoor scantresources,disorder,violence,inadequatehousing, andalackofpublicaccessorgreenspaces,suchasbike lanesandcleanstreetsorsidewalks(e.g.,Kim 2008;Mair etal. 2008;TaylorandRepetti 1997).Accumulatingevi- dencealsosuggeststhatperceiveddisorderand/ordan- gerousnessoftheneighborhoodincreasesdepression, perhapsinpartduetoincreasedfearormistrustamong neighbors(e.g.,RossandJang 2000;RossandMirowsky 2009),socialisolation(e.g.,GeisandRoss 1998;Rossand Mirowsky 2009),feelingsofpowerlessness(e.g.,Anesh- enselandSucoff 1996;GeisandRoss 1998),orfearand anxietyamongstneighbors(e.g.,Hilletal. 2005).Inshort, unkempt,rundown,disorderly,anddisorganizedneigh- borhoodenvironmentscanbedemoralizing,distressing, threatening,andisolatingtoresidentsandcansendthe AmJCommunityPsychol(2015)56:342–356 343 123messagethatnoonecaresabouttheirwellbeingorthatof theneighborhood(RossandMirowsky 2009;Sampson 2013;WilsonandKelling 1982).Allofthesefactorsmight, inturn,contributetodepression. Despitethedeleteriouseffectsofneighborhoodson negativeoutcomes,itisalsopossiblethatneighborhoods may mitigate theeffectsofstressors—includingviolence suchasIPV—byorganizingcommunityresidentstogether andfosteringsocialcohesionand/orinterconnectedness amongthem.Thereissomeevidencesupportingneigh- borhoodprotectiveeffectsagainstmentalhealthoutcomes, butthisbodyofresearchprimarilyfocusesonneighbor- hoodcohesionandcollectiveefﬁcacy.Intheirreview,Mair etal.(2008)found11studiesthatreportedthatpositive interactionsbetweenneighbors—includingincreasedsocial connectionsandsupportfromresidents,increasedcom- municationbetweenresidents,andreducedsocialisola- tion—servedasprotectivefactorsagainstdepression.Ross andcolleagues(KimandRoss 2009;RossandJang 2000) havedemonstratedthatneighborhoodsocialtiesandsocial supportreducedepressionandotherpsychologicaldistress, includingperceivedpowerlessness,fear,andmistrust(see alsoKim 2010).Neighborhoodsocialcohesionhasalso beenlinkedtoreduceddepression(Mairetal. 2010);Ahern andGalea(2011)suggestthatsocialcohesionwithina neighborhoodmayhelpresidentsexertsocialcontrol, whichcanprovidemoresupportbetweenneighbors,reduce thenumberofstressorsthatresidentsperceivetoexist withintheneighborhood,andbufferormitigatethestres- sorswhichdooccur.Further,theynotethatcohesion betweenresidentsmayfostercommunication—whichcan keepresidentsknowledgeableabouttheircommunity— improvethelocalservicesandresourcesthatareavailable, andincreaseapersonalsenseofcontrol,whichinturn,can alleviatedepressivesymptoms(Staffordetal. 2011).Areas whereresidentsknoweachother,communicate,and interactare,infact,perceivedtofosterbettermentalhealth amongresidents:forinstance,Torontoresidentsperceived conceptssuchas‘‘interactionbetweenneighbors,’’ ‘‘knowingneighbors,’’‘‘socialcohesion,’’and‘‘communi- cationbetweenresidents’’tobeimportantpiecesof neighborhoodsupportwhichwerethoughtberelatedto positivementalhealth(Burkeetal. 2009).Indeed,Samp- son(2013)suggeststhatresidentsinneighborhoodsofhigh collectiveefﬁcacy(i.e.,wheretherearesharedexpecta- tionsaboutnormsandbehavior)arebetterabletoachieve commongoalsandaremorelikelytoengagein(and beneﬁtfrom)sociallyaltruisticbehaviors. 1 Suchbeneﬁts areunsurprisinglyassociatedwithbettermentalhealth— collectiveefﬁcacy,forinstance,hasbeenfoundtoreduce depressionamonglong-termLatinoimmigrantsinthe UnitedStates(Vegaetal. 2011)aswellasamongolder adults(AhernandGalea 2011),andisassociatedwithother sociallyaltruisticbehaviorssuchasbystanderintervention (Edwardsetal. 2014). Itisnecessarytoaccountforneighborhooddisadvantage whenexaminingtheimpactofneighborhoodprotective factors,however,becausedisadvantagecanimpedethe qualityandquantityofthehelpingmechanismswhichare available(Goodmanetal. 2009;Sampson 2003).Inother words,themechanismsdescribedabovethatmightmiti- gatedepressionamongresidentsmaybelesslikelytobe presentindisadvantagedneighborhoods.Forinstance, scholarshavesuggestedthattheformationandextentof socialtiesbetweenresidentsmaybelowerindisadvan- tagedareas(seeKornhauser 1978;ShawandMcKay 1942) potentiallybecauseoftheirfear,mistrust,andperceptions ofdangerousness(RossandMirowsky 2009;Rossetal. 2001).Havingfeworweaksocialtiesbetweenneighbors mightreducehelp-seekingbehaviorsamongvictimsof partnerviolence. Additionally,someoftheliteratureonneighborhood protectiveeffectsondepressionandmentalhealthout- comeshasbeenlimitedinotherways.Forinstance,data thatisreportedbyparticipantsbutaggregatedtothe neighborhoodlevel(e.g.,Echeverriaetal. 2008;Mairetal. 2010;Riosetal. 2012)risksconﬂatingoutcomeswith predictorvariables(i.e.,neighborhoodmeasures).While theresearchbasehasexaminedbothneighborhood-level tiesandsupports,therehavealsobeenstudieswhichfocus ontiesandsupportivemechanismsattheindividual-level onlyoruseaggregatedmeasures(e.g.,Echeverriaetal. 2008;GeisandRoss 1998;Riosetal. 2012;RossandJang 2000).Further,manystudieshaveutilizedcross-sectional designs(Mairetal. 2008),whichcannotascertainwhether neighborhoodsupportivemechanismsprecedeincreased depressionorwhetherpeopleseekoutsupporttohelpwith theirdepression.Mostimportantlytothecurrentstudy, researchhasneglectedtoexaminetheeffectsofneigh- borhoodsupportivemechanismsontheIPV—depression relationshipspeciﬁcally. NeighborhoodSupportsMayMitigatetheEffect ofPartnerViolenceonLaterDepression WeexaminewhethertherelationshipbetweenIPVand subsequentdepressiondependsinpartupontheneighbor- hoodinwhichitoccurs.Basedonpreviousresearchand theory,wemightexpectthatneighborhoodsupportive mechanisms,suchasstrongsocialcohesion,collective efﬁcacy,orsocialnetworksbetweenresidentswillboth 1 However,thebeneﬁcialeffectsofcollectiveefﬁcacyonvarious outcomesmaybelimitedorhinderedwhennormssuchaslegal cynicismaretakenintoaccount(Emeryetal. 2011;KirkandMatsuda 2011). 344 AmJCommunityPsychol(2015)56:342–356 123directlydecreasedepressionandalleviatetheeffectsofIPV ondepression,despitethelevelofdisadvantagethatexists withinaneighborhood.Althoughwedonotdirectlytest thesemechanisms,wesuspectthattheseneighborhood protectivefactorswillincreasethesupportthatisavailable fromotherstothevictim,leadingtolowerdepressionandan attenuatedimpactofIPVonthisoutcome.Indeed,Kirstetal. (2015)foundthatTorontoresidentswithhighperceived socialsupportandlargersocialnetworkstodrawuponwere lesslikelytoexperienceIPV.Wright(2012)suggestedthat victimsofpartnerviolencemayrelyonsocialsupportfrom othersforemotionalandﬁnancialhelpwhenabuseoccurs, andthatthesesupportsystemsmayaidethevictiminleaving theabusiverelationshiptemporarilyorpermanently.In addition,supportivenetworkswithothersmayprovidevic- timswithadviceregardingsafeplacestogoortheservices thatareavailabletotheminordertoescapetheviolence (HadeedandEl-Bassel 2006;Moe 2007;Wright 2012).In short,neighborhoodsupportivemechanismsmayprovide optionsforvictimstorelyonwheninneed.Second,we believethatcommunicationwithresidentsintheneighbor- hoodmayincreasethelikelihoodthatvictimstellothers abouttheirvictimization(Browning 2002),andthisopen communicationmayhelpvictimscope(Cokeretal. 2002b) aswellasincreasethelikelihoodthatotherswilllearnabout thevictimizationandpotentiallyintervene(Wright 2012; WrightandBenson 2011).Havingmoreavenuesavailableto victimsforcopingwiththeviolenceorseekinghelpto alleviatetheabusemayreducethestresscausedbythe victimization.Finally,supportiveneighborhoodsmay reduceresidents’feelingsofisolationand/ormistrust,which inturnmayreducevictims’feelingsofpowerlessness. Weexaminethesepossibilitiesinthecurrentstudyand focusonthreeresearchquestions:(1)whatarethelong-term effectsofexperiencingintimatepartnerviolenceonthe likelihoodthatfemaleslaterreportdepressivesymptoms, whilecontrollingforotherimportantindividualandneigh- borhoodcovariates?(2)whatarethedirecteffectsof neighborhoodsupports(e.g.,collectiveefﬁcacy,social cohesion,andsocialnetworkswithothers)onfemales’ depressivesymptoms,accountingforindividualcovariates andneighborhooddisadvantage?and(3)dotheseneigh- borhoodsupportivemechanismsalleviatetheimpactofIPV ondepressivesymptoms3yearslater,aftercontrollingfor relevantindividualandneighborhoodcovariates? Methods Data DataforthisstudyweredrawnfromtheProjecton HumanDevelopmentinChicagoNeighborhoods (PHDCN;Earlsetal. 2002),amulti-component,multi- wavestudywhichallowsfortheexaminationofhow neighborhoodcontextmayimpactindividual-levelout- comes.Thisstudyreliesondatafromthreeseparate componentsofthePHDCN:theLongitudinalCohort Study(LCS),theCommunitySurvey,andthe1990U.S. Census.PHDCNresearchersdividedChicago’s847cen- sustractsinto343geographicallycontinuousneighbor- hoodclusters(NCs).These343NCswerethenstratiﬁed bysevencategoriesofracial/ethnicdiversityandthree levelsofsocio-economicstatus,and80NCswerethen selectedviaprobabilitysampling.Withinthese80NCs, multiplecohortsofeligibleyouthandtheirprimary caregivers(93%ofwhomwerefemale)wereselectedfor inclusionintheLCS.Thecurrentstudyutilizesdata collectedfromfemalecaregivers(hereafterreferredtoas respondents)duringtheﬁrst(1994–1997)andsecond (1997–2000)wavesoftheLCS. Neighborhoodmeasuresrelatedtocollectiveefﬁcacy, cohesion,andsocialinteractionsortiesweretakenfrom thePHDCNCommunitySurveywhilemeasuresfor neighborhooddisadvantagewereabstractedfromthe1990 UnitedStatesCensus. 2 TheCommunitySurvey(con- ductedin1994–1995)sampledresidentsfromtheoriginal 343NCsandaskedabouttheirperceptionsofneighbor- hoodsocialprocesses,includingsocialnetworksand interactionsbetweenresidents,organizationalgroups, values,andsoforth.Usingathree-stagesamplingdesign, cityblocksweresampledwithineachNC,dwellingunits weresampledwithinblocks,andoneadultresidentwas sampledwithineachdwellingunit.Thus,participantsof theCommunitySurveywerelargelyindependentofthose drawnfortheLCS. 3 Thecurrentstudyincludes2959 femalerespondentslivingin80neighborhoodclusters (hereafterreferredtoasneighborhoods)whoweremar- ried,cohabitating,orinadatingrelationshipatwaveone andansweredquestionsrelatedtodepressivesymptomsat wavetwo. 4 2 StaffattheInter-universityConsortiumforPoliticalandSocial ResearchcalculatedNC-linkedU.S.Censusmeasuresinorderto ensuretheconﬁdentialityofthesubjectsofthePHDCN. 3 AlthoughtheCommunitySurveycollectedinformationfromall 343NCsaboutneighborhoodconditionsviainterviewswiththese residents,thisstudyfocusesonlyonthose80NCsinwhichthe participantsoftheLCSwerenested. 4 Toarriveatthe2959,atotalof292casesweredeleteddueto missingdata.Theonlysigniﬁcantdifferencebetweenouranalysis sampleandtheeligiblesampleoffemalecaregiversinarelationship wasthatouranalysissamplehadslightlyfewerHispanicwomen (p\.05).Therewerenosigniﬁcantdifferencesonthemain independentvariableofinterestoranyothercontrolvariables. AmJCommunityPsychol(2015)56:342–356 345 123Measures DepressionSymptoms Thirteendepressionmeasureswereadaptedfromtheshort formoftheCompositeInternationalDiagnosticInterview (UM-CIDI;Kessleretal. 1998)andwereself-reportedby thefemalerespondentsatwavetwo.Respondentswereﬁrst askedquestionsregardingwhethertheyhadfeltsad,blue, ordepressedfortwoormoreweeksinarowduringthe pastyear.Ifrespondentsansweredafﬁrmatively,theywere askedfollow-upquestionsregardingtheperiodoftime theyexperienceddepressivesymptomology(i.e.,during the2-weekperiodwhentheyfeltsadorblue,didtheylose interestinmostthings;feeltiredorhavelowenergy; experienceaweightchangeof10ormorepounds;have troublefallingasleeporconcentrating;experiencefeelings ofworthlessness;thinkaboutdeath).Respondentswere alsoaskedthesamefollow-upquestionsregardingany 2-week(ormore)periodoftimeinthepastyearwhenthey lostinterestinmostthingslikehobbies,work,oractivities thattheyusuallyenjoyed.This13-itemscalehadarelia- bilityof.902. Topredicttheoddsofexperiencingdepressivesymp- toms,weusedamultivariate,multilevelRaschmodel (Raudenbushetal. 2003).Thethree-levelmodelnests depressionitemresponseswithinpersonswithinneigh- borhoods.Thelevel-1model(itemswithinpersons)pro- ducesalatentvariablethatrepresentseachperson’s susceptibilityfor depressionsymptoms (i.e.,theirlikeli- hoodofexperiencingdepressivesymptoms).Thiscontin- uousvariableisassumedtobenormallydistributedona logitmetricandistheoutcomeforthelevel-two(respon- dent-level)andlevel-three(neighborhoodlevel)models (Osgoodetal. 2002). IntimatePartnerViolence SevereIPV wasassessedusingtheConﬂictTacticsScale forParentsandSpouses(Straus 1979)andreﬂectsthe prevalenceofseveremale-partner-perpetratedaggression againstthefemalerespondentinthepastyear.Respondents reportedwhethertheirmalepartnerusedanyofsixforms ofsevereaggressionorviolenceagainsttheminthepast year:hitwithaﬁst,hitwithsomething,beatup,choked, threatenedwithaknifeoragun,orusedaknifeoragun (0 = no;1 = yes).Thismeasurereﬂectspartnervictim- izationatwaveone. NeighborhoodVariables Drawingfrompriorresearch(Cerdaetal. 2008;Molnar etal. 2004, 2008), concentrateddisadvantage was operationalizedasaprincipalcomponentsfactoranalysis includingthepercentageofresidentsinaneighborhood whowerelivingbelowthepovertyline,receivingpublic assistance,andunemployed(alpha = .805).Highervalues reﬂectgreatereconomicdisadvantage.Becauseofthe possibleconfoundingeffectsofdisadvantageonneigh- borhoodsocialsupportmechanisms,concentrateddisad- vantageislargelyincludedasacontrolmeasure. 5 Weusetwomeasuresthatreﬂectthedegreeoftrustand supportbetweenneighborsinthecommunity.Following Sampsonetal.(1997),neighborhood collectiveefﬁcacy wasbasedon10itemsfromadultsparticipatinginthe CommunitySurveyandreﬂectsthedegreeofsocial cohesionandinformalsocialcontrolbetweenneighbors. Residentswereaskedhowstronglytheyagreed(onaﬁve- pointLikertscalerangingfrom‘‘stronglydisagree’’to ‘‘stronglyagree’’)that:peoplearoundherearewillingto helptheirneighbors;thisisaclose-knitneighborhood; peopleinthisneighborhoodcanbetrusted;peopleinthis neighborhoodgenerallydon’tgetalongwitheachother (reversecoded);andpeopleinthisneighborhooddonot sharethesamevalues(reversecoded).Residentswerealso askedﬁveitemsregardingthelikelihood(assessedona ﬁve-pointLikertscalefrom‘‘veryunlikely’’to‘‘very likely’’)thattheirneighborswouldinterveneif:children wereskippingschoolandhangingoutonastreetcorner; childrenwerespraypaintinggrafﬁtionalocalbuilding; childrenwereshowingdisrespecttoanadult;aﬁghtbroke outinfrontoftheirhouse;andtheﬁrestationclosestto theirhomewasthreatenedwithbudgetcuts.Following Sampsonetal.(1997)andothers(Browningetal. 2004; Morenoffetal. 2001),theseitemswerecombinedintoa singlemeasureofcollectiveefﬁcacyusingathree-level itemresponsemodel. 6 Thelevel-onemodeladjustedthe within-personcollectiveefﬁcacyscoresbyitemdifﬁculty, missingdata,andmeasurementerror.Thelevel-twomodel estimatedneighborhoodcollectiveefﬁcacyscoresadjusting forthesocialcompositionofeachneighborhood.Inpar- ticular,potentialbiasesinperceptionsofeachconstruct 5 Wecontrolforneighborhooddisadvantageinmultilevelanalyses because,relativetootherstructuralconditionssuchasresidential mobilityorethnicheterogeneity,disadvantagehasbeenfoundtobe themostconsistentinﬂuenceonbothIPVanddepression(Kim 2008; Mairetal. 2008;PinchevskyandWright 2012). 6 Itemresponsemodelingtechniquesavoidthelossofdatafrom missingresponsestoasetofquestionsorindicators(Osgoodetal. 2002),takeitemdifﬁcultyintoaccount(i.e.,thatsomeindicatorsof neighborhoodconstructsmaybemoredifﬁcultandlessprevalentthan others),andallowsimultaneousestimationoftheimpactofindivid- ual-levelinﬂuences(e.g.,age,gender)onperceptionsofthese constructs(Sampsonetal. 2005).Theitemresponsemodelsusedin thisstudyultimatelyprovidetheneighborhood-levelofcollective efﬁcacy(or,socialcohesion,orsocialinteraction)aftertheseissues havebeenaccountedfor. 346 AmJCommunityPsychol(2015)56:342–356 123resultingfromcharacteristicsrelatedtogender,marital status,homeownership,ethnicityandrace,residential mobility,yearsintheneighborhood,age,andsocioeco- nomicstatuswerecontrolledatlevel-two.Finally,the level-threemodelallowedeachneighborhoodcluster’s meancollectiveefﬁcacyscoretovaryrandomlyarounda grandmean.TheempiricalBayesresidualfromthelevel- threemodelconstitutestheneighborhoodlevelofcollec- tiveefﬁcacyaftercontrollingforitemdifﬁcultyand neighborhoodsocialcompositionandwasthereforeusedas the‘true’neighborhoodscoreoncollectiveefﬁcacy.The internalconsistencyofthisscaleattheneighborhoodlevel was.847. Ithasbeensuggestedthatcohesionbetweenresidents mayuniquelyimpacttheirmentalwell-being(Stafford etal. 2011),andKirstetal.’s(2015)ﬁndingssuggestthere areuniquecontributionsofdifferentformsofneighbor- hoodsocialcapital(e.g.,collectiveefﬁcacy,perceived socialsupport,individualnetworksize)withregardtoIPV. Therefore,wewishedtoseparateouttheeffectsofsocial cohesionfromsocialcontrol,asbotharecapturedinthe overallcollectiveefﬁcacymeasuredescribedabove. 7 We createdaseparatemeasureof socialcohesion,apartfrom thecollectiveefﬁcacymeasure,andmodeledthemsepa- ratelyintheanalyses.Socialcohesionwasbasedonﬁve itemsaskedoftheCommunitySurveyparticipants;resi- dentswereaskedhowstronglytheyagreed(from‘‘strongly disagree’’to‘‘stronglyagree’’)that:peoplearoundhereare willingtohelptheirneighbors;thisisaclose-knitneigh- borhood;peopleinthisneighborhoodcanbetrusted; peopleinthisneighborhoodgenerallydon’tgetalongwith eachother(reversecoded);andpeopleinthisneighbor- hooddonotsharethesamevalues(reversecoded).Inde- pendentlyfromcollectiveefﬁcacy,socialcohesionwas alsooperationalizedusingathree-levelitemresponse model.Asdescribedabove,theempiricalBayesresidual fromthethree-levelmodelconstitutestheneighborhood scoreonsocialcohesionaftercontrollingforsocialcohe- sionitem-difﬁcultyandneighborhoodsocialcomposition (e.g.,respondentcharacteristicssuchasage,race,etc.).The neighborhood-levelinternalconsistencyofthisscalewas .826. Oneitemtappingthelevelofsocialinteractionbetween neighborswasalsoincludedinthisstudy.Adultspartici- patingintheCommunitySurveywereaskedfourquestions abouttheir socialinteractions withotherneighborhood residents.Respondentswereaskedhowoften(onafour- pointLikertscale,from‘‘never’’to‘‘often’’)theyand peopleintheneighborhood:dofavorsforeachother;ask eachotheradviceaboutpersonalthings;haveget-togethers whereotherpeopleintheneighborhoodareinvited;and visitineachother’shomesoronthestreet.Similartothe otherneighborhoodsocialprocessesvariables,athree-level itemresponsemodelwasusedtocreatethesocialinter- actionsscale(seeBrowningetal. 2004).Theinternal consistencyofthisscaleattheneighborhoodlevelwas .734. Finally,threemeasureswereincludedtoassessthe existenceoffriendandfamilysocialtiesintheneighbor- hood.ThesemeasureswerealsotakenfromtheCommu- nitySurvey. Anyfriends and anyfamily werecomputed basedonresidents’answerstowhethertheyhadany friendsandanyrelativesorin-laws,respectively,livingin theirneighborhood(excludingthosefriendsorfamily memberswithwhomtheylived).Theseresponseswere thenaggregatedtotheneighborhood-level.Finally,the averagenumberoffriendsandfamily representsthe numberoffriendsandfamily/in-lawslivingintheneigh- borhood(eachonascaleof0–4,indicatingnoneto10or morefriendsorfamily/in-laws)averaged,andaggregated totheneighborhood-level. ControlVariables Additionalrespondent-levelfactorsdemonstratedinprior researchtobeassociatedwiththeoddsofdepressionand intimatepartnerviolencewerealsoincludedinanalyses (Stithetal. 2004).Suchvariablesincludedemographic predictors(age,race/ethnicity,married[versusdatingor cohabitingwithapartner],householdsalary,andemploy- mentstatus),opportunityvariables(isolation),andprior lifehistories(stressfullifeevent,priordepression). Importantly,priordepression(ayes/noindicatorself-re- portedbytherespondent)wasassessedatwaveone,while thedepressivesymptomsoutcomewasassessedatwave two,whichallowedustocontrolforthetemporalordering betweencurrentandpriordepression.Table 1 provides additionalinformationaboutthesecontrolvariables. StatisticalAnalyses FollowingRaudenbushetal.(2003),Sampsonetal.(2005), andothers(ZimmermanandMessner 2010, 2011),we 7 Thesameﬁveitemsthatmeasuresocialcohesionarealsoincluded inthecollectiveefﬁcacymeasure.Webelievethisoverlapis conceptuallytolerableforthepurposesofourinquiry.First,wewere interestedintheeffectofcollectiveefﬁcacyonbothdepressionand theIPV—depressionrelationship,andthus,neededtoincludethe measureofcollectiveefﬁcacyasithasbeenexaminedinprior research(e.g.,Sampsonetal. 1997).Additionally,therehasbeen somerecentattentiontotheimportanceofsocialcohesionwithregard todepression(Mairetal. 2010)aswellasbyitselfasafacilitatorof positiveneighborhoodbehavior(e.g.,informalsocialcontrol,see Warner 2014).Wewereinterestedinitsuniqueeffects—apartfrom collectiveefﬁcacy—andthereforechosetoincludeaseparate measureofsocialcohesioninouranalyses.Collinearitydidnot presentaproblem,aswemodeledcollectiveefﬁcacyandsocial cohesionseparately. AmJCommunityPsychol(2015)56:342–356 347 123employamultilevellogisticregressionmodeltopredictthe oddsthatarespondentlivinginagivenneighborhoodwill reportexperiencingdepressivesymptoms.Thisallowsus toutilizeall34,766responsestothewavetwodepression symptomitemsprovidedbyall2959ofthefemalesliving in80neighborhoodsinoursample.Thus,ouranalytic techniqueincludesanyfemalewhorespondedtoatleast onedepressionsymptomitemaskedatwavetwo.This techniqueeffectivelyavoidsthelossofdataduetomissing itemresponses(Osgoodetal. 2002)andtakesitemdifﬁ- cultyintoaccount(i.e.,somedepressionsymptomsareless prevalentthanothers)(Sampsonetal. 2005). ThemultivariatemultilevelRaschmodelisathree-level modelinwhichdichotomousdepressionitemsarenested withinpersons,whoarenestedwithinneighborhoods (Raudenbushetal. 2003).Thelevel-oneoutcomeisthe log-oddsofrespondingafﬁrmativelytoitem i of m - 1 depressionitems,by j person,livingin k neighborhood. Table1 Descriptiveinformationfordependentandindependentvariables MeanSDMin– Max Dependentvariable DepressionSymptomsThree-levelitemresponsemodelbasedon13indicatorsofdepression(wave2).09.290to1 Individual-levelvariables SevereIPVSixitemmeasureindicatingwhetherthemalepartnerusedsevereaggressionagainstthe femalerespondentinthepastyear(hitwithaﬁst,hitwithsomething,beatup,choked, threatenedwithaknifeorgun,orusedaknifeorgun)(wave1) .15.350to1 AgeAgeoffemalerespondent(wave1)32.727.9015.02to 80.93 HispanicFemalerespondentisHispanic(wave1).48.500to1 AfricanAmericanFemalerespondentisAfricanAmerican(wave1).31.460to1 OtherraceFemalerespondentisanotherrace/ethnicity(wave1).04.190to1 Caucasian(ref)FemalerespondentisCaucasian(wave1).16.370to1 MarriedFemalerespondentismarried(wave1).63.480to1 HouseholdsalaryMaximumhouseholdsalaryrangingfromlessthan$5000toover$50,000(wave1)4.011.951to7 UnemploymentFemalerespondentisunemployed(wave1).48.500to1 IsolationFive-itemscalereportedbythefemaleindicatinghersocialisolation(feelingalone,not havingfriendstotalkto,eachonascaleof1-3,indicatingverytruetonottrue)(wave1) (reliability = .613) -.001.00 -.92to 3.31 StressfullifeeventFemales’exposuretostressfullifeeventswherebyafriend,familymember,or acquaintancewashurt(e.g.,shotatorraped)orkilledbyaviolentact(waves1and2) .48.500to1 PriordepressionFemalerespondentreportedpriordepression(wave1).12.330to1 Neighborhood-levelvariables Concentrated disadvantage Principalcomponentsfactoranalysisusingthreeitems(reliability = .805)fromthe1990 Census:thepercentageofresidentsbelowpoverty,householdsreceivingpublic assistance,andresidentsunemployed .001.00 -1.51to 2.35 CollectiveefﬁcacyThree-levelitemresponsemodelbasedon10indicatorsofsocialcohesionandinformal socialcontrolreportedbyadultresidentsintheCommunitySurvey(reliability = .847) -.01.22 -.46to .64 SocialcohesionThree-levelitemresponsemodelbasedon5indicatorsofsocialcohesionandtrustbetween neighborsreportedbyadultresidentsintheCommunitySurvey(reliability = .826) -.00.20 -.46to .64 SocialinteractionsThree-levelitemresponsemodelbasedon4indicatorsofinteractionsbetween neighborhoodsreportedbyadultsintheCommunitySurvey(reliability = .734) -.00.17 -.52to .63 AnyfriendsWhetheradultsintheCommunitySurveyreportedhavinganyfriendslivingintheir neighborhood.ResponseswereaggregatedtotheNC-level .83.08.62–.98 AnyfamilyWhetheradultsintheCommunitySurveyreportedhavinganyrelativesorin-lawslivingin theirneighborhood.ResponseswereaggregatedtotheNC-level .45.170–.86 Averagenumberof familyandfriends Numberofrelativesorin-lawsandfriendslivingintheneighborhoodasreportedbythe adultsintheCommunitySurvey(eachonascaleof0–4,indicating0–10ormore). ThesemeasureswereaveragedandthenaggregatedtotheNC-level 1.52.35.96–3.02 Basedon34,766responsesacross2959respondentswithin80neighborhoodclusters 348 AmJCommunityPsychol(2015)56:342–356 123Thisvariablelocatesitemseveritiesonthelogitscale (Raudenbushetal. 2003).Thus,thelevel-onemodel adjuststhewithin-personpropensityfordepressivesymp- tomsbyitemseverity,missingdata,andmeasurement error.Thelevel-oneinterceptservesastheoutcomeforthe level-twoandlevel-threemodelsandisassumedtobe normallydistributedonalogitscale. Thelevel-twomodelexaminestheeffectsofperson- levelcorrelates(e.g.,age,maritalstatus,severeIPV)onthe level-oneintercept(susceptibilityfordepressivesymp- toms),whilealsocontrollingforitemseveritiesatlevel- one.Alloftheperson-leveleffectsweregrand-meancen- tered.Fortheneighborhooddirecteffectsmodels,the effectsofallcovariateswereassumedtobeﬁxedacross neighborhoods;however,theanalysesexaminingthecross- levelinteractionsallowedtheeffectofexperiencingsevere IPVtovaryacrossneighborhoods(p\.05). Thelevel-threemodelallowsestimationofthesuscep- tibilityofdepressivesymptomsacrossneighborhoods.The level-threeintercepts-as-outcomesmodelexaminesthe directeffectsofneighborhooddisadvantageandsupportive mechanismsonthelevel-twointercept(i.e.,thesuscepti- bilityofexperiencingdepressionsymptoms,controllingfor person-levelcorrelatesatlevel-twoanditemseveritiesat level-one).Thelevel-threeslopes-and-intercepts-as-out- comesmodelallowsforexaminationofthecross-level interactionbetweenneighborhoodsupportivemechanisms andthelevel-tworelationshipbetweenexperiencingsevere IPVandsusceptibilityforexperiencingdepressionsymp- toms,whileadjustingfortheindividualandneighborhood maineffects.Multicollinearitywasnotaprobleminanyof themodels(tolerancevalueswereabove.40). Results Tables 2 and 3 presenttheresultsofourthreeresearch questions.Table 2 presentsthemaineffectofexperiencing severeIPVfromone’spartneratwaveoneondepressive symptomsapproximately3yearslater,controllingfor neighborhoodvariablesaswellasindividual-levelfactors, suchaspriordepressionandmaritalstatus.Theseanalyses donotincludethecross-levelinteractionbetweenneigh- borhoodsupportivemechanismsandtheIPV—depression relationship.Theresultsoftheindividual-levelmodels suggestthat,asexpected,experiencingsevereabusefrom one’spartnersigniﬁcantlyincreasesthelikelihoodthat femaleswillreportsubsequentdepressivesymptoms; importantly,thiseffectissigniﬁcantevenaftercontrolling fortheimpactofpriordepression.Additionally,struggling withpriordepression(comparedtonotreportingprior depressivesymptoms),experiencingastressfullifeevent (versusnotexperiencingsuchevents),andhavinghigher householdsalaries(asopposedtolowersalaries)increased thelikelihoodofexperiencingdepressivesymptomsacross allmodels.Beingmarried(asopposedtodatingor cohabitingwithapartner)protectedwomenagainstexpe- riencingdepressivesymptomsacrossallmodels. InthebottomhalfofTable 2,weprovidetheneigh- borhooddirecteffectsondepressivesymptoms,controlling forthemaineffectofneighborhooddisadvantageand respondent-compositionalfactorsinallanalyses.Resultsof theneighborhooddirecteffectsindicatethatmanyofthe neighborhoodsupportivemechanismsare,infact,protec- tiveagainstexperiencingdepressivesymptoms,evenwhen neighborhooddisadvantageandindividual-levelprior depressionaretakenintoaccount.Speciﬁcally,women whoresideinneighborhoodscharacterizedbyhigherlevels ofcollectiveefﬁcacy(Model1),socialcohesion(Model2), socialinteractionsbetweenresidents(Model3),and neighborhoodsinwhichmanypeoplehadfriendsand familymembers(Model6)werelesslikelytoreport experiencingdepressivesymptomsthanwomenresidingin neighborhoodscharacterizedbylesscollectiveefﬁcacy, cohesion,networks,andfewerfriendsorfamilymembers onaverage.Neighborhoodconcentrateddisadvantagewas alsosigniﬁcantly(thoughmodestly)associatedwitha lowerlikelihoodofdepressivesymptomswhencollective efﬁcacy(Model1)andsocialcohesion(Model2)werealso includedintheanalyses,thoughitwasnotdirectlyrelated todepressivesymptomsinanyoftheothermodels. InTable 3,weincludethecross-levelinteractions betweenneighborhoodsupportsandtheIPV—subsequent depressionrelationship.Thesearereportedintherows belowtheSevereIPVvariable.Theseinteractionsare assessedsimultaneouslywiththeindividualandneigh- borhoodmaineffects.Nosubstantivechangesintheindi- vidualorneighborhoodmaineffectswereobservedwhen thecross-levelinteractionwasincludedinTable 3.As mentionedabove,theimpactofexperiencingsevereIPV fromone’spartnerinfactvariedacrossneighborhoods,and wasallowedtovaryforthecross-levelinteractionmod- els(asindicatedbytheitalicizedcoefﬁcients).Thepurpose ofexaminingthecross-levelinteractionswastodetermine ifneighborhoodsupportivemechanisms,suchascollective efﬁcacy,explainedthisvariation,perhapsbymitigatingthe impactofIPVondepressivesymptoms.Wefoundvery littleevidencethatneighborhoodsupportivemechanisms alleviatedtheimpactofIPVonsubsequentdepression,and theonlytwosigniﬁcantmoderatingeffectswefoundwere somewhatcounterintuitive.Model5inTable 3 indicates thattheimpactofexperiencingsevereIPVonlater depressivesymptomswassigniﬁcantlystrongerforwomen whoresidedinneighborhoodscharacterizedbygreater proportionsoffamilymembersintheneighborhood(as reportedbyrespondentsoftheCommunitySurvey,notthe AmJCommunityPsychol(2015)56:342–356 349 123respondentsofIPV),whileModel6demonstratesthatthe inﬂuenceofIPVondepressivesymptomswasalsostronger forwomenlivinginneighborhoodscharacterizedbyhigher averagenumbersoffamilyandfriendslivinginthesame neighborhood(thoughthiseffectwasmodest,reaching signiﬁcanceatthe p B .10level). Figure 1 depictsthecross-levelinteractiondisplayedin Model5ofTable 3,whileFig. 2 depictsthecross-level interactioninModel6.Asshown,therelationshipbetween IPVandthelikelihoodofexperiencingdepressivesymp- tomswassteepandpositiveinneighborhoodscharacter- izedbyhigh(onestandarddeviationabovethemean) familyassociates(Fig. 1)andfriendsandfamilymembers (Fig. 2)intheneighborhood.TheimpactofIPVon depressivesymptomsweakened(asindicatedbytheslope tiltinginthenegativedirection)inareascharacterizedby low(onestandarddeviationbelowthemean)family members(Fig. 1)andfamilyandfriends(Fig. 2).Thus, althoughweexpectedthatneighborhoodsupportive mechanisms,suchastieswithothers,wouldpotentially shieldvictimsfromexperiencingdepressivesymptoms,we foundtheoppositeeffect–insomeneighborhoods,par- ticularlythosecharacterizedbyhighlevelsoffamily members,theimpactofbeingavictimofpartnerviolence onlaterdepressivesymptomswasactuallyworse. Discussion Manyscholarsagreethatneighborhoodcontextisimpor- tantwhenitcomestobothintimatepartnerviolenceand depression(orothermentalhealthproblems)(e.g.,Benson etal. 2003;Goodmanetal. 2009;KimandRoss 2009),and anumberofstudieshaveexaminedthenegativeordetri- mentalimpactsofneighborhoodswithregardtothese outcomes(e.g.,Mairetal. 2008;Miles-Doan 1998). Table2 Directeffectsofintimatepartnerviolenceandneighborhoodsupportivemechanismsondepressionsymptoms Model1Model2Model3Model4Model5Model6 b(SE)b(SE)b(SE)b(SE)b(SE)b(SE) Intercept -1.87(.06) -1.87(.06) -1.87(.06) -1.87(.06) -1.86(.06) -1.86(.06) Individual-leveleffects Age.00(.01).00(.01).00(.01).00(.01).00(.01).00(.01) Hispanic.25(.15).24(.16).26(.15).30(.15).32(.15).30(.15) AfricanAmerican.06(.21).08(.21).05(.21).05(.21).07(.21).03(.22) Otherrace.07(.28).08(.29).07(.28).08(.28).07(.27).06(.28) Married -.31 (.13) -.30(.13) -.31(.13) -.32(.13) -.32(.13) -.32(.13) Householdsalary.07 (.03).07(.03).07(.03).07(.03).07(.03).07(.03) Unemployment.09(.14).09(.14).10(.14).09(.14).08(.14).09(.14) Isolation.06(.07).06(.07).06(.07).06(.07).07(.07).06(.07) Stressfullifeevent.38 (.11).38(.11).38(.11).38(.11).38(.11).38(.11) Priordepression1.07 (.18)1.08(.18)1.08(.18)1.08(.18)1.08(.18)1.08(.18) SevereIPV.50 (.15).50(.16).51(.16).50(.15).50(.15).50(.15) Neighborhooddirecteffects Concentrateddisadvantage -.15 (.08) -.16(.07) -.07(.08) -.09(.08) -.07(.08) -.07(.08) Collectiveefﬁcacy -.59* (.24)–––––––––– Socialcohesion–– -.93**(.31)–––––––– Socialinteraction–––– -.61(.28)–––––– Anyfriends–––––– -.75(.60)–––– Anyfamily–––––––– -.25(.31)–– Averagenumberoffamilyand friends –––––––––– -.23(.11) Variancecomponents Individuallevelintercept7.7947.8037.7997.7947.7907.793 Neighborhoodintercept.205.173.195.207.215.206 Basedon34,766responsesacross2959respondentswithin80neighborhoodclusters ** p B .01;* p B .05; p B .10 350 AmJCommunityPsychol(2015)56:342–356 123However,lessattentionhasbeenpaidtothepotential protectiveorhealth-promotingeffectsthatsometypesof neighborhoodsprovidefortheirresidents(forexceptions, seeforinstance,AhernandGalea 2011;Kim 2010;Kim andRoss 2009;Mairetal. 2008, 2010;RossandJang 2000;Staffordetal. 2011).Additionally,theinterplayof partnerviolence,neighborhoodsupportivemechanisms, anddepressivesymptoms,toourknowledge,hasnotbeen assessed,limitingourunderstandingforhowtheIPV— depressionrelationshipoperateswhenneighborhoodcon- textisconsidered.Ourstudysoughttoadvancethisliter- aturebyexploringwhetherneighborhoodsupportive mechanismsalleviatedthedetrimentalimpactofIPVon subsequentdepressivesymptoms.Ourresultssuggestthat theimpactofviolentrelationshipsonlatermentalhealth outcomesmaydependinpartonneighborhoodcharacter- istics,particularlysocialties,butcontrarytoourexpecta- tions,thepresenceofhighlevelsofneighborhoodties Table3 Effectsofintimatepartnerviolenceandneighborhoodsupportivemechanismsondepressionsymptoms,andcross-levelinteractions betweenneighborhoodsupportivemechanismsandtheIPV—depressionrelationship Model1Model2Model3Model4Model5Model6 b(SE)b(SE)B(SE)b(SE)b(SE)b(SE) Intercept -1.86(.06) -1.86(.06) -1.86(.06) -1.86(.06) -1.86(.06) -1.85(.06) Individual-leveleffectsandcross-levelinteractions Age.00(.01).00(.01).00(.01).00(.01).00(.01).00(.01) Hispanic.22(.15).21(.15).23(.14).27(.15).30(.14).27(.14) AfricanAmerican.06(.20).08(.20).06(.21).05(.21).08(.21).05(.21) Otherrace.01(.28).02(.28) -.01(.27).02(.28) -.02(.26) -.03(.27) Married -.33 (.13) -.32(.13) -.33(.13) -.34(.13) -.34(.13) -.34(.13) Householdsalary.07 (.03).07(.03).07(.03).07(.03).07(.03).07(.03) Unemployment.08(.14).09(.14).10(.14).08(.14).08(.14).09(.14) Isolation.06(.07).06(.07).06(.07).06(.07).06(.07).06(.07) Stressfullifeevent.37 (.11).37(.11).37(.11).38(.11).37(.11).37(.11) Priordepression1.09 (.17)1.09(.17)1.09(.17)1.10(.17)1.10(.17)1.09(.17) SevereIPV .50 (.16) .50 (.17) .51 (.16) .48 (.16) .44 (.15) .48 (.16) 9 Collectiveefﬁcacy.03(.81)–––––––––– 9 Socialcohesion––.24(1.04)–––––––– 9 Socialinteraction––––1.02(.79)–––––– 9 Anyfriends–––––– -1.14(1.64)–––– 9 Anyfamily––––––––1.91(.91)–– 9 Averagenumberoffamily andfriends ––––––––––.63 (.34) Neighborhoodmaineffects Concentrateddisadvantage -.16 (.08) -.16(.07) -.07(.08) -.10(.08) -.07(.08) -.07(.08) Collectiveefﬁcacy -.60(.24)–––––––––– Socialcohesion–– -.92**(.32)–––––––– Socialinteraction–––– -.65(.28)–––––– Anyfriends–––––– -.74(.62)–––– Anyfamily–––––––– -.25(.31)–– Averagenumberoffamilyand friends –––––––––– -.22(.11) Variancecomponents Individual-levelintercept7.6137.6227.6227.6167.6187.620 Neighborhoodintercept.207.175.202.211.220.214 SevereIPVintercept1.8911.8721.8181.8281.6611.729 Basedon34,766responsesacross2959respondentswithin80neighborhoodclusters Thesigniﬁcanceofitalicvalueindicatesrandomlyvaryingcoefﬁcient ** p B .01;* p B .05; p B .10; AmJCommunityPsychol(2015)56:342–356 351 123appearedtostrengthenratherthanattenuatetherelation- shipbetweenIPVanddepressivesymptoms.Wedidﬁnd, however,thatneighborhoodsupportivemechanisms,par- ticularlythoserelatedtocollectiveefﬁcacy,cohesion,and socialinteractionswithneighbors,alleviatedthelikelihood ofexperiencingsymptomsofdepressionevenaftercon- trollingfortheeffectsofpriordepression.Wediscussour ﬁndingsbelow. Ourﬁrstgoalwastoexaminethelong-termimpactof IPVondepressivesymptomsusinglongitudinaldata,as fewstudieshaveassessedtheprospectiveimpactofviolent relationshipsonsuchanoutcome(Campbell 2002;Fletcher 2010;Johnsonetal. 2014).Wefoundthatexperiencing severeIPVsigniﬁcantlyincreasedthelikelihoodthat womenreporteddepressivesymptomsyearslater.The relationshipbetweenIPVanddepressionwasrobust, maintainingasigniﬁcanteffectdespitehighlyspeciﬁed modelswhichincludedseveralotherimportantcovariates, suchaspriordepressionand,perhapsevenmoreimpor- tantly,whilesimultaneouslyinvestigatingcontextual inﬂuences(Goodmanetal. 2009).Giventhestrengthofour multilevellongitudinalmodels,ourresultsdemonstratethat IPVisasigniﬁcantriskfactorforsubsequentmentalhealth problemsthatisindependentfromtheeffectsofother predictorsofdepression,suchaspriordepression,stressful lifeevents,andisolation.ThatIPVmaintainedastrong longitudinaleffectondepressioninthesemodelssuggests thatitremainsacentralpublichealthconcern(seealso Bonomietal. 2006;CaetanoandCunradi 2003;DeMaris andKaukinen 2005;Golding 1999). -4 -3 -2 -1 0 1 High Proporon of Family Members in Neighborhood Average Proporon of Family Members in Neighborhood Low Proporon of Family Members in Neighborhood Likelihood of Depression Symptoms Severe Inmate Partner Violence 0 1 Fig.1 Therelationship betweensevereintimatepartner violenceanddepression symptomsbyneighborhood proportionoffamilyties. Controlsforneighborhood concentrateddisadvantageand allrespondent-levelcovariates -3 -2 -1 0 High Levels of Family and Friends in Neighborhood Average Levels of Family and Friends in Neighborhood Low Levels of Family and Friends in Neighborhood Likelihood of Depression Symptoms Severe Inmate Partner Violence 0 1 Fig.2 Therelationship betweensevereintimatepartner violenceanddepression symptomsbyaveragenumber offamilyandfriendsin neighborhood.Controlsfor concentrateddisadvantageand allrespondent-levelcovariates 352 AmJCommunityPsychol(2015)56:342–356 123Ournextgoalsweretoexamineseveralneighborhood- levelprotectivefactors—particularlythosemeasuring socialconnectionsandcohesivenessamongneighbors— whichhavebeensuggestedasmechanismsthatmight alleviatementalhealthsymptomsand/ortheeffectsof abuse(AhernandGalea 2011;Burkeetal. 2006;Mairetal. 2008;O’Campoetal. 2005;Staffordetal. 2011).We soughttoexamineboththedirectandmoderatinginﬂu- encesoftheseneighborhoodcharacteristicswhilecontrol- lingforimportantindividual-levelcovariates.Consistent withpriorresearchreportingadirectinﬂuenceofneigh- borhoodsondepression(e.g.,Kim 2010;Mairetal. 2008), wefoundthatneighborhoodsupportivemechanismsdo appeartoprotectindividualsfromexperiencingsymptoms ofdepression.Wesuspectthatneighborhoodfactorssuch ascollectiveefﬁcacy,socialcohesionbetweenneighbors, socialinteractionsamongresidents,andhavingmore familyandfriendswhoresideinone’sneighborhoodare protectiveagainstnegativementalhealthsymptoms becausetheyprovidesupport,increasecommunication, andlowerfeelingsofisolation(e.g.,Browning 2002; Sampson 2013;Vegaetal. 2011).Assourcesofsupport, thesetiesmightalsoreducetheeffectsofstress,increase feelingsofcontrol,fosteraltruisticbehaviors,orreduce feelingsoffearormistrust,allofwhichinturnmight alleviatefeelingsofdepression(AhernandGalea 2011; KimandRoss 2009;RossandJang 2000;Sampson 2013; Staffordetal. 2011). However,wealsofoundthatsomeneighborhoodsup- portivemechanismsarenotnecessarilyprotectiveagainst depression,particularlywhenpartnerviolenceiscon- cerned.Livinginneighborhoodswheregreaterproportions offamilymembers(relativesorin-laws)residedactually exacerbatedthenegativeimpactofIPVondepressive symptomsamongfemales.Likewise,wefoundthatwomen wholivedinneighborhoodscharacterizedbyhigherpro- portionsoffamilyandfriendtiesweremoreat-riskfor depressivesymptomsfollowingIPVthanwomenwhose neighborhoodswerecharacterizedbyfewersuch‘‘sup- ports.’’Whilethe‘‘family’’elementmaybelargelydriving thislastﬁnding,westillconsiderbotheffectsnoteworthy andunexpected.Althoughneighborhoodsocialties—par- ticularlybetweenfamilyand/orfriends—islargelydeemed aprotectivemechanismformanyoutcomes(e.g.,Bellair 1997;WellmanandWortley 1990),includingbothpartner violence(e.g.,Agoffetal. 2007;Wright 2012)and depression(e.g.,KimandRoss 2009),wefoundthe oppositetobetrue,atleastregardingtheimpactofIPVon depressivesymptoms.Afewpossibilitiesmayaccountfor thesecontradictoryﬁndings.RecallthattheCommunity Surveysampledrespondentsfromneighborhoodswho werenotnecessarilyparticipantsoftheLCS—thus, ‘‘family’’tiesintheneighborhoodlikelydonotreﬂectthe victim’sownfamilymembers—theysimplyreﬂectthat suchtiesexistwithintheneighborhood.Inthiscase, womenwhoarebeingvictimizedmaynothavetheirown familymemberslivingcloseby.Indeed,itispossiblethat theIPVvictimsintheseneighborhoodsactuallymayhave feltdoublyisolatediftheyhadnolocalfamilyorfriends whileeveryonearoundthemwasstronglytiedintoalocal familial(orfriendship)network.Wesoughttotakethis possibilityintoaccount,however,byincludingthe females’levelofisolationasanindividual-levelcontrol. Thatitwasnotsigniﬁcantinourmodelssuggestsanother explanationiswarranted.Forinstance,wedonotknowif thevictimizedfemaleactuallyutilizedthesetiesforhelp. Whilesome(Browning 2002;VanWyketal. 2003;Wright andBenson 2011)havesuggestedthatmoresocialtiesmay makeitmorelikelythattheviolencewillbecomepublic knowledge,itmaybethatthesetiesweresimplynot activatedorsoughtoutforhelp(WrightandBenson 2011). SinceIPVisoftentimesconsideredveryprivateinnature (Strausetal. 2006),thisiscertainlyapossibility.Itisalso possiblethatbeingsurroundedbymorefamilyties(evenif notone’sownfamilymembers)inone’sneighborhoodputs pressureonwomentostayintherelationship,especiallyif familiesintheareadonotcondonedivorceorseparation, evenfromaviolentpartner(Agoffetal. 2007).Alterna- tively,itispossiblethatinthecontextofpartnerviolence, somesocialsupportsarenothelpful,andmayinfactbe harmfulbysupportingorcondoningtheuseofviolence withinrelationships.Weareunabletoexaminewhetherthe neighborhoodtiesmeasuredherearesupportiveofIPV, andthereforecannottestthispotentialexplanation.Ifthe socialtiesintheneighborhoodswerebetweenresidents whoweresupportiveofviolenceinrelationships,thiscould explaintheexacerbatingeffectfoundbetweenneighbor- hoodsocialtiesandtheimpactofIPVonlaterdepression. Wecannotbesurewhichexplanation,ifany,iscorrect giventheavailabledata,butweencouragefutureresearch toattempttodisentanglethecomplexrelationshipthat appearstoexistbetweenneighborhood-levelsocialties (especiallywithfamilyorrelatives)andpartnerviolence. Asidefromtheeffectoffamilyandfriends,wefoundno otherevidencethatneighborhoodsupportivemechanisms (collectiveefﬁcacy,socialcohesion,socialinteraction) moderatedtheimpactofIPVonsubsequentdepression. Perhapstheneighborhoodsupportivemechanismsexam- inedherearetoogeneraltoalleviatetheeffectsofIPVon mentalhealth—itispossiblethatotherneighborhood constructsgearedmorespeciﬁcallytoviolenceandmental healthsupport,intervention,and/orprevention(e.g., counselingservices,shelters,etc.)wouldbemorerelevant. Futureresearchshouldconsiderthepotentialforsuch neighborhoodsupportivemechanismstoeasetheeffectsof speciﬁcviolencestressorssuchasIPVandabuseon AmJCommunityPsychol(2015)56:342–356 353 123negativementalhealthoutcomessothataggregatepre- ventionandinterventionscanbeimplemented(seealso Cunradi 2010). Limitations Unfortunately,ourstudyisnotwithoutlimitations.We reliedsolelyondatafromfemalecaregiversanddidnot includemalesaspotentialvictimsofIPV.Therefore,wedo notknowifourﬁndingswouldgeneralizetofemales’ violenceperpetratedagainstmales.Similarly,ourstudy doesnotexaminedatingviolencebyadolescents.Whilewe examinedtheimpactoftheprevalenceofsevereIPVon females’depression,wedidnotexaminetheimpactofthe frequencyorchronicityofsuchviolence.Itispossiblethat morefrequentormoresevereviolence,whileperhapsmore detrimentaltodepression(Campbell 2002),couldbe impactedbyneighborhoodfactorsdifferently.Finally,our datawerecollectedfromwomeninonecity(Chicago) duringthemid-1990sandintotheearly2000s;although thePHDCNisahighlyrespecteddataset,wecannotensure thatourresultswouldgeneralizetootherpopulationsor timeperiods. Ourstudyaddstoburgeoningevidencethatneighbor- hoodcontextimpactsvariousinterpersonaloutcomes,in bothdirectandmoderatingways.Itappearsthatneigh- borhoodsupportivemechanismsmaydirectlyprotect femalesfromnegativementalhealthproblemssuchas experiencingdepressivesymptoms.Theseneighborhood supportivemechanisms,thoughlessoftenconsidered,are importanttoexamineforpreventionandinterventionpur- poses,andsuggestthatjustasneighborhoodscanbe detrimental,theycanalsobepositiveandprotective. Neighborhoodcollectiveefﬁcacy,cohesion,andinteraction betweenneighborsmightincreasefeelingsofsupport, belongingness,investment,communication,andother positiveattributes,whichmayinturnreducedepressive symptoms.Webelieveitisimportantforresearchersto continuetoexaminethepositiveandpotentiallymitigating inﬂuencesofneighborhoods,aswellastheirdetrimental effects,inordertobetterunderstandforwhomandunder whichcircumstancesviolentrelationshipsandmental healthareinﬂuencedbycontextualfactors. References Agoff,C.,Herrera,C.,&Castro,R.(2007).Theweaknessoffamily tiesandtheirperpetuatingeffectsongenderviolence:A qualitativestudyinMexico. ViolenceAgainstWomen,13(11), 1206–1220. Ahern,J.,&Galea,S.(2011).Collectiveefﬁcacyandmajor depressioninurbanneighborhoods. AmericanJournalof Epidemiology,173(12),1453–1462. Aneshensel,C.S.,&Sucoff,C.A.(1996).Theneighborhoodcontext ofadolescentmentalhealth. JournalofHealthandSocial Behavior,37(4),293–310. Bellair,P.E.(1997).Socialinteractionandcommunitycrime: Examiningtheimportanceofneighbornetworks. Criminology, 35(4),677–704. Benson,M.L.,Fox,G.L.,DeMaris,A.,&VanWyk,J.(2003). Neighborhooddisadvantage,individualeconomicdistressand violenceagainstwomeninintimaterelationships. Journalof QuantitativeCriminology,19(3),207–235. Bonomi,A.E.,Thompson,R.S.,Anderson,M.,Reid,R.J.,Carrell, D.,Dimer,J.A.,&Rivara,F.P.(2006).Intimatepartner violenceandwomen’sphysical,mental,andsocialfunctioning. AmericanJournalofPreventiveMedicine,30(6),458–466. Browning,C.R.(2002).Thespanofcollectiveefﬁcacy:Extending socialdisorganizationtheorytopartnerviolence. Journalof MarriageandFamily,64(4),833–850. Browning,C.R.,Feinberg,S.L.,&Dietz,R.D.(2004).Theparadox ofsocialorganization:Networks,collectiveefﬁcacy,andviolent crimeinurbanneighborhoods. SocialForces,83(2),503–534. Burke,J.,O’Campo,P.,&Peak,G.L.(2006).Neighborhood inﬂuencesandintimatepartnerviolence:Doesgeographic settingmatter? JournalofUrbanHealth,83(2),182–194. Burke,J.,O’Campo,P.,Salmon,C.,&Walker,R.(2009).Pathways connectingneighborhoodinﬂuencesandmentalwell-being: Socioeconomicpositionandgenderdifferences. SocialScience andMedicine,68,1294–1304. Caetano,R.,&Cunradi,C.(2003).Intimatepartnerviolenceand depressionamongWhites,Blacks,andHispanics. Annalsof Epidemiology,13,661–665. Caldwell,J.E.,Swan,S.C.,&Woodbrown,V.D.(2012).Gender differencesinintimatepartnerviolenceoutcomes. Psychologyof Violence,2(1),42–57. Campbell,J.C.(2002).Healthconsequencesofintimatepartner violence. Lancet,359(9341),1331–1336. Campbell,J.C.,Kub,J.E.,&Rose,L.(1996).Depressioninbattered women. JournaloftheAmericanMedicalWomen’sAssociation, 51(3),106–110. Cerda,M.,Sanchez,B.N.,Galea,S.,Tracy,M.,&Buka,S.L. (2008).Estimatingco-occuringbehavioraltrajectorieswithina neighborhoodcontext:Acasestudyofmultivariatetransition modelsforclustereddata. AmericanJournalofEpidemiology, 168(10),1190–1203. Coker,A.L.,Davis,K.E.,Arias,I.,Desai,R.A.,Sanderson,M., Brandt,H.M.,&Smith,P.H.(2002a).Physicalandmental healtheffectsofintimatepartnerviolenceformenandwomen. AmericanJournalofPreventiveMedicine,23(4),260–268. Coker,A.L.,Smith,P.H.,Thompson,M.P.,McKeown,R.E., Bethea,L.,&Davis,K.E.(2002b).Socialsupportprotects againstthenegativeeffectsofpartnerviolenceonmentalhealth. JournalofWomen’sHealth&Gender-BasedMedicine,11(5), 465–476. Cunradi,C.(2010).Neighborhoods,alcoholoutletsandintimate partnerviolence:Addressingresearchgapsinexplanatory mechanisms. InternationalJournalofEnvironmentalResearch andPublicHealth,7,799–813. DeMaris,A.,&Kaukinen,C.(2005).Violentvictimizationand women’smentalandphysicalhealth:Evidencefromanational sample. JournalofResearchinCrimeandDelinquency,42(4), 384–411. Earls,F.J.,Brooks-Gunn,J.,Raudenbush,S.W.,&Sampson,R.J. (2002). ProjectonhumandevelopmentinChicago 354 AmJCommunityPsychol(2015)56:342–356 123neighborhoods(PHDCN).AnnArbor,MI:Inter-university CorsortiumforPoliticalandSocialResearch. Echeverria,S.,DiezRoux,A.V.,Shea,S.,Borrell,L.N.,&Jackson, S.(2008).Associationsofneighborhoodproblemsandneigh- borhoodsocialcohesionwithmentalhealthandhealthbehav- iors:Themulti-ethnicstudyofatherosclerosis. Health&Place, 14(4),853–865. Edwards,K.M.,Mattingly,M.J.,Dixon,K.J.,&Banyard,V.L. (2014).Communitymatters:Intimatepartnerviolenceamong ruralyoungadults. AmericanJournalofCommunityPsychology, 53,198–207. Emery,C.R.,Jolley,J.M.,&Wu,S.(2011).Desistancefrom intimatepartnerviolence:Theroleoflegalcynicism,collective efﬁcacy,andsocialdisorganizationinChicagoneighborhoods. AmericanJournalofCommunityPsychology,48,373–383. Fletcher,J.(2010).Theeffectsofintimatepartnerviolenceonhealth inyoungadulthoodintheUnitedStates. SocialScienceand Medicine,70,130–135. Geis,K.J.,&Ross,C.E.(1998).Anewlookaturbanalienation:The effectofneighborhooddisorderonperceivedpowerlessness. SocialPsychologyQuarterly,61(3),232–246. Golding,J.M.(1999).Intimatepartnerviolenceasariskfactorfor mentaldisorders:Ameta-analysis. JournalofFamilyViolence, 14(2),99–132. Goodman,L.A.,Smyth,K.F.,Borges,A.M.,&Singer,R.(2009). Whencrisescollide:Howintimatepartnerviolenceandpoverty intersecttoshapewomen’smentalhealthandcoping? Trauma, Violence,&Abuse,10(4),306–329. Hadeed,L.F.,&El-Bassel,N.(2006).SocialsupportamongAfro- Trinidadianwomenexperiencingintimatepartnerviolence. ViolenceAgainstWomen,12(8),740–760. Hill,T.D.,Ross,C.E.,&Angel,R.J.(2005).Neighborhood disorder,psychophysiologicaldistress,andhealth. Journalof HealthandSocialBehavior,46(2),170–186. Johnson,W.L.,Giordano,P.C.,Longmore,M.A.,&Manning,W. D.(2014).Intimatepartnerviolenceanddepressivesymptoms duringadolescenceandyoungadulthood. JournalofHealthand SocialBehavior,55(1),39–55. Kessler,R.C.,Andrews,G.,Mroczerk,D.,Ustun,B.,&Wittchen, H.-U.(1998).TheWorldHealthOrganizationcompositeinter- nationaldiagnosticinterviewshort-form(CIDI-SF). Interna- tionalJournalofMethodsinPsychiatricResearch,7(4), 171–185. Kim,D.(2008).Bluesfromtheneighborhood?Neighborhood characteristicsanddepression. EpidemiologicReviews,30, 101–117. Kim,J.(2010).Neighborhooddisadvantageandmentalhealth:The roleofneighborhooddisorderandsocialrelationships. Social ScienceResearch,39(2),260–271. Kim,J.,&Ross,C.E.(2009).Neighborhood-speciﬁcandgeneral socialsupport:Whichbufferstheeffectofneighborhood disorderondepression? JournalofCommunityPsychology, 37(6),725–736. Kirk,D.S.,&Matsuda,M.(2011).Legalcynicism,collective efﬁcacy,andtheecologyofarrest. Criminology,49(2),443–472. Kirst,M.,Lazgare,L.P.,Zhang,Y.J.,&O’Campo,P.(2015).The effectsofsocialcapitalandneighborhoodcharacteristicson intimatepartnerviolence:Aconsiderationofsocialresources andrisks. AmericanJournalofCommunityPsychology,55, 314–325. Kornhauser,R.R.(1978). Socialsourcesofdelinquency:An appraisalofanalyticmodels.Chicago,IL:UniversityofChicago Press. Latkin,C.A.,&Curry,A.D.(2003).Stressfulneighborhoodsand depression:Aprospectivestudyoftheimpactofneighborhood disorder. JournalofHealthandSocialBehavior,44(1),34–44. Mair,C.,DiezRoux,A.V.,&Galea,S.(2008).Areneighbourhood characteristicsassociatedwithdepressivesymptoms?Areview ofevidence. JournalofEpidemiologyandCommunityHealth, 62(11),940–946. Mair,C.,DiezRoux,A.V.,&Morenoff,J.D.(2010).Neighborhood stressorsandsocialsupportaspredictorsofdepressivesymp- tomsintheChicagoCommunityAdultHealthStudy. Health& Place,16(5),811–819. Miles-Doan,R.(1998).Violencebetweenspousesandintimates:Does neighborhoodcontextmatter? SocialForces,77(2),623–645. Moe,A.M.(2007).Silencedvoicesandstructuredsurvival:Battered women’shelpseeking. ViolenceAgainstWomen,13(7),676–699. Molnar,B.E.,Cerda,M.,Roberts,A.L.,&Buka,S.L.(2008). Effectsofneighborhoodresourcesonaggressiveanddelinquent behaviorsamongurbanyouths. AmericanJournalofPublic Health,98(6),1086–1093. Molnar,B.E.,Miller,M.J.,Azrael,D.,&Buka,S.L.(2004). Neighborhoodpredictorsofconcealedﬁrearmcarryingamong childrenandadolescents:Resultsfromtheprojectonhuman developmentinChicagoneighborhoods. ArchivesofPediatrics andAdolescentMedicine,158(7),657–664. Morenoff,J.D.,Sampson,R.J.,&Raudenbush,S.W.(2001). Neighborhoodinequality,collectiveefﬁcacy,andthespatial dynamicsofurbanviolence. Criminology,39(3),517–559. O’Campo,P.,Burke,J.,Peak,G.L.,McDonnell,K.A.,&Gielen,A. C.(2005).Uncoveringneighbourhoodinﬂuenceonintimate partnerviolenceusingconceptmapping. JournalofEpidemiol- ogyandCommunityHealth,59(7),603–608. Osgood,D.W.,McMorris,B.J.,&Potenza,M.T.(2002).Analyzing multiple-itemmeasuresofcrimeanddevianceI:Itemresponse theoryscaling. JournalofQuantitativeCriminology,18, 267–296. Pinchevsky,G.M.,&Wright,E.M.(2012).Theimpactof neighborhoodsonintimatepartnerviolenceandvictimization. Trauma,Violence,&Abuse,12(2),112–132. Raudenbush,S.W.,Johnson,C.,&Sampson,R.J.(2003).A multivariate,multilevelRaschmodelwithapplicationtoself- reportedcriminalbehavior. SociologicalMethodology,33(1), 169–211. Rios,R.,Aiken,L.S.,&Zautra,A.J.(2012).Neighborhoodcontexts andthemediatingroleofneighborhoodsocialcohesionon healthandpsychologicaldistressamongHispanicandnon- Hispanicresidents. AnnalsofBehavioralMedicine,43(1), 50–61.doi:10.1007/s12160-011-9306-9. Ross,C.E.(2000).Neighborhooddisadvantageandadultdepression. JournalofHealthandSocialBehavior,41(2),177–187. Ross,C.E.,&Jang,S.J.(2000).Neighborhooddisorder,fear,and mistrust:Thebufferingroleofsocialtieswithneighbors. AmericanJournalofCommunityPsychology,28(4),401–420. Ross,C.E.,&Mirowsky,J.(2009).Neighborhooddisorder, subjectivealienation,anddistress. JournalofHealthandSocial Behavior,50(1),49–64.doi:10.1177/002214650905000104. Ross,C.E.,Mirowsky,J.,&Pribesh,S.(2001).Powerlessandthe ampliﬁcationofthreat:Neighborhooddisadvantage,disorder, andmistrust. AmericanSociologicalReview,66(4),568–591. Ross,C.E.,Reynolds,J.R.,&Geis,K.J.(2000).Thecontingent meaningofneighborhoodstabilityforresidents’psychological well-being. AmericanSociologicalReview,65(4),581–597. Sampson,R.J.(2003).Theneighborhoodcontextofwellbeing. PerspectivesinBiologyandMedicine,46(3),S53–S73. Sampson,R.J.(2013). GreatAmericancity:Chicagoandthe enduringneighborhoodeffect.Chicago,IL:Universityof ChicagoPress. Sampson,R.J.,Morenoff,J.D.,&Raudenbush,S.W.(2005).Social anatomyofracialandethnicdisparatiesinviolence. American JournalofPublicHealth,95(2),224–232. AmJCommunityPsychol(2015)56:342–356 355 123Sampson,R.J.,Raudenbush,S.W.,&Earls,F.J.(1997). Neighborhoodsandviolentcrime:Amultilevelstudyof collectiveefﬁcacy. Science,277(5328),918–924. Shaw,C.R.,&McKay,H.D.(1942). Juveniledelinquencyandurban areas:Astudyofratesofdelinquentsinrelationtodifferential characteristicsoflocalcommunitiesinAmericancities.Chicago, IL:UniversityofChicagoPress. Stafford,M.,McMunn,A.,&DeVogli,R.(2011).Neighbourhood socialenvironmentanddepressivesymptomsinmid-lifeand beyond. Ageing&Society,31(6),893–910. Stith,S.M.,Smith,D.B.,Penn,C.E.,Ward,D.B.,&Tritt,D. (2004).Intimatepartnerphysicalabuseperpetrationandvictim- izationriskfactors:Ameta-analyticreview. Aggressionand ViolentBehavior,10(1),65–98. Straus,M.A.(1979).Measuringintrafamilyconﬂictandviolence: Theconﬂicttactics(CT)scales. JournalofMarriageand Family,41(1),75–88. Straus,M.A.,Gelles,R.J.,&Steinmetz,S.K.(2006). Behindclosed doors:ViolenceintheAmericanfamily.NewBrunswick,NJ: TransactionPublishers. Taylor,S.E.,&Repetti,R.L.(1997).Healthpsychology:Whatisan unhealthyenvironmentandhowdoesitgetundertheskin? AnnualReviewofPsychology,48,411–447. VanWyk,J.A.,Benson,M.L.,Fox,G.L.,&DeMaris,A.(2003). Detanglingindividual-,partner-,andcommunity-levelcorrelates ofpartnerviolence. Crime&Delinquency,49(3),412–438. Vega,W.A.,Ang,A.,Rodriguez,M.A.,&Finch,B.K.(2011). NeighborhoodprotectiveeffectsondepressioninLatinos. AmericanJournalofCommunityPsychology,47(1–2),114–126. Warner,B.D.(2014).Neighborhoodfactorsrelatedtothelikelihood ofsuccessfulinformalsocialcontrolefforts. JournalofCriminal Justice,42,421–430. Wellman,B.,&Wortley,S.(1990).Differentstrokesfromdifferent folks:Communitytiesandsocialsupport. AmericanJournalof Sociology,96(3),558–588. Wilson,J.Q.,&Kelling,G.L.(1982).Thepoliceandneighborhood safety:Brokenwindows. AtlanticMonthly,127,29–38. Wright,E.M.(2012).Therelationshipbetweensocialsupportand intimatepartnerviolenceinneighborhoodcontext. Crime& Delinquency.doi:10.1177/0011128712466890. Wright,E.M.,&Benson,M.L.(2010).Immigrationandintimate partnerviolence:Exploringtheimmigrantparadox. Social Problems,57(3),480–503. Wright,E.M.,&Benson,M.L.(2011).Clarifyingtheeffectsof neighborhooddisadvantageandcollectiveefﬁcacyonviolence ‘‘behindcloseddoors’’. JusticeQuarterly,28(5),775–798. Xue,Y.,Leventhal,T.,Brooks-Gunn,J.,&Earls,F.J.(2005). Neighborhoodresidenceandmentalhealthproblemsof5-to 11-yearolds. ArchivesofGeneralPsychiatry,62(5),554–563. Zimmerman,G.M.,&Messner,S.F.(2010).Neighborhoodcontext andthegendergapinadolescentviolentcrime. American SociologicalReview,75(6),958–980. Zimmerman,G.M.,&Messner,S.F.(2011).Neighborhoodcontext andnonlinearpeereffectsonadolescentviolentcrime. Crimi- nology,49(3),873–903. 356 AmJCommunityPsychol(2015)56:342–356 123Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.