Religiosity,PsychologicalResources,and - PDF document

Religiosity,PsychologicalResources,and PhysicalHealth J OONMO S ON DepartmentofSociology NationalUniversityofSingapore J OHN W ILSON DepartmentofSociology DukeUniversity Variousexplanationshavebeengivenforthepositiveassociationbetweenreligiosityandphysicalhealth.Using datafromtwowavesoftheNationalSurveyofMidlifeintheUnitedStates(1995,2005)andretrospectivedata ontheimportanceofreligioninthehomeinwhichrespondentswereraisedweÞndthatpsychologicalresources, operationalizedbymeasuresofemotionalandpsychologicalwell-being,mediatetheeffectofthisearlyexposure toreligionbutonlyonself-ratedhealthandphysicalsymptomatology;chronicillnessesandhealthlimitationson activitiesofdailylivingareunaffected. Keywords: religiosity , health , well-being . I NTRODUCTION Ithaslongbeenarguedthatreligiouspeopleenjoybetterhealth.Onlyinthelasttwo decades,however,hasthisargumentundergonesustainedscientiÞcscrutiny.Mostoftheresearch hastargetedmentalhealth,wherethepossibleinßuenceofreligiosityismoreintuitive(Kelley- MooreandFerraro2001:S366;Krause,Ellison,andMarcum2002:22).Lessattentionhasbeen paidtophysicalhealthwherebiologicalchangesareinvolved.Althoughforthemostpartstudies haveshownthatreligiositydoeshavesalutaryeffectsonphysicalhealth,expertsnotecontinuing inconsistenciesbetweenstudies.Theyalsocallformoretheorytestinginthisarearatherthan speculationaboutpossiblemechanisms. ResearchonReligionandPhysicalHealth Numerousscholarlyreviewshaveconcludedthatreligiouspeopleenjoybetterphysical health,includingÒglobalself-ratingsofhealth;individualitemsandratingscalesassessingfunc- tionalhealthanddisabilitylimitation;physicalsymptomatology;theincidenceandprevalence ofcancer,bothoverallandsitespeciÞc;theincidenceandprevalenceofcoronaryheartdisease, hypertension,andcerebrovasculardiseaseÓ(LevinandChatters2008:160).OmanandThoren- son(2005:454)describetherelationbetweenreligionandphysicalhealthasÒrobust.ÓMyers (2008:336)declaresthatÒreligiousinvolvementrivalsnonsmokingandexerciseeffectsÓasa predictorofphysicalhealthandlongevity. Andyettherearealsodissentingvoices.Powellandcolleaguesdiscoveredthatnoneof threeÒwell-controlled,prospectivestudiesoftheelderly ... foundanyrelationshipbetweenthe religiousvariableandthedevelopmentofdisabilityÓ(Powell,Shahabi,andThorenson2003:43). InastudybyKoenigandVaillant(2009)thepositiveeffectofchurchattendanceonhealth disappearedovertime.Parketal.(2008)foundthatserviceattendancewasassociatedwithlower ratesofincreaseinthenumberofintermediateactivitiesofdailyliving(IADL)activities(e.g., preparingmeals)buthadnoeffectonbasicactivitiesofdailyliving(BADL)suchasdressing CorrespondenceshouldbeaddressedtoJoonmoSon,DepartmentofSociology,NationalUniversityofSingapore,Faculty ofArtsandSciences,AS1#04Ð24ArtsLink,Singapore117570.E-mail:socioson@nus.edu.sg JournalfortheScientiÞcStudyofReligion(2011)50(3):588Ð603 C  2011TheSocietyfortheScientiÞcStudyofReligion RELIGIOSITY,PSYCHOLOGICALRESOURCES,HEALTHandundressing.Benjamins(2004)foundthatreligiousattendanceamongtheÒoldestoldÓhelpeddelaytheonsetoffunctionalimpairmentÞveyearslaterbutreligioussalienceincreasedfunctionalimpairment.Kelly-MooreandFerraro(2001)detectednoeffectofreligiousserviceattendanceonfunctionallimitations.ArecentstudyofrespondentsintheWisconsinLongitudinalStudy,whoweretrackedfrom1993(whentheywere53Ð54yearsold)until2004(whentheywere64Ð65yearsold),foundnoassociationbetweeneitherchurchattendanceorreligiousimportanceandself-ratedhealth(BrennerandSiegl2008).TheseinconsistenciesledEllisonetal.(2010:327)todescribetheevidenceforalinkbetweenreligionandphysicalhealthasÒinconclusiveÓandtobemoanÒanabsenceofcumulative,well-replicatedÞndingsonthetopicofreligionandphysicalhealth.ÓMusickandWorthen(2010:250)observethattheconnectionbetweenreligionandself-ratedhealthÒhasnotbeenwelldocumentedsince2000,ÓaddingthatitÒdeservesmoreattentionintheliterature.ÓInlightofthispreviousresearchtherearetwosolidreasonsforundertakinganotherstudyofthistopic.TheÞrstisempirical.CanthesalutaryeffectsofreligiononarangeofphysicalhealthmeasuresfoundinpreviousstudiesbereplicatedusinganationallyrepresentativesampleoftheU.S.adultpopulation?Thesecondistheoretical.CantheassociationbetweenreligionandphysicalhealthbeexplainedbythekindsofmechanismsthathavebeensuggestedbyexpertsintheÞeld?ExplanatoryMechanismsExplainingwhyreligionmighthavepositiveeffectsonphysicalhealthmeansindicatingwhatmightlinkthemtogether.Forthis,amediationmodelisnecessary,guidedbythepropositionÒthatreligionleadstocertainsocialandpsychologicaloutcomes,suchasself-esteemorsocialsupport,thatinturntendtobetterhealthÓ(MusickandWorthen2010:254).Potentialmediatorsfallintothreebroadcategories(EllisonandLevin1998;Ellisonetal.2010;Flannelly,Ellison,andStrock2004;LevinandChatters1998).TheÞrstisbehaviorsandlifestyles.ThistheorypostulatesthattherelationbetweenreligionandmorbidityÒis,toalargeextent,explainedbytheencouragementthatreligionorspiritualityprovidesforlivinghealthierlifestylesÓ(Powell,Shahabi,Thorenson2003:42).Manycongregationsrunprogramsandprovidecounselingser-vicestoprotectthehealthofmembersandpromotehealthylifestyles.Religiousinvolvementcanalsoinhibitbehaviorsinjurioustohealth,suchasexcessivealcoholortobaccoconsumptionanddrugabuse,anddiscourageriskybehaviorsthatendangerhealthorcausestress.Forexample,religiousyouthinAmericaarelesslikelytoengageinhealthcompromisingbehaviorandmorelikelytotakecareofthemselves(WallaceandForman1998).Highlevelsofreligiousobser-vanceleadtogreateruseofpreventivehealthcaresuchasmammogramsandPapsmears(IdlerThesecondcategoryofmechanismsissocialnetworksandsocialsupport.ÒChurchesandsynagogues,aswellasthesmallgroupstheyengender,mayenhanceperceptionsofsupportbyfosteringasenseofcommunitythatleadsindividualmemberstofeelloved,caredfor,valuedandintegratedÓ(EllisonandLevin1998:706).Morethanothervoluntaryassociations,congre-gationsÒofferamixofstrongandweaktiesandcanthusofferabroadrangeofbothemotionalandpractical,instrumentalsupportsÓ(Idler2009:140).ThishelpsexplainwhyÒchurch-basedemotionalsupportoffsetstheeffectsofÞnancialstrainonself-ratedhealthinlaterlifeÓwhereassupportfromseculargroupsdoesnothavethesameeffect(Krause2006:S40).Thethirdcategoryofmechanismsispsychologicalfactors,sometimesreferredtoasÒre-sourcesÓbecausetheymakeachievinggoodhealtheasier(George,Ellison,andLarson2002:195).Religiouspeoplearelesslikelytoexperiencenegativepsychologicalstates.Ameta-analysisof147studiesfoundanaveragecorrelationbetweenreligiousinvolvementanddepressionofÐ.10, JOURNALFORTHESCIENTIFICSTUDYOFRELIGIONincreasingtoÐ.15forstudiesofstressedpopulations(Smith,McCullough,andPoll2003).versely,religiouspeoplearemorelikelytoenjoypositivepsychologicalstates.Manysuggestionshavebeenmadeastowhythisshouldbeso,including:religiousbeliefsinstillthefeelingthatadi-vinebeinglovesyouandthatyouhaveapersonalrelationshipwithadivineother,whichenhancesself-worth,efÞcacy,andmastery;supportfromfellowchurchmembersenhancesself-appraisal;religiousteachingsandpracticeshelpmanagenegativeaffectsuchasanger,fear,andhostility,andencouragepositiveemotionssuchashopeandoptimism;abeliefthatGodÕswillisexpressedineventsinstillsasenseofpurposeinandcontroloveroneÕslife;andreligiousguidancefostersafeelingofcalm,reassurance,andabilitytocopewithstressstemmingfromillnesses,portrayingthemasopportunitiesforspiritualgrowthoraspartofalargerplan(EllisonandLevin1998:707;Krause2010;MusickandWorthen2010:254;OmanandThoresen2005:446;Ryff,Singer,andLove2004:95).Psychologicalresources,inturn,haveapositiveeffectonphysicalhealth(Kaplan2007:105;Krause2010:9).Self-esteemispositivelyrelatedtobetterself-ratedhealth(Schieman2002:637)andfunctionalability(RietzesandMutran2006:S49).Conversely,depressioninolderadultsin-creasestheriskoflaterphysicaldisability(Pennixetal.1999).Psychologicalresourcescaneveninducebiologicalchanges.Frequentchurchgoershavelowerallostaticload,themeasureofthecumulativewearandtearofvariousphysiologicalsystemsasmeasuredbybiologicalmarkers,andpsychologicalfactorscouldwellaccountforsomeofthis(Ellisonetal.2010;Maselkoetal.2007).Eudemonicwell-being(purposefullifeengagement)ispositivelyrelatedtolowinßammationresponse,lowerlevelsofglycosylatedhemoglobin,lowerwaist-to-hipratios,andlowertotal/high-densitylipoprotein(HDL)cholesterolratios(Ryff,Singer,andLove2004:1390).Hedonicwell-being(positivemoods,contentmentwithlife)alsohasbiologicalbeneÞts(Huppert2009;PressmanandCohen2005;Saloveyetal.2000).Forexample,onestudyshowedanegativeeffectofemotionalwell-beingontheincidenceofstrokeinolderadults(Ostiretal.2001),anothershowedthatpositivemoodhelpsbufferagainstcardiovascularresponsetostress(Fredricksonetal.2000),andanothershowedthatpositiveaffectandemotionsinßuencethesecretionofthestresshormonecortisol(Laietal.2005).Ingeneral,prolongedreactivitytostressisharmfultoimmunefunctionandtootherphysiologicalprocesses,whilearapidrecoveryfromstressisbeneÞcialforhealth.Insummary,thereisgoodreasontobelieve,onthebasisofpriorresearchandtheoryaboutpossiblemechanisms,thattheeffectofreligiononphysicalhealthcanbepartlyexplainedbypsychologicalresources.AnalyticalPlanAlthoughthemediationmodelisoftenreferredtoinreligionandhealthstudies,ithasrarelybeentested.Therearesomenotableexceptions.KoenigandVaillant(2009:123)hypothesizedthatregularchurchattendancewouldprotectpeopleagainstpoorphysicalhealthbecauseitwouldÒincreaseoneÕslevelofcontentment,includingthroughincreasedsocialsupport,amorepositiveoutlookonlife,[and]increasedhopeandencouragement.ÓTheyfoundthatÒbettermoodsÓex-plainedsomeoftheeffectofchurchattendanceonsubsequentphysicalhealth.However,theirstudywaslimitedtomales.IdlerandKasl(1997)useda12-yearfollow-upofasampleofAmericanstolookattheinßuenceofchurchattendanceonfunctionaldisability.Theyhypoth-esizedthatsomeoftheeffectofreligiousattendanceonfunctioningwouldbeexplainedbyhigherlevelsofpsychologicalwell-beingoffrequentchurchgoers.Althoughtheyfoundthatonlyoptimismmediatedthereligiouseffectonlaterfunctioningthestudyneverthelesssupportsthetheorythatpsychologicalresourcescanhelpexplainwhyreligionaffectsphysicalhealth.ThisstudywasconÞnedtotheelderly.Insum,althougheachstudyhasitsowndatalimitations,theySeealsoEllison,Burdette,andHill(2009),HackneyandSanders(2003),andShreve-NeigerandEdelstein(2004). RELIGIOSITY,PSYCHOLOGICALRESOURCES,HEALTHbothsupportthegeneralideathatpsychologicalresourcescouldexplaintheinßuenceofreligiononphysicalhealth.ThepreferredmethodoftestingthemediationmodelistousethreewavesofdatafromalongitudinalsurveywherethesecondwavemediatingvariableisconsequentupontheÞrstwaveandantecedenttothethirdwave.TheNationalSurveyofMidlifeintheUnitedStates(MIDUS)hasonlytwowavesofdata,collectedin1995and2005,butin1995respondentswereaskedanumberofquestionsaboutwhatthingswerelikewhentheyweregrowingup,oneofwhichaskedthemtorecallhowmuchimportancewasattachedtoreligioninthehome.RatherthancallthisvariableÒparentalreligiosityÓasinRossi(2001:302),weusethetermhomereligiositytorefertothefactthatitmeasuresthesalienceofreligioninthehomeinwhichtherespondentwasraised.WetreattheseretrospectivedataasTime1becausetheyrefertoastatethatclearlyexistedpriorto1995.Wetreat1995asTime2,thestageatwhichtomeasurethemediatingfactor,and2005asTime3,thetimetomeasuretheoutcomevariable.Wehypothesizethathomereligiosityhasapositiveeffecton1995psychologicalresources,whichinturnhaveapositiveeffecton2005physicalhealth.AlthoughthesubjectsinMIDUSrangedinagefrom25to74in1995whentheywereaskedaboutreligioninthefamilyoforiginwealreadyknowthatthereligiosityofoneÕsparentscanhavelong-lastingconsequences.KrauseandEllison(2007:122)foundthatrecalledparentalreligioussocializationpracticeshadapositiveeffectontheself-esteemofadultsaged66ormore.Inaddition,becauseweareabletomeasuretherespondentÕsreligionin1995wecanestimatebothmediatedandunmediatedeffects.Thefamilyoforiginreligiosityprobablyinßuences1995religion,whichinturncouldhaveapositiveeffecton2005physicalhealthregardlessofthepathwaythroughpsychologicalresources.Nevertheless,ifthemediationhypothesisisvalid,partoftheinßuenceofhomereligiosityon2005physicalhealthwillbefunneledthroughpsychologicalresourcesin1995.Toestimatethemediatingeffectsofpsychologicalresourcesweusethreescalesofwell-beingincludedinMIDUS.TheÞrstofthescalesiscalledÒhedonicwell-beingÓbecauseitisrootedintheideaofpleasure,happiness,contentment,thesatisfactionofhumanappetites.ThesecondscaleiscalledÒeudemonicwell-beingÓbecauseitreferstopurposefullifeengagement(Ryff,Singer,andLove2004).Thethirdisasocialwell-beingscaletheinspirationforwhichcomesfromadeÞnitionofhealthissuedbytheWorldHealthOrganization(WHO)in1948inwhichreferencewasmadetoÒsocialÓaswellasphysicalandmentalwell-being(WHOMIDUScontainsawiderangeofhealthmeasuresandinthisstudywesubjectthemalltoanalysis.Self-ratedhealthasksrespondentstogiveanoverallassessmentoftheircurrentphysicalhealth,rangingfrompoortoexcellent.AskingrespondentstogiveanoverallassessmentoftheirhealthisawidelyusedmeasureofhealthstatusandisconsideredtobeanaccuratemeasureofapersonÕsphysicalhealthstatus(Bjorner,Fayers,andIdler2005:314).Themeasureispredictiveofchronicdiseaseincidence,recoveryfromillness,andfunctionaldecline(IdlerandKasl1995:S35;ShieldsandShooshtari2001:37).Becausemoststudiesonthistopichaveusedonlyself-ratedhealthastheoutcomemeasure(Kelley-MooreandFerraro2001:S366),otherhealthmeasureshavebeensomewhatoverlooked.Forexample,Benjamins(2004:357)observesthatÒonlyasmallnumberofstudiesconcentrateonfunctionallimitations.ÓWethereforedrawuponMIDUSformeasuresoffunctionallimitations,usingascaleofBADLmeasuringlimitationsondailyroutineactivitiessuchasbathingoneselfandanotherscaleonintermediateactivitiesofdailyliving(IADL)measuringlimitationsonmorechallengingactivitiessuchasclimbingstairsorwalkingseveralblocks.Inadditiontothesehealthmeasureswealsolookat:physicalsymptomatology,whichmeasurestheincidence(inthepastmonth)ofphysicaldiscomfortssuchasheadaches,hotßashes,andproblemssleeping;andthe JOURNALFORTHESCIENTIFICSTUDYOFRELIGIONexperienceofchronicmedicalproblemssuchasconstipation,ulcers,asthma,highbloodpressure,anddiabetes.Weusethenationalrandom-digit-dialingsamplefromMIDUS,atwo-wavepanelsur-vey.Eligiblerespondentswerenoninstitutionalized,English-speakingadultsinthecoterminousUnitedStatesbetweenagesof25and74.ThebaselinenationalRDDsamplewasselectedin1995fromworkingtelephonebanks.Malesbetweenagesof65and74wereoversampled.Therespondentsparticipatedinacomputer-assistedtelephoneinterviewandalsocompletedtwoself-administeredquestionnairebookletsmailedtotheirhouseholds.Thesampleconsistsof3,487respondents.Theresponserateestimatesare70percentforthetelephoneinterview,86.8percentforthecompletionoftheself-administeredquestionnaires,and60.8percentforthecombinedresponse(i.e.,.700Afollow-upsurveyoftheoriginalMIDUSsamplewasconductedbetween2004and2006.Themortality-adjustedretentionrateofthenationalRDDsampleis71percent.Multivariatelogitregressionofattritionrevealedthatthosewhofailedtorespondtothesecondwaveweremorelikelytobenonwhitemaleswithloweducationandincomelevel.Inlightoftheattrition,weemploymultiply-imputeddatathroughoutouranalysesusingMplus6.1,astructuralequationmodeling(SEM)package(Arbuckle1996;Graham2009;PeughandEnders2004;Rubin1976;Schafer2003).ThisprocedureproducesparameterestimatesbyaveragingthesetofanalysesontheÞvemultiply-imputeddatasets,theirstandarderrorsbeingcalculatedonthebasisoftheaverageofthestandarderrorsoverthesetofanalysesandthebetween-analysisparameterestimationvariation(MuthenandMuthen2007).TheimputeddatasetswerealsoweightedtocorrectforunequalstratiÞedprobabilitiesofhouseholdandwithin-householdrespondentselectionatthebaseline.ThesampleweightpoststratiÞedthesampletomatchtheproportionsofadultsinthe1995CurrentPopulationSurveyinregardtoage,gender,race,education,maritalstatus,MSA(i.e.,metropolitanandnonmetropolitanstatisticalareas),andregion(Northeast,Midwest,South,andWest).TheÞnalsamplecountofthemultiply-imputeddatasetsis3,257excluding228respondentswhodiedbetweenthetwowavesandtwowhoarenotcoveredbytheweightvariable.VariablesTable1providesabriefdescriptionofeachofthevariablesusedintheanalysistogetherwithresponsecategories,meansscores,standarddeviations,andrange.ESULTSAsshowninTable1respondentsreportedsigniÞcantlyincreasednumberofphysicalsymp-toms,BADL,andIADLacrossthetwowaves(.001),buttheirself-ratedhealthandchronicmedicalproblemsdidnotchangesigniÞcantly.AllthreeobservedindicatorsofthepsychologicalresourcesindicatesigniÞcantlossesin10years(.001).ThemeanforhomereligiosityÑ2.16ona0Ð3scaleÑisquitehigh,indicatingthatmostrespondentsrecalledlivinginahomewherereligionwasimportant.Theaverageageofrespondentsin1995was43,55percentwerewomen,86percentwerewhite,67percentweremarried,and69percentwereworkingfull-time.AdditionalinformationontheconÞrmatoryfactorloadingsofreligionacrosstwowavesisprovidedinFigureAofAppendixS1alongwithdetailsfortheindividualitemsinthethreepsychologicalresourcesscales(TableA).AppendixS1maybeaccessedatwileyonlinelibrary.com. RELIGIOSITY,PSYCHOLOGICALRESOURCES,HEALTHTable1:Variablesintheanalyses(multiply-imputeddata[3,257],sampleweighted) MeasureMeasureDescription Endogenousmeasure(2005)Religion(factor)Asecond-orderlatentfactorofpublicreligion(i.e.,churchattendance,religiousmeetings,andactivities),religiousidentiÞcation(sixindicators:e.g.,religiosity,importanceofreligion),andreligiouscoping(i.e.,religioushealing,religiousdecisionmakingofdaily-lifematters).HealthoutcomesSelf-ratedhealthAnitemaskingtherespondent:ÒIngeneral,wouldyousayyourphysicalhealthis?Ó(1poor,2fair,3good,4verygood,5excellent;mean1.02;range:1Ð5).PhysicalsymptomsAnine-itemsummatedscaleaskingwhether,inthepastmonth,therespondenthadexperiencedsymptomsofheadaches,lowerbackaches,sweatingalot,andsoforth(0notatall,1amonth,2severaltimesamonth,3onceaweek,severaltimesaweek,5almosteveryday;mean7.17;range:0Ð45).ChronicmedicalproblemsA29-itemsummatedscaleaskingif,duringthepast12months,therespondenthadexperiencedorbeentreatedforconditionssuchasasthma,constipation,diabetes,stroke,arthritis,orvaricoseveins(0no,1yes;meanrange:0Ð29).BADLAtwo-itemsummatedscalemeasuringhealthlimitationsonbathingordressingoneselfandwalkingoneblock(0notatall,1alittle,2some,3alot;meanrange:0Ð6).IADLAseven-itemsummatedscalemeasuringhealthlimitationsonintermediateactivitiesofdailylivingsuchasliftingorcarryinggroceries,climbingseveralßightsofstairs(0notatall,1little,2some,3alot;mean6.07;range:Psychologicalresources(factor)Asecond-orderlatentfactorofthefollowingthreewell-beingmeasures(seeAppendixTableAforspeciÞcindicatorsofeachwell-beingmeasure).Emotionalwell-beingThesummatedscaleconsistsofpositiveaffectandlifesatisfaction.Theformerismeasuredbysixquestionsinwhichtherespondentwasaskedhowmuchtimeduringthepast30daystheyfelt(1noneofthetime,2alittle,3most,5all).Forthelatter,therespondentwasaskedtoÒratetheirlifeoverallthesedaysÓonascalerangingfrom0to10(mean10.73;range:1Ð33).Psychologicalwell-beingThesummatedscale,derivedfromtheworkofRyff(1989),measureshowmuchindividualsseethemselvesasthrivingintheirpersonallife(mean13.42;range:1Ð104).Socialwell-beingThesummatedscaleisanÒappraisalofoneÕscircumstancesandfunctioninginsocietyÓ(Keyes1998:122)(mean11.23;range:1Ð81).Exogenousmeasure(1995)Seeabovefordescriptionsofthemeasuresexceptthehomereligiosity. JOURNALFORTHESCIENTIFICSTUDYOFRELIGIONTable1( MeasureMeasureDescription Religion(factor)HealthoutcomesSelf-ratedhealthMean.97;range:1Ð5PhysicalsymptomsMean7.01;range:0Ð44ChronicmedicalproblemsMean2.51;range:0Ð21BADLMean.95;range:0Ð6IADLMean5.17;range:0Ð21Psychologicalresources(factor)Emotionalwell-beingMean5.47;range:1Ð40Psychologicalwell-beingMean13.60;range:1Ð99Socialwell-beingMean12.46;range:1Ð89HomereligiosityAretrospectiveitemasking,Òhowimportantwasreligioninyourhomewhenyouweregrowingup?Ó(0notatallimportant,notveryimportant,2somewhatimportant,3veryimportant;mean.87;range:0Ð3).Controls(1995)AgeAcontinuousvariableinyears(meanrange:20Ð74).FemaleAdichotomousvariable(1female,0male;mean.50;range:0Ð1).WhiteAdichotomousvariable(1white,0other;mean.35;range:0Ð1).MarriedAdichotomousvariable(1married,0notmarried;.47;range:0Ð1).EducationAvariableindicatingthehighesteducationalgradeoftherespondent(1somegradeschooltosomehighschool,GEDorhighschooldiploma,3somecollege[nobachelorÕsdegree],4bachelorÕsdegreeormoreadvanceddegree;mean.96;range:1Ð4).IncomeA31-categorymeasureofpersonalincomeinthepastyear9.76;range:1Ð31).EmployedAdichotomousvariableindicatingemploymentstatusinthepastyear(1workedfull-time[35hours/week],0[workedpart-time(lessthan35hours/week),noworkorworkedlessthansixmonthsinthepastyear,orfull-timestudent];mean.46;range:0Ð1). WeusetheSEMpathanalysistotestthehypothesizedpsychologicalmediationofhomereligiosityonhealthoutcomes.IntheÞrststageofSEMtestingwefoundthatsocialwell-beinghadnoeffectonanyhealthoutcome.Wethereforedroppedsocialwell-beingfromfurtheranalysis.Psychologicalandemotionaleffectswerefoundforself-ratedhealthandphysicalsymptomsbutnottheotherthreehealthmeasures(i.e.,chronicmedicalproblems,BADL,andIADL).Wethereforereportonlythemediatorypathsfromhomereligiositytopsychologicalandemotionalwell-beingtoself-ratedhealthandphysicalsymptoms.First,Figure1-1testsifhomereligiosityaffectsthe2005self-ratedhealththroughthe1995psychologicalwell-being.EventhoughhomereligiosityisnotsigniÞcantlyrelatedtothe1995self-ratedhealth,itaffectspositivelythepsychologicalwell-beingoftherespon-dentsin1995(.10),andpsychologicalwell-beinginturnsigniÞcantlyenhances2005self-ratedhealth(.07).Apartfromthemediationeffect,the1995self-ratedhealthalsoaffects RELIGIOSITY,PSYCHOLOGICALRESOURCES,HEALTHFigure1-1SEMpathanalysisofreligion,psychologicalwell-being,andself-ratedhealth 2005 Religion2 2005 Self-ratedhealthR2 Psychological 1995 Self-ratedhealthR2 Religion2 Psychological Home 3,257.Allestimatesarestandardized.ModelÞts:CFI(comparativeÞtindex)=.98,TLI(Tucker-Lewisindex).96,RMSEA(rootmeansquareerrorofapproximation).03.The1995and2005religionvariablesareconÞrmatorylatentvariableswiththreeindicatorsateachwave:publicreligion,religiousidentiÞcation,andreligiouscoping.MLM(maximumlikelihoodparameterestimateswithstandarderrorsandamean-adjustedchi-squareteststatistic)estimatorappliedduetothenonnormalityofendogenousmeasures.TheanalysesemployedÞveweightedmultiply-imputeddatasets..001(twotailed).Figure1-2SEMpathanalysisofreligion,psychologicalwell-being,andself-ratedhealth Religion2 1995 Self-ratedhealthR2 Emotional 2005 Religion2 2005 Self-ratedhealthR2 Emotional Home :SeefullnotetoFigure1-1.the2005psychologicalwell-beingsothatcross-laggedeffectsbetweenself-ratedhealthandpsychologicalwell-beingareidentiÞed.NotethatthemediatorypathremainssigniÞcantevenaftercross-sectionalcorrelationsbetweenthe1995self-ratedhealthandpsychologicalwell-being)andbetweenthe2005self-ratedhealthandpsychologicalwell-being(.08)aretakenintoaccount.HomereligiosityisalsosigniÞcantlyrelatedtothe1995religionbutthisdoesnotyieldhealthbeneÞtsbecausethe1995religiondoesnotaffectthe2005self-ratedhealth.Figure1-2employsemotionalwell-beinginsteadofpsychologicalwell-being.Themediationhypothesisissupportedagain:homereligiositysigniÞcantlyincreases1995emotionalwell-being),whichinturnpromotesthe2005self-ratedhealth(.06 JOURNALFORTHESCIENTIFICSTUDYOFRELIGIONFigure2-1SEMpathanalysisofreligion,psychologicalwell-being,andphysicalsymptoms Religion2 Physical symptoms Psychological 2005 Religion2 Physical symptoms Psychological -.15*** .10** Home 3,257.Allestimatesarestandardized.ModelÞts:CFI(comparativeÞtindex).98,TLI(Tucker-Lewisindex).96,RMSEA(rootmeansquareerrorofapproximation).03.The1995and2005religionvariablesareconÞrmatorylatentvariableswiththreeindicatorsateachwave:publicreligion,religiousidentiÞcation,andreligiouscoping.MLM(maximumlikelihoodparameterestimateswithstandarderrorsandamean-adjustedchi-squareteststatistic)estimatorappliedduetothenonnormalityofendogenousmeasures.TheanalysesemployedÞveweightedmultiply-imputeddatasets..001(twotailed).Figure2-2SEMpathanalysisofreligion,psychologicalwell-being,andphysicalsymptoms Religion2 Physical symptoms Emotional .07 1995 Religion2 Physical symptoms Emotional Home :SeefullnotetoFigure2-1.Inthenextstep,weusephysicalsymptomsastheendogenousmeasuresreplacingself-ratedhealth(Figure2-1).Thesamemediatorypathworks:homereligiosityincreasesthe1995psychologicalwell-being(.10)andthe1995psychologicalwell-beingdecreasesthenumberof2005physicalsymptoms(Ð.07).Inaddition,becausehomereligiositydecreasesthe1995physicalsymptomssigniÞcantlyitislikelythathomereligiosityhasanindirecteffectonthe2005physicalsymptoms.Thepatternisrepeatedwhenemotionalwell-beingisemployedinsteadofpsychologi-calwell-beinginFigure2-2.Thatis,themediatorypathfromhomereligiositytoemotional RELIGIOSITY,PSYCHOLOGICALRESOURCES,HEALTHTable2:Standardizedtotal,direct,andindirecteffectsofhomereligiosityonhealth HealthOutcomes:Self-RatedHealth,PhysicalSymptoms BasedonanalysisrepresentedinFigures1-1and1-2Homereligiosityto2005self-ratedhealth(mediator:1995psychologicalwell-being)HRto2005SH(total)2005SH(direct)1995PsyW 2005SH(indirect)1995SH2005SH(indirect)1995R2005SH(indirect)Homereligiosityto2005self-ratedhealth(mediator:1995emotionalwell-being)HRto2005SH(total)2005SH(direct)1995EW 2005SH(indirect)1995SH2005SH(indirect)1995R2005SH(indirect)BasedonanalysisrepresentedinFigures2-1and2-2Homereligiosityto2005physicalsymptoms(mediator:1995psychologicalwell-being)HRto2005PS(total)2005PS(direct)1995PsyW 2005PS(indirect)1995PS2005PS(indirect)1995R2005PS(indirect)Homereligiosityto2005physicalsymptoms(mediator:1995emotionalwell-being)HRto2005PS(total)2005PS(direct)1995EW 2005PS(indirect)1995PS2005PS(indirect)1995R2005PS(indirect) :HRhomereligiosity,SHself-ratedhealth,PSphysicalsymptoms,PsyWpsychologicalwell-being,emotionalwell-being,Rreligion.TheSEMsoftware(Mplus)doesnotprovideindirecteffectsestimateswhenusingmultiply-imputeddata;thus(standardized)coefÞcientswereaveragedacrossÞvemultiply-imputeddata;werealsoaveragedacrosstheÞvedatasetstoproducetwo-tailed-valuesfor.001(twotailed).well-beingtothe2005physicalsymptomsturnsouttobesigniÞcant.Also,becausehomereli-giosityreducesthe1995physicalsymptoms,ithaslong-termeffectsonphysicalsymptomsin2005.NotethatmodelÞtsaresatisfactorygiventhatcomparativeÞtindexandTucker-Lewisindexvaluesaregreaterthan.95androotmeansquareerrorofapproximationvaluesaresmallerthan.05inallthepathanalyses(Yu2002).Thesevencontrolmeasuresweretakenintoaccountforallendogenousvariablesin1995and2005.Consideringthatourmaininterestliesinthemediatoryroleofpsychologicalresourcesbetweenparentalreligiosityandhealthoutcomes,indirecteffectsneedtobetested.BecausetheSEMpackagedoesnotprovideestimatesofthetotal,direct,andindirecteffectswhenmultiply-imputeddatasetsareused,weaveragedthestandardizedstructuralcoefÞcientsand-scoresforthepathsfromhomereligiositytothetwohealthoutcomesviathetwopsychologicalresourcesmeasures,runningeachoftheÞvemultiply-imputeddatasets.ThetoptwopanelsinTable2showthetotal,direct,andindirecteffectsofhomereligiositywhentheÞnalendogenousmeasureisthe2005self-ratedhealthandarebasedonFigures1-1and1-2.TheresultsconÞrmthatpsychologicalandemotionalwell-beingarebothsigniÞcantmediatorsbetweenhomereligiosity JOURNALFORTHESCIENTIFICSTUDYOFRELIGIONandthe2005self-ratedhealth.NotethatbecausetheotherthreepathsofdirectandindirecteffectsarenotsigniÞcantthetotaleffectsarenotsigniÞcanteither.Thenexttwopanelsatthebottomofthetablereportthetotal,direct,andindirecteffectsofhomereligiosityonthe2005physicalsymptomsandarebasedonFigures2-1and2-2.Themediationhypothesisissupported.Inaddition,theresultsalsoconÞrmthathomereligiosityisasigniÞcantsuppressorofthe2005physicalsymptomsthroughitseffectonthe1995physicalsymptoms;inotherwords,thosewhowereraisedinhomeswherereligionwasconsideredimportanthadsigniÞcantlyfewerphysicalsymptomsin1995,whichinturnmeanttheyhadfewerphysicalsymptomsin2005.ThemagnitudesofsuchindirecteffectsintermsofthestandardizedcoefÞcientsaresmall;however,theyarenotignorableconsideringthatwearedealingwithlife-longeffectofreligiousupbringingonhealthoutcomesthroughpsychologicalWetestedthehypothesisthattheinßuenceofreligiononphysicalhealthcanbeexplainedbythefactthatreligiouspeoplepossessmorepsychologicalresourcesandenjoybetterphysicalhealth.Wecombinedretrospectiveandlongitudinaldata,usingaquestionaskedintheÞrstwaveofMIDUSabouttheimportanceofreligioninthehomewhenrespondentswereyoung,measuresoftheirpsychologicalresourcesintheÞrstwave,andmeasuresoftheirphysicalhealth10yearslater,controllingfortheireffectsin1995.Thestudythereforemakesacontributionthatisbothempirical,inthatitaddstothebodyofknowledgeonthereligion-physicalhealthlink,andtheoretically,inthatithelpsexplainwhyreligionhaslong-termpositiveeffects.Wetestedformediationeffectsofthreetypesofpsychologicalresources.Oneofthese,socialwell-being,wasunrelatedtofuturephysicalhealthandwasdroppedfromtheanalysis.Thefactthatthesocialwell-beingscaledidnotpredictphysicalhealthinwhateverformwasinterestingbecausemanyscholarshavearguedforthehealthbeneÞtsofsocialintegrationandthesocialwell-beingscaleis,insomeparts,apsychologicalmeasureofsocialintegration.Indeed,thisisthenamegiventooneofthesubsetsofitemsthatincludesitemssuchasÒmycommunityisasourceofcomfort.ÓThissuggeststhatsocialintegrationmightnothavepsychologicalconsequencesforhealthbutthatitoperatesonphysicalhealthmainlythroughmoreinstrumentalmethodssuchasprovidingsupportintimesofillnessandenforcingnormsofhealthbehavior.Whiletherecanbenodoubtthatreligiouscongregations,aswellasthesmallgroupstheyengender,fosterasenseofcommunityÒthatleadsindividualmemberstofeelloved,caredfor,valued,andintegratedÓ(EllisonandLevin1998:706),itshouldnotbeassumedthatthisÒsenseofcommunityÓtranslatesintobetterphysicalhealth.Bothemotionalandpsychologicalwell-beingmediatedtheeffectofhomereligiosityonhealthbutonlyfortwooftheoutcomes:self-ratedhealthandphysicalsymptoms.Thequestionarisesastowhyhomereligiosityhadnoeffectontheotherhealthoutcomesthroughpsychologicalresources.Oneansweristhatself-ratedhealthandphysicalsymptomshavealargerpsycholog-icalcomponentthantheothermeasures.Itiswidelyacknowledgedthatself-ratedhealthisacombinationofÒrealÓphysicalconditionsandtheassessmentsofthoseconditionsÑanindicatorofhowhealthdisordersinßuenceoverallwell-being(Zimmeretal.2000:467).Inotherwords,peopleÕsratingsoftheirhealthÒarebasedonmorethanphysicalstatus,ÓoftenincludingfactorssuchasÞtnessandgeneralwell-being(ShieldsandShooshtari2001:37).Thismeansthatevenaftercontrollingforactualvariationinhealthstatuspsychologicalstateshaveanimpactonoverallself-assessments(TesslerandMechanic1978:258).Researchershavealsodiscernedapsycho-logicalcomponenttophysicalsymptomatology:Òthereisconsiderableevidencelinkingpositiveaffecttoreportsoffewersymptoms,lesspain,andbetterhealthbutthereisreasontothinkthatthisassociationmaybedrivenprimarilybyPA(positiveaffect)inßuencesonhowpeople RELIGIOSITY,PSYCHOLOGICALRESOURCES,HEALTHperceivetheirbodiesratherthanaffect-elicitedchangesinphysiologicalprocessesÓ(PressmanandCohen2005:939).ExperimentsconÞrmthis:ÒtheconsistenteffectofinducedmoodonsymptomreportingdoessuggestthatthereareconditionsunderwhichmoodcansystematicallyaltersymptomreportsÓ(Saloveyetal.Thehealthmeasuresnotaffectedbypsychologicalresources,namely,healthlimitationsondailyactivitiesandchronicmedicalproblems,couldbemoreinsulatedfrompsychologicalbias.TheyaresomewhatmorespeciÞc,moreconcrete,and,perhaps,moreobjective.Thiscouldexplainwhyreligionhasnoeffectonthemeitherdirectlyorindirectlythroughpsychologicalresources.ÒItisentirelypossiblethatgoingtoreligiousservicesandactivitieshasmanyhealthbeneÞtsthatincludelowerdepressionandbettersubjectivehealthÑbutthatattendancehasnoeffectonwhetherarespondentcanwalkupaßightofstairsÓ(Kelley-MooreandFerraro2001:S371).Andyetresearchhasshownthatevenwhenpeoplerespondtochronicillnessquestionstheirresponsesarebiasedbytheirpsychologicalstatesandtraits(PressmanandCohen2005:938).Forexample,higherpositiveaffectresultsinfewerreportedsymptomsamongpatientssufferingfromLymedisease,lupus,multiplesclerosis,coronaryheartdisease,andupperrespiratoryinfection(PressmanandCohen2005:939).Itmightthereforebeamatterofdegree.Allself-reportedhealthstatusesaresubjecttobiasbutsomemorethanothers.Thegoldstandardfortestingthepsychologicalmediationmodelshouldthereforebetheuseofthird-partyreportsontherespondentÕshealthstatus.Withoutthis,themostwecanconcludeisthatpsychologicalresourcesmakeadifferencetohowpeopledealwiththeirhealthstatusand,toanunknownextent,theiractualhealthstatus.OnebeneÞtofusinglongitudinaldataisthatreciprocaleffectscanbeestimated,thushelpinganswerthequestionwhetherreligionÒcausesÓhealthorviceversa.Wecanalsoseeifthepsychologicalresourcesarethecauseortheeffectofphysicalhealth.AsdisplayedintheSEMÞgures,the1995religionhasnoeffectoneitherself-ratedhealthorphysicalsymptomsnordotheearlierhealthmeasureshaveanyeffectonthe2005religion.Emotionalandpsychologicalwell-beinghavepositiveeffectsonbothself-ratedhealthandphysicalsymptomsbutonlyself-ratedhealthhasareciprocaleffect:thebetterthehealthofrespondentsin1995thebettertheirpsychologicalwell-beingin2005.Thisisstrikingevidenceofthereciprocaleffectbetweenpsychologicalwell-beingandphysicalhealthasratedbytherespondentbutfurtherconÞrmation,perhaps,thatself-ratedhealthhasapsychologicalcomponent.Inlightofthefactthatthe1995religionhadnodirecteffectonthe2005healthweexperi-mentedwithdifferentmeasurementsofreligiosity.Itisoftennotedbyscholarsinthisareathatonedimensionofreligionhasadifferenteffectonhealththananother(Ainley,Singleton,andSwigert1992).UsingthethreedimensionsofreligionidentiÞedearlier(publicreligion,religiousidentiÞcation,andreligiouscoping)weestimatedmodelsforallÞvehealthoutcomes.Innomodeldid1995religionaffectahealthoutcome.First,ourstudydesigncallsforaÒÞrstwaveÓthatmeasuresreligionpriortopsychologicalresources.MIDUSdoesnotprovideinformationonreligionpriorto1995exceptthequestionaskingrespondentsifreligionwasimportantinthehomewhentheyweregrowingup.ItwouldhavebeenpreferabletohaveameasureoftherespondentÕsownreligiositypriorto1995.ThereissomepossibilitythattheeffectofbeingraisedinareligioushomewastransientanddidnotsurviveadolescenceandisthereforenotanaccuratemeasureoftherespondentÕsreligiositypriorto1995.Andthehomereligiosityitemcouldbesubjecttorecallbiasbecausecurrentchurchgoersaremorelikelytoremembertheirhomesbeingreligious(PlochandHastings1998).However,thepathcoefÞcientfromparentalreligiositytorespondentÕsreligiosityislargeandhighlysigniÞcant),conÞrmingwhatmanyotherstudieshaveshownabouttheintergenerationaltransmissionofreligiosity. JOURNALFORTHESCIENTIFICSTUDYOFRELIGIONSecond,theabsenceofanymajorchangesinthehealthoftherespondentsinMIDUSlimitsthepotentialforsigniÞcantresultsinastudywiththisdesign.ThemeanIADLscorerosefrom3.53to5.08ona0Ð21scalebetween1995and2005.Inthesametime,themeanforchronicillnessrosefrom2.36to2.45onascaleof0Ð29.Thismightexplainwhyneither1995religionnor1995psychologicalresourceshadanyeffectonchangesinthreeofthehealthoutcomes.However,thiscannotbethewholestorybecauseself-ratedhealthdeclinedonlybyaverysmallamount,fromameanof3.51to3.49.Mostrespondentssaidtheywereinverygoodorexcellenthealthinbothwaves.Andyetpsychologicalresourceshadapositiveeffectonself-ratingsofhealth.Giventhehighinitialscoresitislikelythatreligion(throughpsychologicalresources)ishelpingmaintaingoodhealthandstavingofftheonsetofill-healthratherthanimprovingit.Third,wetestedonlyonemediationmodel,featuringtheroleofpsychologicalresources.Asindicatedearlier,thereareseveralpossiblemechanismslinkingreligiontophysicalhealth,includinghealthbehaviorsandsocialnetworks.TheseareundoubtedlyimportantbutourSEManalyseswouldhavebecomeextremelycomplicatedhadweinsertedadditional1995mediators(e.g.,ameasureoffrequencyofexercise,aquestiononsmoking,anitemonsocialnetworks)notleastbecausepathstoandfromthesefactorsandexisting1995measureswouldneedtobeestimated.Futurestudiesinthisareausingthismethodshouldsubstitutehealthbehaviorsorsocialnetworksforpsychologicalresources.Fourth,wecannotgointodetailwithrespecttothephysicalhealthmeasures.ThetermÒphysicalhealthÓcoversaverywiderangeofconditionsandalthoughMIDUSenablesustobreakdownhealthoutcomesintobroadcategoriesitishighlylikelythattheinßuenceofreligiononhealthisoccludedbythefacttheyareeachquiteheterogeneous.Forexample,itisassumedthatreligionaffectshealthbyshapinglifestylesandbehaviorsandtheseinturnhaveastrongereffectonsomehealthconditionsthanonothers.Religiousproscriptionsmighthaveastrongereffectondiseasesassociatedwithbehaviorssuchasovereating,drugabuse,heavysmokinganddrinking,andsexualpromiscuity.Inshort,religionmightÒtargetÓcertainillnessesbetterthanothers.Similarly,religiouscounseling,whereitisprovided,mighttargetcertainconditions,suchashypertension,betterthanothers.Inshort,ÒtheeffectofreligiononmorbiditymayvaryacrossspeciÞcdiseasesanddisordersÓ(EllisonandLevin1998:715).Thisstudysetouttomaketwocontributionstotheresearchonreligionandphysicalhealth.TheÞrstwasempirical.Inalarge,nationallyrepresentativesampleofadultAmericansaged25Ð75witha10-yearfollow-updoesreligiosityhaveapositiveeffectonphysicalhealth?Thesecondwastheoretical.Ifreligiositydoeshaveapositiveeffectonhealthwhatcouldexplainit?Thestudyisnoteworthyinitsuseofmultiplemeasuresofpsychologicalresourcesandhealthstatus.Theresultssuggestthattheassociationbetweenreligionandphysicalhealthisnotasrobustassomeotherwritershavestated.Instead,itreplicatespreviousstudiesthatfoundnoconnectionbetweenreligionandfunctionaldisability.Itindicatesclearlytheneedtoconsiderdiversemeasuresofhealthstatus.Itsuggeststhatpsychologicalresourceswillmediatetheeffectofreligionwhenhealthmeasureshaveanobviouspsychologicalcomponentandindicatestheneedtotestthismediationhypothesisonhealthoutcomesthatareobjective,third-partyassessmentsordiagnoses.Futurework,besidesconsideringarangeofphysicalhealthoutcomesandusingobjectivemeasures,shouldexperimentwithdifferentmechanisms,suchassocialnetworks,andshouldalsopayattentiontothepossibleinteractionbetweenreligiousinvolvementandotherfactorsknowntoinßuencephysicalhealth.Forexample,Schieman,Nguyen,andElliot(2003:208)ÞndthatreligiosityisaresourcethatbolstersmasterybutonlyamongpeoplewithgreatereducationalandÞnancialresources.Thisinteractionneedstobeembeddedinthemediationmodel.Thepowerof RELIGIOSITY,PSYCHOLOGICALRESOURCES,HEALTHpsychologicalresourcestomediatetheeffectofreligiononphysicalhealthmightwelldependonothercharacteristicsoftheindividual.Ainley,StephenC.,RoyceSingleton,Jr.,andVictoriaL.Swigert.1992.Agingandreligiousparticipation:Reconsideringtheeffectsofhealth.JournalfortheScientiÞcStudyofReligionArbuckle,JamesL.1996.Fullinformationestimationinthepresenceofincompletedata.InAdvancedstructuralequationmodeling:Issuesandtechniques,editedbyGeorgeA.MarcoulidesandRandallE.Schumacker,pp.243Ð77.Mahwah,NJ:LawrenceErlbaumAssociates,Inc.Benjamins,MaureenR.2004.Religionandfunctionalhealthamongtheelderly:Istherearelationshipandisitconstant?JournalofAgingandHealthBjorner,Jakob,PeterFayers,andEllenIdler.2005.Self-ratedhealth.InAssessingqualityoflifeinclinicaltrials:Methodsandpractice,editedbyPeterM.FayersandRonD.Hays,pp.309Ð24.Oxford,NY:OxfordUniversityPress.Brenner,PhilipandEricaSiegl.2008.Assessinglongitudinalrelationshipsbetweensocialfactorsandhealth.CDEWorkingPaperNo.2008Ð11.Madison,WI:UniversityofWisconsin.Ellison,ChristopherG.,AmyM.Burdette,andTerranceD.Hill.2009.Blessedassurance:Religion,anxiety,andtranquilityamongUSadults.SocialScienceResearchEllison,Christopher,RobertHummer,AmyBurdette,andMaureenBenjamins.2010.Race,religiousinvolvementandhealth.InReligion,familiesandhealth,editedbyChristopherG.EllisonandRobertA.Hummer,pp.321Ð48.NewBrunswick,NJ:RutgersUniversityPress.Ellison,ChristopherG.andJeffreyS.Levin.1998.Thereligion-healthconnection:Evidence,theoryandfuturedirections.HealthEducation&BehaviorFlannelly,KevinJ.,ChristopherG.Ellison,andAdrienneL.Strock.2004.Methodogicalissuesinresearchonreligionandhealth.SouthernMedicalJournalFredrickson,BarbardL.,RobertaA.Mancuso,ChristineBranigan,andMicheleM.Tugade.2000.Theundoingeffectofpositiveemotions.MotivationandEmotionGeorge,LindaK.,ChristopherG.Ellison,andDavidB.Larson.2002.Explainingtherelationshipsbetweenreligiousinvolvementandhealth.PsychologicalInquiryGraham,JohnW.2009.Missingdataanalysis:Makingitworkintherealworld.AnnualReviewofPsychologyHackney,CharlesH.andGlennS.Sanders.2003.Religiosityandmentalhealth:Ameta-analysisofrecentstudies.JournalfortheScientiÞcStudyofReligionHuppert,FeliciaA.2009.Psychologicalwell-being:Evidenceregardingitscausesandconsequences.AppliedPsychology:HealthandWell-BeingIdler,Ellen.2009.Healthandreligion.InThenewBlackwellcompaniontomedicalsociology,editedbyWilliamCockerham,pp.133Ð59.NewYork:JohnWileyandSons.Idler,EllenandStanislavKasl.1995.Self-ratingsofhealth:Dotheyalsopredictchangeinfunctionalability?JournalofGerontologyÑÑ.1997.ReligionamongdisabledandnondisabledpersonsII:Attendanceatreligiousservicesasapredictorofthecourseofdisability.JournalofGerontologyKaplan,Howard.2007.Self-referentconstructsandmedicalsociology:Insearchofanintegrativeframework.JournalofHealthandSocialBehaviorKelly-Moore,JessicaandKennethFerraro.2001.Functionallimitationsandreligiousserviceattendanceinlaterlife:Barrierand/orbeneÞtmechanism?JournalofGerontologyKeyes,CoreyL.M.1998.Socialwell-being.SocialPsychologyQuarterlyKoenig,LauraB.andGeorgeE.Vaillant.2009.Aprospectivestudyofchurchattendanceandhealthoverthelifespan.HealthPsychologyKrause,Neal.2006.Exploringstress-bufferingeffectsofchurch-basedandsecularsocialsupportonself-ratedhealthinlaterlife.JournalofGerontologyÑÑ.2010.Religionandhealth:Makingsenseofadisheveledliterature.JournalofReligionandHealthKrause,NealandChristopherEllison.2007.Parentalreligioussocializationpracticesandself-esteeminlatelife.ReviewofReligiousResearchKrause,Neal,ChristopherEllison,andJackMarcum.2002.Theeffectsofchurch-basedemotionalsupportonhealth:Dotheyvarybygender?SociologyofReligionLai,JulianC.L.,AliceM.L.Chong,SamuelM.Y.Ho,OswaldT.Siu,PhilD.Evans,RainbowT.H.Ho,PlatoChan,CharlesC.Chan,andSikHungNg.2005.Optimism,positiveaffectivityandsalivarycortisol.BritishJournalofHealthPsychology JOURNALFORTHESCIENTIFICSTUDYOFRELIGIONLevin,JeffreyandLindaM.Chatters.1998.Researchinreligionandmentalhealth:AnoverviewofempiricalÞndingsandtheoreticalissues.InHandbookofreligionandmentalhealth,editedbyHaroldKoenig,pp.33Ð50.SanDiego,CA:AcademicPress.ÑÑ.2008.Religion,agingandhealth:Historicalperspectives,currenttrends,andfuturedirections.JournalofReligion,SpiritualityandAgingMaselko,Joanna,LauraKubzinsky,IchiroKawachi,TeresaSeeman,andLisaBerkman.2007.Religiousengagementandallostaticloadamonghighfunctioningelderly.PsychosomaticMedicineMusick,MarcandMeredithWorthen.2010.ReligionandphysicalhealthamongUSadults.InReligion,familiesand,editedbyChristopherEllisonandRobertHummer,pp.248Ð72.NewBrunswick,NJ:RutgersUniversityen,LindaK.andBengtO.Muthen.2007.MplususerÕsguide.LosAngeles,CA:MuthenandMuthMyers,David.2008.Religionandhumanßourishing.InThescienceofsubjectivewell-being,editedbyMichaelEidandRandyJ.Larsen,pp.323Ð43.NewYork:GuilfordPress.Oman,DougandCarlThoreson.2005.Doreligionandspiritualityinßuencehealth?InHandbookonthepsychologyofreligionandspirituality,editedbyRaymondF.PaloutzianandCrystalL.Park,pp.435Ð59.NewYork:GuilfordOstir,GlenV.,KyriakosS.Markides,M.KristenPeek,andJamesS.Goodwin.2001.Theassociationbetweenemotionalwell-beingandtheincidenceofstrokeinolderadults.PsychosomaticMedicinePark,NanSook,DavidKlemmack,LucindaRoff,MichaelParker,HaroldKoenig,PatriciaSawyer,andRichardAllman.2008.ReligiousnessandlongitudinaltrajectoriesineldersÕfunctionalstatus.ResearchonAgingPennix,Brenda,SuzanneLeveille,LuigiFerrucci,JacquesvanEijk,andJackGuralnick.1999.Exploringtheeffectofdepressiononphysicaldisability.AmericanJournalofPublicHealthPeugh,JamesL.andCraigK.Enders.2004.Missingdataineducationalresearch:Areviewofreportingpracticesandsuggestionsforimprovement.ReviewofEducationalResearchPloch,DonaldandDonaldHastings.1998.Effectsofparentalchurchattendance,currentfamilystatus,andreligioussalienceonchurchattendance.ReviewofReligiousResearchPowell,LyndaH.,LeilaShahabi,andCarlE.Thorenson.2003.Religionandspirituality:Linkagestophysicalhealth.AmericanPsychologistPressman,SarahD.andSheldonCohen.2005.Doespositiveaffectinßuencehealth?PsychologicalBulletinReitzes,DonaldandElizabethMutran.2006.Selfandhealth:Factorsthatencourageself-esteemandfunctionalhealth.JournalofGerontologyRossi,Alice.2001.Developmentalrootsofadultsocialresponsibility.InCaringanddoingforothers,editedbyAliceRossi,pp.227Ð320.Chicago,IL:UniversityofChicagoPress.Rubin,DonaldB.1976.Inferenceandmissingdata.Ryff,CarolD.1989.Happinessiseverything,orisit?Explorationsonthemeaningofpsychologicalwell-beging.JournalofPersonalityandSocialPsychologyRyff,CarolD.,BurtonH.Singer,andGayleDienbergLove.2004.Positivehealth:Connectingwell-beingwithbiology.PhilosophicalTransactionsoftheRoyalSocietyofLondonSalovey,Peter,AlexanderJ.Rothman,JerushaB.Detweiler,andWayneT.Steward.2000.EmotionalstatesandphysicalAmericanPsychologistSchafer,JosephL.2003.Multipleimputationinmultivariateproblemswhentheimputationandanalysismodelsdiffer.StatisticaNeerlandicaSchieman,Scott.2002.Socioeconomicstatus,jobconditions,andwell-being:Self-conceptexplanationsforgender-contingenteffects.SociologicalQuarterlySchieman,Scott,KimNguyen,andDianaElliott.2003.Religiosity,socioeconomicstatusandthesenseofmastery.SocialPsychologyQuarterlyShields,MargotandShahinShooshtari.2001.Determinantsofself-perceivedhealth.HealthReportsShreve-Neiger,AndreaK.andBarryA.Edelstein.2004.Religionandanxiety:Acriticalreviewoftheliterature.PsychologyReviewSmith,TimothyB.,MichaelE.McCullough,andJustinPoll.2003.Religiousnessanddepression:Evidenceforamaineffectandthemoderatinginßuenceofstressfullifeevents.PsychologicalBulletinTessler,RichardandDavidMechanic.1978.Psychologicaldistressandperceivedhealthstatus.JournalofHealthandSocialBehaviorWallace,JohnM.andTyroneA.Forman.1998.ReligionÕsroleinpromotinghealthandreducingriskamongAmericaÕsHealthEducationandBehaviorWorldHealthOrganization(WHO).1948.PreambletotheconstitutionoftheWorldHealthOrganization.Geneva,Switzerland:WorldHealthOrganization.Yu,Ching-Yun.2002.EvaluatingcutoffcriteriaofmodelÞtindicesforlatentvariablemodelswithbinaryandcontinuousPh.D.dissertation.LosAngeles,CA:UniversityofCalifornia,DepartmentofEducation.Zimmer,Zachary,JoseÞnaNatividad,Hui-SengLin,andNapapornChayovan.2000.Across-nationalexaminationofthedeterminantsofself-assessedhealth.JournalofHealthandSocialBehavior RELIGIOSITY,PSYCHOLOGICALRESOURCES,HEALTHUPPORTINGNFORMATIONThefollowingsupportinginformationisavailableforthisarticle:AppendixS1.TableA:ScalesofpsychologicalresourcesandFigureA:ConÞrmatoryfactorloadingsofreligionacrosstwowaves.SupportingInformationmaybefoundintheonlineversionofthisarticleatwileyonlinelibrary.Pleasenote:Wiley-Blackwellisnotresponsibleforthecontentorfunctionalityofanysupportinginformationsuppliedbytheauthors.Anyqueries(otherthanmissingmaterial)shouldbedirectedtothecorrespondingauthorforthearticle.