Upload
others
View
3
Download
0
Embed Size (px)
Citation preview
RELIGIONANDPOLITICSINAMERICA:
ANEMPIRICALTESTOFINDIVIDUALISTVERSUSSOCIALMODELS
ThomasA.Hirschl([email protected])isProfessorofDevelopmentSociologyatCornellUniversity.HereceivedhisPh.D.attheUniversityofWisconsin‐Madison,andhaspublishedwidelyonsocialclassandthelifecourse.Fax:607/254‐2896;phone:607‐255‐1688;address:DepartmentofDevelopmentSociology,WarrenHall,Ithaca,NY14853.JamesG.Booth([email protected])isProfessorofStatisticsandComputationalBiologyatCornellUniversity.HereceivedhisPh.D.attheUniversityofKentuckyandisafellowofAmericanStatisticalAssociation. LelandL.Glenna([email protected])isAssistantProfessorofRuralSociologyatPennsylvaniaStateUniversity.HereceivedaPh.D.insociologyattheUniversityofMissouri,andaMasterofDivinity(MDiv)fromHarvardUniversity.BrandnQ.Green([email protected])isPh.D.CandidateinRuralSociologyatPennsylvaniaStateUniversity.
ABSTRACT
ThispaperexaminesempiricaldataonvotingbehaviorinU.S.Presidentialelections,andweinterpretthefindingsusingasocialmodel.Theanalysisproceedsthroughthreestages:(1)wereviewcontemporarysocialscienceliteratureonU.S.Presidentialvoterchoice,(2)weproposeasocialmodelofvoterchoicewherereligiouseffectsonvoterchoicearecontextualizedbytheindividual'srelationshiptosociety,and(3)weempiricallyanalyze2008Presidentialelection,buildinguponanearlierpaperanalyzingthe1980‐2000elections.The2008analysispresentedhereincludesanewmeasureofreligiousidentitythatfitsthedatabetterandbolstersthecaseforthesocialmodel.Inaddition,thealternativemeasureisvalidatedbyaseparatesurvey.Theanalysisfindsthattheaffectofreligiousidentityonvoterchoiceencompassesasocialdimensiondefinedbytheindividual’srelationshiptosociety,andoperationalizedbyrace,class,andgender.Keywords:socialclass,voterchoice,religion,ideology,andpolitics.Acknowledgements:LisaDundonreadearlydraftsofthispaper,andprovidedvaluablefeedback.
2
TheroleofreligioninU.S.politicsisalong‐standingsocialsciencequestion
thatbeginswiththefoundingoftherepublicandensuingdebatesaboutthe
separationofchurchandstate.1Thisquestionre‐surfacedinrelationtothereligious
right'spoliticalpartisanshipthatgatheredmomentumduringthe1970s,and
reachedanapexinPresidentGWBush's2000and2004electoralvictories.
AttemptsbyDemocraticPartyactiviststocounterwitha"religiousleft"appearto
havebeenlessthansuccessful,perhapsbecausetheDemocraticcoalition
encompassesasecularelementthatviewsreligionasarusetoswayvotersaway
fromeconomicconcerns.2Thispartisancontextraisesanewsocialsciencequestions
aboutthestrengthandcharacterofreligiousinfluenceswithinU.S.politics.
ThispaperanalyzestheroleofreligiousidentityinU.S.Presidentialvoter
choiceinordertoaddressthefollowingtwoquestions.First,isreligiousidentity,in
fact,aninfluenceonvoterchoice?Second,totheextentinfluenceisfound,whatis
thecharacterofsaidinfluence?Doestheconcept"voterpreference"bestdefinethe
influence,oristheinfluencemoreeffectivelyconceptualizedinsomecollective,or
social,sense?
Inansweringthesequestionsweupdateandbuilduponanearlierpaperthat
comparedtherelativeeffectivenessofanindividual‐levelvoterpreferencemodel
againstasocialmodelinformedbyclassicaltheory.3Inthepreviouspaper,
empiricaltestswereconductedforU.S.Presidentialelectionsinanationalsample
coveringthe1980to2000elections.Thecurrentpaperisbaseduponaseparate
nationalsamplethatrecoversvoterchoiceforthe2008Presidentialelection,and
includesanalternativemeasureofreligiousidentitydesignedbythestudyauthors.
3
Thetestresultssuggestthatthealternativemeasureimprovesmodelfit.Finally,we
validatethealternativemeasureinaseparatesurveyofuniversityundergraduates,
andfindthatthemeasurehasvaliditywithinthisalternativecontext.
CURRENTRESEARCHONRELIGIOUSIDENTITYANDVOTERCHOICE
Muchoftheempiricalworkonreligionandpoliticspresumesthatreligious
identitycanbemeasuredsolelybyindividualvariables.Suchapproachestendto
deployone,ormore,ofthreedomainsofreligiousidentity:denominational
affiliation,frequencyofchurchattendanceor“religiouscommitment,”andreligious
beliefs.4Differentresearchersmayemphasizeoneormoreofthesedomainsover
theothers.ForexampleManzaandBrooksdeploya“multicategorydenominational
scheme”5toclassifyindividuals’self‐identifieddenomination,andthenmodelthe
extentthatthesecategoriespredictvoterchoiceandpartisanship.Muchoftheir
workisderivedfromanalyzesoftheNationalElectionStudies,andamongtheir
findingsareapost‐1980compositionalshiftof“evangelicalProtestants”and
CatholicstowardtheRepublicanParty,ashiftof“mainlineProtestants”towardthe
DemocraticParty,butlittle,orno,netreligious/politicalpolarization.Intheir
analysesManzaandBrooksregardthevalidityofdenominationalaffiliationinmuch
thesamewaythatscholarsanalyzingsocialclassregardoccupationalclassification
schemes.Thisissomewhatsurprisinggiventhesymbolic/ideationalcharacterof
religionincomparisontotherelativelymaterialcharacterofwork,andthe
functionaldivisionoflaborinsociety.
4
Otherresearchersobjecttoemphasizingdenominationalaffiliationto
measurereligiousidentity,arguingthatitfailstocapturethefullimportofreligious
behavior,andinparticulartheeffectofreligiousbeliefsthat“transcend
denominationaldoctrine."6StudiesbyLaymanutilizingtheNationalElectionSurvey
suggestthatreligiouspoliticalcleavagehas,infact,widenedconsiderablysince
1980aroundtheaxisof“doctrinalorthodoxy”definedbyacompositescoreof
beliefsabouttheBibleandwhetherrespondentsbelievetheyare“bornagain.”7In
accordwithManzaandBrooks,Laymanfindslittleornopost‐1980political
polarizationindenominationalaffiliation,howeverindividualswhoespouseliteral
interpretationsoftheBible,and/orbelievetheyare“bornagain,”havebecome
strongerRepublicanvotersincomparisontoindividualsespousingmoreliberaland
secularbeliefswhohavebecomestrongerDemocraticvoters.
FindingsparalleltoLayman’sarearticulatedwithinananalysisofanational
samplecollectedbyDriskell,EmbryandLyon.8Thisstudytestsfortheeffectsof
religiononpoliticalparticipation,andthereligiousmeasuresincludedenomination
aswellasbeliefs,specificallyquestionsaboutsocialjusticeandtheworldlyroleofa
deity.Thetestresultssuggestthatwhenindividualsbelievethatadeityintervenes
inworldlyaffairstheyarelesslikelytoparticipatepolitically;whenreligiousbeliefs
incorporateconcernsaboutsocialjustice,politicalparticipationismorelikely.The
multivariatetestresultsareconsistentwiththepropositionthattheeffectofbelief
isindependentoftheeffectofdenomination,andthestudyauthorscontendthat
relyingsolelyupondenominationunderstates,andmayevendistort,thestrengthof
thelinkbetweenreligiousidentityandpoliticalparticipation.Whencomparing
5
findingsfromtheDriskelletal.studytoLayman’sstudiesandtheBrooks/Manza
studies,itisapparentthatmeasurementdecisionsinfluenceresearchfindingsabout
religiouseffectsonpolitics.
Thereareadditionalapproachestomodelingreligionandpolitics,for
examplethe“civilreligion”modelthatfusespoliticswithreligion,ascribinga
transcendentgoaltopoliticalleadership,especiallywithregardtotheU.S.President
asthehighestelectedofficial.9Thisapproachcanbeoperationalizedusingsurvey
questionstoidentifycivilreligiousvoterswhomaybeinclinedtofavorcandidates
espousingcivilreligiousconcerns.Wimberly’sstudyofvotersandpublicofficialsin
Raleigh,NorthCarolinafoundevidencethatthisperspectiveexplainedacomponent
ofthe1972Presidentialvote(RichardNixonwastheputative“civilreligion”
candidate).10Yetanotherapproachinvestigatesreciprocalcausationbetween
religionandpolitics,andreportsevidencethatreligiousattendanceisaffectedby
politicalideologicalconservativism/liberalism,andvice‐versa,overtime,inacross‐
lagged,paneldesignutilizingtheNationalElectionStudy.11Thisstudysuggeststhat
religiousattendanceaffectspartyidentification,andthatpartyidentificationaffects
religiousattendance,raisingthepossibilitythatreligiousattendanceresultsfrom
theinteractionofsecularandsacredmotivationsand/orbehaviors.
Anissueencounteredbyresearchersattemptingtospecifyreligiouseffects
onpoliticsisthattheseeffectsmaynotbeuniformacrossindividuals.Forexample,
inastudyusingtheGeneralSocialSurvey,GreeleyandHoutreportevidencethat:12
6
LiteralinterpretationsoftheBibleandfrequentreligiouspractice
pushAfricanAmericanstowardtheDemocratsandwhitestowardthe
Republicans.Literalismintensifiesthediametricallyopposedpolitical
orientationsinthetwogroups;itpullsthemfurtherapartpolitically.
InthisinstanceAfricanAmericansandwhitesprovidingthesameresponsesto
GeneralSocialSurveyquestionsregardingbeliefsabouttheBible,andfrequencyof
churchattendance,haveopposedpoliticalpartisanship.Herereligionappearsasa
forcethatunitesindividualsofthesameraceinpartisanship,whilesimultaneously
dividingtheracesagainstoneanother.
Allofthestudiescitedinthisliteraturereviewmodeltheeffectofreligionon
politicsusing,implicitlyorexplicitly,rationalchoicetheoriesofindividualbehavior
andbeliefs.Totheextentthatstructuresaretheorized,theyaretreatedasthe
unintendedconsequencesofindividualactors.Religionasasocialstructureis
conceptualizedasanaggregationofindividualpreferencesthatservesasaresource
thatenablespeopletoactontheirbeliefs.13Incontrasttothisperspective,classical
theoristspositasocialtheoryofreligionandpoliticsthatisbeyondtherealmof
individualpreferences.
DURKHEIMANDMARXONRELIGION,IDEOLOGY,ANDSOCIALBEHAVIOR
InTheElementaryFormsofReligiousLifeDurkheimanalyzed
hunter/gathererarchivalrecordsandconcludedthatsocietyisthe“source”of
religion,andthatreligiousbeliefsandritualsservetoidealizethematerialand
7
mentaldependencyofindividualsuponsociety.14Farfrombeingtheaggregationof
individualpreferencesforDurkheim,religionprovidesthesharedcategoriesof
thoughtthatsustainssocialcohesion.
Durkheim’sdefinitionofreligionextendsbeyondtheboundariesofformal
religiousorganizations,andisrelatedtoMarx’sconceptionofideologyandsocial
reproduction.Marxviewedformalreligionasanindicatorofnegativesocial
conditions,abeliefsystemthatoppressedpeopleturntothewaysickindividuals
turntoopium.15NeverthelessDurkheimsharesMarx’smaterialisticontological
assumption,andacknowledgesthisinhisstatementthatthe“materialsubstrate”
setsparametersforintellectuallife.16
MarxandEngel’stheoryofsociety17(i.e.,“historicalmaterialism”)directly
impliessocialdependencyofindividuals,andtheirvariouswritingsdemonstrate
thatindividualstendtoexperiencethisdependencyinideologicalorreligious
terms.17Volume1ofCapitalisrepletewithreligiousreferencesandmetaphors.For
example,thetheoryofcommodityfetishismisintroducedwiththefollowing
sentences:“Acommodityappears,atfirstsight,averytrivialthing,andeasily
understood.Itsanalysisshowsthatitis,inreality,averyqueerthing,aboundingin
metaphysicalsubtletiesandtheologicalniceties.”18ThispassagereflectsMarx’s
themethatreligiousandideologicalunderstandingsanimatepopularconceptions
aboutsociety,andinthisinstancefacilitatethefetishismofcommoditieswhere
socialrelationshipsaretransposedontorelationshipsbetweenobjects.
Marx’stheoryofcommodityfetishismistheoreticallyparalleltoDurkheim’s
treatmentofthetoteminhunter/gatherersociety.Inbothinstancessocialrelations
8
intrinsictosocialreproductionaretransposedontothings.Thistheoreticalparallel
illustratestheircommonpropositionthatreligionand/orideologyrepresent
idealized,popularknowledgesystemsthatfacilitatesocialreproduction.ForMarx,
religionandideologyobscurethematerialinterestsofsocialrelationships,and
impedethetransformationofcapitalistsocialrelations.ForDurkheim,religious
beliefsandritualsperpetuatesocialcohesion.BothMarxandDurkheimdefine
religionand/orideologyasknowledgesystemsthatreflectandidealizethe
dependenceofindividualsuponsociety.
Recognizingthesocialdimensionofreligionoffersastarkcontrastto
rationalchoicetheories,whichtendtoattributereligiousmotivationtoindividual‐
levelpreferences,andfailtoconsiderhow“religiouscapital”maybeapportioned
unevenlyandindependentlyofindividualpreferences.However,awholesale
adoptionofasocialtheoryofreligionisnotwithoutpitfalls.Durkheim’sandMarx’s
theoriesofreligiontreatindividualbehaviorasthemereexpressionofthesocial
structure.Whatiscalledforisatheorythatincorporatestheindividual’s
dependencyonsocietywhilestillrecognizingthatreligiousexpressionandits
politicalconsequencesarenotpredetermined.
ASOCIALTHEORYOFRELIGIOUSIDENTITYANDVOTERCHOICE
ThemodelweproposebearsarelationshiptoBourdieu’sconceptofsocial
capitalthatislinkedtotheunequaldistributionofpower.Forexample,Bourdieu
describeshowlanguagebecomesamarkerofclass,sinceitisanindicatorofasocial
group’sabilitytoengageincollectiveactiontoachievesocialreform.19Similarly,we
9
willargue,avoter’sreligiousbeliefmaybelinkedtoaparticularpartisan
preference,contingentuponthelocationofthatvoterwithinthestratificationorder.
Socialclass,race,genderrepresentdifferentialmaterialandideologicalinfluences
onhowindividualsexperiencesociety,anddifferentiallypatternreligiousinfluences
onpoliticalpartisanship.
Severalimplicationsfollowfromourproposedsocialtheoryofreligion,
ideology,politics,andsociety.First,religious/ideologicalbehaviorreflects
individualandcollectiveagencyinthecontextofsocialdependency,arelationship
thatindividualswilltendtoapprehendinsymbolicterms.Hencethestandard
measurementapproachthatattributesreligiousmotivationsolelytoindividual‐
levelpreferencesismisguided.Somesocialconstructthatrepresentsthis
dependencyiscalledfortomodelthesocialcharacterofreligion/ideology.Second,
totheextentthatreligiouseffectsaretheorizedsolelyintermsofindividual
preferences,theclassicalmodelisobviated.Researchthatdeploysatheoryof
individualpreferencestointegratecontextualmeasureswithindividualmeasuresof
religiousbehavioristheoreticallyincongruouswiththesocialtheoryoutlinedin
priorparagraphs.20
Inanearlieranalysiswemodeledreligiousidentityintermsofitsdifferential
expressionacrossthestratificationorderdefinedbyrace,class,andgender.Our
rationaleisthatindividualsexperiencesocietydifferentially,dependingupontheir
sociallocation,andthatthesethreedimensionsrepresentareasonablefirst
approximationoftheAmericanpeoples’stratificationexperience.21Second,that
politicsallocatesreal,aswellasthesymbolic,resources,andthatanindividual’s
10
abilitytoforwardclaimsonpoliticallydistributedresourcesisrelatedtohis/her
locationinthestratificationorder.Hencetheindividual’sreligiousidentityin
relationtopoliticsandwilltendtoberelatedtohis/herpositionwithinthe
stratificationorder.Thisleadsustohypothesizethattheeffectofreligiousidentity
onvoterchoiceiscontingentonsociallocation.Instandardindividualistmodels,
conversely,religiousidentityishypothesizedhaveanindependenteffectonvoter
choice,andthisindependenteffectisbelievednottobecontingentonrace,gender,
orsocialclass.Totheextentthatreligiousidentityeffectsonvoterchoiceare
independentofrace,class,andgender,theindividualistmodelissupported,andthe
proposedsocialmodelisnotsupported.
DATASOURCESANDMEASUREMENTRATIONALE
InthispaperweextendandbuilduponanalysisoftheGeneralSocialSurvey
(GSS),anationally representative, repeat cross-section sample of the English speaking,
non-institutional population age 18 and over.22 The GSS study analyzed voter choice for
the six Presidential elections during the period 1980 to 2000, and the data were derived
from 17 annual waves of data comprising 5,543 respondents with complete sets of
independent and dependent variables.
The analysis reported in this paper is based upon two new data sources. First, the
2009 Cornell National Social Survey (CNSS) was conducted by the Survey Research
Institute at Cornell University. The survey is a random sample of 1,000 households
within the United States, selecting one household member age 18 years of age and older.
The survey contains questions about Presidential voter choice in 2008, and repeats GSS
11
survey questions used to construct the independent and dependent variables in the GSS
analysis. The third data source is a survey of students taking “Introduction to Sociology”
at Cornell University during the spring, 2010, and represents an attempt to validate a new
biblical authority scale described in following paragraphs.
ThedependentvariablefortheGSS/CNSSanalysesisthebinaryvoting
preferencefortheRepublicanversusDemocraticPresidentialcandidate.We
excludenonvotersandthirdpartycandidatesfromtheanalysisinordertofocus
uponcandidatechoiceforthetwomajorpoliticalparties.Othervariablesonthe
righthandsideincluderelativefamilyincomeusedtoproxytheindividual’ssocial
class.Althoughincomeisnotthefavoredapproachformeasuringsocialclass,itis
widelyacknowledgedtobeimplicatedinclassposition.23Werecodefamilyincome
intoquartileswherequartile1isthebottom25percent.Othervariablesinthe
analysisarerace(whiteversusblack)andgender.IntheGSS,blacksarethesole
minoritygroupsufficientlylargeformultivariatestatisticalanalysis.
TheGSS/CNSSindependentvariablereligiousidentityisoperationalizedby
cross‐classifyingtwomeasures:biblicalauthorityandreligioustradition.Biblical
authorityisoperationalizedbytheanswertothismultiple‐choicequestioninthe
GSS:24
1.WhichofthesestatementscomesclosesttodescribingyourfeelingsabouttheBible?a.TheBibleistheactualwordofGodandistobetakenliterally,wordforword.b.TheBibleistheinspiredwordofGodbutnoteverythinginitshouldbetakenliterally,wordforword.c.TheBibleisanancientbookoffables,legends,history,andmoralpreceptsrecordedbymen.
12
ItisnotnecessarythattherespondentactuallyreadtheBibleorhasdirecttextual
knowledgeofit.Rather,thevariablemeasureswhethertheindividualpossesses
“feelings”thattheBibleisthe“literal”wordofGod,versustheinspiredwordofGod,
orabookoffables.Thebiblicalliteralismresponse(a)inparticular,isakintoMarx's
commodityfetishismandDurkheim'stotemism,insofarasitrepresentsthe
projectionofabeliefsystem("yourfeelings")ontoanobject,inthiscasethe
ChristianBible.
Thereareseveralreasonswhybiblicalauthorityisanappropriatevariableto
operationalizereligiousidentity.First,Riesebrodt’scomparativestudysuggeststhat
biblicalauthorityisakeycomponentofChristianfundamentalisminAmerican
society,andspecificallyidentifiesfundamentalismwithbiblicalliteralism.25
RespondentswhostatethattheBibleistheliteralwordofGodarethereforeina
leveragedpositionvis‐à‐vistheRepublicanParty’sstrategyofattracting
fundamentalistsandreligiousconservativesduringthestudyperiod.26
BecausetheGSSanalysisdemonstratedthatthesocialandpolitical
implicationsofbiblicalbeliefdifferaccordingtoreligioustradition,weincludethis
variableinourCNSSmeasureofreligiousidentity.TheGSS/CNSSmeasureof
religioustraditionisderivedfromtheGSSquestion,“Whatisyourreligious
preference?”27,andisusedtocreateameasurethatcorrespondstothemajortwo
religioustraditionsinAmericansociety:ProtestantandCatholic,versusallother
religioustraditionsandnon‐religiousindividuals.IntheGSSanalysiswesubdivided
BaptistsfromotherProtestantsbecauseBaptististhelargestdenominationwithin
thelargesttradition;howeverwefoundnosubstantivedifferencesbetweenBaptists
13
andotherProtestants,andthereforedonotsubdivideProtestantsintheCNSS
analysis.
TheGSSbiblicalbeliefscaleperformedasanticipatedintheanalysis,
howeverthestudyauthorsdevelopedquestionsaboutitsinterpretation.Ingeneral
weareunsurehowtointerpretthevariousresponsesbeyondhypothesesdescribed
bytheresearchliterature.Withtheintentionofcreatingamoreaccurateand
encompassingscalewescriptedadditionalbiblicalauthorityquestionsthatare
askedontheCNSSsurvey.Thesequestionsaredescribedinfollowingparagraphs.
ANALYSIS
ResultsfromtheGeneralSocialSurvey
Hirschl,Booth,andGlennausedatafromtheGSStoinvestigatethelink
betweenvoterchoiceandreligiousidentityinsixPresidentialelectionsfrom1980
to2000.Theyuselogisticregressionmodelstoassesstheassociationbetweenthe
choiceof4,754whitevotersbetweenRepublicanandDemocraticPresidential
candidatesandseveralpredictorsincludingtheirlevelofbiblicalauthorityas
measuredbyresponses“a”,“b,”or“c”totheGSSbiblicalbeliefquestiondescribedin
earlierparagraphs.Blackvoterswerenotmodeledduetotheroughly9tooneodds
ofblacksvotingDemocraticacrosseachofindependentvariablecategories.
Specifically,let denotetheprobabilityofvotingRepublican(versus
Democrat)forapersonofsex (female=1,male=2),religioustraditionj(1=Baptist,
2=otherProtestant,3=Catholic,4=other),andbiblicalauthoritylevelk(a=1,b=0,c=‐
1),inincomequartilel.OnemodelconsideredbyHirschletal.is
14
(1),
where“ln”denotesnaturallogarithm, (category1isthe“baseline”
categoryofeachfactor),and isanindicatorforCatholics.Thevariables,LandI,
areinteger“scores”thatallowfor(linear)trendsinthelogoddswithincreasing
levelsofbiblicalauthorityandincome.Thetwointeractiontermsinmodel(1)allow
theslopeofthetrendwithincreasinglevelsofbiblicalauthoritytovary
systematicallywithincreasingincome,andforthebaselineslope(i.e.inthefirst
incomequartile)tobedifferentforCatholics.Thisismostparsimoniouslogistic
modelconsideredbyHirschletal.thatisconsistentwiththeGSSdataintermsof
themodeldevianceor statistic.28Thereisasignificantdeteriorationinthefitif
anyofthemodelcomponentsareremoved.
Themodelimpliesaconstantgenderdifferenceregardlessoftheother
factorsrepresentedbytheparameter .Hirschletal.reportanestimated
value, ,whichimpliesthattheoddsamalevotesRepublicanareabout35%
higherthanthoseforafemale(since ),controllingforreligioustradition,
income,andbiblicalauthorityresponse.Differencesbetweenthereligioustradition
categoriesarealsoconstantwiththeexceptionofCatholics.Inparticular,Hirschlet
al.foundthatdifferencesinthevotingpatternsofBaptistsandotherProtestantsare
negligible.
15
TheanalysisoftheGSSdatainHirschletal.indicatesthatthebiblical
authorityresponseishighlypredictiveofvotingpreferenceamongProtestants,but
muchlesssoamongCatholics.Forexample,theyestimatedtheoddsthatafemale
Protestant,inthehighestincomequartile,votedRepublicantobe0.97ifshegave
response“c”(fables)tothebiblicalauthorityquestion.Theseoddsessentially
doubleto1.90ifshegaveresponse“b”(spiritual),anddoubleagainto3.70ifshe
gaveresponse“c”(literal).
ComparisonwiththeCornellNationalSocialSurvey
TheCNSSisasurveyofadults,aged18andover,whoareresidentsofthe
continentalUnitedStates,administeredbytheSurveyResearchInstituteatCornell.
Weanalyzedresponsesofthe1,000adultssurveyedin2009.Thissurveyincluded
fivequestionsconcerningbiblicalauthorityincludingquestion1fromtheGSS
discussedinpriorparagraphs(hereafterreferredtoasbiblicalauthorityquestion
1).Respondentswerealsoaskedwhomtheyvotedforinthe2008Presidential
election(McCain/PalinorObama/Biden).
WhitevotersinterviewedbytheCNSSfavoredMcCainoverObamaby1.1to
one,andblackvotersfavoredObamaby30toone.Thismarginmakesitimpractical
tomodelblackvoters,andrepresentsandevenhighermarginthanwasfound
duringthe1980to2000electionswheretheDemocraticmarginamongblackvoters
was9toone.ConsistentwiththeGSSdata,biblicalliteralismwithintheCNSSwas
foundtobehigheramongblackvoterscomparedtowhitevoters:inboththeCNSS
16
andtheGSS,approximatelyoneinthreeblacksandoneinfourwhitesanswered“a”
toquestion1.
Wefitmodel(1)tothe449whiterespondentswithcompletedataonall
relevantquestions,withtheonlymodificationbeingthatBaptistswerenot
separatelyidentifiedfromotherProtestants.Theresidualdevianceforthefittothe
ungroupeddatais523.8(df=441).Aftergroupingtherespondentsinto72
categoriesformedbycross‐classificationbygender,religioustradition,income
quartile,andbiblicalauthoritylevel,thesamemodelhadaresidualdevianceof78.9
(df=62).Themodelfitstotheungroupedandgroupeddataareidenticalintermsof
parameterestimatesandstandarderrors.However,thefittothegroupeddatacan
beusedtoassesslack‐of‐fit,althoughadmittedlysomeofthecellcountsaftercross‐
classificationaresmallandsothevalidityofthechi‐squaredtestissomewhat
questionable.Withthiscaveatthefitofmodel(1)totheCNSSdataappearstobe
reasonable.Wefurthernotethatthetwointeractiontermsinthemodelareonly
borderlinesignificant,presumablybecauseofthesmallersamplessizeintheCNSS.
ThegeneralpatternsofassociationintheCNSSdata(e.g.asmeasuredbythe
signsofthecoefficients)aresimilartothosereportedbyHirschletal.basedonthe
GSSdata.However,therearesomestrikingdifferencesintermsofthemagnitudes
oftheassociations.Forexample,theestimatedgenderdifferencefromtheCNSS
datais whichtranslatesintoanodds‐ratiocomparingtheoddsofamale
votingRepublicantothoseofafemale,controllingforotherfactors,of2.32.
Moreover,theresponsetothequestiononbiblicalauthorityisevenmorepredictive
ofvoterpreference(particularlyamongnon‐Catholics).Theestimatedoddsthata
17
femaleProtestant,inthehighestincomequartile,votedRepublican,is0.53tooneif
shegaveresponse“c”(fables)tothebiblicalauthorityquestion.Theseoddsare
multipliedbyafactoroffiveto2.53tooneifshegaveresponse“b”(spiritual),and
byanotherfactoroffiveto12.15tooneifshegaveresponse“a”(literal).Inlater
paragraphsweprovideinterpretationaboutwhythedirectionoftheCNSSgender
andbiblicalauthoritycoefficientsarethesameastheGSScoefficients,howeverthe
slopesareamplified.Wenextconsiderhowthenewbiblicalauthorityquestions
informthemodel.
ACompositeMeasureofBiblicalAuthority
Inadditiontobiblicalauthorityquestion1,participantsintheCNSSwere
askedwhethertheyagreed(yesorno)withthefollowingfourstatementsaboutthe
Bible:
2.TheBibleshouldhelpguidepoliticaldecisions.
3.TheBibleistobereadliterally.
4.TheBibleiswithoutcontradiction.
5.TheBibleisanauthoritativedocumentwhichhasmoralrulesI
mustfollow.
Thus,theCNSSdatacontainsfivecategoricalresponsesrelatingtorespondents’
viewsofbiblicalauthority.
Inprincipleitispossibletoextendmodel(1)byaddingtermsassociated
withdummyvariablesforeachbinary(yes/no)response,butwiththepossibilityof
multi‐wayinteractionsaswellasmulticollinearity,thisapproachisnotveryfruitful.
18
Insteadwecreateacompositebiblicalauthorityindexbasedonallfivecategorical
variablesrelatingtofeelings/beliefsabouttheBiblethatcanbeusedinmodel(1)in
placeofthethreelevelscorebasedonQuestion1alone.
(Figure1abouthere.)
Thecompositeindexforeachindividualisdefinedastheir1stprinciple
coordinatescorefromamultiplecorrespondenceanalysisofthefivecategorical
variables.Multiplecorrespondenceanalysis(MCA)canbethoughtofastheanalog
ofprinciplecomponentsanalysisforcategoricalvariables.29Themultiple
correspondencemapisaplotoftheresponsesonthefirsttwoprincipledimensions.
ThismapisgiveninFigure1fortheresponsestothequestionandstatements
concerningviewsabouttheBible.Themapsuggestsanorderingoftheresponseson
a1‐dimensionalcontinuousscale,namelythe1stprinciplecoordinate.Thus,for
example,responses“a”and“c”toQuestion1are,respectively,atthepositiveand
negativeextremes,whileresponse“b”isinthemiddleofthescale.A“no”response
toStatement5isclosetotheresponse“c”forQuestion1atthenegativeendofthe
scale.Ontheotherhanda“yes”responsetoStatement5isinthepositivepartofthe
scale,butsomedistancefromthepositiveextreme.
Asavalidationexperimentweaskedstudentsin“IntroductiontoSociology”
(DSOC1101)atCornellUniversityduringthespringof2010torateeachofthe11
responsestoQuestion1,andStatements2‐5,concerningbiblicalauthority.The
studentswereaskedtousea5‐pointLikertscalewith1representingaresponse
19
indicatingtherespondentisa“non‐Christian/atheist/religiouscynic,”and5
indicatingan“evangelicalChristian/biblicalliteralist”respondent.Theaverage
ratingsgivenbythestudentsoftheelevenresponseswereremarkablyconsistent
withtheprinciplecoordinatescoresobtainedbyMCA.Theseaveragesarebasedon
108studentswhogavelogicallycoherentscores.Forexample,studentswho
assignedascoreofonetoeveryresponsewereeliminated.
(Table1abouthere.)
Table1showshowthestudentsratedtheelevenresponses,sortedaccording
toaverageLikertscore.1ThisorderingisthesameasthatbasedontheMCA1st
principlecoordinatewiththeexceptionthattheorderofresponses4.yand3.yare
switched.However,thescoresforthesetworesponsesarealmostidenticalbothin
theCNSSdataandthestudentsurvey.Infact,thespacingsofthescoresfromthe
twodatasetsareremarkablysimilar.Themaindifferenceisthatthestudentsrate
response“b”toquestion1closertothehigher(biblicalliteralist)endofthescale.
ResultsforCNSSusingtheCompositeMeasure
Table2givestheparameterestimatesandstandarderrorsobtainedby
fittingmodel(1)totheCNSS(whitevoter)datausingthesimplemeasureofbiblical
authoritybasedontheresponsestobiblicalauthorityquestion1alone,andalso
usingthecompositemeasurediscussedabovebasedonresponsestoquestion1and
20
statements2‐5.Thecompositemeasurehasbeenscaledsothat‐1istheaverage
scoreforparticipantswhogaveresponse“c”(fables)toquestion1,and1isthe
averageforthosewhogaveresponse“a”(literal).Thiswasdonesothatthescales
arecomparableinthetwomodelfits.Theparameterestimatesaresimilarinthe
twofits,butkeymodelcomponentssuchasgender,andthetwointeractionsterms
involvingthebiblicalauthorityscore,areconsiderablymoresignificantwhenthe
compositescoreisused.Moreover,theimprovementinthemodelfitisquite
dramaticwiththecompositescore,theresidualdeviancebeing479.7(df=437),
comparedwith523.8(df=441)withthesimplescore.Thecompositescore
effectivelyusesupfouradditionaldegreesoffreedombecauseitisbasedonfive
responsesratherthanone.
(Table2andFigure2abouthere.)
Figure2providesagraphicaldescriptionofmodel(1)deployingthe
compositebiblicalauthorityscale;thecompositescaleisalsothex‐axisinFigure2.
Notethatthelinesextendsbeyond‐1and+1becausetheseintegervaluesaresetto
theaveragevalueofarespondentwhoanswered“c”and“a,”respectively,tobiblical
authorityquestion1.Infacttherearerespondentswithmoreextremescoresthan
theaveragevaluesforrespondentsanswering“c”and“a,”andhencethegraphlines
extendbeyond‐1and+1.ThegraphlinesrepresentProtestant(red),Catholic
(green),andother(blue),andmenaresymbolizedbysquares,womenbytriangles.
Thedensityofsquaresandtrianglesrepresentssampledensity.They‐axisisthe
21
naturallogodds,andcomputedtobesymmetricalaboutzero.Valuesof‐1and+1
represent2.72toonepreferencefortheDemocratsandRepublicans,respectively;
valuesof‐2and+2represent7.39toonepreferencetheDemocratsand
Republicans,respectively;andvaluesof‐4and+4represent54.60toonepreference
fortheDemocratsandRepublicans,respectively.
SeveralfeaturesofFigure2arenoteworthy,inparticulartheextremesplit
withintheProtestanttradition.Atlowlevelsbiblicalauthority(Mean=‐1),andat
allincomelevels,ProtestantsleantowardtheDemocratsbyanoddsof
approximatelytwotoone;howeverathighlevelsofbiblicalauthority(Mean=+1)
ProtestantsarestrongRepublicanpartisans,andthepartisanshipoddsintensifyas
incomeincreases.Forexample,Protestantmenare7tooneRepublicanvotersatthe
lowestincomequartile,and50tooneatthehighestincomequartile.Catholics
followadifferentpattern.AtthelowestincomequartileCatholicsarepartisan
neutralatalllevelsofbiblicalauthority;asincomeincreasestheslopebecomesless
flat,andapartisansplitemergesalongthelinesofbiblicalauthoritywithinthe
third,andespeciallythefourth,incomequartiles.Finally,the“other”group,whichis
aresidualcollectionofreligiousandnonreligiousidentities,displayabiblical
authoritypolarizationpatternsimilartoProtestants,butnotasextreme.Thuswe
surmisethatincomeintensificationofreligiousidentityisgeneralthroughoutthe
sample:theslopesofthethreelinesincreaseacrosseachoftheincomequartiles,
suggestingthattheeffectofreligiousidentityonvoterchoiceiscontingenton
income,heretakenasanindicatorofsocialclass.
22
(Figure3abouthere.)
ROCcurvesshowingthepredictiveaccuracyofmodel(1)withsimpleand
compositescoresrespectivelyareplottedinFigure3.Thecurvesareplotsofthe
(in‐sample)sensitivity(theprobabilityapersoniscorrectlypredictedtovote
Republican)againstoneminussensitivity(theprobabilityapersonisincorrectly
predictedtovoteRepublican)astheclassificationcutoffvariesbetween1and0.
(ThemodelclassifiesapersonashavingvotedRepublicanatcutoffδifπ>δ,whereπ
istheirfittedprobabilityofhavingvotedRepublican.)TheROCcurveforthe
compositemeasureliesalmostuniformlyabovethatforthesimplemeasure
indicatingsuperiorsensitivityandspecificityatnearlyallvaluesofδ.Thus,it
appearsthatthecompositescoreisamuchmorerefinedmeasureintermsof
capturingtheassociationbetweenbiblicalauthorityandvoterchoice.
IntegratedModelingProcedures
Thestatisticalanalysisdescribedintheprevioussectioninvolvedtwo
distinctstages.Atthefirststageweconstructedanindexofbiblicalauthorityusing
MCA.Thisindexwasthenusedasapredictorofvotingpreferenceinalogistic
regressionmodel.Wealsoconductedasinglestage,integrated,analysisinwhichan
indexwasconstructedaspartofthemodelfittingprocessasfollows.Definethe
indexforagivensubjectasaweightedaverage, ,oftheir5
responses,where istheresponsetoquestion1,codedas1forresponse“a”,0for
“b”and‐1for“c”,andwhere istheresponsetoStatementsm=2,3,4and5,
23
coded‐1for“no”and1for“yes.”Thecoefficientsintheweightedaveragecanbe
estimatedwhilesimultaneouslyfittingmodel(1)byiterativelyupdatingthetwo
setsofparameters.Thisprocessresultedinanevenfurtherimprovedfit
( ,df=437),howeverthequalitativeconclusionsfromthefitted
parametersareessentiallyunchanged.TherawdataandRcodeforthismodelare
availablefromtheauthors.30
DISCUSSION
Inthispaperwesetouttoanalyzetherelationshipbetweenreligiousidentity
andvoterchoice,andre‐testasetofempiricalfindingsidentifiedbyanearlier
analysis.Theearlieranalysisfoundthatthattheaffectofreligiousidentityonvoter
choiceisnotindependent,butratheriscontingentontheindividual’sgender,class,
andracestatus.Thepresentanalysisisconsistentwiththeearlierstudy,although
samplesizelimitationspreventafullexplorationofthenuancesofgender
dependency.31Thepresentanalysisdeploysanewbiblicalauthorityscalethatis
foundtoprovideabettermodelfitincomparisontothescaleusedintheearlier
analysis.
Beforeinterpretingthefindings,wefirstdelineatewhatwebelievetobethe
twomostsignificantlimitationsofthepresentstudy.Thefirstisthatthestudy
samplesizeis445whitevotersand62blackvoters;allbuttwooftheblacksreport
votingforObama.Thislimitsthenumberofparametersthatcanbemodeledforthe
whitesub‐sample,andreiteratesthestrongDemocraticleaningamongblacksfound
inthepriorstudyofsixPresidentialelections.Whilethewhitesamplesizerestricts
24
thecomplexityofthemodelsthatcanbeconsidered,itpresentsanopportunityfor
empiricaltestingfocusedontherevisedbiblicalauthorityscale.Additionallythe
revisedbiblicalauthorityscaleisvalidatedbyasurveyofuniversity
undergraduates.
Asecondlimitationofthepresentstudyisthatitmeasuresoneperiodof
timeandonePresidentialelection,andmaynotberepresentativeofothertime
periods/Presidentialelections.Howeverbecausethe2008modelresultsare
generallyconsistentwiththepreviousstudyofsixPresidentialelections,webelieve
wearemoreorlessjustifiedingeneralizingthe2008findingstothelonger,prior
periodoftime.
InapriorstudywedeployedabiblicalauthorityscaleavailableintheGSS
becauseitcametheclosest,amongavailableconstructs,tosatisfyingtheoretical
specificationsidentifiedbyourtheoryofChristianfundamentalismandthe
ideologicalfunctionoftheChristianBiblewithinAmericansociety.Inthepresent
analysisweproposewhatwebelievetobeanimprovedbiblicalauthorityscale,and
empiricalanalysiswithinanationallyrepresentativesurveysuggeststhatit
comprisesahierarchical,one‐dimensionalscale.Thescalehierarchyisconfirmedby
asurveyofuniversityundergraduates,suggestingithasrobustvalidity.Thuswe
findevidencethatthescaleisasociallyvalidmeasureofbiblicalauthority.
Asecondfeatureofthenewbiblicalauthorityscaleisthatitdramatically
improvesstatisticalfitwithinamodelofPresidentialvoterchoice,incomparisonto
theGSSscale.Itbetterexplainstherelationshipbetweenreligiousidentityandvoter
choice(seetheROCcurvesinFigure3,andrelativemodeldeviancesdescribedin
25
priorparagraphs),andthisimprovedfitsuggeststhatbiblicalauthorityisa
componentofreligiousidentityinfluencinglargenumbersofvoters,andespecially
Protestantvoters,whenevaluatingvoterchoice.Furthermore,whenthescaleis
estimatedaspartofasingle‐stagefittingprocedure,themodelfitisfurther
improved,providingadditionalevidencethatbiblicalauthorityrelatedreligious
identitypredictsvoterchoicedecisionsinU.S.Presidentialelections.
Thissetofempiricalfindingsisconsistentwiththesocialmodelproposedby
DurkheimandMarxwherereligionisasociallyderivedentitythatfacilitatessocial
cohesion.Specifically,itisapparentthatbiblicalbeliefservestounitesocially
definedgroupsinpoliticalpartisanship.Thisunityisnotderivedfromsacredtext,
butratheristransposedontothesacredtextbysocialgroupspossessingshared
experiences,ideologies,andmateriallocations.
Ifonedisagreeswiththisinterpretation,thenaquestionposedbythe
obtainedrelationshipbetweenreligiousidentityandvoterchoiceisthecharacterof
analternativeinterpretation.Doestheobtainedrelationshipsolelyreflectthe
agencyofindividualvoterchoice,ordoesasocialelemententerin?Second,ifthere
isasocialelement,whatisthecharacterofthisputativesocialelement?In
answeringthisquestionwerelyupontheconceptofsocialdependencyimpliedby
Durkheim’sandMarx’sanalysesofreligion,andreasonthatindividuals’experience
ofsocialdependencywillvary,dependingupontheirlocationinsociety.Specifically
weproposethatthepartisanconsequencesofreligiousidentityarecontingentupon
race,class,andgender.
26
ManysocialanalystsviewraceasthewidestdividewithintheAmerican
stratificationorder,andthevoterchoiceanalysishereinsuggestsitisalsothe
widestdivideinpoliticalpartisanshiplinkedtoreligiousidentity.Whereassocial
class,gender,andreligiousidentitysplitwhitevoterpartisanship,theblackvoteis
stronglyDemocraticacrosseachofthesedimensions.Whitebiblicalliteralistsare
strongRepublicanvoters,andblackbiblicalliteralistsandareequallystrong
Democraticvoters,notbecausethesetwogroupsreaddifferentBibles,butbecause
theirexperiencesofsocietyaredifferent,andhencetheirreligiousidentitiesand
politicalpartisanshiparecoloredbytheexperienceofrace.
Socialclassisalsoacontingencyintherelationshipbetweenreligious
identityandvoterchoice.Asincomerises,doctrinalpolarizationwithinreligious
traditionssharpens(seeFigure2).Thustheexperienceofsocialclassleadsto
differingpoliticalconsequenceswithinthesamereligioustraditionsandbiblical
beliefsets.
Inthisanalysiswelackthesamplesizetoexploregendernuancesidentified
bythepreviousstudywherebiblicalbeliefandincomewereco‐contingenciesinthe
effectofgenderonvoterchoice.Thepresentanalysisidentifiesaconstanteffectof
genderonvoterchoice,andwenotethatthiseffectamplifiedin2008,comparedto
the1980–2000Presidentialelections.Themagnitudeofthe2008gendereffect
overshadowsthe7percentgendergapreportedbytheexitpolls32becausethe
computedgendereffect,unliketheexitpollmargin,isnetofothervariablesin
model(1).Ourresultssuggestthathigh‐income,biblicalliteralist,Protestant
women,votedRepublicanbya12toonemargin.Thusbiblicalauthorityandsocial
27
class,inthecontextofaneteffectthatfavoredtheDemocraticcandidate,splitwhite
women’spartisanshipinthe2008Presidentialelections.
Incomparingthepresentpaperwiththepreviousone,severaldifferences
emerge,firstthatthe2008electionapparentlyamplifiedtheblackoddsofvoting
Democratic.ThisishardlysurprisingsincetheDemocraticcandidatewasan
African‐American,andwhosecandidacywasgenerallyhailedassymbolizingthe
forwardmarchofcivilrights.Second,the2008electionamplifiedtheoddsamong
whitemen,amongthewhiteupperclass,andamongwhitebiblicalliteralists,of
votingRepublican.Thuswesurmisethe2008electionwaspolarizingacrossrace,
gender,socialclass,andreligiouslines,incomparisontothe1980–2000
Presidentialelections.
Ourfindingregarding2008voterpolarizationisdistinctfromotherresearch
findingsregardingreligiouspolarization,forexamplethestudiesbyLaymanand
Driskelletal.describedinearlierparagraphs.Intheseotherstudiesreligious
identityistreatedasanindependentinfluenceonpoliticsthatderivesfrom
variationinindividualvoterpreferences.Inthepresentstudyreligiousidentityis
conceivedasnotonlyembeddedwithintheindividual;itisalsoembeddedwithin
society.Religiousidentitythereforeservesasamarkerforthepartisan
relationshipsofstratificationgroupswithinthesocialandpoliticalorder,reflecting
theeffortsofthesegroupstoobtainpoliticallydistributedresources,toseekredress
forperceived,orreal,injustices,etc.
Thefindingsreportedbythispapersuggestthattheaffectofreligious
identityonvoterchoiceencompassesasocialdimension,andisnotsolelyan
28
individualchoice.Thesocialdimensionisdefinedbytheindividual’srelationshipto
society,andcanbeoperationalizedusingstratificationcategories.Thusaparticular
religiousidentityisfoundtotriggerdifferentvoterchoices,contingentonthe
individual’srace,class,andgenderstatus.Finally,thefindingsareconsistentwith
theinsightsofDurkheimandMarxwhocharacterizereligionasapopular
worldviewrelatedtosocialcohesionandtosocialreproduction.
29
Notes
1.IsaacKramnickandR.LaurenceMoore,TheGodlessConstitution:AMoralDefense
oftheSecularState,(NewYork:W.W.Norton,2005).
2.ThomasFrank,What’sthematterwithKansas:HowConservativesWontheHeart
ofAmerica,(NewYork:MetropolitanBooks,2004);GeoffreyLayman,TheGreat
Divide:ReligiousandCulturalConflictinAmericanPartyPolitics,(NewYork:
ColumbiaUniversityPress,2001);KevinPhillips,AmericanTheocracy:ThePeriland
PoliticsofRadicalReligion,Oil,andBorrowedMoneyinthe21stCentury,(NewYork:
Viking,2006).
3.ThomasA.Hirschl,JamesG.BoothandLelandL.Glenna,“TheLinkBetweenVoter
ChoiceandReligiousIdentityinContemporarySociety:BringingClassicalTheory
BackIn.”SocialScienceQuarterly90no.4(2009):927‐944.
4.AndrewGreeleyandMichaelHout,TheTruthAboutConservativeChristians,
(Chicago:TheUniversityofChicagoPress,2006);DavidKnoke,“Religion,
StratificationandPolitics:Americaninthe1960s,”AmericanJournalofPolitical
Science18no.2(1974):331‐345;GeoffreyLayman,TheGreatDivide:Religiousand
CulturalConflictinAmericanPartyPolitics;JeffManzaandClemBrooks,“The
ReligiousFactorinU.S.PresidentialElections,1960‐1992,"TheAmericanJournalof
Sociology103no.1(1997):38‐81;StratosPatrikios,“AmericanRepublican
Religion?DisentanglingtheCausalLinkBetweenReligionandPoliticsintheUS,”
PoliticalBehavior30no.3(2008):367‐389;DarrenE.Sherkat,andChristopherG.
Ellison,“RecentDevelopmentsandCurrentControversiesintheSociologyof
30
Religion,”AnnualReviewofSociology25(1999):363‐394;andRobertD.Woodberry
andChristianS.Smith,“FundamentalismEtAl:ConservativeProtestantsin
America,”AnnualReviewofSociology24(1998):25‐56.
5.ManzaandBrooks,“TheReligiousFactorinU.S.PresidentialElections,1960‐
1992,"p.40;seealsoClemBrooksandJeffManza,“AGreatDivide:Religionand
PoliticalChangeinU.S.NationalElections,1972‐2000,”TheSociologicalQuarterly45
no.3(2004):421‐450;_____,“GroupSize,Turnout,andPoliticalAlignmentsandthe
DevelopmentofU.S.PartyCoalitions,1960‐1992,”EuropeanSociologicalReview15
no.4(1999):369‐389;_____,“TheGenderGapinU.S.PresidentialElections:When?
Why?Implications?,"TheAmericanJournalofSociology103no.5(1998):1235‐
1266.
6.GeoffreyLayman“ReligionandPoliticalBehaviorintheUnitedStates:TheImpact
ofBeliefs,Affiliations,andCommitmentFrom1980to1994.”ThePublicOpinion
Quarterly61no.2(1997):288‐316,p.310
7.ibid,p.291;seealsoLayman,TheGreatDivide:ReligiousandCulturalConflictin
AmericanPartyPolitics.
8.RobynDriskell,ElizabethEmbry,andLarryLyon,“FaithandPolitics:The
InfluenceofReligiousBeliefsonPoliticalParticipation,”SocialScienceQuarterly89
no.2(2008):294‐314.
9.Bellah,RobertN.,“CivilReligioninAmerica,”Daedalus117no.3(1988):97‐118.
10.RonaldC.Wimberly,“CivilReligionandtheChoiceforPresident:Nixonin’72.”
SocialForces59no.1(1980):44‐61.
31
11.StratosPatrikios,“AmericanRepublicanReligion?DisentanglingtheCausalLink
BetweenReligionandPoliticsintheUS,”PoliticalBehavior30no.3(2008):367‐
389.
12.GreeleyandHout,TheTruthAboutConservativeChristians,p.72.
13.SeeAndrewGreeleyforaconciseexplanationoftherationalchoiceperspective,
“ColemanRevisited:ReligiousStructuresasaSourceofSocialCapital,”American
BehavioralScientist40no.5(1997):587‐594.
14.EmileDurkheim,TheElementaryFormsofReligiousLife,(NewYork:Oxford
UniversityPress,2001),p.313.
15.AndrewM.McKinnon,"ReadingOpiumofthePeople:Expression,Protest,and
theDialecticsofReligion,"CriticalSociology31nos.1‐2(2005):15‐38.
16.Durkheim,TheElementaryFormsofReligiousLife,pp.318‐319.
17.KarlMarxandFrederickEngels..“PrefacetoAContributiontotheCritiqueof
PoliticalEconomy,”Pp.502‐506inKarlMarxandFrederickEngelsSelectedWorks,
Volume1,(Moscow:ProgressPublishers,1969);_____,“Feuerbach.Oppositionofthe
MaterialisticandIdealisticOutlooks,”Pp.16‐80inKarlMarxandFrederickEngels
SelectedWorks,Volume1,(Moscow:ProgressPublishers,1969);seeespeciallythe
descriptionofFrenchpeasantsinKarlMarx,The18thBrumaireofLouisBonaparte
[electronicresource],(London:ElectricBookCo,2001).
18.KarlMarx,Capital:ACritiqueofPoliticalEconomy,Volume1,(NewYork:
InternationalPublishers,1970),p.71.
19.PierreBourdieu,LanguageandSymbolicPower,translatedbyGinoRaymondand
MatthewAdamson,(Cambridge,MA:HarvardUniversityPress,1991),p.127.
32
20.MarkD.Regenerus,DavidSikkinkandChristianS.Smith,“Votingandthe
ChristianRight:ContextualandIndividualPatternsofElectoralInfluence,"Social
Forces77no.4(1999):1375‐1401.
21.DavidB.Grusky(ed.),SocialStratification:Class,Race,andGenderinSociological
Perspective,(Boulder,CO:WestviewPress,2003).
22.Hirschl,BoothandGlenna,“TheLinkBetweenVoterChoiceandReligious
IdentityinContemporarySociety:BringingClassicalTheoryBackIn.”
23.G.WilliamDomhoff,WhoRulesAmerica?:PowerandPolitics,4thEdition,(Boston:
McGrawHill,2002);RobertPerrucciandEarlWysong,TheNewClassSociety,3rd
Edition,(Lanham,Md:Rowman&Littlefield,2007);ErikO.Wright,“TheContinuing
RelevanceofClassAnalysis–Comments,”TheoryandSociety25no.5(1996):693‐
716.
24.JamesA.Davis,TomW.SmithandPeterV.Marsden,GeneralSocialSurveys,
19722004[CumulativeFile],(AnnArbor,MI:Inter‐universityConsortiumfor
PoliticalandSocialResearch,retrievedDecember12,2005,
http://webapp.icpsr.umich.edu/cocoon/ICPSR‐SERIES/00028.xml),p.197
25.MartinRiesebrodt,PiousPassion:TheEmergenceofModernFundamentalismin
theUnitedStatesandIran,(Berkeley:UniversityofCaliforniaPress,1993);PeterL.
Berger,“TheClassStruggleinAmericanReligion,”TheChristianCentury25
February(1981):194‐199.
26.JohnC.Danforth,“IntheNameofPolitics,”NewYorkTimes,March30,2005;
Phillips,AmericanTheocracy:ThePerilandPoliticsofRadicalReligion,Oil,and
BorrowedMoneyinthe21stCentury.
33
27.Davis,SmithandMarsden,GeneralSocialSurveys,19722004[CumulativeFile],
p.169.
28.Agresti,A.,CategoricalDataAnalysis,(JohnWiley&Sons,1990),Section3.3.2.
29.MichaelGreenacreandJörgBlasius,MultipleCorrespondenceAnalysisand
RelatedMethods,(BocaRaton,LA:Chapman&Hall/CRC,2006).
30.RDevelopmentCoreTeam,R:ALanguageandEnvironmentforStatistical
Computing,RFoundationforStatisticalComputing,(Vienna,Austria:http://www.R‐
project.org,2010).
31.Hirschl,Booth,andGlenna,“TheLinkBetweenVoterChoiceandReligious
IdentityinContemporarySociety:BringingClassicalTheoryBackIn,”seeModel2.
32.CenterforAmericanWomenandPolitics,TheGenderGap:VotingChoicesin
PresidentialElections,2008,DownloadedfromtheInternet4August2010,
http://www.cawp.rutgers.edu/fast_facts/voters/gender_gap.php#Facts.
34
Table1:RespondentRatingsofBiblicalAuthorityQuestion1andStatements2–5,108StudentsEnrolledinIntroductorySociology,CornellUniversity,Fall,2009*Response Q1.c S5.n S2.n S3.n S4.n Q1.b S5.y S2.y S4.y S3.y Q1.aAverageScore 1.37 1.48 1.70 1.96 2.34 3.56 4.41 4.51 4.73 4.75 4.89
*SeearticletextforexplanationofscoresderivedfromLikertscalemeans.
35
Table2:FitsofModel(1)totheCNSSDataUsingaSimpleMeasureofBiblicalAuthorityBasedSolelyonQuestion1,andaCompositeMeasureBasedonResponsestoQuestion1andStatements2‐5.***,**,*,and#indicatesignificanceatthe0.001,0.01,0.05and0.1levelsrespectively.
SimpleBiblicalAuthorityScore
CompositeBiblicalAuthorityScore
Parameter Estimate Std.Error Estimate Std.Error
Intercept 0.674* 0.306 0.435 0.325Gender=female
(male)
‐0.840*** 0.219 ‐0.923*** 0.231
Rel=Catholic(Protestant) ‐0.633* 0.254 ‐0.636* 0.283
Rel=Other(Protestant) ‐0.871** 0.287 ‐0.757* 0.306
BiblicalAuthority
(linearslope)0.458 0.356 0.754* 0.341
Income(linearslope)
0.064 0.098 0.245* 0.114
CatholicXBiblicalAuthority
‐0.771# 0.405 ‐0.987** 0.362
IncomeXBiblicalAuthority
0.278# 0.142 0.318* 0.143
36
Figure1:MultipleCorrespondenceMapofResponsestoQuestionandStatementsConcerningBiblicalAuthority*
*Thedatalabelscorrespondtoresponsestobiblicalauthorityquestion1andstatements2–5,e.g.,S5.ndenotesa“no”responsetostatement5,etc.Thefirstprinciplecoordinateexplains96.8%ofinertia.
37
Figure2:FittedNaturalLogOddsofVotingRepublicanVersusDemocratic(y‐axis),byBiblicalAuthority(compositescale;x‐axis),Income,Gender,andReligiousTradition
38
Figure3:ROCcurvesBasedonPredictionofRepublicanVoteUsingModel(1)*
*Thebluecurveisforthemodelfitusingthesimple“biblicalauthority”scorebasedonQuestion1,andtheredcurveisforthemodelfitusingthecompositescorebasedonQuestion1andStatements2‐5.