Incongruent Voting or Symbolic Representation? Asymmetrical Representation in Congress, 2008–2014

15 Pages·2020·0.552 MB·English

by  Adam Cayton and Ryan Dawkins

Article
Incongruent Voting or Symbolic
Representation? Asymmetrical
–
Representation in Congress, 2008 2014
AdamCaytonandRyanDawkins
Theelectoralconnectionincentivizesrepresentativestotakepositionsthatpleasemostoftheirconstituents.However,onvotesfor
whichwehavedata,lawmakersvoteagainstmajorityopinionintheirdistrictononeoutofeverythreehigh-profilerollcallsinthe
U.S.House.Thisrateof“incongruentvoting”ismuchhigherforRepublicanlawmakers,buttheydonotappeartobepunishedfor
itathigherratesthanDemocratsonElectionDay.Why?Researchinpoliticalpsychologyshowsthatcitizensholdbothpolicy-
specific and identity-based symbolic preferences, that these preferences are weakly correlated, and that incongruous symbolic
identity and policy preferences are more common among Republican voters than Democrats. While previous work on
representation has treated this factasanuisance, weargue thatitreflects tworeal dimensions ofpolitical ideologythat voters
usetoevaluatelawmakers.UsingfouryearsofCCESdata,district-levelmeasuresofopinion,andtheroll-callrecord,wefindthat
both dimensions of ideology matter for how lawmakers cast roll calls, and that the operational-symbolic disconnect in public
opinionleadstodifferentkindsofrepresentationforeachparty.
W
hydomembersofCongresssometimescastroll- More puzzling is the fact that Republicans tend to vote
callvotesthatareincongruentwiththeopinions against district opinion more frequently than Democrats
oftheirconstituents? Astrongelectoralconnec- even though the logic of the electoral connection should
tion between lawmakers and those they represent forms apply equally to both parties (Clinton 2006). Neverthe-
thefoundationofdemocraticrepresentationandcreatesa less, Republican lawmakers are not systematically voted
powerful incentive for lawmakers to follow district opin- outofofficeathigherratesthanDemocrats.
ion.Whenlawmakersbehaveinwaysthatareoutofstep Whatexplainsthis?Drawingonresearchfrompolitical
withtheirdistrict, citizens can,anddo,vote themoutof psychology, we argue that past scholarship on roll-call
office (Canes-Wrone, Brady, and Cogan 2002; Mayhew voting and representation has incorrectly conceptualized
1974).Despitethiselectoralincentive,membersofCon- constituency preferences by dismissing symbolic attach-
gress frequently cast votes contrary to the preferences of menttoideologicallabelsasasourceofrealattitudesused
their constituents, even on the most high-profile issues. to evaluate legislative position taking. Indeed, political
AlistofpermanentlinkstoSupplementalMaterialsprovidedbytheauthorsprecedestheReferencessection.
*DatareplicationsetsareavailableinHarvardDataverseat:https://doi.org/10.7910/DVN/2IYSI8
AdamCaytonisAssistantProfessorofPoliticalScienceattheUniversityofWestFlorida([email protected]).Hisresearchfocuses
onlegislativerepresentation,onhowfactorsotherthanpolicypreferencesinfluencepolicymaking,andhowinstitutionsinteract
withthepoliticalenvironment.HehaspublishedinLegislativeStudiesQuarterly,PoliticalResearchQuarterly,andState
Politics&PolicyQuarterly.
RyanDawkins isAssistantProfessorofPoliticalScienceattheUnitedStatesAirForceAcademy([email protected]).
Hisresearchisfocusedprimarilyinpoliticalpsychology,publicopinion,andelectoralbehavior.Heisinterestedinattitudes
towardpoliticalinstitutions,representation,andinter-grouprelations.HisworkhasbeenpublishedinPoliticalPsychology,
ElectoralStudies,UrbanAffairsReview,andStatePolitics&PolicyQuarterly.
TheywouldliketothinkAdamZelizerandJeffHardenfortheirthoughtfulcommentsonearlierdraftsofthispaper.Theyalso
thanktheanonymousreviewerswhoofferedchallengingandinsightfulsuggestions.Theirinsightsgreatlyimprovedthepaper.
doi:10.1017/S1537592720003746
©TheAuthor(s),2020.PublishedbyCambridgeUniversityPressonbehalfoftheAmericanPoliticalScienceAssociation 1
Downloaded from https://www.cambridge.org/core. Serials Section, on 22 Dec 2020 at 19:36:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1537592720003746
Article | Incongruent Voting or Symbolic Representation?
psychology research has shown that operational policy preferences.Conversely,Republicansaremorelikelythan
preferencesandsymbolicideologyaredistinctdimensions Democratstoapproveofrepresentativeswhovoteinline
ofideologythatarenotnecessarilystronglycorrelatedwith withtheirideologicalidentity,eveniftheysometimesvote
one another (Converse 1964; Ellis and Stimson 2012). against their preferred policy outcome. In other words,
Mason (2018a) calls these two dimensions “issue-based Republicanidentifiersrewardsupportforin-grouployalty
ideology” and “identity-based ideology.” We argue that totheconservativeteambutDemocraticidentifiersreward
bothdimensionsofideology—issue/policyandidentity— supportfortheirindividualpolicypositions.
influence public evaluations of elected representatives’ Second,weusetheCooperativeCongressionalElection
roll-callvotes.Moreover,becausepeopleoftenholdpolicy Study(CCES)datafrom2008-2014tocreatedistrict-level
preferences that do not match their symbolic, identity- aggregatemeasuresofsymbolicandoperationalopinionto
based attitudes, a lawmaker’s roll-call decision can create estimate models that predict roll-call voting. Consistent
internal conflict in how constituents evaluate their per- withourtheory,wealsofindthatbothconstituents’policy
formance. A single roll-call vote can either satisfy both opinionsandsymbolicattitudespredictroll-callbehavior.
dimensionsofaconstituent’spreferences,neitherdimen- Among districts represented by Republicans, though,
sion,oronlyone. operational opinion has a weaker association with roll-
When these two preferences diverge, we argue that call decisions than it does among districts represented by
rank-and-file Democrats reliably prioritize policy prefer- Democrats,whilethereverseistrueforsymbolic,identity-
ences over symbolic attachments, but rank-and-file based attitudes. The district’s symbolic attitudes predict
Republicanstendtoreconciletheconflictinfavoroftheir roll-call voting for Republicans in Congress more than
symbolic attachments to their ideological identity. These theydoforDemocrats.Wearguethatthesefindingshave
differencesinthebehaviorofDemocratsandRepublicans, important implications for how we conceptualize repre-
weargue,areafunctionofthe“sourcecues”thatcitizens sentationalcongruence,ourunderstandingoftheinternal
receive from political elites—cues that help structure the dynamics of the parties in American legislatures, and the
opinionofrank-and-filepartisans.Duetotheideological way scholars can integrate recent advances in political
compositionofeachparty’scoalition,theelitecuesdiffer psychologyandpolicyrepresentation.
systematically between Democrats and Republicans. On
theonehand,Republicanidentifiersoverwhelminglyalso The Policy Congruence Tradition of
identifythemselvesasconservatives—i.e.,theyembracea Representation
conservative identity—but they hold diverse operational An established line of research going back to Miller and
policy preferences. On the other hand, Democrats are a Stokes(1963)hasexploredtherelationshipbetweencon-
diverse coalition of ideological identities, but all largely stituency opinion and roll-call voting. From the begin-
endorseliberalpolicypositions. ning, this work has produced divergent findings, with
Asaresult,RepublicanandDemocraticlawmakersface some authors finding relatively high levels of responsive-
different incentives when it comestothe way theyframe ness(Erikson1978)andsomefindingverylittle(Bafumi
politicalissuesandengageinroll-callbehavior.Republican and Herron 2010; Jacobs and Shapiro 2000). However,
lawmakersaremorelikelythanDemocratstocastroll-call mostworkhasfoundthattherelationshipiscontingenton
votes that are incongruent with district opinion on high- a host of different things, including lawmaker, district,
profilepolicyissuesbecausetheirconstituentsoftenvalue issue, and contextual factors (Achen 1978; Bailey and
symbolic loyalty to “conservatism” more than they care Brady 1998; Butler and Nickerson 2011; Gerber and
about the content of the public policy being advanced, Lewis2004;Griffin2006;HurleyandHill2003;Jackson
whiletheoppositeistrueforDemocraticlawmakers.Asa andKing1989;MillerandStokes1963).Thesefindings
result,lawmakersofbothpartiesarefollowingtheirelect- have led researchers to conclude that lawmakers some-
oral incentives, but they serve districts that demand dif- times engage in instructed delegate representation, but
ferentpatternsofrepresentation.Ourargumentultimately sometimestheyfollowdifferentmodels(Hill,Jordan,and
standsincontrasttorecentworkdepictingRepublicansas Hurley2015).
more motivated by ideology and Democrats by identity Interpretingthediversityoffindingsiscomplicatedby
(seeGrossmannandHopkins2016). thefactthatscholarsinthisareausedifferentmeasuresof
Totestthistheoryofroll-callrepresentation,werelyon constituentopinionanddisagreeoverhowtoestimateits
two types of evidence. First, we use individual data from relationship to roll-call voting. Measures of statistical
the Cooperative Congressional Election Studies (CCES) covariation between constituency opinion and roll-call
from2008to2014toshowthatpartisansfromeachparty patternsarethemostcommonmeasureofrepresentation,
reactdifferentlytoroll-callbehaviorwhenevaluatingtheir evenassomehavequestionedthevalidityofthisapproach
representative’s performance. We show that Democrats on both statistical and theoretical grounds (Achen 1977;
are more likely than Republicans to approve of represen- Hill,Jordan,andHurley2015;Matsusaka2001).Thereis
tatives who cast votes in line with their specific policy also disagreement over the proper way to conceptualize
2 PerspectivesonPolitics
Downloaded from https://www.cambridge.org/core. Serials Section, on 22 Dec 2020 at 19:36:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1537592720003746
district opinion. Some conceive of it as the mean or symbolic dimension grounded by in-group identity
median position on a one-dimensional ideological scale (Carmines and D’Amico 2015; Ellis and Stimson 2012;
that summarizes many issue positions (Bafumi and Her- Free and Cantril 1967; Jost, Federico, and Napier 2009;
ron 2010; Gerber and Lewis 2004; Tausanovitch and PoppandRudolph2011).Eventhoughbothdimensions
Warshaw2013),whileothersarguethatmeasuringopin- are often politically salient, they are only marginally
iononspecificissuesseparately isamorevalidcharacter- correlated,whichmeanspeopleroutinelyholdconflicting
izationofmassattitudes(AhlerandBroockman2018). attitudes across the two dimensions, producing an
However, what prior studies have in common is that “operational-symbolic disconnect” (Ellis and Stimson
they define constituent opinion as average preferences 2012). The disconnect results from the fact that while a
about policy, with the debate boiling down to whether it majority of the American public embraces a conservative
ismoreappropriatetoaggregateissuepositionsormeasure social identity, only about one-third consistently endorse
themseparately.Thisdebateisimportantforclarifyingthe conservativepolicypositions.Bycontrast,alargemajority
natureofpolicypreferencesamonggroupsofvoters,butit of the American public endorses liberal policy positions
does not address the fact that voters’ operational policy but only about one-third identifies as liberal (Ellis and
preferences may not be the only attitudes they use to Stimson2012).
evaluatetheirmemberofCongress’s(MC)roll-callvotes. The most prominent feature of this disconnect, how-
Symbolic ideology—i.e., identity-based ideology—can ever,isitsasymmetry—thatis,symbolicconservativesare
also influence how people want their representative to far more likely to express liberal policy preferences than
behave. symbolic liberals are to endorse conservative policies. In
Indeed, one of the most consistent findings in public fact,accordingtoEllisandStimson(2012)whereasthere
opinionresearchisthatpeoplelackcoherentbeliefsystems is little conflict among liberals between their symbolic
(Converse1964).Nevertheless,mostpeoplestilldescribe identificationandtheiroperationalpreferences,two-thirds
themselvesusingideologicallabels(ConoverandFeldman of symbolic conservatives experience conflict with their
1981).Scholarsconceptualizetheadoptionoftheselabels operationalpolicypreferencesoneconomicissues,cultural
asaformofsocialidentity,wherebypeopleformsymbolic issues,orboth.1
attachments to ideological groups without consciously While research on policy representation has long
endorsing an entire policy program (Conover and Feld- focused on the effects of public opinion about policy,
man1981;Devine2015;EllisandStimson2012;Kinder political psychology suggests that other attitudes derived
andKalmoe2017;MalkaandLelkes 2010).These ideo- from a person’s ideological identity might affect public
logicalattachmentsareusuallyinheritedfromotherareas reactions to lawmaker behavior as well. These often
of people’s lives such as their family members, religious incompatiblepolicyandsymbolicattitudesnotonlycreate
affiliations, peers, or elsewhere, and they serve important ambivalenceamongcitizenswhentheycomeintoconflict,
socialandcognitivefunctions(Devine2015;Jost,Feder- but these cross-pressured constituents can also create
ico, and Napier 2009). In addition to helping people incentives for lawmakers to vote against their district’s
navigate the complexities of the political world, these statedpolicypreferences.Aswewillexplaininmoredetail,
attachmentshelpthemcopewithuncertainty,findmean- these incentives are more prevalent among Republican
ingintheirlives,anddevelopasenseofbelongingintheir membersoftheHousethanDemocraticmembers.
communities.
As a result, ideological labels are not simply an error- A Theory of Symbolic and Operational
laden shorthand for policy preferences; they represent a Representation
distinct dimension of political attitudes (Mason 2018a).
Groupidentificationisaformofsocialidentitytiedtoself- HowVotersEvaluateRollCalls
esteemthatleadsonetoperceivecommongroupinterests Partisan identification is the most important concept in
andtohavemorefavorableattitudestowardsmembersof American politics (Bafumi and Shapiro 2009; Bartels
their perceived in-group (Tajfel 1978). Ideological labels 2002;Campbelletal.1960;Johnston2006).Itrepresents
inAmericanpoliticsconstituteagroupidentitythatleads adeeply rooted socio-psychological attachment tooneof
people to favor legislation simply because they learn that the two major parties, and it serves as most people’s
members of their group support it (Devine 2015; Malka primary political identity. It structures the way people
and Lelkes 2010), even if they may not personally agree make sense of their political reality, and it is the primary
with the policy. As such, ideological group identification psychological construction that tethers citizens to the
canhaveanindependenteffectonsupportforpoliciesand American political system. In other words, “partisanship
evaluationsofrepresentatives. pullstogetherconceptuallynearlyeveryaspectofelectoral
Theimplicationofthislineofresearchisthatideologyis politics”(AchenandBartels2017,268).
amulti-dimensionalconstructcomposedofanoperational Even though partisan identities are socialized through
dimension that reflects citizens’ policy preferences and a family and social networks early in life, their ideological
3
Downloaded from https://www.cambridge.org/core. Serials Section, on 22 Dec 2020 at 19:36:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1537592720003746
Article | Incongruent Voting or Symbolic Representation?
content is largely elite driven. Following Converse’s Second,ideologicalidentityismoresalientforconserva-
(1964) notion of sociological constraint, Noel (2012, tives than liberals. One’s conservative ideological identity
158) explains that “the content of ideology is the result more commonly overlaps with their partisan, racial, and
of some small set of idea organizers who define what it religious identities than for those who espouse a liberal
means to be liberal or conservative,” and “voters merely identity, giving conservatives a simpler identity structure,
respond to this organization” as ideas trickle down from or a “mega identity” (Mason 2018b), which increases the
theelitetothemasslevels.Amongthese“ideaorganizers” salience of identity conflict and reduces tolerance (Brewer
areinterestgroups,opinionleaders,andactivistswhoform and Pierce 2005). Liberal identifiers are overwhelmingly
coalitions around policy programs, and they push these Democrats,butmostDemocratsarenotliberalsandliberal
programs as part of an extended party organization that identityisnotascloselyassociatedwithracial,religious,and
selectscandidatesduringnominatingcontests(Bawnetal. otheridentities.Inotherwords,Republicanidentifiersarea
2012).Throughout thelasthalfofthetwentiethcentury farmorehomogeneousgroupthanDemocraticidentifiers.
into the early twenty-first century, the policy content of This research is bolstered by findings that show that
modern liberalism is largely derived from the active state conservative identity is stronger than liberal identity, the
interventionsoftheNewDeal,theGreatSociety,andthe contentofwhichis“primarilyareactionagainstliberalism
Civil Rights eras. By contrast, the modern conservative and its associated social groups” (Devine 2015, 510;
movementdevelopeditspolicyprograminreactiontothe Zschirnt 2011). Similarly, the increased emphasis on
growthofthemodernwelfarestate.Thethreepillarsthat identity-based ideology also makes conservative identifiers
define modern conservatism are a reverence for small more sensitive to elite social cues that signal what newly
government, a hawkish foreign policy built around a politicized policies are deemed consistent with the pre-
strongmilitary,andtraditionalculturalvalues. scribedmeaningofconservatism(BarberandPope2019).
Politicians adopt party platforms to mobilize these Third, other differences between liberal and conserva-
ideological groups into viable electoral coalitions. The tive identifiers should lead Republicans to favor identity
starkest example of this process occurred in the mid- overpolicy.Forexample,accordingtomoralfoundations
twentieth century, when the political realignment of the theory—which identifies patterns that recent research
South motivated liberal elites to realign themselves into suggests might be socialized rather than innate (Smith
the Democratic Party and conservative elites started to etal.2017)—conservativeidentifiersaremorelikelythan
identify as Republicans more consistently (Levendusky liberals to value binding moral foundations like loyalty,
2009). The rank and file gradually adopted the parties’ conformity,andpurityinpolitics,aswellasotheraspects
preferredpositionsandsymbols,decidedtoswitchparties, oflife(Graham,Haidt,andNosek2009).Theyalsotend
or were aged-out and replaced by younger, more sorted tobe more sensitive tothe violation ofgroup normsand
cohorts. As a result, today the term “liberal” is virtually values(Haidt2012).Thesetendencies shouldleadmany
synonymouswith“Democrat,”whiletheterm“conserva- conservativestofavorlawmakerswhosidewiththeirtribe
tive”isindeliblylinkedtotheRepublicans. and retain ideologically pure voting records, even when
While rank-and-file Democrats and Republicans theydisagreewiththespecificpolicy.Conversely,liberals,
shouldbothvaluerepresentativeswhobehaveinlinewith for whom loyalty and purity are less salient, tend to
boththeirsymbolicattitudesandpolicypreferences,each embrace individualizing moral foundations and should
party will ultimately favor one over the other for three placelessvalueonideologicalconformityforitsownsake
reasons.First,theasymmetrythatexistsintheoperational- fromtheirrepresentatives.Instead,theseliberalidentifiers,
symbolic disconnect in public opinion poses a different who overwhelmingly identify at Democrats, ought to
problem for leaders of each party. For Republicans, prefer congruence with their opinions on specific issues.
because conservative identifiers have diverse policy opin- Perhaps another waytophrase itis thatconservatives are
ions, opinion leaders and strategists alike have strong morewillingthanliberalstoputthegroup’swishesahead
incentives to appeal to conservative identity and to min- oftheirown.
imize the salience of policy differences. These identity- As a result of these three factors, rank-and-file Demo-
basedideologicalappeals,focusedonsymbolism,in-group crats are more likely to both receive and respond to
loyalty,andout-groupthreat(Mason2018a),arethekey operational policy messages from their party, while
for maintaining a viable electoral coalition that can sup- Republicans are more likely to both receive and respond
port a conservative government in power. By contrast, tosymbolicappealstoconservativeidentity.Moreover,as
because appeals to liberal identity cannot rally a viable partymessagesaredisseminateddownwardandabsorbed
electoral coalition, Democratic strategists and opinion by rank-and-file partisans, the positive feelings that par-
leaders have strong incentives to downplay symbolic tisans have toward their party ultimately extends to the
appealstoliberalismandemphasizethesalienceofliberal messages as well, defining what liberalism and conserva-
policy goals, particularly goals that appeal to a diverse tism mean in practice (Lenz 2013; Levendusky 2009;
coalition(GrossmannandHopkins2016). Malka and Lelkes 2010; Mondak 1993). As a result,
4 PerspectivesonPolitics
Downloaded from https://www.cambridge.org/core. Serials Section, on 22 Dec 2020 at 19:36:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1537592720003746
Democraticvotersshouldcaremoreaboutpolicyappeals lawmakers should be motivated by both their district’s
and have greater attachment to their party’s policy posi- operationalandsymbolicpreferences.
tionsthanRepublicanvoters,whowillbemoreattachedto
H4:Districtsymbolicandoperationalconservatismeachhavea
conservative symbols and identity. These theoretical positiveeffectonthelikelihoodthatamemberofCongress
expectations are supported by past research showing that
casts a conservative roll call holding the other factor
in fact Democratic lawmakers make more policy based constant.
appeals to voters, while Republican lawmakers tend to
Second, even though both dimensions have an inde-
makesymbolicappeals(Vavreck2001).
pendent effect on lawmaker behavior, congressional par-
We are left with an understanding of constituent ideo-
ties can help minimize the electoral risk to their party
logicalpreferencesconsistingoftwoequallyreal—andoften
members and help their potential challengers win in the
incongruent—attitudes: operational preferences and sym-
nextelectioncyclebycultivatinganideologicalreputation
bolic attachments. Voters should genuinely value both
thatwillleadtoafavorablepartybrandcapableofgarner-
policy congruence and fidelity to the ideological in-group
ing electoral majorities (Cox and McCubbins 2005; Lee
fromtheirrepresentatives.However,becausetherankand
2016). These brands are achieved through coordinated
file of each party have been socialized to value different
messagingstrategiesinwhichpartyleadersareincentivized
behaviorsfromtheirrepresentatives,theyshouldplacemore
tobolstertheparty’sideologicalreputationusingdifferent
weight on different dimensions of ideology when they
dimensionsofideology.
evaluatetheirrepresentatives’roll-callvotes.
On the one hand, Republican members of Congress
Thislogicleadstothreehypotheses:
understand that they can win a majority by appealing to
H1:Symbolicideologicalandpolicycongruencearebothasso- symbolic attachment to conservative identity, but not
ciatedwithhigherindividualevaluationsoflawmakerper- necessarily a shared policy program, and the way to
formanceholdingtheotherconstant. maintain its support is to demonstrate loyalty to conser-
H2:Rank-and-fileRepublicanwillreward(punish)theirrepre- vative symbols. They also understand that conservative
sentatives more when they cast votes that are congruent identity is far more popular than liberal identity. When
(incongruent)withtheirideologicalidentitythanrank-and-
casting roll-call votes on policy questions lawmakers are
fileDemocrats.
taking operational positions, but they have control over
H3:Rank-and-fileDemocratswillreward(punish)theirrepre- howtheypresentthosedecisionstoconstituents.Repub-
sentatives more when they cast votes that are congruent
(incongruent)withtheirpolicyopinionthanrank-and-file licanlawmakerswanttobeabletocrediblyclaimthatthey
have“conservativevotingrecords,”thatthey“standwith”
Republicans.
conservativeactivistgroups,andtheywantopinionleaders
in their partisan coalition to praise them. To accomplish
Roll-CallRepresentation this,theytakeoperationallyconservativepositionsthatare
Thosesocialforcesshapingthewaycitizensevaluateroll- oftenunpopular,butrelyoncommunicationsstrategyto
call votes also structure the incentives facing electorally representthesevotesasrootedinconservativesymbolsand
mindedmembersofCongress.WhenMCscastarollcall, oppositiontoliberalout-groups.
they want to avoid backlash and minimize the electoral On the other hand, Democrats represent districts that
penalty for their position. Roll-call decisions become are symbolically diverse, including many self-identified
complicated when key supporters disagree. The liberals, moderates, and even conservatives. As such,
symbolic-operational disconnect embedded in public Democraticlawmakers,whocampaignmoreonconcrete
opinion (Ellis and Stimson 2012) implies that legislators policypositions, mustbefarmoreattentive tothepolicy
canbecross-pressuredbytheirdistrictswhenthemajority opinions of the majority in their district. More than
simultaneouslyholdsattitudesthatcouldleadittoeither Republicans,Democrats shouldbeawareofwhichissues
approve or disapprove of a vote for or against a policy. are,andarenot,partoftheirelectoralappealandwillvote
When, as often happens, voters are caught between their accordingly,sometimeseventakingtheconservativepos-
issuepreferencesandtheirideologicalidentity,lawmakers ition on a roll-call vote when they perceive it to be more
areoftenuncertainaboutwhichdimensionvoterswilluse popularintheirdistrict.BecauseDemocratscampaignon
toevaluatetheirrollcalls.Thisuncertaintymeansthatthe policy, by emphasizing policy benefits tokey constituen-
district’spredominant symbolic ideologyandoperational cies instead of making broad symbolic appeals, Demo-
opinionbothaffecttherepresentative’sroll-calldecisions. cratic members of Congress are less constrained by the
Previousresearchhastendedtotreatoneasaproxyforthe needtobeconsistentlyinstepwiththeliberalposition.
other, but we argue that they are conceptually different,
H5:Republican-helddistrictsshouldbesymbolicallyconserva-
andlawmakersrespondtoboth.Thisunderstandingyields tive but operationally diverse while Democratic districts
a number of testable expectations. First, it suggests that shouldbesymbolicallydiversebutoperationallyliberal.
5
Downloaded from https://www.cambridge.org/core. Serials Section, on 22 Dec 2020 at 19:36:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1537592720003746
Article | Incongruent Voting or Symbolic Representation?
H6:Theeffectofadistrict’soperationalpreferencesonroll-call DataandMeasures
votingwillbestrongeramongDemocraticlawmakersthan
Representative approval.2 To investigate the differential
amongRepublicanlawmakers.
effects of operational and symbolic opinion on roll-call
H7:Theeffectofadistrict’ssymbolicideologyonroll-callvoting
behavior we use the CCES from four years: 2008, 2010,
willbestrongeramongRepublicanlawmakersthanamong
Democraticlawmakers. 2012,and2014.Thedependentvariableinouranalysisis
representativeapproval.TheCCESroutinelyasksrespond-
entswhethertheyapproveordisapproveoftheirrepresen-
Data and Methods
tative’s performance in office. The question produces a
Totestourexpectations,weconducttwostudies.Thefirst
four-point measure where respondents can either strongly
is an analysis of representative approval using public
orsomewhatapprove/disapprove.Themeasurerangesfrom
opinionandroll-calldatafromthe2008–2014Coopera-
0-3, with 0=strongly disapprove and 3=strongly approve.
tiveCongressionalElectionStudy(CCES).Thisallowsus Theaverageresponseisμ=1.57andthestandarddeviation
to see if each type of congruence has a different effect is σ=0.993, which translates into the average respondent
depending on a person’s partisanship. The second study
somewherebetweensomewhatdisapprovingandsomewhat
uses measures of district-level operational and symbolic
approvingoftheirrepresentative.
ideology, aggregated from the CCES, and high-profile
roll-call votes about which the CCES directly measures
publicopiniontoseeifrepresentativesofeachpartyfollow Operational and symbolic congruence. We created two
differentdimensionsofideology. measuresofideologicalcongruenceusingtheroll-callvotes
The CCES is a two-wave internet panel study run by forwhichtheCCESasksrespondentstheiropinion.These
Stephen Ansolabehere and Brian Schaffner, and it was roll-callvotesareamongthemostconsequentialandhigh-
founded with the intent to specifically study district-level profile each session. The first measure of congruence
publicopinion,publicperceptionsofmembersofCongress, capturestheproportionofvoteseachmemberofCongress
and congressional representation. The pre-election wave cast that are the same as the roll-call preferences of the
takes place at the beginning of October, while the post- respondent. For example, if the respondent favored the
election wave occurs shortly after the November elections. Affordable Care Act but their MC voted against it, then
Becauseitisanonlineopt-insurvey,theCCESemploysa that vote was incongruent with that respondent’s policy
samplematchingmethodologyforcreatingarepresentative preferences. As a result, it is coded as 0. However, if the
samplefromnon-randompoolsofrespondents.Thematch- respondentopposestheAffordableCareAct,thenitwould
ing procedure attempts to approximate the random-digit be coded as 1. We repeat this process for all the roll-call
dialing used by Pew Research Center. Unlike most survey votes in the dataset and average them to create a propor-
studies, which have modest sample sizes (1,000-2,000 tion of policy congruence between each respondent and
observations),theCCEStradesonitsunusuallylargesample their member of Congress (μ=0.549 σ=0.301). Zero
sizes, which allows for relatively large samples from each signifies complete incongruence and 1 denotes complete
congressionaldistrict.Itsothermajoradvantageisthatitasks congruence.
its large sample to weigh in on several individual roll -all The second measure is symbolic congruence, which
questions, making it possible to compare individual-/ capturestheproportionofthevotesarespondent’srepre-
district-levelopinionwiththevotesofrepresentatives. sentative takes that are congruent with the respondent’s
ideological identity. These are votes that the respondent
Study1:PublicEvaluations ofLawmaker
may not necessarily personally agree with as a matter of
Roll-Call Behavior
policy,buttheyarevotesinthepolicydirectionassociated
For the first part of our analysis, we investigate whether withthegrouptowhichtheyidentify.Weclassifyvotesas
members of Congress are rewarded for the asymmetrical conservativeiftheyopposeorreducefederalintervention
roll-callstrategythatweexpect.Ourtheoryofoperational in the economy, increase federal activity on national
and symbolic representation is driven by the idea that the security, promote traditional morality, or support free
emphasisDemocraticlawmakersplaceonoperationalopin- trade, with the opposite positions being liberal. We then
ionandRepublicansonsymbolicideologyisdrivenbythe classify conservative roll calls as symbolically congruent
preferences of Democratic and Republican voters respect- withrespondentswhoidentifyasconservativeandliberal
ively.Thatistosay,theyarerespondingtothedemandsof rollcallsassymbolicallycongruentwiththosewhoidentify
the constituencies they represent via the electoral connec- asliberal.Forexample,takearespondentwhoidentifiesas
tion. To see if this is in fact true, we want to see how a conservative but supports the Affordable Care Act. If
RepublicansandDemocratsassesstheperformanceoftheir their member of Congress votes against the ACA, then
individual representatives inthe Housewhen those mem- thatvoteiscodedasbeingsymbolicallycongruentwiththe
berscastvotersthatareincongruentwithboththeiroper- conservativerespondent,eventhoughitwasincongruent
ationalpolicyopinionandtheirsymbolicideology. with their stated policy preferences (due to the
6 PerspectivesonPolitics
Downloaded from https://www.cambridge.org/core. Serials Section, on 22 Dec 2020 at 19:36:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1537592720003746
respondent’s support for the ACA). As with policy con-
gruence, the measure ranges from 0–1 (μ=0.506 Table 1
σ=0.354).Asexpected,thecorrelationbetweensymbolic Predictors of Representative Approval
andoperationalcongruenceislow,just0.237. Education −.005**
(.002)
Female .072**
Controls. Since our theory is conditional on partisanship,
(.006)
wearealsoincludingastandardseven-pointpartisaniden- Nonwhite .096**
tification measure, ranging from strong Democrat (0) to (.007)
strongRepublican(6).Inadditiontoourkeyindependent Age .002**
(.000)
variablesofinterest,wealsocontrolledforastandardsetof
Income −.001
demographic controls, including age, gender, education,
(.001)
and race. We controlled for these demographics because Republican(7-Point) −.013**
theyareoftenassociatedwithattitudestowardspoliticians, (.002)
symbolic attachments, and issue positions. As such, we PolicyCongruence 1.094**
(.017)
wanted to account for the most well-established demo-
SymbolicCongruence .596**
graphicconfounders.Acontrolforhouseholdincomewas
(.015)
also included because past research suggests that wealthier RepublicanxPolicyCongruence −.075**
individuals are more likely to prefer and to receive oper- (.004)
ationalpolicyrepresentation(Harden2015). RepublicanxSymbolicCongruence .150**
(.004)
YearFixed-Effects(Relativeto2008)
Results 2010 −.158**
(.009)
TheCCESrepresentsanesteddatastructurewithrespond-
2012 .005
ents clustered within congressional districts. As such, we (.010)
estimated a linear multilevel random-intercept model. 2014 .006
Coefficientsinlinearmulti-levelmodelsareinterpretedjust (.009)
likeOLScoefficients,buttheyalsoletusestimateaseparate Constant .504**
(.018)
intercept for each congressional district. This approach
VarianceComponents
allows us to model the clustered nature of the data— Variance,Constant .013**
namely, spatial autocorrelation. Because we are interested (.011)
in how Democrats and Republicans value policy and Variance,Residual .712**
symboliccongruencedifferently,weintroducedinteraction (.003)
termsbetweenourcongruencemeasuresandtherespond- Waldχ2 41383.65**
ent’spartisanidentification.Wealsoinclude dummiesfor N 98,359
eachcongressionaltermtodealwithanytemporalvariation
Note:LinearMultilevelmodel.**denotesp<.05.Stderrorsare
inrepresentativeapprovalacrosssurveyyears.3 inparentheses.
Table 1presents our results predicting the approval of
individualmembersofCongress.Interestingly,age,being
nonwhite, and being female are all positively related to respondent’s partisanship, meaning the relative influence
representative approval. Women are more likely to ofeacharedifferentforDemocratsandRepublicans.The
approveoftheircongresspersonthanmen,theoldregister interaction between partisanship and policy congruence
higher levels of approval than the young, and people of (β=-0.075, SE=0.004, p<0.001) shows that it is more
color have higher evaluations than white Americans. stronglyassociatedwithapprovalamongDemocratsthan
Levels of political interest are also positively associated Republicans. Figure 1a shows the conditional slopes for
withrepresentativeapproval.However,neithereducation the effect of policy congruence on lawmaker approval
norincomehadastatisticallysignificantrelationshipwith across the range of the standard seven-point measure of
approval.Thissuggeststhatonceonecontrolsforpolitical partisanship.Asfigure1areveals,thedifferencesbetween
interest,socio-economicstatushasnoeffectonapproval, DemocraticidentifiersandRepublicanidentifiersbecome
which is interesting given that high SES Americans are moreextremeastheyembracetheirpartisanidentitymore
betterrepresentedthaneveryoneelse(Bartels2008;Miler strongly. For example, the marginal effect of policy con-
2010). gruence on representative approval for strong Democrats
Moving to the direct tests of our hypotheses, both (1.098) is nearly twice the size of the marginal effect for
policy and symbolic congruence between the respondent strongRepublicans(0.647).
and their congressperson predict higher representative Turning to the substantive effect of operational con-
approval, but those effects are conditional on the gruence on Democratic and Republican approval, when
7
Downloaded from https://www.cambridge.org/core. Serials Section, on 22 Dec 2020 at 19:36:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1537592720003746
Article | Incongruent Voting or Symbolic Representation?
Figure1
Effectofoperationalandsymbolicrepresentationonapprovalofrepresentativeperformance
strong Democrats are faced with a representative who disapprove and somewhat disapprove. However, when
votesinwaysthatare100%incongruentwiththeirpolicy thatsamerepresentative’svotesare100%congruent, the
preferences, holding all else equal, then their predicted predicted level of approval for strong Democrats is 2, or
approval is a mere 0.903 on a scale ranging from 0 to somewhat approve. That is more than a one standard
3. This means the average lawmaker evaluation among deviationincreaseinrepresentativeapproval.Ontheother
strong Democrats is somewhere between strongly hand, Republicans put far less weight on their
8 PerspectivesonPolitics
Downloaded from https://www.cambridge.org/core. Serials Section, on 22 Dec 2020 at 19:36:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1537592720003746
Congresspersonvotinginlinewiththeirpolicyopinions. structures for how to represent the preferences of their
Strong Republicans evaluate their representative’s per- constituents. While Democrats have a stronger incentive
formancemorehighly(1.317)whentheirrepresentative’s to follow their constituents’ policy preferences, Repub-
votes are 100% incongruent with their own policy opin- lican representatives are incentivized to follow voters’
ionscomparedtostrongDemocratsinthesamesituation. symbolic preferences, even if it means betraying their
Moreover,eventhoughstrongRepublicansevaluatetheir district’s policy preferences to do so. After all, if Repub-
representativesaboutthesameasstrongDemocratswhen lican representatives were completely faithful a person’s
theirvotingrecordmatchestheirpolicypreferences100% policypreferences,theycouldexpectaperformanceevalu-
of the time (1.96), the min-to-max change is only an ation of about 1.9 out of 3, all else being held equal.
increase of about two-thirds of a standard deviation in However, if they were completely faithful to that same
representativeapproval. garden-variety constituent’s symbolic preferences, even if
Inotherwords,whileitistruethatbothDemocratsand they occasionally voted against that person’s policy pref-
Republicansrewardtheirrepresentativeswhentheirvotes erences, they would be rewarded with an evaluation of
are congruent with their policy preferences, Democrats roughly2.3outof3.
reward them a lot more and are far less approving of Eventhoughthisevidenceisobservational,itisconsistent
representatives whose votes are not congruent with their withpreviousexperimentalevidenceinpoliticalpsychology
ownpolicypreferences.Thisfindingisconsistentwithour showingthatRepublicansaremorelikelythanDemocratsto
argumentaboutdifferencesinpolicyandsymbolicrepre- accept policies when their partisan or ideological in-group
sentationbetweenthepartiesandshowsthatDemocratic supportsthem(Bullock2011;BarberandPope2019).Itis
lawmakershaveagreaterincentivetofollowtheirdistrict’s alsoconsistentwithsimilarexperimentalevidenceshowing
policy opinion. Simply put, Democratic representatives thatRepublicansaremorelikelythanDemocratstopunish
are punished more by Democratic constituents for not their representatives when they take positions that are at
followingtheirpolicypreferencesthanRepublicanrepre- odds with the rest of the party (Arceneaux 2008). The
sentatives are punished by Republican constituents for differing effects across the parties were either viewed as
doing the same thing. Even though survey data makes anomalous or idiosyncratic by these scholars, but they fit
causal identification difficult, this finding is consistent ourargumentaboutasymmetriesinpartisanbehavior.
withpastexperimentalevidenceshowingthatDemocrats
are more responsive to persuasive appeals that include Study 2: Explaining Roll-Call Voting
policy-based ideological cues than Republicans, which
DataandMeasures
wasnotedasananomalybyBullock(2011).
These partisan differences are even more stark when To investigate the differential effects of operational and
lookingattheeffectofsymboliccongruence.Indeed,the symbolic opinion on roll-call behavior we again use the
conditionaleffectofsymboliccongruence—thatis,when Cooperative Congressional Election Study (CCES) com-
lawmakersvoteinwaysthatarecongruentwitharespond- moncontentandroll-calldatafrom2008–2014.Thekey
ent’s ideological identity—is even stronger than policy independent variables are average policy conservatism on
congruence(β=0.147,SE=0.004,p<0.001).Asourtheory the issue and average symbolic conservatism at the district
predicts,theeffectofsymboliccongruenceisfarandaway level.4 Policy conservatism is the proportion of constitu-
stronger for Republicans than Democrats. The essential entssupportingtheconservative positionontheroll call.
pattern we observed with operational congruence is Symbolicconservatismisaverageself-placementonafive-
reflected here, except in the reverse. Figure 1b presents pointideologicalscale,thestandardmeasureofideological
thisresultvisually.Whilethemarginaleffectofsymbolic identity (Conover and Feldman 1981; Devine 2015), at
congruenceforstrongRepublicansis1.496,themarginal thedistrictlevelrescaledsothattheminimumvalueiszero
effect for strong Democrats is less than half that size and the maximum is one. Higher values indicate more
(0.613). conservative identifiers in the district. Roll-call votes are
Toputthisintoclearerperspective,thepredictedmin- coded such that the conservative position has a value of
to-max change from complete symbolic incongruence to oneandtheliberalpositionavalueofzero.Refertoonline
completecongruenceforstrongDemocratsisonlyabouta appendixAforalistofroll-callvotesandsurveymeasures
two-thirds standard deviation increase in representative used in this analysis, distributions of sample sizes by
approval,muchsmallerthanthemorethanoneandone- congressionaldistrict,andanumberofrobustnesschecks.
half standard deviation difference in approval between Thesetwomeasuresallowustocapturetheuniqueeffect
strong Republicans whose representative is completely of each dimension of a district’s attitudes on roll-call deci-
incongruent symbolically as opposed to completely con- sions.Thismeasurementstrategymeansthateachlawmaker
gruent. appearsinthedatasetmultipletimesineachcongressional
WefindstrongsupportfortheideathatRepublicanand termbecausetheunitofanalysisistheroll-calldecisionwith
Democratic lawmakers face very different incentive adifferentpolicyopinionmeasureforeachvote.Toensure
9
Downloaded from https://www.cambridge.org/core. Serials Section, on 22 Dec 2020 at 19:36:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1537592720003746
Article | Incongruent Voting or Symbolic Representation?
thatourresultsarenotanartifactofinflatedstatisticalpower Contrastthisdistributionwiththesecondpanelshow-
we use clustered standard errors (by district) in the pooled ing that while Democratic districts span the full range of
modelsandestimateseparatemodelsforeachcongressional symbolic ideology with most of them near the middle of
term.Wealsoestimatemixedeffectslogisticregressionwith thescale,almostallRepublican-helddistrictsaresymbol-
randomslopesandinterceptsforeachcongressionaldistrict icallyconservative.Thisclusteringtotherightmeansthat
anddummyvariablesforeachcongressionalterm.Relevant casting a conservative roll call nearly always puts Repub-
coefficientsfrombothofthesemodelsarepresentedinthe licansinstepwiththesymbolicattitudesoftheirdistricts,
onlineappendix. butonlyputsthemincongruencewithitspolicyopinion
half of the time. For Democrats, by contrast, a reliably
liberalvotemayormaynotmatchthesymbolicattitudes
Results
of most of their constituents, so they need to be more
We begin our analysis by looking at some descriptive sensitivetodistrictpolicyopiniononsalientissues.
statistics.Weexpectthatlowerratesofcongruenceamong Figure 2 also lends credibility to the premise that
Republicans stem from the fact that the two parties RepublicansaremorelikelythanDemocratstorepresent
represent different kinds of districts. Figure 2 demon- districtswhereprevailingpolicyopinionisnotcongruent
strates this by presenting kernel density plots of district withsymbolicandpartisanattachments.Inotherwords,it
policyandsymbolicconservatismintheroll-calldata.The suggests that the symbolic-operational disconnect that
toppanelshowsthatwhilebothpartiesrepresentdistricts researchers have found at the individual level is reflected
thatspanthefullrangeofpolicyopinion,theDemocratic at the aggregate congressional district level as well. This
districtsareconcentratedslightlytotheleftofcenterand disconnectsuggeststhatDemocraticandRepublicanlaw-
Republican districts appear evenly split between liberal makers face very different incentives given the compos-
and conservative opinion on the roll calls in the study, itionoftheirrespectiveelectoratesacrosseachideological
consistentwithH5. dimension—i.e.,incentives thatencourageDemocrats to
be more congruent with their districts’ prevailing policy
preferences in order to maintain their electoral coalition
Figure2 andRepublicanstobemoresymbolicallycongruent.
Distributionofdistrictpolicyopinionand Figure3showstheproportionsofeachparty’srollcalls
symbolicidentitybylawmakerparty in our data that are congruent with the prevailing policy
opinion and symbolic ideology in their districts. The
parties are nearly mirror images of one another, with
Democrats voting consistent with policy opinion 71%
ofthetimeandwithsymbolicideology56%ofthetime.
Figure3
Operationalandsymboliccongruenceonroll
callvotingbyparty
.8 Policy Congruence
Symbolic Congruence .75
.71
.7
.6 .56 .57
.5
n
o
orti .4
p
o
Pr
.3
.2
.1
0
Democrats Republicans
10 PerspectivesonPolitics
Downloaded from https://www.cambridge.org/core. Serials Section, on 22 Dec 2020 at 19:36:23, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms.
https://doi.org/10.1017/S1537592720003746