bullock elite

7/31/2019 Bullock Elite

1/20

American Political Science Review Vol. 105, No. 3 August 2011

doi:10.1017/S0003055411000165

Elite Influence on Public Opinion in an Informed ElectorateJOHN G. BULLOCK Yale University

An enduring concern about democracies is that citizens conform too readily to the policy views ofelites in their own parties, even to the point of ignoring other information about the policies inquestion. This article presents two experiments that undermine this concern, at least under one

important condition. People rarely possess even a modicum of information about policies; but when they

do, their attitudes seem to be affected at least as much by that information as by cues from party elites.The experiments also measure the extent to which people think about policy. Contrary to many accounts,they suggest that party cues do not inhibit such thinking. This is not cause for unbridled optimism aboutcitizens ability to make good decisions, but it is reason to be more sanguine about their ability to useinformation about policy when they have it.

Most people are unfit for self-governance:Scholars since Thucydides have expressed thisfear, and social science has done more to con-

firm it than to allay it. Two findings seem to especiallyimpeach the publics fitness for democracy. The first isthat most people are awash in ignorance of politics(Kinder 1998, 78589). Their ignorance of policy is es-

pecially acute (Delli Carpini and Keeter 1996, 7986;Lewis-Beck et al. 2008, 17781). The second findingis that most people conform readily to the wishes ofauthority figures even when those wishes are extreme(Browning 1992; Milgram 1974). This latter findinghas a cousin in research showing that party identifi-cation powerfully shapes peoples views and that itseffects are strongest among the best informed (Green,Palmquist, and Schickler 2002, chap. 8, Zaller 1992).Collectively, these findings have helped to give rise to acommon claim about the way democracy really works:Even when people know about important attributes ofpolicies, they neglect that knowledge and mechanically

adopt the positions of party leaders as their own.No one believes that this claim holds true for every-one. And some disagree that it holds on average in theAmerican electorate (e.g., Key 1966; Nie, Verba, and

John G. Bullock is Assistant Professor, Department of Political Sci-ence and Institution for Social and Policy Studies, Yale University,77 Prospect Street, P.O. Box 208209, New Haven, CT 06520 ([email protected]).

The experiments reported in this article can be replicated withmaterials that are available at http://bullock.research.yale.edu/elite.Available at the same URL are files that can be used to reproducethe figures and tables in the article. Yale Universitys Institution

for Social and Policy Studies funded the experiments reported inthis article. I received helpful comments from Kevin Arceneaux,Jonathan Bendor, Will Bullock, Geoffrey Cohen, Joshua Cohen,David Darmofal, Nick Dragojlovic, Johanna Dunaway, Chris El-lis, Simon Jackman, Alan Jacobs, Stephen Jessee, Christopher Kam,Jon Krosnick, Gabriel Lenz, Matt Levendusky, Robert Luskin, NeilMalhotra, Yotam Margalit, Nora Ng, Ben Nyblade, Celia Paris,Limor Peer, Huib Pellikaan, Paul Sniderman, Elizabeth Suhay, AlexTahk, John Zaller, and the anonymous reviewers. David Brady, JohnC. Bullock, Will Bullock, Felicia Cote, Philip Garland, Laurel Har-bridge, Frank Markowitz, Sonia Mittal, Paul Piff, Louis Rinaldi, andCissy Seguija helped with Experiment 1 on many occasions. CharlesBullock, Celia Paris, Rachel Silbermann, and Julia Weatherseedhelped with Experiment 2. Geoffrey Cohen shared materials fromhis experiments, and Kevin Arceneaux, Adam Berinsky, Cindy Kam,and Wendy Rahn shared data from theirs. I thank all of them.

Petrocik 1976). But the modern student of public opin-ion cannot escape the claim that cue-based processingof messages about policy predominates over evalua-tion of their content (Iyengar and Valentino 2000, 109).Citizens neglect policy information in reaching evalu-ations even when they are exposed to it; instead, theyuse the [party] label rather than policy attributes in

drawing inferences (Rahn 1993, 492). And even whencitizens are well informed, they react mechanicallyto political ideas on the basis of external cues abouttheir partisan implications and typically fail to rea-son for themselves about the persuasive communica-tions they encounter unless those communications areextremely clear (Zaller 1992, 45). Cohen (2003) sum-marizes this view in the title of his article on politicaldecision making: Party over Policy: The DominatingImpact of Group Influence on Political Beliefs.

From a normative standpoint, this claim is dour.Facts about policy are the currency of democratic cit-izenship (Delli Carpini and Keeter 1996, 811), and

traditionally, the greatest concern about elite influenceon public opinion has been that it causes people to holdpositions that they would not hold if they knew morefacts (e.g., Kuklinski and Hurley 1994). But if peopleignore facts about policy even when exposed to suchfacts, there is little reason to expect that facts will helpthem to make better decisions or protect them frommanipulation by elites.

In spite of numerous claims about the relative influ-ence of policy attributes and position-taking by partyelites, direct evidence is slight because few studies di-rectly compare the effects of these variables. Those thatdo make such comparisons use policy descriptions thatare short and vaguefor example, decrease services a

medium amount. This article presents two studies thatpermit comparison of party-cue effects to the effects ofmore substantial policy descriptions. Of course, peopleoften express their views without prior exposure to rel-evant policy details. But much interest hinges on howparty cues and policy details would influence peopleif they were exposed to more than a few of the latter.Examining that counterfactual condition is the point ofthis article.

The results suggest that position-taking by partyelites affects even those who are exposed to a wealthof policy information. Butcontrary to some previ-ous claimsthe effects of such position-taking are

496


2/20

American Political Science Review Vol. 105, No. 3

generally smaller than the effects of policy informa-tion. The experiments also include extensive measuresof the attention that subjects pay to policy, and theysuggest that when people are exposed to both partycues and policy information, the cues do not reducetheir attention to the information. If anything, theyenhance it. To the extent that party cues have largeeffects in nonexperimental settings, it may be because

citizens often know nothing else about the policies andcandidates that they are asked to judge.I begin by reviewing theory and evidence about the

effects of policy information and party cues. The nexttwo sections introduce experiments that permit directcomparison of these effects. The following section re-visits previous studies in light of the findings from theseexperiments. Both previous studies and those reportedhere are rooted in American politics, and the next sec-tion considerswhat we canlearnfromrelevantresearchin other countries. The final section concludes with sug-gestions for future research.

THEORY AND PRIOR EVIDENCEA cue is a message that people may use to infer otherinformation and, by extension, to make decisions. Partycues come in two forms. They may reveal a party af-filiation: Obama is a Democrat. Or they may link aparty to a stand on an issue:The Republicans voted fortax cuts. Policy information is explicitly about the pro-visions and immediate consequences of policies: thislegislation will loosen Medicaid eligibility standardsor that bill will increase co-payments for Medicaidrecipients. People often use party cues to make infer-ences about policies, but party cues are not themselvespolicy information in the sense intended here.1

Many studies have considered the effects of cues andinformation on voters views. For example, one line ofresearch asks how general knowledge of politics mod-erates the connection between values and vote choice(e.g., Althaus 2003, pt. 2; Zaller 1992). Another askswhether cues lead voters astray or help them to act asthough theywereinformed(e.g.,Cutler 2002;Kuklinskiand Quirk 2000; Lau and Redlawsk 2006; Lupia andMcCubbins 1998). But research on the specific effectsof party cues is relatively rare; I return to this pointlater in thetext. Andgeneral political knowledge, whilecorrelated with exposure to descriptions of policy, is adifferent variable. Most importantly, little of this re-search speaks directly to the question at hand, which

is about the relative effects of party cues and policydescriptions on people who are exposed to both.

That said, there is a prominent generalization aboutpeople who are exposed to both types of information:

1 A few political scientists define party cues or partisan cuesmore broadly than I do. For example, Squire and Smith (1988)examine an experiment in which California residents were askedwhether theywould vote to recallcertainjudges.Some residentswererandomly assigned to hear the name of the governor who appointedthe judges. The governors name may be important, but it is not aparty cue by the definition given here.

They will be far more affected by party cues. Thus,Rahn (1993, 492) writes that people use the [party] la-bel ratherthan policy attributes even when exposed tosuch attributes. Cohen (2003, 808) contends that evenwhen one knows about important attributes of a pol-icy, ones attitude toward the policy depends almostexclusively upon the stated position of ones own po-litical party. McGuire (1969, 198) writes that a citizen

is a lazy organism who relies heavily on source cuesand tries to master the message contents only when itis absolutely necessary. And Popkin writes that theMichigan approach emphasized that no informationcould be used, even if obtained, when voters identifiedwith a party (Popkin 1994, 55, emphasis in original).These claims are only weakly qualified: Their scope istypically not limited to particular issues or to particularkinds of people.2

But there have always been countervailing claims.Key (1966) is adamant that voters are responsible,by which he largely means responsive to policy consid-erations. Erikson, MacKuen, and Stimson (2002) andAnsolabehere, Rodden, and Snyder (2008) also mount

general arguments about voter responsiveness to pol-icy, while Aldrich, Sullivan, and Borgida (1989) makethe case for responsiveness to policy in foreign affairs.And Butler and Stokes (1974, esp. chap. 14) make aqualified argument that party ID itself is influenced bypolicy preferences. These views imply a public whosepolicy views are more than adjuncts to partisan feeling.They are hard to reconcile with the claim that peo-ples policy attitudes depend almost exclusively onmessages from party elites.

Claims about the relative power of party cues andpolicy information are often grounded in dual-processtheories of attitude change. These theories hold that

persuasion can occur through systematic or heuris-tic information processing (Eagly and Chaiken 1993;see also Petty and Cacioppo 1986). Systematic process-ing is effortful; it entails checking messages for internalconsistency and against ones existing stock of knowl-edge. Heuristic processing is passive; it occurs throughthe use of simple decision rules rather than throughevaluation of policy content. Dual-process theories im-ply that heuristic processing is more likely when peoplelack motivation or ability to scrutinize the messagesthat they receive. This suggests that party cues willhave greater effects on policy attitudes: Cues arewidelythought to be processed heuristically (e.g., Kam 2005;Rahn 1993), but few people are motivated to scrutinize

information about policies, and fewer still possess theknowledge that is typically required to evaluate argu-ments about policies (Converse 2000; Delli Carpini andKeeter 1996).

A further claim is that party cues reduce attention topolicy information even among people who have beenexposed to it. This claim is consistent with research

2 Popkins characterization of the Michigan school may be toostrong. Compare it to the treatment of voting in Campbell et al.(1960, chap. 8).

497


3/20

Elite Influence on Public Opinion August 2011

on cues as information shortcuts, but most of thatresearch focuses on whether cues make people lesslikely to seek information about policy, not on whethercues make people less likely to use information thatthey already have (e.g., Downs 1957, chaps. 1112;Popkin 1994, chaps. 23). Cues might reduce atten-tion to policyeven when people have descriptions ofpolicy in handbecause they permitpeople to be confi-

dent of their views with less effort (Petty and Cacioppo1986) or because they are clearer guides than policycontent to ingroup-consistent views (e.g., Kruglanskiand Webster 1996, 26465; Mackie, Gastardo-Conaco,and Skelly 1992, 14546, 150). The implications are thesame in either case: Party cues will lead people to beless affected by policy information, and perhaps to beaffected in the wrong ways by superficial understand-ings of policies.

While dual-process models suggest that cue effectsmay outweigh policy effects, they also suggest that theweight of these influences on any particular persondepends on personal characteristics. Notably, the dual-process emphasis on motivation suggests a moderator:

need for cognition, the extent to which people en-joy thinking. Because need for cognition is a stabledisposition, it is a poor measure of cognitive effort inany particular situation. People low in need for cog-nition sometimes scrutinize messages, and people highin need for cognition often give them little thought.Still, people do vary in their general tendency to thinksystematically, and need for cognition captures thisvariation (Cacioppo et al. 1996). The straightforwardprediction is that people who are high in need for cog-nition should be more affected by policy information,which require a modicum of thinking to evaluate. Asecond hypothesis, somewhat less straightforward, is

that people who are high in need for cognition shouldbe less affected by party cues. Later, I consider theevidence for these claims.

In spite of dual-process-based reasons to expect thatparty-cue effects generally outweigh policy effects, theevidence is equivocal. Exposure to party cues is difficultto measure in nonexperimental studies. And compar-ing the effects of party cues tothose of policywhen peo-ple are exposed to both requires research designs thatexpose people to both types of stimuli. Only six pub-lished studies (discussed later) fit this description, andthey vary on several important dimensions. The mostsignificant variationmay lie in their findings: Acrossthesix studies, party cues have average effects on attitudes

of between 3% and 43% of the range of the attitudescales. Policy-information manipulations have averageeffects of between 1% and 28%. Variation this greatmakes generalization difficult.

That said, there are two important respects in whichthese studies vary little. One is the amount of policyinformation provided to subjects. Of the six studies inwhich both policy and party cues are manipulated, fiveprovide no more than three-sentence descriptions ofpolicies, and the sixth offers one to two short para-graphs. The most typical policy descriptions in thesestudies are brief and vague: for example, increase theeconomic status of women (Riggle et al. 1992, 76)

or decrease services a medium amount (Tomz andvan Houweling 2009, 88). Variation on this dimensionis relevant because systematic processing is thoughtto be more likely when people are exposed to mes-sages that are detailed and unambiguous (Chaiken andMaheswaran 1994; Petty et al. 1993). The relative in-fluence of cues from party elites may therefore dependon variation along this dimension.

A second respect in which prior studies vary little istheir reliance on highly indirect measures of depth ofprocessing. For example, subjects in Mackie, Gastardo-Conaco, and Skelly (1992) read a message contain-ing a strong argument about an issue. If they lateragree with the argument, they are assumed to haveprocessed the message systematically. If they disagree,they are assumed to have processed it heuristically.The possibility that subjects might think intently aboutthe argument and yet disagree with it is ruled out byassumption. Similarly indirect inferences about depthof processing are common in political research (Kam2005; Rahn 1993).3 But without more direct measures,it is hard to be confident that cognitive effort is affectedby exposure to cues.

Measures of stable traitsfor example, political so-phistication and need for cognitionare more com-mon (e.g., Kam 2005; Mondak 1993). But because theyare stable, they cannot be used to test hypotheses aboutshort-term variation in depth of processing that mightbe induced by party cues. Moreover, the record of needfor cognitionthe best-established measure of the ten-dency to think systematicallyis puzzling. In the onlyprevious test of the connection between need for cog-nition and party-cue influence, Kam (2005) finds nomoderating effects. This result is compatible with Bizeret al. (2002) and Holbrook (2006), whose analyses of

American National Election Studies data suggest thatneed for cognition does not moderate the effects ofpolicy information. But it is difficult to reconcile any ofthese results with psychological studies suggesting thatneed for cognition moderates the influence of sourcecues and other kinds of messages (e.g., Cacioppo et al.1996).

This article presents two experiments that isolate theeffects of both policy descriptions and position-takingby party elites. In each experiment, subjects read abouta debate modeled on the heated 2005 debate in Mis-souri over health care for the poor. Each experimentexposes subjects to substantial policy information andcontainsdirect measuresof processing depth. Together,

the experiments permit direct evaluation of the claimthat party cues outweigh the effects of policy informa-tion among people whoare exposed to sizable amountsof the latter. They also permit evaluation of the ex-tent to which party cues reduce attention to policyinformation.

3 Cohen (2003) is an exception. He uses analysis of subjects open-ended comments to argue that cues do notdecreaseand may increasedepth of processing. See also Rahn (1990).

498


4/20


TABLE 1. Design of Experiment 1

Expand benefits Reduce benefits

No cues Some legislators supportchanges; othersoppose them

Some legislators supportchanges; others opposethem

Democrats support cues Democratic legislatorssupport changes;

Republican legislatorsoppose them

Democratic legislatorssupport changes;

Republican legislatorsoppose them

Democrats oppose cues Democratic legislatorsoppose changes;Republican legislatorssupport them

Democratic legislatorsoppose changes;Republican legislatorssupport them

Note:Experiment 1 had a 3 2 factorial design. Each subjectread about legislation that would expand orreduce state-provided health-care benefits. In the Democrats support condition, Democratic legislatorssupported the changes while Republican legislators opposed them. In the Democrats oppose condition,Democratic legislators opposed the changes while Republican legislators supported them. In the nocues condition, subjects read about support for and opposition to the proposed changes, but thepositions were not linked to political parties.

EXPERIMENT 1

Subjects, all partisans, received a detailed newspa-per article about health care for the poor in Wis-consin. It contrasted the existing health-care regimewith changes that had just been passed by thestate House of Representatives. It also offered ar-guments from supporters and opponents of thechanges. The supplemental online Appendix (avail-able at http://www.journals.cambridge.org/psr2011009)includes a summary of the arguments and the text ofeach version of the article.

The experiment included a manipulation of policy:

Subjects were randomly assigned to read that the pro-posed changes would expand or curtail health-carebenefits. It also included a manipulation of party cues:Some subjectsreceivedno partycues,while others weretold that Democratic legislators either supported oropposed the policy changes. In these last two cuecondi-tions, Republican legislators opposedtheir Democraticcounterparts. Table 1 summarizes the experimental de-sign.

Participants, Design, and Procedure

A nonprobability sampleof 2,473 subjectswho had pre-viously identified as Democrats or Republicans were

recruited by Survey Sampling International to partici-pate in a study of reactions to news media in differentstates. Of these, 50% identified with the DemocraticParty and 50% with the Republican Party. The studywas fielded from December 16, 2008 through Decem-ber 26, 2008.

The SSI sample appears to resemble the populationof U.S. partisansin mostrespects, including age, gender,and region of residence. (See Figure A1 of the onlineAppendix.) The outlier, as with most Internet samples,is the proportion of people who report having no post-high-school education: 19% of the sample age 25 orolder fit this description, against 41% of American par-

tisans age 25 or older. But the online Appendix showsthat subjects median level of education is identical tothemedian of all U.S. partisans, andit suggeststhat low-education subjects are more affected by policy descrip-tions when exposed to them. (See Figure A2, whichalso suggests that party-cue effects are approximatelyequal for low- and high-education subjects.) In short,the samples nonrepresentativeness on education is notlikely to affect the analyses sharply. And to the extentthat it does affect them, it probably causes them tounderstate the power of policy descriptions.

All subjects were presented with a newspaper articleand asked to read it carefully, as most of the questions

that follow will be about your reactions to it.4 The arti-cle was closely modeled on an Associated Press articleabout Medicaid cuts that were passed by Missourislegislature (Lieb 2005). It contained between 627 and647 words, depending on the condition to which thesubject was assigned. This makes it longer than the av-erage article in low-circulation newspapers but shorterthan the average article in high-circulation newspapers(Project for Excellence in Journalism 2004).5 As withthe Associated Press article, much of the article thatsubjects read was devoted to the policy provisions ofthe bill that the legislature was considering.

Policy Treatment. The status quo health-care policy

was held constant across all versions of the article. Itprovided coverage for single parents of two if theyearned less than $1,334 per month. Under it, Medicaid

4 The prompt does not seem to have induced high levels of at-tention: Most subjects did not correctly answer three basic fac-tual questions about the policy described in the article. See thetwo sections on Depth of Processing and Figures A3 and A8 inthe supplemental online Appendix (available at http://www.journals.cambridge.org/psr2011009).5 The Project for Excellence in Journalism last analyzed the lengthof newspaper articles in 2004.

499


5/20


costs had tripled in the past twelve years, and theyaccounted for nearly one-third of Wisconsins budgetat the time of the articles publication. The status quowas contrasted with changes that would either restrictor expand health care for the poor. In one condi-tion, changes would reduce coverage for 100,000 of thestates one million Medicaid recipients by tightening el-igibility standards. In another, changes would increase

coverage for the same number of recipients by loosen-ing eligibility standards. For brevity, these conditionsare herelabeled the conservativeand liberalpolicyconditions; they were not so labeled in the articles thatsubjects read. The article included many more detailsabout the status quo and the proposed alternatives,e.g., details about co-payments and disability coverage.Table A2 of the online Appendix provides an extensivesummary.

Party-cue Treatment. In the first paragraph of ev-ery article, the proposed changes were said to havepassed the House by an 8771 vote. In one condition,the parties were not identified. In another, 90% of

Democratic legislators supported the proposed pol-icy changes, whether liberal or conservative; 90% ofRepublican legislators opposed the changes. In the fi-nal condition, 90% of Democratic legislators opposedthe proposed policy changes, whether liberal or con-servative; 90% of Republican legislators supportedthe changes. For brevity, these last two conditionsare here labeled Democrats legislators support andDemocrats legislators oppose. They were not so la-beled in thearticlesthat subjects read, whichgave equalattention to the stands of each party.

Post-treatment measures. After reading the article,subjects reported their attitudes toward the policy

changes on a seven-category scale ranging from dis-approve strongly (coded as 1) to approve strongly(coded as 7). They also answered three factual ques-tions about thepolicy; these questions were designed totest whether subjects had paid attention to the article.Finally, they answered six items designed to measureneed for cognition, all derived from similar items thathad high factor loadings in the battery developed byCacioppo, Petty, and Kao (1984). These items formeda reliable battery ( = .81) and were summed andrescaled to form an index that ranges from 0 to 1. Thetext of all items is reported in the online Appendix.

Randomization Checks. By chance, a greater pro-

portion of Democrats than Republicans was assignedto the conservative policy condition (54.9% against48.1%). Because the effects of party cues and policyare analyzed separately for members of each party,this difference does not affect the results reported inthe next section. Success of random assignment to theparty-cue condition was gauged by regressing it on as-signment to policy condition (liberal or conservative),age, education, gender, and region of residence. Simi-larly, assignment to the policy condition was regressedon party-cue condition, age, education, gender, andregion. The 2 statistics from these regressions weresmall, suggesting that the randomizations worked as

intended. (Results from each regression are reportedin the online Appendix.)

Results

Figure 1 presents the main results. As expected,Democrats were more supportive of liberal policychanges when Democratic legislators supported them

(mean attitude rating = 5.15) and less supportive whenDemocratic legislators opposed them (M= 4.64); thedifference is significant at p = .004, one-tailed. (Be-cause there are clear expectations about the directionsof cue effects, significance tests for such effects areone-tailed unless otherwise noted.) Similar patternsemergein the opposite directions, of coursefor Re-publicans reading about liberal changes. They were lesssupportive when Republican legislators opposed thechanges (M= 3.42); more supportive when Republi-can legislators supported the changes (M= 4.48). Thisdifference, too, is unlikely to have occurred by chance(p < .001).

The patterns held when subjects read about con-servative policy changes. Democrats were more sup-portive of the conservative changes when Demo-cratic legislators supported the changes (M= 2.45)than when Democratic legislators opposed the changes(M= 2.05). And Republicans were more supportiveof the conservative changes when Republican legisla-tors supported those changes (M= 3.42), less support-ive when Republican legislators opposed the changes(M= 2.77). These differences, too, are unlikely to haveoccurred by chance (p = .004 and p < .001, respec-tively). It appears, then, that party cues affect eventhose whoare exposed to a wealth of information abouta specific policy. But how much?

By conventional standards, not much. The largesteffect of party cues is depicted in the upper right-hand corner of Figure 1: Republicans reading aboutliberal policy changes had a mean attitude of 4.48 whenRepublican legislators supported those changes, 3.42when Republican legislators opposed those changes.This is a shift of 18% on the 17 attitude scalesizablebut not extraordinary. And as the left-hand panel ofFigure 2 shows, the average effects of party cues aresmaller. The mean absolute attitude change caused byexposing subjects to Democratic legislators supportcues rather than Democratic legislators oppose cuesis 0.65 points, or 11% of the range of the seven-point at-titude scale. This is substantial, but it is swamped by theaverage absolute effect of exposing subjects to detailsabout a liberal rather than a conservative policy: 1.68points, covering 28%of the scale. Note that the averagedifference caused by changes in policy (1.68) exceedseven the greatest difference caused by a change in cues(4.48 3.42 = 1.06).6

6 Figure 1 also suggests that cue and policy effects depend lit-tle on whether the cues are stereotypical (e.g., Democrats sup-port expansion of benefits) or counterstereotypical (e.g., Democratsoppose expansion of benefits). The sole exception lies with

500


6/20


FIGURE 1. Effects of Cues and Policy Direction

Dem. legislators oppose

No cues

Dem. legislators support

2 3 4 5

Dem. subjects (N = 677)

conservative policy


No cues


2 3 4 5

GOP subjects (N = 596)

conservative policy


No cues


2 3 4 5

Dem. subjects (N = 557)

liberal policy


No cues


2 3 4 5

GOP subjects (N = 643)

liberal policy

Note:All panels plot mean attitude toward the proposed policy changes. Responses range from 1 (disapprove strongly) to 7 (approvestrongly). Black lines are 95% confidence intervals. The results show that both party cues and policy affected attitudes. The effect ofpolicy was greater on average and greater for Democratic than for Republican subjects.

FIGURE 2. Mean Attitude Differences by Changes in Party Cues, Party ID, and Policy

Average Attitude Differences

Dem. legislators support vs. no cues

Dem. legislators oppose vs. no cues

Dem. legis. support vs. Dem. legis. oppose

Subjects party ID: Dem. vs. Republican

Policy direction: liberal vs. conservative

0.5 1.5 2.5

All subjects(N = 2473)

0.5 1.5 2.5

Dem. subjects(N = 1234)

0.5 1.5 2.5

GOP subjects(N = 1239)

Note:Each row plots the average of absolute differences between different groups attitudes toward the proposed policy changes. Forexample, the middle row of the left-hand panel shows that, on average, exposing subjects to Democratic legislators support cuesinstead of Democratic legislators oppose cues changed attitudes by 0.65 points on the seven-point attitude scale. In each row, blacklines are 95% confidence intervals. The top three rows show that changes in cue condition have slight to middling effects on attitudes.

The average difference between Republicans and Democrats, displayed in the fourth row of the left-hand panel, is greater. The greatesteffect is caused by exposing subjects to liberal rather than conservative policy changes, but this result masks a large difference betweenDemocratic and Republican responsiveness to policy.

Republican subjects who read about a benefit-expanding health-care policy. For these subjects, the effect of counterstereotypicalDemocrats oppose cues (0.94)was greaterthanthe effect ofstereo-typical Democratssupport cues (0.11). The difference betweentheeffects is 0.83 (95% CI: [.19, 1.18]).

The average effects depicted in the left-hand panelof Figure 2 mask differences between Democratic andRepublican subjects. Among Democratic subjects, theaverage effect of exposure to the Democratic legisla-tors support cues instead of the Democratic legisla-tors oppose cues was 0.45 points on the 17 scale;among Republican subjects, it was 0.85 points. Thepartisan difference in policy effects was starker: 2.64

501


7/20


points for Democratic subjects against 0.71 for Repub-lican subjects. These differences were unexpected; Ex-periment 1 was not designed to investigate differencesbetween Republicans and Democrats and cannot shedmuch more light on them. I revisit this finding in thediscussion of Experiment 2.

On average, Republicans disapproved of both theliberal and the conservative policies. But they disap-

proved less of the liberal policy. The right-hand panelsof Figure 1 make this clear: Averaging over all cueconditions, the mean Republican attitude toward theliberal policy is 3.79; for the conservative policy, it is3.09. (For the difference, p < .001, two-tailed.) Thisresult does not speak directly to the influence of partycues or policy information, but in light of Republicanopposition to the national health-care plan that wasenacted in March 2010, it is striking. I return to it in thediscussion of Experiment 2.

Need for Cognition. Although the preceding analy-ses distinguish between Democratic and Republicansubjects, they still conceal much variation in policy at-

titudes. For example, the median Democratic rating ofthe liberal policy changes was somewhat approvewhen Democratic legislators supported these changes,but fully 21% of Democratic subjects disapproved ofthe policy. Similarly, the median Republican ratingof the policy changes under the same conditions wasslightly disapprove, but 17% of Republican subjectsapproved somewhat or strongly.

To better understand this diversity of responses, Iconsider the effects of need for cognition by estimatingthe model

Policy attitude

= 0+1(Democratic legislators support)

+2(Democratic legislators oppose)

+3(liberal policy changes)

+4(need for cognition)

+5(Democratic legislators support

need for cognition)

+6(Democratic legislators oppose

need for cognition)

+7(liberal policy changesneed for cognition)+ . (1)

Policy attitude is scored from 1 to 7, where highervalues indicate more positive attitudes toward theproposed policy changes. Democratic legislators sup-port, Democratic legislators oppose, and liberalpolicy changes are scored 1 for subjects who wereassigned to these conditions, 0 for other subjects. Needfor cognition is scaled to range from 0 to 1. And

iid N(0, 2) is a vector of disturbances.Table 2, which reports OLS estimates of the model,

shows that need for cognition moderated party-cue ef-

TABLE 2. Need for Cognition Moderates theEffects of Policy in Experiment 1

Democratic Republicansubjects subjects

Intercept 2.51 .28 1.80 .36Democratic legislators

support0.74 .38 0.14 .45

Democratic legislatorsoppose

0.30 .38 0.74 .46

Liberal policy changes 2.09 .32 2.09 .37Need for cognition 0.34 .46 1.89 .58Democratic legislators

support need forcognition

1.43 .61 0.59 .73

Democratic legislatorsoppose need forcognition

0.00 .62 0.13 .74

Liberal policy changes need for cognition

0.95 .51 2.27 .60

Standard error ofregression

1.63 1.88

R2 .41 .09

Number of obser vations 1,163 1,183Note:Each column reports OLS estimates and standard errorsfor thecoefficientsin Equation (1). Thedependent variable is atti-tude toward the proposed policy changes, which is measured ona seven-point scale; higher values indicate a more positive atti-tude. The party-cues variables (Democratic legislators supportand Democratic legislators oppose) and the policy variable(Liberal policy changes) are scored 0 or 1. Need for cognitionranges from 0 to 1. The interactions in the last row of estimatessuggest that need for cognition strongly moderates the effectsof policy. It does not consistently moderate the effects of partycues. These patterns hold under other model specifications: seeTable A4 of the online Appendix.

fects only inconsistently (consistent with Kam 2005)

but that it heavily moderated policy effects. The es-timated coefficient of liberal policy changes needfor cognition represents the expected effect of a shiftfrom the low to the high extreme of need for cognitionamong those who read about liberal policy changes, netof the effect that would have been expected under anycircumstances from the increase in need for cognition.This effect is stronger among Republicans than amongDemocrats, but in both cases it is large and of the ex-pected sign: among Democrats, it increases approvalof liberal policy changes ( 7 = 0.95,p = .03); amongRepublicans, it decreases approval ( 7 = 2.27, p

bullock elite

Documents