Abstract
We investigated how judgments of political messengers depend upon what would benefit one’s preferred candidate. In Study 1a, participants were asked to evaluate the warmth and competence of the writer of a pro- or anti-Obama political message for the 2012 presidential election (Obama/warm; Romney/competent). When judging the messages, warmth was emphasized by Democrats and competence by Republicans. Study 1b replicated these effects for messages about Romney as well. Study 2 examined the 2004 presidential election where perceptions of the party candidates’ warmth and competence reversed (Bush/warm; Kerry/competent). There competence was emphasized by Democrats and warmth by Republicans. Study 3 showed that varying the warmth and competence of each party’s prospective candidates for the 2016 election influences whether warmth or competence is emphasized by Democrats or Republicans. Thus, differences between Republicans and Democrats in emphasizing warmth or competence reflect a dynamic motivated cognition that is tailored to benefit their preferred candidate.
If you’re not a Democrat when you’re 20, you’re heartless. If you’re not a Republican when you’re 40, you’re brainless.
This is a modern version of an oft-quoted statement that originated with mid-19th-century historian and statesman François Guizot. It conveys a sentiment widely held in our society, stereotyping Republicans as lacking feelings for others and Democrats as lacking grounded realism. This perception is typically attributed to the different kinds of philosophical liberalism that each party emphasizes. Republican candidates tend to emphasize the importance of economic liberalism, grounding their arguments in their faith in the free market’s ability to get results (e.g., more economic growth, improved education) and pursuing policies that they hope will achieve those ends (e.g., tax cuts, school vouchers). In contrast, Democratic candidates tend to emphasize social liberalism, grounding their arguments in an ethic of care for others (e.g., alleviating poverty, universalizing health care) and pursuing policies that they hope will achieve those ends (e.g., welfare programs, national health insurance). This, the argument goes, leads to Republicans having a perceived competence advantage and Democrats having a perceived warmth advantage.
However, society also maintains stereotypes that directly contradict those above. For example, a (possibly apocryphal) statement attributed to Republican Senator Alan Simpson seems to suggest the opposite dynamic: “There are two parties in America, the Stupid Party and the Evil Party. I’m a proud member of the Stupid Party.” The stereotype here calls to mind Republican politicians who claim to hold folksy, uncomplicated ethical views of the average American—while deprecating Democrats as being out of touch and opposed to family values—or Democratic politicians attempting to cultivate a reflective persona with an emphasis on intelligence—while painting their Republican opponents as backward and anti-science. In this case, Republicans are perceived as having a warmth advantage, whereas Democrats are perceived as having a competence advantage.
Our research objective was to determine which of these popular warmth/competence maxims transfers to how Democrats and Republicans actually evaluate other citizens who express political beliefs. Fiske, Cuddy, and Glick (2007) outlined two major dimensions of social judgment: warmth and competence. Warmth designates traits that facilitate likability and pro-sociality such as friendliness and trustworthiness. Competence designates traits that facilitate agency and intellectualism such as intelligence and skillfulness. Combinations of these two continua of judgment have been shown to underlie a great deal of our social cognition (Fiske, Xu, Cuddy, & Glick, 1999; Judd, James-Hawkins, Yzerbyt, & Kashima, 2005).
The above discussion of various stereotypes surrounding Democrats and Republicans shows that it is unclear whether warmth or competence is the perceived strength of either party. This ambiguity leads to two basic hypotheses:
These two hypotheses are based on a general individual difference model of political cognition between Republicans and Democrats, a subject that has received a considerable amount of attention among psychologists (Cornwell & Higgins, 2013; Janoff-Bulman, Sheikh, & Baldacci, 2008; Jost, Federico, & Napier, 2009; Jost, Glaser, Kruglanski, & Sulloway, 2003). According to this view, differences between Republicans and Democrats are based upon distinctions that are stable across election cycles.
There is, of course, a third hypothesis that rests not on how Republicans and Democrats differ from one another, but in how they are the same. This third hypothesis proposes that members of each party emphasize whichever dimension they believe benefits their candidate of choice. Rather than being the product of individual differences, social judgments would instead be the product of the motivation to emphasize whichever dimension is seen as an advantage to one’s party given the current political candidates running for office. There is related research in the field of moral psychology that individuals adopt such a perspective to attain moral “victory” in argument (Haidt, 2001, 2012), so it may be the case that this dynamic applies to political arguments as well. Thus, formally described, Hypothesis 3 is as follows:
While most research on political differences has focused on how political leaders are viewed (e.g., Castelli, Carraro, Ghitti, & Pastore, 2009; Todorov, Mandisodza, Goren, & Hall, 2005), the strongest test of these hypotheses would test the degree to which individuals extend their judgments to non-politicians. Specifically, we were interested in whether everyday Democrats and Republicans use different means for judging the character of their fellow citizens who write an opinion message about a Democratic or Republican political candidate. Research on “perceiver effects” has shown that the dimensions of ratings of targets often depend on the characteristics of the perceiver (Srivastava, Guglielmo, & Beer, 2010; Wood, Harms, & Vazire, 2010). For example, if one group sees itself as more warm than competent, it will emphasize warmth more than competence when differentiating groups, emphasizing the ingroup’s advantage in warmth over outgroup members rather than paying attention to any difference in competence (Oldmeadow & Fiske, 2010). Recent research has shown that political differences produce strong ingroup/outgroup dynamics (Iyengar & Westwood, 2014), so it stands to reason that the effects found regarding previous intergroup effects would also apply to political groups.
In this research, we aim to show that, when judging others, perceivers will emphasize the dimension of social cognition (either warmth or competence) most important (and thus accessible) to them when considering political judgments, whether that importance is relatively chronic (i.e., holding true across election cycles) or politically dynamic (i.e., holding true only within particular election cycles). Generally speaking, perceivers should like those with whom they agree more than those with whom they disagree, as attitude similarity has been linked to liking (Byrne, 1961), but, critically, these differences should be most pronounced along the social judgment dimension (either warmth or competence) preferred by perceivers as a function of its benefit for the candidate that they prefer.
The most recent presidential election cycle appears to follow the stereotypes first mentioned above: The Democratic candidate for president, Barack Obama, was generally seen as more likable and trustworthy than his Republican opponent, Mitt Romney, who was generally seen as more capable on issues related to the economy and job growth (Jones, 2012). Thus, we might expect Democratic voters to emphasize warmth in their judgments and Republican voters to emphasize competence more in their judgments. Study 1a tested this possibility.
Study 1a
Method
Participants
One hundred twenty-five participants were recruited from the Mechanical Turk subject pool for the sum of US$1. There were 89 Democrats and 36 Republicans. No data were collected on sex differences. As we were unsure of the size of our effects, our aim was to have a relatively large number of participants (approximately 30) in each of the four conditions to have enough power for our analysis. Participants were randomly assigned to the four conditions of the study. This study was conducted prior to the 2012 election.
Procedure
Participants were first asked to supply politically relevant information about themselves, indicating with which party they most identify (Democrat or Republican). 1 After completing this initial questionnaire, participants were told that they would be randomized to a “writer” group or a “reader” group. The former would be asked to write a short political message about a political candidate giving their opinion of the candidate and their reasoning for that opinion. The latter would be asked to read a statement from another participant and answer questions about him or her. In fact, all participants were assigned to the “reader” condition and received randomized prepared messages.
Participants then viewed one of four messages, which they believed were being provided by other participants. As it is difficult to provide positive or negative evaluations of political candidates without implicitly suggesting either warmth or competence as the dimension of judgment, half of the messages used warmth-related reasoning and the other half used competence-related reasoning. This provided us with the ability to control for the content of the message and thus examine only the variable of interest: the message’s evaluation (pro- or anti-Obama). The conditions were as follows: a pro-Obama warmth message, an anti-Obama warmth message, a pro-Obama competence message, and an anti-Obama competence message. The precise wording of these messages is available in the supplemental materials.
Following the presentation of the message, participants were asked to rate the traits of the writer associated with warmth: friendly, well-intentioned, trustworthy, warm, good-natured, sincere, and tolerant, as well as the traits of the writer associated with competence: competent, confident, capable, efficient, intelligent, skillful, and independent. Each of these was rated on a 7-point scale from 1 (not at all) to 7 (very much so). Participants were then asked to guess the political party of the writer of the message: Republican or Democrat. Finally, participants were asked to respond to some demographic questions, and debriefed.
Our approach was to examine the degree to which message content (whether the message was one congruent or incongruent with the participants’ partisan preferences) predicted judgments of warmth (high vs. low) more strongly or judgments of competence (high vs. low) more strongly. Our prediction was that the dimension that benefited their own candidate would be emphasized more when making judgments of the message writer as if that was the dimension that really mattered when judging others. Specifically in this case, if the dimension of warmth is an advantage for Obama and competence is an advantage for Romney, then Democrats will be motivated to emphasize warmth as the trait dimension that matters when judging others, whereas Republicans will be motivated to emphasize competence as the dimension that matters when judging others.
The emphasis of Democrats on the importance of warmth should be reflected in their judgments of the warmth of the writer, with the writer being seen as warm if Obama is described positively and being seen as not warm if Obama is described negatively (e.g., “This writer must be a cold fish if he/she can’t see Obama’s warmth”). The same should be true for Republicans and competence (e.g., “This writer must be dense if he/she can’t see that Romney is far more competent than Obama”). Relatedly, as the difference in emphasis between the two groups arises from a distinction between ingroups and outgroups, the favored judgment continuum (either warmth or competence) should also best predict participants seeing the writers as belonging to a political ingroup rather than an outgroup.
Results
The warmth-related and competence-related items for the traits of the writer were each aggregated into a scale for overall warmth (α = .92) and competence (α = .92). They were also significantly correlated (r = .72, p < .001), so whenever we analyzed one as a dependent variable we controlled for the other in all of the following analyses. We also controlled for the message reasoning (either warmth- or competence-based) in each of the following analyses. There were no overall differences between Republicans or Democrats with respect to the perceived warmth or competence of writers in general (t < 1).
To test party effects on warmth and competence ratings, two multiple regressions were run for each party—one predicting warmth, the other predicting competence. The first included dummy variables for pro- versus anti-Obama messages and warmth versus competence messages, as well as competence as a covariate. The second was identical except that warmth was included as a covariate instead of competence. Democrats rated writers as more warm, controlling for competence, for the pro-Obama message than the anti-Obama message, β = .44, t(85) = 7.30, p < .001, 95% confidence interval (CI) = [0.848, 1.482], but when controlling for warmth, the effect of message content on competence dropped to a trend, and actually reversed direction, β = −.15, t(85) = −1.59, p = .11, 95% CI = [−0.775, 0.085]. We found the opposite to be true for Republicans, who rated writers of pro-Obama messages as significantly less competent, controlling for warmth, than writers of anti-Obama messages, β = −.49, t(32) = −4.35, p < .001, 95% CI = [−1.862, −0.675], whereas their ratings of warmth, controlling for competence, also dropped to a trend and reversed direction, β = .27, t(32) = 1.63, p = .11, 95% CI = [−1.333, 0.146]. To determine whether party was having a significant impact on these results, we needed to standardize the direction of the effect of the variables across party (i.e., create a variable that increased warmth and competence for both Republicans and Democrats; the terms above run in opposite directions). Thus, we created an “agreement” dummy variable (equal to 1 when the message contained pro-perceiver’s candidate/anti-opponent of perceiver’s candidate or 0 when the message contained the reverse) and interacted it with political party (removing the “pro- vs. anti-” dummy variable due to potential collinearity). These two sets of results led to a significant interaction between the message (either “agreement” or “disagreement”) and the political party of the participant (either Republican or Democrat) predicting warmth, controlling for competence, β = .66, t(119) = 5.95, p < .001, 95% CI = [1.154, 2.305], and predicting competence, controlling for warmth, β = −.59, t(119) = −4.59, p < .001, 95% CI = [−2.071, −0.822]. 2 These interactions showed that, when evaluating writers of agreeable versus disagreeable political messages, Democrats emphasized warmth more than Republicans, and Republicans emphasized competence more than Democrats. The results are visualized in Figure 1. A breakdown of the warmth and competence residual effects is included in Table 1.
Ratings of warmth minus ratings of competence across message content (pro-perceiver’s candidate/anti-perceiver’s candidate’s opponent or vice versa—we used “agree” and “disagree” as shorthand) by political party of the perceiver (Republican or Democrat).
Breakdown of Results of Message Content (Pro-perceiver’s Candidate/Anti-perceiver’s Candidate’s Opponent) Predicting Warmth, Controlling for Competence, and Competence, Controlling for Warmth, in a Variety of Subsamples Within the Study 1a Sample, Controlling for Condition.
Note. The numbers indicate standardized beta coefficients.
p < .05. **p < .01. ***p < .001.
Democrats and Republicans also differed as to whether the perceived warmth or competence of the writer predicted whether they identified the writer as a member of a political ingroup or outgroup. We created a “same-party” dummy variable using the party identification of the participants and the predicted party of the writers so that it was 1 when the party was the same (Democrats seeing writers as Democrats; Republicans seeing writers as Republicans) and 0 when different (Democrats seeing writers as Republicans; Republicans seeing writers as Democrats). This variable was used as the outcome variable of a multiple logistic regression, which included warmth and competence ratings as predictors and the warmth versus competence condition dummy variable as a covariate. As expected, we found that higher ratings of a writer’s warmth, controlling for competence, were associated with a greater likelihood that a Democrat participant identified that writer as a Democrat rather than a Republican (odds ratio [OR] = 48.09, z = 3.86, p < .001, 95% CI = [6.712, 344.546]), whereas higher ratings of a writer’s competence, controlling for warmth, predicted a greater likelihood of identifying that writer as a Republican rather than a Democrat (OR = 0.27, z = −2.15, p = .03, 95% CI = [0.080, 0.888]). The same was true for Republican participants though associations were reversed, with greater competence ratings of a writer, controlling for warmth, predicting a greater likelihood of identifying that writer as a Republican rather than a Democrat (OR = 22.80, z = 2.83, p = .005, 95% CI = [2.610, 199.189]), whereas higher warmth ratings of a writer, controlling for competence, actually predicted a greater likelihood of identifying that writer as a Democrat rather than a Republican (OR = 0.14, z = −2.00, p = .05, 95% CI = [0.022, 0.967]). Unfortunately, there were not enough Republican observations to fit a logistic regression model while controlling for all experimental conditions (though we controlled for the warmth vs. competence dummy above), so it is possible in this study that these effects would be mediated by message content. However, as these effects replicate in the subsequent studies while controlling for experimental condition, we decided to report them here.
Discussion
Taken together, these results suggest that warmth ratings of the writer of the message are relatively more important for Democrats, and competence ratings of the writer are relatively more important for Republicans, consistent with Hypotheses 1 and 3. However, an alternative interpretation is possible given that the statements were specifically about the Democratic candidate for President, Barack Obama. Perhaps perceived warmth is the dominant judgment continuum when it comes to judging individuals making statements (pro- or anti-) about a member of one’s own party, and perceived competence is the dominant judgment continuum for judging individuals making statements (pro- or anti-) about a member of the opposing party. In Study 1b (also conducted before the 2012 election occurred), we examined this possibility by including statements about Mitt Romney, the Republican nominee for president. Even though Study 1b included individuals making statements about the Republican Romney, thereby changing the party membership of the candidate who is the topic of half of the statements, we expected to find the same pattern as in Study 1a. Because the election concerned the same candidates (Obama and Romney), Republican participants should still emphasize competence and Democratic participants should still emphasize warmth (still consistent with Hypotheses 1 and 3). We also included new follow-up questions to examine how voting preferences would be influenced by a candidate’s good intentions (warmth) versus effectiveness (competence).
Study 1b
Method
Participants
One hundred eighty-three participants were recruited from the Mechanical Turk subject pool for the sum of US$1. As the effects in the previous study were relatively large, we only aimed to have an average of 20 to 25 participants in each condition (approximately 180 total) in this study. The participants consisted of 86 females and 97 males, 142 identified as Democrats, 41 identified as Republicans. As females, across all types of messages, saw the writers as more competent compared with males, β = .16, t(181) = 2.23, p = .03, 95% CI(diff) = [−0.854, −0.052], sex differences were included in all of the analyses below. Participants were randomly assigned to the eight conditions of the study.
Procedure
The procedure for Study 1b was identical to that used in Study 1a, except that four additional conditions were added where the statement was targeting Mitt Romney instead of Barack Obama. The precise wording of these messages is available in the supplemental materials. In addition to the follow-up questions posed in the first study, the participants were also asked three questions targeting differences in prioritizing warmth (good intentions) or competence (getting results). Participants were asked, (a) “Do you think it’s more important for a candidate’s policies to have good intentions or to effectively get results?” (b) “How likely would you be to vote for a less effective, but more well-intentioned candidate?” and (c) “How likely would you be to vote for a more effective, but less well-intentioned candidate?” Each of these questions was answered on a 7-point scale. For the first question, 1 indicated much more importance on intentions, while 7 indicated much more importance on results. For the latter two questions, 1 indicated not at all likely, and 7 indicated very likely. We also asked participants for whom they intended to vote in the November at the beginning of the study, limiting the choice to either Barack Obama or Mitt Romney.
As we have no theoretical reason to expect that one candidate (Obama or Romney) or one type of message (positive or negative) would have a larger impact on social judgment, we collapsed across conditions to examine content either congruent or incongruent with participants’ partisan preferences (pro-Obama or anti-Romney for Democrats; anti-Obama or pro-Romney for Republicans).
Results
As in Study 1a, the warmth (α = .93) and competence (α = .93) scales showed strong internal reliability and were highly correlated (r = .77, p < .001). There were no overall differences in the perceived warmth or competence of writers (controlling for experimental condition) between Republicans or Democrats (t < 1). The other results of this study were generally consistent with those from Study 1a regarding the emphasis of Republican participants on competence and Democratic participants on warmth.
Similar to Study 1a, we ran two multiple regressions for each party—one predicting warmth, the other competence. The first included dummy variables for the pro- versus anti- condition, the Obama versus Romney condition, the interaction of these two variables, the warmth versus competence condition dummy variable, a dummy variable for sex differences, and competence as a covariate. The second was identical except for the inclusion of warmth as a covariate instead of competence. By looking at interaction effects of the Obama versus Romney and pro- versus anti- conditions, we found that, consistent with Study 1a, Democrats saw writers of pro-Obama and anti-Romney messages as more warm, controlling for competence, β = .30, t(135) = 3.84, p < .001, 95% CI = [0.487, 1.522], but found that there was no significant interaction effect predicting competence, controlling for warmth, t(135) < 1. For Republicans, the opposite was true, seeing writers of anti-Obama and pro-Romney messages as more competent, controlling for warmth, although this effect was only a trend, likely owing to the limited statistical power due to the small number of Republicans in the sample, β = −.29, t(34) = −1.47, p = .15, 95% CI = [−2.352, 0.381]. For warmth, controlling for competence, the effect was non-significant and close to zero, t(34) < 1.
Looking at our “agreement” variable (created in the same manner as in Study 1a), once again, we ran two multiple regressions, including the “agreement” variable, political party, the warmth versus competence message content dummy variable, the interaction of “agreement” with political party, and a dummy variable for sex differences. The first predicted warmth with competence as a covariate; the second predicted competence with warmth as a covariate. Like Study 1a, these differences produced a significant interaction between the message content (our “agreement” variable) and the political party of the participant (Republican or Democrat) predicting warmth, controlling for competence, β = .23, t(176) = 2.44, p = .02, 95% CI = [0.129, 1.223], and predicting competence, controlling for warmth, β = −.21, t(176) = −2.15, p = .03, 95% CI = [−1.170, −0.051]. 3 These results are illustrated in Figure 2. A breakdown of the effects is included in Table 2.
Ratings of warmth minus ratings of competence across message content (pro-perceiver’s candidate/anti-perceiver’s candidate’s opponent or vice versa—we used “agree” and “disagree” as shorthand) by political party of the perceiver (Republican or Democrat).
Breakdown of Results of Message Content (Pro-perceiver’s Candidate/Anti-perceiver’s Candidate’s Opponent) Predicting Warmth, Controlling for Competence, and Competence, Controlling for Warmth, in a Variety of Subsamples Within the Study 1b Sample, Controlling for Condition Sex Differences.
Note. The numbers indicate standardized beta coefficients.
p < .05. **p < .01. ***p < .001.
Also, importantly, we replicated the effects from Study 1a regarding how Democrat versus Republican participants’ ratings of the warmth and competence of a writer predicted their belief that this writer belonged to a political ingroup or outgroup. In this study, we were able to confirm that these effects held true while controlling for experimental condition. As in Study 1a, we created an identical “same-party” variable based on participant party and predicted party of the writer of the political message. This was predicted in a multiple logistic regression using warmth, competence, the three dummy variables associated with experimental condition (pro- vs. anti-, Obama vs. Romney, and warmth vs. competence), and a dummy variable for sex differences. Democrat participants were more likely to identify writers as Democrats rather than Republicans when rating the writers as high in warmth, controlling for competence (OR = 4.02, z = 3.42, p = .001, 95% CI = [1.810, 8.905]), but there was no effect of competence, controlling for warmth (z < 1). The opposite was true for Republican participants, for whom greater ratings of competence, controlling for warmth, were marginally significantly predictive of perceiving the writer as a Republican rather than a Democrat (OR = 4.60, z = 1.71, p = .09, 95% CI = [0.802, 26.373]), whereas warmth, controlling for competence, had no effect (z < 1). These results replicate the patterns from Study 1a and show that they were not simply an effect of the target candidate’s political membership in relation to the participant (i.e., political candidate as a member of one’s own vs. the other party).
Interestingly, the additional questions added to this study indicate what this difference in emphasis on warmth versus competence may predict in terms of choice of political candidates. Multiple regressions (including the three dummy variables associated with experimental condition and a dummy variable for sex differences as covariates) showed that Republicans and Democrats significantly differed on the importance of a candidate having good intentions versus his or her ability to get results, with Democrats favoring the former and Republicans favoring the latter, β = −.32, t(177) = −4.54, p < .001, 95% CI = [−1.702, −0.670]. Similarly, Republicans indicated that they would be significantly more likely to vote for a candidate who is more effective, but less well-intentioned than Democrats, β = −.29, t(177) = −4.03, p < .001, 95% CI = [−1.596, −0.547], and Democrats indicated that they would be more likely than Republicans to vote for a candidate who is more well-intentioned, but less effective, β = .29, t(177) = 3.98, p < .001, 95% CI = [0.505, 1.500]. As one might expect, relative emphasis on getting results over having good intentions predicted a lower likelihood of intending to vote for Barack Obama over Mitt Romney (OR = 0.58, z = −4.57, p < .001, 95% CI = [0.458, 0.732]). These results indicate that the warmth/competence emphasis in judging fellow voters translates into voting preferences for political candidates as well.
Discussion
These results are again consistent with Hypotheses 1 and 3 above, ruling out Hypothesis 2. But what remains is distinguishing between Hypotheses 1 and 3. As noted in the introduction, Obama was generally perceived as having a warmth-related advantage in the election, while Romney was generally perceived as having a competence-related advantage. Perhaps, then, the judgments of writers of political messages were influenced by the specific qualities of the presidential candidates in that election cycle. The question, then, is would the same effect hold true in an election cycle where the roles were reversed? Hypothesis 1 predicts “yes,” but Hypothesis 3 predicts “no.” In Study 2, we examine the above effects for the 2004 election cycle in which George W. Bush, the Republican candidate, was generally perceived as having a warmth-related advantage, whereas John Kerry, the Democratic candidate, was generally perceived as having a competence-related advantage (Benedetto, 2004).
Study 2
Method
Participants
We recruited 311 participants from Amazon’s Mechanical Turk to participate in the study for the sum of US$1.50. We dramatically increased our sample size in anticipation of some attrition due to participant suspicion, and to avoid having too few Republican participants (a potential issue with Study 1b above). Our goal was to have approximately 20 to 25 participants in each condition following the attrition (again, approximately 180 total, though we ended up with far less attrition than expected). We asked individuals only to participate if they were of voting age at the time of the election. We excluded 16 participants who indicated that they were currently younger than 26 years of age (the age one would need to be to have been 18 in November 2004 given the time the study was done) for not following this direction. We also excluded an additional 20 participants who indicated that they did not believe that another participant wrote the political message in 2004. In order to test our hypothesis about the effects of the campaign season, we needed to limit our analyses to those who were able to recall the events of 2004 relatively well, 4 so we also excluded those who answered greater than one standard deviation below the mean 5 to the question, “How closely did you follow the campaign in 2004?” This left us with 233 participants (133 females and 100 males) consisting of 149 Democrats and 84 Republicans. There were no sex differences for any of the variables of interest in our study.
Procedure
The procedure for Study 2 was identical to Study 1b except that participants were told that the statements they were reading were from individuals completing a different study that took place in 2004. The statements also replaced references to Mitt Romney and Barack Obama with George W. Bush and John Kerry, respectively. We also included the question above to assess participants’ recall of the events in the 2004 election season and asked participants for whom they voted in 2004, limiting the choice to George W. Bush and John Kerry. Finally, as the procedure involving messages from 2004 may be more implausible to participants than our previous procedures, we asked an open-ended question about suspicion at the end of the study, where participants could indicate whether they thought the “writer” was an actual person. As noted above, those who indicated that they thought the messages were written by the experimenter were excluded from the analysis. All other aspects of the study remained exactly the same.
Results
As in the previous two studies, ratings of warmth (α = .93) and competence (α = .94) both displayed a high degree of internal reliability and were highly correlated (r = .77, p < .001). To test for the effects of interest in this study, we ran multiple regressions identical to those used in Study 2, with the exception that we did not include a dummy variable for sex differences as a covariate. Interestingly, now that the targets of the messages were changed, the effects from our previous studies reversed, as predicted by Hypothesis 3 but not Hypothesis 1. A significant interaction between the pro- versus anti- and Bush versus Kerry dummy variables, using the warmth versus competence dummy variable as a covariate, showed that Democrats rated writers of pro-Bush and anti-Kerry messages as less competent, controlling for warmth, compared with writers of pro-Kerry or anti-Bush messages, β = −.30, t(143) = −3.09, p = .002, 95% CI = [−1.487, −0.327], whereas the effect on warmth, controlling for competence, was only a trend, β = −.17, t(143) = −1.61, p = .11, 95% CI = [−1.108, 0.114]. The opposite was true among Republicans, who rated writers of pro-Bush or anti-Kerry messages as significantly more warm, controlling for competence, compared with anti-Bush or pro-Kerry messages, β = .27, t(78) = 2.01, p = .05, 95% CI = [0.004, 1.156], whereas the effect on competence, controlling for warmth, was non-significant, t(78) < 1. As in the previous two studies, in a multiple regression including a dummy variable for warmth versus competence message content as a covariate, these differences produced a significant interaction between message content (the “agreement” variable created in an identical manner used in Study 1b) and political party of the participant (Republican or Democrat) for competence, controlling for warmth, β = .32, t(227) = 2.17, p = .03, 95% CI = [0.044, 0.898]. 6 The interaction predicting warmth, controlling for competence, was not significant, perhaps because Republicans did not display significant attitude differences toward writers making pro- or anti- statements about Kerry. When the analysis is limited to those who read pro- or anti-statements about Bush, the above interaction is significant for warmth, controlling for competence, β = −.72, t(125) = −3.67, p < .001, 95% CI = [−1.536, −0.460]. 7 Consistent with Hypothesis 3, these results reverse those of Studies 1a and 1b for Obama and Romney, showing that for Kerry and Bush, when evaluating writers of agreeable versus disagreeable political messages, Democrats emphasized competence more than Republicans, and Republicans emphasized warmth more than Democrats. These results are available in Figure 3 and the residual effects within subsamples in Table 3.
Ratings of warmth minus ratings of competence across message content (pro-perceiver’s candidate/anti-perceiver’s candidate’s opponent or vice versa—we used “agree” and “disagree” as shorthand) by political party of the perceiver (Republican or Democrat).
Breakdown of Results of Message Content (Pro-perceiver’s Candidate/Anti-perceiver’s Candidate’s Opponent) Predicting Warmth, Controlling for Competence, and Competence, Controlling for Warmth, in a Variety of Subsamples Within the Study 2 Sample, Controlling for Condition.
Note. The numbers indicate standardized beta coefficients.
p < .05. **p < .01. ***p < .001.
Effects around indicating the writer as a member of a political ingroup or outgroup were also reversed in this study. We used the “same-party” variable, created in the same manner as in previous studies, as the outcome variable for a multiple logistic regression, which included warmth, competence, and the three condition dummy variables as covariates. Democrats were more likely to see the writer as a Democrat rather than a Republican when they were rated as higher in competence, controlling for warmth (OR = 2.63, z = 3.04, p = .002, 95% CI = [1.411, 4.906]), whereas for warmth, controlling for competence, the effect was only marginally significant (OR = 1.63, z = 1.63, p = .10, 95% CI = [0.907, 2.933]). For Republicans, however, participants were significantly more likely to see the writer as a Republican rather than a Democrat when rated as higher in warmth, controlling for competence (OR = 2.26, z = 2.03, p = .04, 95% CI = [1.027, 4.961]), whereas the effect of competence, controlling for warmth, was non-significant (z < 1).
With regard to the follow-up questions, interestingly, as in the previous study, Republicans gave more support to a hypothetically more effective, but less well-intentioned politician compared with Democrats, β = −.21, t(228) = −3.26, p = .001, 95% CI = [−1.140, −0.281], and Democrats gave more support to a hypothetically more well-intentioned, but less effective candidate, β = .17, t(228) = 2.60, p = .01, 95% CI = [0.137, 0.988]. In the question directly comparing support for the two hypothetical candidates, Republicans, compared with Democrats, trended significantly more toward a candidate who gets results over a candidate with good intentions, β = −.23, t(228) = −3.63, p < .001, 95% CI = [−1.166, −0.2345]. Intriguingly, and contrary to expectations, a greater emphasis on effectiveness versus intentions predicted a lower likelihood of having voted for John Kerry in 2004 (OR = 0.82, z = −2.09, p = .04, 95% CI = [0.684, 0.988]). We explore this effect in more detail in the general discussion.
Discussion
Study 2 shows that when making judgments of writers’ political comments about the actual candidates in the specific election of 2004, participants responded in terms of the party advantage for that particular election (warmth advantage for Republican Bush and competence advantage for Democrat Kerry). However, it is notable that the later results for a hypothetical matchup of unknown candidates are in the same direction as those in Study 1b even as the results are reversed regarding warmth-competence judgments of writers making political comments on the actual presidential candidates in the specific 2004 election. This will be considered more fully in the general discussion.
As a whole, the results from Study 2 support Hypothesis 3 more than Hypothesis 1: The relative importance of warmth versus competence when judging fellow citizens’ positive or negative statements about the actual presidential candidates in a specific election track how a relative emphasis on warmth or competence would favor participants’ desired candidate in that specific election. In brief, the results support the conclusion that there is motivated cognition from Republicans and Democrats in the service of supporting their party candidate.
However, there is another possible factor that needs to be considered. In each study, the incumbent (i.e., the candidate who had previously won a presidential election) inspired a disparate emphasis on warmth over competence and the challenger inspired an emphasis on competence over warmth. 8 Research has shown that incumbency can have important effects on how candidates are judged (Kritzer & Eubank, 1979; Kinder, Peters, Abelson, & Fiske, 1980). To address this issue, and to strengthen the case for Hypothesis 3 over Hypothesis 1, the final study, Study 3, examined potential 2016 presidential candidates who have never previously been elected to that office.
Study 3
In addition to addressing the issue described above, Study 3 was designed to specifically pit Hypothesis 1 and Hypothesis 3 against one another. To accomplish this, candidates were selected that exemplified a Democrat-warmth/Republican-competence contest or a Republican-warmth/Democrat-competence contest, thereby varying where each party had an advantage. If Hypothesis 3 is correct, then the extent to which warmth or competence is emphasized by Republicans or Democrats should vary for the two different contests.
To select the prospective 2016 candidates, we ran a pilot study of 102 Mechanical Turk participants, asking them to rate 10 prospective 2016 presidential candidates: Democrats Hillary Clinton, Joe Biden, Elizabeth Warren, Bernie Sanders, and Andrew Cuomo; Republicans Mike Huckabee, Chris Christie, Rand Paul, Jeb Bush, and Marco Rubio. As we were interested in warmth and competence perceived advantages, we examined the ratings of Democrat participants for the Democratic candidates and those of Republican participants for the Republican candidates. The results are in Figure 4.
Graphs from the pilot study indicating ratings of warmth (x-axis) and competence (y-axis) for prospective 2016 candidates
The Republicans with the greatest competence advantage were Chris Christie and Rand Paul. We opted to use Chris Christie as this would produce a bigger warmth difference with the Democratic candidate as Rand Paul was rated relatively highly on both dimensions. The only Republican with a warmth advantage was Mike Huckabee. On the Democrat side, the candidate with the strongest competence advantage was Hillary Clinton, but because we did not want to introduce an additional potential confound (gender; see Suitner & Maass, 2008), and all of the Republican candidates are male, we opted for the second-highest competence-advantaged candidate, Andrew Cuomo. The Democrat with the strongest warmth advantage was Joe Biden. Thus, our matchups were determined to be Joe Biden versus Chris Christie for the Democrat-warmth/Republican-competence comparison, and Andrew Cuomo versus Mike Huckabee for the Democrat-competence/Republican-warmth comparison.
Method
Participants
Five hundred participants were recruited from Amazon’s Mechanical Turk to participate in the full study for the sum of US$1.50. Given the attrition in the previous study, we picked our number with the goal of once again having an average of approximately 20 to 25 participants in each condition (approximately 360 total). The sample consisted of 250 females, 248 males, and 2 of unspecified gender; 157 Republicans and 343 Democrats. As in Study 2, 22 participants indicated that they did not believe that the message they were reading was provided by a real person whom they were judging, and were therefore excluded from the analysis. Analogous to the Study 2 question regarding closely following the 2004 campaign, as the 2016 candidates have not been given a substantial amount of media attention yet, we wanted to ensure that those participating were at least somewhat familiar with the candidates in question. Therefore, as in Study 2 we excluded participants scoring below the midpoint (4) of a scale asking participants to what extent they were familiar with the candidates under discussion. This left a total of 356 participants: 187 females, 168 males, 1 unspecified gender; 126 Republicans and 230 Democrats. As there were significant differences in overall assessment of warmth, t(353) = −2.83, p = .005, 95% CI(diff) = [−0.593, −0.107], and competence, t(353) = −2.56, p = .01, 95% CI(diff) = [−0.550, −0.072], with males’ ratings being higher in both categories, M(warmth) = 4.86; M(competence) = 4.89, compared with females, M(warmth) = 4.51; M(competence) = 4.58, we controlled for sex differences in all of our analyses below.
Procedure
The procedure in this study was identical to the previous studies with the exception that there were twice as many conditions. Half of the participants were randomized into the Cuomo versus Huckabee contest, and the other half were randomized into the Biden versus Christie contest. All of the questions following that randomization were identical to previous studies, interchanging the 2016 candidate names for the names used in the previous studies. We also asked a slightly differently worded question to get at familiarity: “To what extent are you familiar with the candidacies of Mike Huckabee and Andrew Cuomo?” or “To what extent are you familiar with the candidacies of Chris Christie and Joe Biden?” depending on condition. This item was rated on a scale from 1 (not at all familiar) to 7 (very familiar). As noted above, as in Study 2, we excluded those who scored below the midpoint of the scale (leaving participants with at least a moderate level of familiarity). We also included the same questions as previous studies regarding preferences for well-intentioned versus effective candidates, and an open-ended question regarding participant suspicion. Because the participants were more familiar with Biden and Christie (M = 5.58) than they were with Cuomo and Huckabee, M = 5.03; t(354) = −5.45, p < .001, 95% CI(diff) = [−0.767, −0.333], we also controlled for familiarity in the analyses below.
Results
As in previous studies, both warmth (α = .93) and competence (α = .92) showed high internal reliability and were highly correlated with one another (r = .70, p < .001). Also as in previous studies, there were no main effects of warmth (t < 1) or competence (t < 1) as a function of whether a participant identified as a Republican or a Democrat.
Those in the Biden versus Christie (warmth and competence advantaged, respectively) condition showed the same effect as those in Studies 1a and 1b. Once again, we ran two multiple regressions for each party in which we interacted the message content (Support vs. opposition × Christie vs. Biden) and included the warmth versus competence dummy variable, sex differences, and familiarity as covariates: the first predicting warmth controlling for competence and the second predicting competence controlling for warmth. Among Democrats, we found a significant interaction of message content predicting warmth controlling for competence, β = .62, t(134) = 6.20, p < .001, 95% CI = [1.111, 2.151], but not for competence controlling for warmth (t < 1). For Republicans, we found the opposite pattern, with the message content interaction significantly predicting competence controlling for warmth, β = −.32, t(61) = −2.69, p = .009, 95% CI = [−1.755, −0.258], but no effect predicting warmth controlling for competence, β = .13, t(61) = 1.00, p = .32, 95% CI = [−0.380, 1.145]. When message content (the “agreement” variable, determined in the same manner as in previous studies) was interacted with political party (Republican vs. Democrat), we found significant interactions between party and content predicting both warmth controlling for competence, β = .32, t(203) = 3.42, p = .001, 95% CI = [0.338, 1.257], and competence controlling for warmth, β = −.27, t(203) = −2.73, p = .007, 95% CI = [−1.161, −0.187]. 9
The results for the Cuomo versus Huckabee (competence and warmth advantaged, respectively) did not completely reverse like the Bush versus Kerry results of Study 2 but there was attenuation. Among Democrats, there was a significant, though attenuated, interaction (Pro- or anti- × Target) of message content predicting warmth controlling for competence, β = .26, t(80) = 2.00, p = .05, 95% CI = [0.003, 1.561], with no interaction predicting competence controlling for warmth, β = .17, t(80) = 1.24, p = .22, 95% CI = [−0.294, 1.274]. Among Republicans, as expected, the effect for Cuomo versus Huckabee was reversed from Christie versus Biden. There was a significant interaction of message content predicting warmth controlling for competence, β = −.37, t(48) = −1.98, p = .05, 95% CI = [−1.763, 0.013], but there was no effect of the interaction of message content predicting competence controlling for warmth (t < 1). These results are shown in Figure 5 and an overview of them is available in Tables 4 and 5. Notably, this latter effect among Republicans is opposite to that predicted by Hypothesis 1.
Ratings of warmth minus ratings of competence across message content (pro-perceiver’s candidate/anti-perceiver’s candidate’s opponent or vice versa—we used “agree” and “disagree” as shorthand) by political party of the perceiver (Republican or Democrat).
Breakdown of Results of Message Content (Pro-perceiver’s Candidate/Anti-perceiver’s Candidate’s Opponent) Predicting Warmth, Controlling for Competence, and Competence, Controlling for Warmth, in a Variety of Subsamples Within the Study 3 “Biden Versus Christie” Subsample, Controlling for Condition, Familiarity, and Sex Differences.
Note. The numbers indicate standardized beta coefficients.
p < .05. **p < .01. ***p < .001.
Breakdown of Results of Message Content (Pro-perceiver’s Candidate/Anti-perceiver’s Candidate’s Opponent) Predicting Warmth, Controlling for Competence, and Competence, Controlling for Warmth, in a Variety of Subsamples Within the Study 3 “Cuomo Versus Huckabee” Subsample, Controlling for Condition, Familiarity, and Sex Differences.
Note. The numbers indicate standardized beta coefficients.
p < .05. **p < .01. ***p < .001.
It appears, then, that the effects on social judgment changed (attenuated or reversed) by switching from a Democrat-warmth/Republican-competence contest to a Democrat-competence/Republican-warmth contest. Was this change revealed in a significant interaction? To answer this question, we interacted contest (Biden vs. Christie or Cuomo vs. Huckabee) with content (the “agreement” variable) and party (Republican or Democrat), including the warmth/competence message content dummy variable and the variables for sex differences and familiarity as covariates, to see if the effects were significantly shifted. We found a significant three-way interaction predicting warmth controlling for competence, β = .25, t(343) = 1.95, p = .05, 95% CI = [−0.004, 1.440], and a non-significant trend in the opposite direction for competence controlling for warmth, β = −.18, t(343) = −1.35, p = .18, 95% CI = [−1.239, 0.231]. 10 The warmth interaction was primarily driven by a significant attenuation among Democrats when the targets are Cuomo and Huckabee rather than Biden and Christie, β = .20, t(222) = 2.13, p = .03, 95% CI = [0.037, 0.934]. The overall cross-party effects appear to be driven by negative messages. If one examines only negative messages, the warmth controlling for competence effect is highly significant, β = .52, t(166) = 2.69, p = .008, 95% CI = [0.400, 2.614], and the competence controlling for warmth effect is marginally significant, β = −.38, t(166) = −1.83, p = .07, 95% CI = [−2.132, 0.079].
With respect to identifying the writer as a political ingroup or outgroup, the results were consistent with Hypothesis 3. Once again, we tested this with a multiple logistic regression predicting the likelihood of seeing the writer as a member of the participants’ own party (“1” for yes; “0” for no) with both warmth and competence controlling for sex differences, familiarity, and the three dummy variables associated with experimental condition. In the Biden/Christie matchup, we found that, among Democrats, warmth, but not competence, significantly predicted seeing the writer as a Democrat (OR = 4.31, z = 4.58, p < .001, 95% CI = [2.308, 8.049]). For Republicans, the result was the reverse, with competence predicting identifying the writer as a Republican (OR = 6.91, z = 3.55, p < .001, 95% CI = [2.379, 20.045]), and warmth actually marginally significantly predicting not Republican (OR = 0.44, z = −1.72, p = .09, 95% CI = [0.171, 1.120]).
As expected, the results for predicting the political party of the writer of the message were all reversed in the Cuomo/Huckabee matchup. In this condition, among Democrats, competence, marginally significantly predicted seeing the writer as a Democrat (OR = 1.77, z = 1.89, p = .06, 95% CI = [0.978, 3.214]). The effect of warmth was marginally significant as well, but weaker (OR = 1.68, z = 1.69, p = .09, 95% CI = [0.919, 3.079]). Among Republicans, the effect was the opposite, with warmth, but not competence (z < 1), predicting seeing the writer as a Republican (OR = 5.46, z = 2.60, p = .009, 95% CI = [1.520, 19.591]). There was a strong enough reversal compared with the Biden/Christie contest to produce significant three-way interactions. That is, when contest (Biden/Christie vs. Cuomo/Huckabee) and party (Republican vs. Democrat) were interacted with warmth and competence in the multiple logistic regression, the three-way interactions produced significant effects predicting whether the participants perceived the writer as an ingroup or outgroup member. This was true both for Warmth × Contest × Party (OR = 13.01, z = 3.57, p < .001, 95% CI = [3.177, 53.263]) and Competence × Contest × Party (OR = 0.09, z = −3.15, p = .002, 95% CI = [0.019, 0.398]). 11
The follow-up questions regarding preferences for more well-intentioned or more effective candidates provided more mixed results. In the Christie/Biden condition, Republicans were significantly more likely to prefer effectiveness over good intentions than Democrats, β = −.21, t(204) = −3.12, p = .002, 95% CI = [−0.991, −0.223]. Republicans were also less likely to prefer a more well-intentioned, less effective candidate compared with Democrats, β = .19, t(204) = 2.72, p = .007, 95% CI = [0.165, 1.038]. With respect to a less well-intentioned, more effective candidate, results of party were non-significant, β = −.10, t(204) = −1.46, p = .15, 95% CI = [−0.726, 0.108]. Also in line with previous studies, a greater emphasis on good intentions versus effectiveness significantly predicts support of a Democratic candidate over a Republican candidate in 2016 for those in the Biden/Christie condition (OR = 0.65, z = −3.26, p = .001, 95% CI = [0.507, 0.844]).
In contrast, party differences predicted none of these variables significantly in the Cuomo/Huckabee condition, intentions versus effectiveness: β = −.11, t(137) = −1.31, p = .11, 95% CI = [−0.832, 0.168]; more effective, less well-intentioned: t < 1; more well-intentioned, less effective: t < 1. Furthermore, the good intentions versus effectiveness variable did not predict 2016 candidate selection in this condition (OR = 0.83, z = −1.42, p = .16, 95% CI = [0.649, 1.071]). Thus, compared with the Biden/Christie condition, the Cuomo/Huckabee condition did not support the effect predicted by Hypothesis 1.
Discussion
The results from this study strengthen Hypothesis 3 while also showing that the pattern of results is not due to incumbency. When a competence-advantaged Republican (Chris Christie) and a warmth-advantaged Democrat (Joe Biden) were put together in a matchup, we found an effect similar to Studies 1a and 1b. However, matching up a competence-advantaged Democrat (Andrew Cuomo) with a warmth-advantaged Republican (Mike Huckabee) did not show this effect, and, when it comes to identifying the party of the writer, showed the opposite effect (as in Study 2). The difference in differences was large enough to produce significant three-way interactions in overall warmth judgments and significant three-way interactions for both warmth and competence with respect to identifying writers as ingroup or outgroup members. All of these results support Hypothesis 3: Both Democrats and Republicans will adopt the judgment dimension that most favors their preferred candidate.
General Discussion
So which stereotype is the correct one? The answer is that it depends on party affiliation and election cycle. Members of both parties emphasize whichever social perception dimension favors their presidential candidate of choice in a specific election when making their judgments. For the 2012 election, when Democrats had a perceived warmth advantage (Obama) and Republicans had a perceived competence advantage (Romney), Democrats perceived support for (vs. attacks on) their candidate and attacks on (vs. support for) their candidate’s opponent as indicative of high (vs. low warmth) more than high (vs. low) competence. In contrast, Republicans perceived support for (vs. attacks on) their candidate and attacks on (vs. support for) their candidate’s opponent as indicative of high (vs. low) competence more than high (vs. low) warmth. Democrats and Republicans traded roles in emphasizing warmth versus competence for the 2004 election where there was a perceived competence advantage for the Democratic candidate (Kerry) and a perceived warmth advantage for the Republican candidate (Bush). When considering potential 2016 candidates, the preferred judgment dimension generally tracked the degree to which either the Republican or Democrat had a warmth or competence advantage in the matchup, rather than showing consistent party effects across the conditions.
These results potentially have wide-ranging implications for interparty relations. Research has associated warmth and competence ratings with a variety of emotional and behavioral consequences (Cuddy, Fiske, & Glick, 2008; Cuddy, Glick, & Beninger, 2011). For example, past warmth/competence research found different emotional patterns associated with warmth and competence judgments. Additional research needs to examine whether these emotional patterns could track election cycle dynamics. More generally, when considering recent research on political psychology (e.g., Cornwell & Higgins, 2013; Rule & Ambady, 2010), it is important to note that this research was likely conducted during the Obama administration and this could influence what was found. It is possible that in future election cycles that return to a more Bush-versus-Kerry dynamic, the results could change and even reverse—producing an apparent failure to replicate from an election cycle effect similar to the reversals we found in the present studies. It is also possible that earlier research in political psychology (e.g., Janoff-Bulman et al., 2008; Jost et al., 2003) could have been influenced by a Bush (warmth) versus Gore or Kerry (competence) dynamic. Importantly, such potential changes as a function of current political dynamics would establish a new type of situational-political moderation—one worth keeping in mind particularly when considering research that highlights biological explanations of differences between Republicans and Democrats (e.g., Amodio, Jost, Master, & Yee, 2007; Fowler & Schreiber, 2008; Hatemi et al., 2011).
Given the novelty of the present research, there are a number of avenues for future research that should be explored to understand better the scope and boundary conditions of the effect we found. One obvious limitation of the present research is that it collapses judgments across conditions irrespective of whether the messenger in that condition wrote something positive about participants’ own candidate or negative about the opponent of their candidate, or wrote something negative about their own candidate or positive about the opponent of their candidate. Future research should investigate in more detail the relative contribution to the effects from pro- versus anti-statements about individuals’ own candidate or the opponent of their candidate. There could be political conditions where such differences will really matter.
Our studies also collapsed across the target of the political messages to produce conditions either congruent or incongruent with partisan preferences. In the case of the two-party elections we examined, this approach is legitimate because the favoring or disfavoring of one candidate implies an opposing view toward that candidate’s opponent. However, if this research were extended to contests with more candidates (such as primary contests or contests in countries with a proportional representation system), it is possible that our effect would be restricted to candidates for which participants have strong positive or negative feelings. Future research should explore this possibility.
Another limitation is that our research took place within the dynamic of an Obama presidency and primarily competence-based Republican opposition. Generally speaking, Republicans appeared to favor a hypothetical candidate emphasizing effectiveness (competence) over good intentions (warmth) with Democrats showing the opposite pattern. Indeed, this is perhaps why there was not a complete reversal of the effect in Study 3 for the Huckabee/Cuomo matchup, and why those favoring results over good intentions indicated that they were more likely to vote for Bush over Kerry in Study 2. It is unclear whether this Republican-competence/Democrat-warmth bias is a fundamental individual difference between the two parties that can then be shifted in accordance with the demands of a particular national presidential contest, or whether this is simply a product of the political dynamics of an Obama presidency in which all of these studies took place. It will be impossible to know for certain until a future presidency in which the current dynamic is reversed with there being, at the time the study is done, either a competence-favored Democrat or a warmth-favored Republican in office.
One final limitation of our research is that it collapsed across a two-party system. Specifically, the question asking for partisan affiliation forced participants to identify as either a Democrat or a Republican (i.e., asking them with which party they identify more), and the dependent measures predicting the partisan affiliation of the message writer and actual election choices were also restricted to either Democrat or Republican. It is unclear whether the dynamics shown in this article would hold true for political independents as well, dependent upon which of the two major candidates receives the most independent support within a particular election cycle. It is also possible that independents would show a different dynamic from Republicans and Democrats, again perhaps because they would be less invested in their preferred candidate.
Another potential future direction is to examine the political dynamics that are responsible for the perception of candidates as being more warm or competent. On the Republican side, at least in the foregoing research, those candidates producing a warmth dynamic (i.e., Bush and Huckabee) were religious conservatives who emphasized their social issue positions while being more open to liberal economic issues like government spending, whereas Republicans emphasizing their economic conservatism (i.e., Romney and Christie) produced a competence dynamic. For Democrats, those seen as emphasizing economic inequality or greater redistribution toward disadvantaged groups (i.e., Obama and Biden) produced a warmth dynamic, whereas those Democrats emphasizing either their foreign policy credentials (i.e., Kerry) or their centrist economic positions (i.e., Cuomo) tended to produce a more competence-based dynamic. These dynamics may be related to warmth and competence flowing from perceived status and competition, respectively (Fiske, Cuddy, Glick, & Xu, 2002), the self-perceived social status of Republicans and Democrats (Oldmeadow & Fiske, 2010), or negative attributes in candidates that produce “rebound” judgments, like insensitivity (producing a compensatory judgment in favor of competence) or disorganization (producing a compensatory judgment in favor of warmth; Holoien & Fiske, 2013; Kervyn, Judd, & Yzerbyt, 2009). These dynamics were not the focus of this research, and so these inferences can only be considered preliminary, but they may be of interest to researchers in political science and political psychology.
In addition, this research may also create an alternative approach to the examination of the roles of warmth and competence in electoral outcomes. Past research has shown that signals of competence in facial features (Castelli et al., 2009; Todorov et al., 2005) and competence more generally (Funk, 1996) are more important in predicting electoral outcomes than warmth-related features. Consistent with this, in our studies most participants preferred a hypothetically “effective” candidate over a hypothetical “well-intentioned” candidate (though this gap was smaller for Democrats than Republicans). However, in each of the past elections examined in this article (i.e., 2004 and 2012), the candidates with the warmth advantage (i.e., Bush in 2004 and Obama in 2012) defeated the candidates with the competence advantage (i.e., Kerry in 2004 and Romney in 2012). Perhaps when one is speaking in terms of idealized candidacies, competence is indeed important, but for high-profile contests in which there is a trade-off, a warmth advantage may actually be more beneficial. The relation of warmth and competence to electoral outcomes bears further empirical examination.
On a more practical note, this research could be useful to political campaigners and ordinary citizens alike. Those running political campaigns who wish to increase their acceptability to the opposing party should work to shore up their perceived vulnerabilities once the relative strengths and weaknesses of each candidate along the warmth and competence continua become apparent. There is also a takeaway message regarding family and friends. The wisdom of “don’t talk politics at the dinner table” now becomes clearer: When the party affiliations of friends and family members are different, the fundamental difference between whether a candidate’s warmth or competence should matter most can provide a unique recipe for conflicts and misunderstandings. It is important to remember that humans are vulnerable to their politically motivated cognitions and that caution should be taken to avoid making critical judgments of those with whom we disagree on politics—whether it be family and friends or our fellow citizens more generally. Understanding how our political climate and party differences affect our motivated social perceptions is the first step to understanding how to diminish the divisions among us.
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Science Foundation through a Graduate Research Fellowship awarded to James Cornwell and by Grant 39429 from the National Institute of Mental Health to E. Tory Higgins.
Notes
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
