Abstract
One potential obstacle to cooperation between political parties is ingroup projection, the tendency for members of subgroups to define superordinate groups based on characteristics of their own ingroups. In five studies spanning 11 years and three presidential administrations, we demonstrated that ingroup projection can be an obstacle that prevents bipartisanship between Republicans and Democrats. Study 1 showed that Americans perceived political ingroups as more prototypical of Americans than outgroups and that the degree of mismatch between the outgroup and the superordinate group was associated with ingroup bias. Studies 2–5 demonstrated that perceiving the outgroup as poorly fitting the prototype of the superordinate group predicted opposition to bipartisan cooperation and a lower likelihood of having engaged in bipartisan behavior (Studies 4 and 5). These studies provide evidence for ingroup projection among American political parties and suggest that it contributes to political polarization.
Bipartisanship is rare, despite its importance for the proper functioning of the United States government. Political cooperation to address critical issues has been difficult to achieve, as demonstrated by recent clashes over health care, immigration policy, and the response to COVID-19. Both actual political polarization (Abramowitz & Saunders, 2008) and perceived polarization (Westfall et al., 2015) have increased over time. In this context, both ordinary citizens and policy makers support policies from their own party and devalue policies from the opposing party. One potential obstacle to cooperation between political parties is ingroup projection, the tendency for members of subgroups to define superordinate groups based on characteristics of their own ingroups.
Interparty relations are some of the most affectively charged in American life. Top-down explanations characterize polarization as elite-driven, with elites modeling ideologically extreme positions (Federico, 2021) and sending cues that partisans follow (Ehret et al., 2018), exacerbated by media’s framing of politics in terms of partisan conflict (Han & Federico, 2017; Levendusky & Malhotra, 2016). Bottom-up explanations view polarization as arising from variation in personality traits and psychological needs, like the need for security and certainty, which lead to opposing political preferences (Jost et al., 2003). There is also evidence of a recursive process of polarization, where each side views the other as extreme and thus feels that bipartisanship is pointless (Van Boven et al., 2018; Westfall et al., 2015).
From a social identity perspective, the lack of cooperation between Republicans and Democrats is not surprising. The parties compete over scarce resources and often lack common goals (Sherif, 1958), and the boundaries dividing them are salient, long-standing, and meaningful, strengthening their impact (Rutchick et al., 2009; Seyle & Newman, 2006; Stangor et al., 1992). However, in the current paper, we examine how differing views of something they actually have in common, their superordinate identity of being American, contribute to greater division. Specifically, we examine partisan differences in definitions of what it means to be an American, and the possibility that these differing conceptions can promote division and inhibit bipartisan cooperation.
The ingroup projection model (Mummendey & Wenzel, 1999; Wenzel et al., 2007) offers one perspective on why perceptions of a shared identity vary among subgroups. This model argues that people perceive the attributes and beliefs of their subgroup as normative for the superordinate group, which makes them perceive their own subgroup as more prototypical of the superordinate group than other subgroups (Wenzel et al., 2007). The model has considerable empirical support (e.g., Doucerain et al., 2018; Ehrke et al., 2014; Lie & Verkuyten, 2012; Machunsky & Meiser, 2014; Sacchi et al., 2013; Steffens et al., 2017; Verkuyten & Martinovic, 2016; Waldzus et al., 2003, 2004; Wang et al., 2018; Wenzel et al., 2003). For example, students, teachers, and bikers (Waldzus et al., 2004; Wenzel et al., 2003) all rated their own group (e.g., business majors) as more similar to the relevant superordinate groups (e.g., students in general) than other subgroups. In turn, outgroup members’ poor fit to this biased standard should lead them to be evaluated negatively. Indeed, negative attitudes toward outgroups are predicted by the relative difference in perceived superordinate group prototypicality between ingroups and outgroups (e.g., Waldzus et al., 2003; Wenzel et al., 2003). Thus, the projection of ingroup characteristics onto the superordinate category has implications for intergroup relations (Wenzel et al., 2007). This theoretical perspective seems especially well suited to understanding polarization and resistance to cooperation between American political parties.
Social identity theory in political psychology has generally focused more on group membership (i.e., which groups a respondent belongs to), rather than the subjective underlying meaning of social identities to explain political attitudes and behavior (e.g., Duck et al., 1995). One important exception is Citrin et al. (1990), who introduced a measure of “Americanism” that comprised six elements (e.g., voting, speaking English, and defending America when it is criticized). They found that Americans’ subjective conceptions of national identity predicted attitudes toward cultural minorities and ethnic change. Still, little attention has been paid to the specific mechanisms by which identity influences political attitudes and behavior, particularly with respect to the substantive meaning of membership in a particular group.
Here, we focus on the meaning of American identity: We examine Republicans’ and Democrats’ beliefs about the meaning of their superordinate group, “Americans,” and the correlates of these beliefs. We argue that ingroup projection contributes to anti-bipartisanship in American politics. To our knowledge, this article is the first test of ingroup projection in American political parties. Beyond examining Republicans’ and Democrats' attitudes toward their respective ingroup and outgroup, we also examine behavioral intentions and reported behaviors with respect to interaction with outgroup members, which is important in the current highly polarized political environment.
In Study 1, we examined whether Republicans and Democrats have different beliefs about what it means to be an American, assessed whether this difference is associated with ingroup projection, and examined the intergroup perceptions related to differences in superordinate identity construal. In Studies 2 and 3, we also examined a behavioral intention important in politics: the desire to cooperate and collaborate with members of other political parties. In Studies 4 and 5, we considered bipartisan behavior in addition to behavioral intentions and intergroup bias.
Study 1
In Study 1, we sought to examine the antecedents and consequences of the different beliefs held by Republicans and Democrats about their superordinate group, Americans. Study 1 tested three predictions derived from the ingroup projection model, first, that perceivers endow the superordinate group with attributes of their own subgroup; second, that perceivers consequently evaluate other subgroups as poorer-fitting members of the superordinate group; and third, that mismatch to the superordinate group’s characteristics is related to more negative evaluations of the outgroup.
Method
Participants were 90 U.S. psychology students. Six were excluded from analyses for no political affiliation; of the remaining 84, 74 identified as Democrats and 10 as Republicans based on the prompt: “Please select the political party you most identify with” (see Table 1 for demographics). The study was conducted in the fall of 2008, before the presidential election. We presented participants with a questionnaire about the meaning of three groups: Democrats, Republicans, and Americans.
Descriptive Characteristics of Samples Across Studies 1–5.
Participants responded to six elements of American identity (e.g., “Supporting the administration”; taken from Rutchick & Eccleston, 2010; see Table 2 for full list). Each participant provided ratings for all three groups. Participants rated each element on a scale ranging from 0 (not at all what it means to be a Democrat/Republican/American) to 6 (an essential part of what it means to be a Democrat/Republican/American). Participants then completed a feeling thermometer, evaluating Democrats and Republicans on a 0–100 cold–warm scale. Last, participants reported demographics, including party affiliation and political beliefs (on a 7-point scale anchored at “very conservative” and “very liberal”). Study materials, post hoc power analyses, and data files are available at https://osf.io/nugs7/?view_only=5213c2e8219241bf95023530f964d20a.
Standardized Regression Coefficients Predicting Ratings on Elements of American Identity From Ratings of Ingroup and Outgroup Identity, Studies 1 and 2.
Note. BCa = Bias-corrected and accelerated. Bootstrap results are based on 1,000 bootstrap samples. Differences between ingroup meaning and outgroup meaning were determined by comparing overlapping confidence interval (CIs). In Study 1, ingroup meaning was a significantly stronger predictor than outgroup meaning for three of the six elements. In Study 2, ingroup meaning was a significantly stronger predictor than outgroup meaning for four of the six elements (see Supplementary Material for these comparisons).
* Coefficient significant at p < .05. ** Coefficient significant at p < .01. *** Coefficient significant at p < .001.
Results and Discussion
To examine the first prediction, that perceivers project ingroup characteristics onto the superordinate group, we conducted a series of multiple regressions, one for each element of identity. In each analysis, the extent to which the element was believed to be an important part of what it means to be American was the outcome, and the corresponding evaluations of the meaning of the ingroup and outgroup were the predictors. This assessed the projection of ingroup attributes onto the superordinate group; if ingroup attributes predict superordinate group attributes more strongly than do outgroup attributes, ingroup attributes are projected onto the superordinate group relatively more.
As predicted, the meaning of the ingroup strongly predicted the meaning of the superordinate group for all six elements (all standardized βs > .38, all ps < .001; the complete list of regression coefficients is presented in Table 2). Moreover, the meaning of the ingroup was a significantly stronger predictor than that of the outgroup for four of the six elements (see Supplemental Material for these comparisons and additional descriptive statistics). Perceivers, then, project specific attributes of their own ingroup onto the superordinate group; that is, Democrats believe that what it means to be an American overlaps considerably with what it means to be a Democrat, and moreover overlaps to a greater extent than with what it means to be a Republican (and vice versa for Republicans’ beliefs).
To examine the second prediction—that, because people project their ingroup’s attributes onto the superordinate group, they then evaluate outgroups as poorer-fitting members of the superordinate group—we computed two composite indices of fit. For each element, we subtracted participants’ rated importance of that element to American identity from its importance to their ingroup identity. The absolute values of these differences were then summed across the elements, creating an index of the degree of mismatch between the perceived meaning of the ingroup and the perceived meaning of the superordinate group (see Wenzel et al., 2003, for a similar procedure). An analogous process generated an index of mismatch between the outgroup and the superordinate group. A comparsison of these indices revealed that the outgroup differed from the superordinate group to a greater degree (M = 12.58, SD = 8.33) than did the ingroup (M = 6.11, SD = 4.15), t(83) = 6.41, p < .001, d = .703 (see Table 3), confirming that the outgroup’s beliefs were perceived to be less prototypical of the superordinate group than the ingroup’s.
Dependent t Tests Examining Subgroup–Superordinate Group Differences, Studies 1–5.
* Coefficient significant at p < .05. ** Coefficient significant at p < .01. *** Coefficient significant at p < .001.
To examine whether the relatively poorer fit of the outgroup to the superordinate group would predict evaluations of outgroup members, we computed an index of in group projection (i.e., the relative prototypicality of the ingroup compared to the outgroup), subtracting the indices of mismatch from one another; this index, then, reflected the degree to which each participant believed their ingroup definition matched that of the superordinate group better than did the outgroup definition (see Wenzel et al., 2007, for a discussion of the conceptualization of relative prototypicality). We then predicted ingroup bias (using differences in ratings of the ingroup and the outgroup on the feeling thermometer measure) from this index. As hypothesized, ingroup projection significantly predicted ingroup bias (β = .381, p < .001). It is possible, however, that strongly identified people with extreme political beliefs are especially likely to both dislike outgroup members and see them as fitting the superordinate group poorly. Thus, we conducted a multiple regression analysis in which we controlled for the extremity of each participant’s political attitudes as measured by distance from the scale midpoint. Ingroup projection significantly predicted ingroup bias (β = .341, p = .001; see Table 4) even after controlling for attitude extremity. 1
Regression Based on Ingroup Projection, Controlling for Attitude Extremity.
* Coefficient significant at p < .05. ** Coefficient significant at p < .01. *** Coefficient significant at p < .001.
One could imagine that either perceiving a large outgroup–superordinate group mismatch (“those Democrats are really unlike real Americans!”) or that perceiving a small ingroup-superordinate group mismatch (“we Republicans are almost exactly like real Americans!”) might be important. Because political party identification in the United States can be considered a morality-based social identity, in that differences between the groups are seen as a matter of right and wrong, group attitudes may be shaped more by outgroup “hate” than ingroup “love” (see Parker & Janoff-Bulman, 2013). Thus, we predicted that bias would primarily be based on the mismatch between the outgroup and superordinate group. We tested this hypothesis by conducting a multiple regression analysis, decomposing ingroup projection into two separate measures: outgroup–superordinate group mismatch and ingroup–superordinate group mismatch (which would be a negative predictor of bias). Outgroup–superordinate group mismatch predicted ingroup bias (β = .333, p = .002), whereas ingroup–superordinate group mismatch did not (β = −.124, p = .222; see Table 5).
Multiple Regression Based on Ingroup, Outgroup–Superordinate Group Mismatch—Controlling for Attitude Extremity (Studies 1–3).
Note. BCa = Bias-corrected and accelerated. Bootstrap results are based on 1,000 bootstrap samples. Across Studies 1–3, the outgroup–superordinate group mismatch was a significantly stronger predictor than that of the ingroup for ingroup bias and antibipartisanship (see Supplementary Material for these comparisons).
* Coefficient significant at p < .05. ** Coefficient significant at p < .01. *** Coefficient significant at p < .001.
The intergroup perceptions observed in Study 1 are consistent with ingroup projection. Members of each subgroup projected their own group’s attributes onto the superordinate group and, correspondingly, evaluated the outgroup as fitting the superordinate group’s attributes more poorly. This relatively poorer fit was in turn associated with ingroup bias.
Study 2
Study 1 demonstrated that subgroup definitions are projected onto the group “Americans,” and that this difference is associated with ingroup bias. In Study 2, we examined a potential consequence of such bias: a reluctance to engage in bipartisan cooperation. We hypothesized that the mismatch between definitions of the outgroup and definitions of the superordinate group would predict negative evaluations of bipartisan activity—a behavior that has important consequences for modern American politics.
Method
Participants were recruited in the summer of 2011 using Mechanical Turk. Of the 174 initially recruited, 64 were removed for incomplete data (45), finishing in 2 minutes or less (14), or no political affiliation (5); of the final 110 participants, 72 were Democrats and 38 were Republicans. Participants completed questionnaires about the meaning of Democrats, Republicans, and Americans, responding to the same elements used in Study 1 (except that the element “supporting American troops in Iraq” was changed to “…Afghanistan”) and rating each element on the same scale. Participants then evaluated Democrats and Republicans on a 0–100 cold–warm scale. Next, participants completed four items assessing bipartisanship (adapted from Hornsey & Hogg, 2000) in which they rated the desirability of working with a Democrat on a bipartisan political participation project (e.g., “I would feel good about working with a Democrat”, “Working with a Democrat would be difficult”) on 9-point scales ranging from 1 (not at all) to 9 (very much). They then responded to the same items with respect to working with a Republican. Last, participants reported demographic and political information.
Results and Discussion
As in Study 1, we found evidence of ingroup projection: the meaning of the ingroup significantly predicted the meaning of the superordinate group for all six elements (all βs > .342, all ps < .001; see Table 2), and ingroup meaning was a significantly stronger predictor than outgroup meaning for three elements. Again, ingroup projection predicted ingroup bias (β = .504, p < .001; see Table 5). This effect was primarily driven by the outgroup’s perceived lack of fit to the superordinate group prototype; multiple regression revealed that outgroup–superordinate group mismatch strongly predicted ingroup bias (β = .528, p < .001), whereas ingroup–superordinate group mismatch did not (β = −.128, p = .121).
To examine the central question of Study 2, whether ingroup projection was associated with attitudes toward bipartisan cooperation, we conducted additional regressions. Here, the dependent variable was the desire to work with ingroup members relative to outgroup members, which would be an obstacle to bipartisan collaboration. This index of antibipartisanship was generated by computing the difference of two composite indices: the desire to work with ingroup members (four items) and outgroup members (four items). High scores indicate stronger relative preference for ingroup members. As hypothesized, ingroup projection predicted antibipartisanship (β = .419, p < .001); outgroup–superordinate group mismatch (β = .527, p < .001) predicted antibipartisanship whereas ingroup–superordinate group mismatch (β = −.011, p = .899) did not.
Study 3
Studies 1 and 2 demonstrated that people project their own subgroup definitions onto the group “Americans” and that this difference is associated with ingroup bias. Study 2 also demonstrated that the mismatch between definitions of the outgroup and definitions of the superordinate group predicts opposition to bipartisan cooperation. These studies used elements of American identity generated using a bottom-up approach (Rutchick & Eccleston, 2010), a method that yielded some elements that could be considered time-sensitive or idiosyncratic (e.g., “supporting American troops in Afghanistan”). It is important to show that the phenomenon depends on the general perception of mismatch on important elements of identity, rather than mismatch on the specific elements participants considered. Thus, in Study 3, a different set of elements of identity was used.
Method
Participants were recruited in the summer of 2012 using Mechanical Turk; recruitment materials indicated that participants should belong to a political party. Of the 171 initially recruited, 51 were removed for incomplete data (46), finishing in less than 2 min (3), or identifying as Libertarian (2). The final sample comprised 88 Democrats and 32 Republicans.
Study 3 substituted the set of elements of Americanism selected by Citrin et al. (1990) for those used in previous studies, but the procedure was otherwise identical to Study 2. As noted, these elements were based on research that was somewhat out of date, but they are less time-sensitive (e.g., “trying to get ahead on one’s own efforts”) than some of the elements used in Studies 1 and 2 (e.g., “supporting American troops in Afghanistan”; see Table 6 for the full list of elements).
Standardized Regression Coefficients Predicting Ratings on Elements of American Identity from Ratings of Ingroup and Outgroup Identity, Studies 3–5.
Note. Bootstrap results are based on 1,000 bootstrap samples. In Study 3, ingroup meaning was a significantly stronger predictor than outgroup meaning for four of the six elements. In Studies 4 and 5, ingroup meaning was a significantly stronger predictor than outgroup meaning across all six elements.
* Coefficient significant at p < .05. ** Coefficient significant at p < .01. *** Coefficient significant at p < .001.
Results and Discussion
All analyses conducted were identical to those of Study 2. For all six elements, ingroup meaning significantly predicted superordinate group meaning (all βs > .288, all ps ≤ .003; see Table 6), and for most elements, ingroup meaning was a significantly stronger predictor than outgroup meaning, consistent with previous studies.
As in previous studies, ingroup projection predicted ingroup bias (β = .219, p = .017; see Table 4); moreover, outgroup–superordinate group mismatch significantly predicted ingroup bias (β = .202, p = .027) whereas ingroup–superordinate group mismatch did not (β = −.129, p = .140; see Table 5). Last, ingroup projection significantly predicted antibipartisanship (β = .269, p = .004). In line with our argument that, because of the nature of these groups, outgroup attitudes should more strongly predict bias, outgroup–superordinate group mismatch significantly predicted antibipartisanship (β = .300, p = .001), but ingroup–superordinate group mismatch did not (β = −.058, p = .512). These findings extend those of Study 2 by demonstrating that relative mismatch on a different set of elements of identity can drive ingroup projection.
Study 4
Studies 2 and 3 suggested that ingroup projection is associated with antibipartisan attitudes, and that this effect was largely driven by the mismatch between conceptions of the political outgroup and superordinate group. Another kind of bipartisanship, dialogue and interaction with the members of political outgroups, may be particularly important for reducing political polarization. Thus, Study 4 examined the link between ingroup projection and participants’ reported and desired bipartisan behaviors.
Method
Participants were recruited in the summer of 2014 using Mechanical Turk; recruitment materials indicated that participants should belong to a political party. Data were collected from 379 participants. After excluding participants who did not complete the study (34), who completed the study in less than 2 minutes (17), and who were not Democrats or Republicans (14), the final sample consisted of 314 participants, of whom 184 were Democrats and 130 were Republicans. As in Study 3, participants rated the importance of various facets of American identity (Citrin et al., 1990; see Table 4) to being a Democrat, a Republican, and an American, and completed a feeling thermometer about Republicans and Democrats.
Participants next reported how often they engaged in various bipartisan behaviors, on a scale from 1 (never) to 9 (frequently). These comprised the following behaviors, all framed with respect to someone with whom the participant disagreed with politically: “acknowledged they made a good point about a political issue,” “intentionally requested their opinion on a political issue,” “held a civil discussion about politics,” “reconsidered your opinion on a political issue as a result of discussing politics,” and “made a compromise during a disagreement over politics.” Collectively, these behaviors capture many of the positive politically relevant interactions that people might have with outgroup members. To evaluate the desirability of working with outgroup members, participants rated the likelihood that they would work with Democrats and Republicans on the same bipartisan behaviors. Last, participants reported demographic and political information.
Results and Discussion
Study 4 replicated previous studies with respect to ingroup projection. For all six elements, ingroup meaning significantly predicted superordinate group meaning (all βs > .483, all ps < .001; see Table 6) and was a significantly stronger predictor than ou-group meaning. Ingroup projection significantly predicted ingroup bias (β = .301, p < .001; see Table 7). Again, outgroup–superordinate group mismatch had a stronger predictive relationship to ingroup bias (β = .285, p < .001) than did ingroup–superordinate group mismatch (β = −.123, p = .015).
Multiple Regression Based on Ingroup, Outgroup–Superordinate Group Mismatch—Controlling for Attitude Extremity (Studies 4 and 5).
Note. Bootstrap results are based on 1,000 bootstrap samples. Across Studies 4 and 5, the outgroup–superordinate group mismatch was a significantly stronger predictor than that of the ingroup for ingroup bias, past bipartisan behavior, and future bipartisan behavior.
* Coefficient significant at p < .05. ** Coefficient significant at p < .01. *** Coefficient significant at p < .001.
To examine bipartisanship, an index was developed by computing the difference of two composites: reported interactions with ingroup members and reported interactions with outgroup members. Ingroup projection significantly predicted less previous bipartisan behavior (β = .298, p < .001). Outgroup–superordinate group mismatch also significantly predicted less bipartisan behavior (β = .310, p < .001) whereas ingroup–superordinate group mismatch did not (β = −.073, p = .174). A parallel index was created for desired future interaction, subtracting desired future interaction with outgroup members from desired future interaction with ingroup members. Again, ingroup projection predicted less future bipartisanship (β = .281, p < .001). Outgroup–superordinate group mismatch predicted less future bipartisanship (β = .290, p < .001) whereas ingroup–superordinate group mismatch did not (β = −.073, p = .163).
Study 5
Study 4 demonstrated that ingroup projection was associated not only with more negative attitudes toward bipartisanship, but also with less bipartisan behavior. To provide further support for the hypothesis that this phenomenon can generalize across temporal and political context, we conducted Study 5, a preregistered replication of Study 4 that used a nationally representative sample and was conducted during a different presidential administration with different issues at the forefront of political discourse.
Method
Participants were recruited in the fall of 2019 using Lucid Theorem. Data were collected from 444 participants, 105 were not Democrats or Republicans and were excluded, yielding 339 participants (183 Democrats and 156 Republicans). The size of the predictive relationship between ingroup projection and future bipartisan behaviors in Study 4 was β = .281; 187 participants would provide 95% power to detect that effect at p < .01, per G*Power (Faul et al., 2007). We increased that to account for attrition and provide the ability to examine Republicans and Democrats separately. 2 The materials measuring the main variables of interest were identical to those used in Study 4.
Results
For all six elements, ingroup meaning significantly predicted superordinate group meaning (all βs > .507, all ps < .001) and was a significantly stronger predictor than outgroup meaning (see Table 6), replicating previous studies. Ingroup projection predicted ingroup bias (β = .415, p < .001; see Table 4), and outgroup–superordinate group mismatch was a stronger predictor of ingroup bias (β = .418, p < .001) than ingroup–superordinate group mismatch (β = −.112, p = .023; see Table 7). Consistent with Studies 2–4, ingroup projection significantly predicted less previous bipartisan behavior (β = .440, p < .001). Outgroup–superordinate group mismatch also significantly predicted less bipartisan behavior (β = .388, p < .001); whereas ingroup–superordinate group mismatch predicted more bipartisan behavior (β = −.226, p < .001). With respect to desired future interaction, ingroup projection predicted less bipartisan behavior (β = .425, p < .001). Outgroup–superordinate group mismatch predicted less future bipartisan behavior (β = .379, p < .001) and ingroup–superordinate group mismatch predicted more future bipartisan behavior (β = −.211, p < .001).
General Discussion
The current studies suggest that ingroup projection may contribute to the difficulty of achieving bipartisan cooperation in American politics. Based on the ingroup projection model (Mummendey & Wenzel, 1999; Wenzel et al., 2007), we hypothesized that each subgroup would perceive the superordinate group based on its own attributes, and that these differences in conceiving the superordinate identity can be problematic for intergroup relations. Indeed, Study 1 demonstrated that people perceived significantly more overlap between what it means to be an American and what it means to be a member of their own party than between Americans and the other party. Furthermore, the degree of mismatch between the outgroup and the superordinate group was associated with ingroup bias. Study 2 demonstrated that perceiving the outgroup as a poorer fit to the prototype of the superordinate group also predicted opposition to bipartisanship, and Study 3 replicated these findings using a different set of elements of American identity. Study 4 demonstrated that ingroup projection is also associated with a lack of bipartisan behavior, and Study 5 replicated this finding. Although these studies were conducted over an 11-year period, spanning three different presidential administrations and vastly different political landscapes, their findings are strikingly consistent and robust.
The current paper extends the extant literature on ingroup projection. First, these findings are the first demonstration of which we are aware that ingroup projection occurs in American politics. Second, these studies consistently showed that larger outgroup–superordinate mismatch (rather than better ingroup–superordinate fit) is a stronger driver of ingroup attitudes and behavior, at least in the context of these highly moralized groups (Parker & Janoff-Bulman, 2013). Third, the current studies go beyond attitudes toward ingroups and outgroups by examining both behavioral intentions and reported behaviors with respect to interaction with outgroup members. Extending ingroup projection to political affiliation and examining behavior, specifically engagement in bipartisanship, has important practical implications for American politics.
The current findings demonstrate that Republicans and Democrats have different ideas about what it means to be American. Without a shared definition of what it means to be American, reducing bias and improving bipartisanship by encouraging them to think of their shared identity is unlikely to be successful (see also Rutchick & Eccleston, 2010). That is, the process of assessing the superordinate category is itself subject to partisan bias, via ingroup projection; the intergroup attitudes and consequent resistance to bipartisanship associated with ingroup projection can create further polarization. This suggests that messages seeking to reduce partisan animosity by highlighting superordinate identities (e.g., “let’s come together as Americans”) are likely to backfire, as they remind partisans that the opposing party does not satisfy their definition of “American.” At the same time, political polarization can catalyze ingroup projection. This feedback loop may help to explain why, despite its desirability, bipartisanship may be difficult to bring about. Indeed, Kessler et al. (2010) demonstrated the bidirectional nature of relative ingroup prototypicality and intergroup attitudes within the context of immigration in Germany. A longitudinal research design allowed for cross-lagged analyses that revealed that relative ingroup prototypicality significantly predicted emotions, prejudice and competitive behavior, but that the reverse direction of causality also held.
The current studies had several limitations. Study 1 relied on an undergraduate student sample that was not representative, particularly in terms of age and political orientation, although Studies 2–5 replicated the observed effects using more representative samples. Studies 1–3 used small sample sizes that are not in keeping with current standards, although Studies 4 and 5 used samples that were ample to detect the large observed effects. In addition, all of the studies are correlational, and thus the causal role of ingroup projection in ingroup bias was not established by these findings. Experimental research has shown a causal link between ingroup perceptions and mental representations of superordinate groups on ingroup projection (Bianchi et al., 2010; Machunsky & Meiser, 2014), but few experiments have examined the link between ingroup projection and intergroup bias. Diversifying perceptions of superordinate group categories does appear to reduce outcomes associated with intergroup bias (Ehrke et al., 2014; Jung et al., 2018; Knab & Steffens, 2018; Mummendey & Wenzel, 1999; Wenzel et al., 2016). However, diversifying perceptions of the superordinate group can also backfire and increase bias due to identity threats (Steffens et al., 2017). More experimental and longitudinal work is necessary to better understand how to effectively mitigate bias associated with ingroup projection.
Despite the presence of pro-bipartisan rhetoric, the American political system is structured such that bipartisanship faces significant obstacles. A two-party system, with a prominent and long-standing group boundary, is already a difficult context in which to form inclusive superordinate identities. Now, political party has arguably become a “mega-identity” that proxies other important identities (e.g., religion, race, education, urban/rural location; Mason, 2018). The current studies show that ingroup projection presents an additional challenge (consistent with Doucerain et al., 2018, and Wenzel et al., 2016), leading to divergent definitions of the superordinate group itself, with serious implications for American politics. Specifically, our findings suggest that people seeking to unite across ideological or party lines should approach outgroup characterizations with caution; explicitly or implicitly characterizing others as not fully American may increase ingroup projection. This then makes the ingroup less likely to engage in bipartisanship, which reinforces and intensifies polarization. One alternative strategy to improve relations between party would be to emphasize common beliefs and behaviors among opposing parties (e.g., Lowe & Muldoon, 2014) as part of belonging to a shared, multifaceted American identity (e.g., Verkuyten & Martinovic, 2016); however, attempts to invoke a common ingroup identity can backfire due to the perceived differences in the definition of American identity (Rutchick & Eccleston, 2010). The lack of bipartisanship that characterizes American politics perpetuates the cycle of polarization and stands in the way of calls for unity.
Supplemental Material
Supplemental Material, sj-docx-1-spp-10.1177_19485506211046788 - Ingroup Projection in American Politics: An Obstacle to Bipartisanship
Supplemental Material, sj-docx-1-spp-10.1177_19485506211046788 for Ingroup Projection in American Politics: An Obstacle to Bipartisanship by Angela C. Bell, Collette P. Eccleston, Leigh A. Bradberry, William C. Kidd, Catherine C. Mesick and Abraham M. Rutchick in Social Psychological and Personality Science
Footnotes
Acknowledgment
We thank Laura E. Wallace for statistical assistance.
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) received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material
The supplemental material is available in the online version of the article.
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.
