This is an open access article distributed under the terms of the Creative Commons
Attribution License (Though not a new phenomenon, populist movements are gaining momentum in many Western democracies and beyond (Albertazzi & Mueller, 2013; Kriesi & Pappas, 2015; Mudde & Rovira Kaltwasser, 2013; Skocpol & Williamson, 2012; Woods & Wejnert, 2014). Prominent examples are the “Tea Party” in the U.S.A., the “Front National” (National Front) in France, and the “Alternative für Deutschland” (Alternative for Germany) in Germany. These movements typically revolve around efforts at cultural, economic, and political restoration (“Get your country back!”). They claim they fight for the “neglected” majority population and against powerful cultural, economic, and political elites or the establishment with the alleged aim of redressing various grievances of “the people.” To recruit supporters and enlarge their base, populist movements thus need to politicize people as, or in the name of, the majority.
In this article, we examine how such majority politicization can be set in motion. Both the psychological literature and the sociological literature suggest that feeling aggrieved as a group is a necessary first step for people to politicize. Unequal treatment, deprivation, or discrimination are typical grievances (Klandermans, 1997; Lalonde & Cameron, 1994; Simon & Klandermans, 2001; Walker & Pettigrew, 1984). They sparked, for example, the labor movement, the civil rights movement, the women’s movement as well as the gay and lesbian movement. Environmental or humanitarian disasters or imminent threats thereof together with authorities’ failures to prevent or adequately deal with them are other classical examples of grievances. They sparked, for example, the environmental movement, the peace movement, and the anti-nuclear power movement (Klandermans, 1997; Kriesi, Koopmans, Duyvendak, & Giugni, 1995). More recently, research by Craig and Richeson (2014, 2015) demonstrated that perceived status threats resulting from demographic shifts in the general population induce White Americans to shift to the conservative pole of the political spectrum. That is, the researchers identified a grievance specifically of people undergoing a change from majority to minority membership.
Our interest lies more generally in the effectiveness of grievances in politicizing people in the name of the majority. From the literature as well as popular themes in public discourses we extracted three types of injustices that seemed particularly relevant to the construal of the majority population as an aggrieved group. We focus on injustices that the majority population may suffer in the cultural, economic, or political spheres of society owing to the perception of excessive influences of the norm of “political correctness,” economic interest groups, or established political parties. While we do not claim that the resulting cultural, economic, or political grievances are perfectly mutually exclusive or that this list of grievances is exhaustive, our selection certainly targets three important, if not the most important, arenas of societal participation and thus typical breeding grounds for societal injustices and the experience of grievances. While economic and political grievances are quite familiar from the arena of traditional politics, the cultural grievance symbolized by the political correctness norm points to the arena of more recent (postmodern) identity politics. At least anecdotal evidence suggests that, similar to the situation in the U.S.A., public(ized) discourses in Germany, where the present research was conducted, increasingly revolve around people’s alleged fears that the political correctness norm would illegitimately restrict the freedom of the majority (e.g., Novotny, Pham, & Schmidt, 2016; Staas, 2017).
We used the experimental method to provide research participants with different injustice frames (Gamson, 1992; Sniderman & Theriault, 2004) and thus manipulate their perception of majority grievances. To gauge the resulting majority politicization, we built on the notion of politicization as strategic involvement and enlistment of wider society or representatives thereof as third parties (Simon & Klandermans, 2001; Thomas, McGarty, & Louis, 2014) and measured research participants’ demands for greater societal support for the interests of the majority population. To demarcate politicization both from more prosocial and more radical forms of engagement, we also measured majority solidarity tapping direct engagement for the interests of the majority population (for a similar distinction between political and civic engagement, see Barrett & Zani, 2015) as well as acceptance of non-legal or violent action as an indicator of radicalization of means (Simon, 2011). Finally, because intergroup polarization is a typical consequence of politicization processes (Azzi, Chryssochoou, Klandermans, & Simon, 2011), we also assessed research participants’ attitudes toward a number of groups likely to be seen as responsible for, or benefitting from, the majority population’s grievances (i.e., homosexuals, Muslims, millionaires, politicians). It is noteworthy that the politicization–polarization link is also in line with the (recalibrated) ingroup projection model (Wenzel, Mummendey, & Waldzus, 2007). Wenzel et al. (2007, p. 337) reason that ingroup projection is actually a “claim” to superiority of the ingroup relative to outgroups (i.e., superior prototypicality with regard to the superordinate group). Taking this reasoning to its logical conclusion, it appears that the superiority claim is basically a political claim in that it is addressed to a superordinate social entity and its constituents (Simon, 2011). When this claim, like all political claims, intensifies in the course of politicization, attitudes toward oppositional groups should deteriorate accordingly.
To summarize, our basic hypothesis was that perceived grievances of the majority population would facilitate majority politicization and, as a further consequence, negative intergroup attitudes. We conducted two experiments to test this hypothesis. Experiment 1 was conducted in the laboratory while Experiment 2 was conducted online. In Experiment 1 we employed a convenience sample consisting predominantly of university students, whereas for Experiment 2 we deliberately recruited sympathizers of a populist party in response to an observation we made in Experiment 1 regarding the role of the alleged political correctness norm as a majority grievance.
Finally, it is important to highlight that the bulk of prior work has traditionally centered on politicization processes among members of clearly demarcated or “marked” social groups, such as workers, women, and various ethnic, religious, or sexual minorities (Azzi et al., 2011; Kelly & Breinlinger, 1996; Klandermans, 1997). This is not to say that members of majority or high-status groups would not engage in political struggles. The growing body of research on environmental or peace activism clearly suggests that they do (e.g., Fielding, McDonald, & Louis, 2008; Louis, Amiot, Thomas, & Blackwood, 2016; Renger & Reese, 2017). However, our experiments uniquely add to the still comparatively underresearched majority perspective in that they focus on politicization that people undergo explicitly in the name of the majority. Note that such majority politicization is to be understood in psychological, not sociological, terms. A person does not need to be a majority member on all possible sociological dimensions in order to be able to politicize in the name of the majority. For example, a lesbian woman may very well evince such majority politicization vis-à-vis issues of immigration or taxation. In fact, it is largely its plasticity or fuzziness in which the mobilizing power of majority politicization lies and which provides the psychological basis for its possible ideological (mis)use.
Experiment 1
Our first test of the hypothesis that perceived grievances of the majority population would facilitate majority politicization and, as a further consequence, negative intergroup attitudes was conducted in the laboratory. We presented research participants with interpretive frames that highlighted either a cultural, an economic, or a political injustice against the majority population. We also included a (no frame) control condition. Since we had no a priori information about the specific composition of our convenience sample, except that participants would be recruited on the university campus and that no psychology students would be admitted, we did not venture specific predictions as to which of the three injustice frames would generate the strongest and most effective grievance perception. In this regard, the experiment was exploratory. However, as a testable corollary of our basic hypothesis, we expected that the strongest grievance perception, be it cultural, economic, or political in nature, would also cause the strongest majority politicization and the strongest deterioration of intergroup attitudes.
Method
Participants and Design
One hundred and seven people (65 women and 37 men, 5 missing gender specifications; 101 with and 5 without German citizenship, 1 missing citizenship specification) were recruited on the university campus to participate in “a survey on political processes in German society.” Each participant was paid €7 (then approximately $7.90 U.S.). Average age was 23 years (SD = 4). The majority of participants (93%) were university students from various faculties or departments except psychology. The remaining participants undertook a vocational training, were employed or self-employed, or provided no pertinent information. On average, participants evinced a center-left political orientation with a mean of 3.42 (SD = 0.79) on a left–right self-rating scale ranging from 1 (very left) to 7 (very right) with a separate no-answer option (can’t or do not want to answer).
The experimental design consisted of one between-subjects variable with four conditions (frame: cultural, economic, political, control), to which participants were randomly assigned. The number of participants in a condition varied between 25 and 29.
Procedure
Data were collected between May 12 and 20, 2016. Up to seven people participated in an experimental session. To ensure independence of observations, each participant sat at an individual computer terminal in a separate cubicle, where he or she received all instructions (including those concerning the experimental manipulation) and completed the dependent measures.
Before the experiment started, participants gave their informed consent. Next, each participant, except those in the control condition, read a short paragraph providing one of three different injustice frames designed to draw participants’ attention to potential grievances of the majority population. While the first two sentences were always identical (“In a democracy, politics should follow the opinion of the majority. However, many people in Germany fear that the opinion of the majority is increasingly disregarded.”), the third sentence contained the critical variation. Depending on condition, it read: “Many suspect that /considerations of ‘political correctness’ (e.g., in order to avoid the impression of discrimination) [cultural injustice frame]/economic interest groups [economic injustice frame]/established parties [political injustice frame]/ have an excessive influence and that therefore it often is not the majority opinion on an issue that determines what policy is pursued.” In addition, participants were asked if they could think of an example for such an influence and provide a brief description thereof. The provision of an example was not mandatory for inclusion in the sample. The purpose of the option to provide an example was merely to supplement the main injustice frame manipulation effected by the different introductory paragraphs described above. Participants in the control condition were simply instructed to “think of the previous day and briefly describe what you have done or experienced yesterday.” Subsequently, participants completed the dependent measures, provided sociodemographic information (age, gender, political orientation, etc.), and were then debriefed, thanked, and paid for their participation.
Dependent Measures
We employed two items to check whether the injustice frames induced the perception of majority grievances (“I am convinced that in our society the opinion of the majority population is increasingly disregarded” and “I am angry about how the majority population is treated;” rSB = .77). The same block also contained two items that assessed identification with the majority population (“I see myself as a member of the majority population” and “I identify with the majority population;” rSB = .76). Unless otherwise noted, these and all of the following ratings were made on 7-point scales ranging from 1 (not at all) to 7 (absolutely), and the order of items within a block was always randomized for each participant.
Majority politicization and majority solidarity were measured in the next block, with the former centering on the mobilization of support for the majority in the larger polity and the latter centering on majority members’ mutual support. Majority politicization was measured with three items (“The German state should show more support for the majority population,” “The German government should speak up more clearly for the rights of the majority population,” and “German politics should turn its attention more to the interests of the majority population;” Cronbach’s α = .91), majority solidarity with two items (“I am willing also to work for the interests of the majority population” and “I think it is important that members of the majority population work for their interests, too;” rSB = .56). Radicalization was assessed with two items in a separate block (“Sometimes non-legal protest actions are the only means to wake up the public” and “Only nonviolent protest actions should be supported” [reverse scored]; rSB = .46).
Subsequently, we assessed participants’ (negative) attitudes toward homosexuals, Muslims, millionaires, and politicians belonging to established political parties, each with nine items (Cronbach’s αs ≥ .73). We used parallel items for the four target groups with group-specific adaptations where necessary. For Muslims as target group, for example, the items read: “I feel uneasy around Muslims,” “There are many commonalities between Muslims and non-Muslims”(reverse scored), “Muslims are not willing to integrate,” “Muslims too often disregard German laws,” “For Muslims, already enough is being done in German society,” “Discrimination against Muslims is often portrayed publicly in an exaggerated way,” “We should all help so that Muslims, too, can feel good in Germany” (reverse scored), “I am gladly willing to also personally assist a Muslim who needs help” (reverse scored), and “Muslims have too much influence in Germany.” Because we suspected that the attitude questions could be experienced as particularly sensitive questions, we added a separate no-answer option (can’t or do not want to answer) to the ratings scales. The selection of this option was treated as a missing value. Items for the different target groups appeared in separate blocks, the order of which was randomized for each participant. We scored all responses in such a way that higher scores indicated a more negative attitude.
Finally, in the last block, participants indicated, on separate rating scales, the extent to which they thought that political correctness, economic interest groups, and established political parties were responsible for the neglect of the opinion of the majority population. All measures were self-designed.
Results
For each dependent measure we conducted a one-way analysis of variance (one-way ANOVA) with frame as a four-level between-subjects variable (cultural, economic, political, control). The means, standard deviations, and ANOVA results (including post-hoc tests with Bonferroni correction) are presented in Table 1.
Table 1
Means and Standard Deviations for the Dependent Measures (Experiment 1)
| Measures | Cultural injustice frame
|
Economic injustice frame
|
Political injustice frame
|
Control condition
|
FFrame (3,103) | ||||
|---|---|---|---|---|---|---|---|---|---|
| n = 29
|
n = 28
|
n = 25
|
n = 25
|
||||||
| M | SD | M | SD | M | SD | M | SD | ||
| Perceived majority grievance | 2.50a | 1.13 | 3.73b | 1.26 | 3.78b | 1.46 | 3.36ab | 1.33 | 5.89*** |
| Majority identification | 4.29 | 1.15 | 4.23 | 1.12 | 4.20 | 1.22 | 4.26 | 1.42 | 0.03 |
| Majority politicization | 3.77a | 1.31 | 4.37ab | 1.33 | 4.84b | 0.99 | 4.20ab | 1.26 | 3.43* |
| Majority solidarity | 4.53 | 1.16 | 4.84 | 1.04 | 5.24 | 0.95 | 4.56 | 1.13 | 2.41† |
| Radicalization | 2.22a | 1.11 | 3.13b | 1.34 | 2.94ab | 1.40 | 2.70ab | 1.09 | 2.78* |
| Negative intergroup attitudes | |||||||||
| Homosexuals | 1.86 | 0.61 | 2.05 | 0.96 | 1.91 | 0.64 | 1.78 | 0.51 | 0.72 |
| Muslims | 2.30 | 0.61 | 2.94 | 1.31 | 2.77 | 1.12 | 2.60 | 1.02 | 1.95 |
| Millionaires | 3.64 | 1.08 | 3.97 | 0.86 | 3.92 | 1.07 | 4.03 | 0.92 | 0.85 |
| Politicians (establ. parties) | 3.58 | 1.06 | 3.89 | 0.83 | 3.74 | 1.12 | 3.75 | 0.90 | 0.45 |
| Perceived responsibility | |||||||||
| Political correctness | 3.76 | 1.33 | 4.54 | 1.67 | 3.88 | 1.59 | 3.52 | 1.71 | 2.07 |
| Economic interest groups | 5.38 | 1.15 | 5.21 | 1.34 | 5.84 | 0.94 | 5.48 | 1.23 | 1.32 |
| Established political parties | 4.34 | 1.34 | 4.68 | 1.33 | 4.28 | 1.17 | 4.48 | 1.42 | 0.48 |
Note. Higher mean scores indicate a higher level of the construct in question. Scores could range from 1 to 7. Means in the same row with completely different subscripts differ significantly from each other (p ≤ .05, post-hoc tests with Bonferroni correction). Where necessary, we conducted tests for unequal variances.
†p = .071. *p ≤ .050. **p ≤ .010. ***p ≤ .001.
Manipulation Check
The ANOVA and the subsequent post-hoc tests revealed that the level of perceived majority grievance was low in the cultural frame condition, but significantly higher – and on a moderate absolute level – in the economic and political frame conditions. The respective mean in the control condition fell between that in the cultural frame condition, on the one hand, and those in the economic and political frame conditions, on the other hand.i Although the low level of perceived majority grievance in the cultural frame condition was not expected, the clear difference between the cultural frame condition and the economic as well as the political frame condition still allowed us to test our prediction that majority politicization would be strongest in the condition(s) with the strongest perceived majority grievances. Because the economic and political frame conditions, which were virtually identical in terms of perceived majority grievance, also did not differ from each other on any of the other dependent measures (see Table 1), we focused in the following analyses on the contrast between the cultural frame condition and the combined economic/political frame conditions.
Majority Identification
Identification with the majority population was moderately high (around the scale midpoint of 4) and virtually on the same level in all four conditions.ii
Majority Politicization
The pattern of means followed closely that for perceived majority grievances (see Table 1). More specifically, contrast analysis confirmed that majority politicization was stronger in the combined economic/political frame conditions than in the cultural frame condition (Mcultural = 3.77, SDcultural = 1.31; Meconomic/political = 4.59, SDeconomic/political = 1.20), t(103) = 2.92, p = .004, Cohen’s d = 0.65; all tests reported in the article are two-tailed.
Majority Solidarity
The pattern of means followed closely that for perceived majority grievances, and contrast analysis verified that majority solidarity was stronger in the combined economic/political frame conditions than in the cultural frame condition (Mcultural = 4.53, SDcultural = 1.16; Meconomic/political = 5.03, SDeconomic/political = 1.01), t(103) = 2.04, p = .044, d = 0.46.
Radicalization
Again, the pattern of means followed closely that for perceived majority grievances, and contrast analysis verified that radicalization was stronger in the combined economic/political frame conditions than in the cultural frame condition (Mcultural = 2.22, SDcultural = 1.11; Meconomic/political = 3.04, SDeconomic/political = 1.36), t(103) = 2.81, p = .006, d = 0.65.
Intergroup Attitudes
The one-way ANOVAs yielded no overall effect of frame on any of the four attitudes, but contrast analysis revealed that anti-Muslim attitudes were stronger in the combined economic/political frame conditions than in the cultural frame condition (Mcultural = 2.30, SDcultural = 0.61; Meconomic/political = 2.86, SDeconomic/political = 1.21), t(78.43) = 2.75, p = .007, d = 0.58, test for unequal variances (for other attitudes, ps ≥ .186). Because this pattern of anti-Muslim attitudes closely followed the patterns we had observed for majority politicization, majority solidarity, and radicalization, we also examined whether these three measures served as parallel mediators (see Figure 1). Mediation was tested using Process (Hayes, 2013). The indirect effect via majority politicization and that via majority solidarity were both significant (bias-corrected bootstrapping based on 5,000 re-samples), but operated in opposite directions. Whereas the indirect effect via majority politicization was positive and in line with our expectation (bindirect = 0.303, 95% confidence interval [CI] [0.099, 0.628]), the indirect effect via majority solidarity was negative (bindirect = -0.160, 95% CI [-0.469, -0.017]). There was no indirect effect via radicalization (bindirect = 0.018, 95% CI [-0.146, 0.204]). The remaining direct effect was nonsignificant (bdirect = 0.399, 95% CI [-0.084, 0.883]).iii
Figure 1
Effect of injustice frame on anti-Muslim attitude as mediated simultaneously by majority politicization and majority solidarity, but not radicalization (Experiment 1). Coefficients represent standardized regression weights. The regression weight from the regression analysis without the mediators is given in parentheses.
*p ≤ .05. **p ≤ .01. ***p ≤ .001.
We also explored the relationships between majority politicization, majority solidarity, and radicalization, on the one hand, and intergroup attitudes, on the other hand, in the entire sample using multiple regression analyses. Gender, age, political orientation, and experimental conditions served as control variables. Preliminary analyses conducted separately for attitudes toward homosexuals, Muslims, millionaires, and politicians as the criterion yielded very similar results, and principal-components analysis on the four intergroup attitudes yielded only one component (56% explained variance). To summarize this pattern, we therefore built an overall index of intergroup attitudes by averaging over the four group-specific attitudes (Cronbach’s α = .73). As can be seen in Table 2, we again observed opposite relationships for majority politicization and majority solidarity. Whereas stronger majority politicization was associated with more negative intergroup attitudes, stronger majority solidarity was associated with less negative intergroup attitudes.iv All other relationships were nonsignificant.
Table 2
Multiple Regression Analysis With Negative Intergroup Attitudes Averaged Over Target Groups as Criterion (Experiment 1)
| Control and predictor variables | b | SE | β | t(86) |
|---|---|---|---|---|
| Gender | 0.071 | .140 | .050 | 0.51 |
| Age | 0.001 | .024 | .006 | 0.06 |
| Political orientation | -0.107 | .090 | -.118 | -1.19 |
| Cultural injustice frame | -0.063 | .193 | -.040 | -0.33 |
| Economic injustice frame | 0.260 | .199 | .165 | 1.31 |
| Political injustice frame | 0.083 | .203 | .050 | 0.41 |
| Majority politicization | 0.256 | .060 | .473 | 4.25*** |
| Majority solidarity | -0.199 | .071 | -.303 | -2.80** |
| Radicalization | 0.029 | .055 | .054 | 0.53 |
Note. Gender was coded 1 for women and 2 for men. Experimental conditions were dummy coded with the control condition as reference category. R2 = .25, R2adj = .17, F(9, 86) = 3.19**, variance inflation factors (VIFs) ≤ 1.8.
*p ≤ .050. **p ≤ .010. ***p ≤ .001.
Perceived Responsibility for Majority Grievances
We observed no effect of frame on any of the three measures (see Table 1).
Discussion
Experiment 1 demonstrated that people politicize in the name of the majority to the extent that they see the majority population as an aggrieved social entity suffering from injustices. More specifically, research participants perceived stronger majority grievances in the economic and political frame conditions than in the cultural frame condition, and we subsequently observed a parallel increase in majority politicization with a medium effect size. Similar increases emerged for majority solidarity and radicalization. This pattern of results fits in with the extant collective action and social movement literature on the role of grievances in politicization processes (e.g., Van Stekelenburg & Klandermans, 2013), but also extends it by bringing the majority population into focus as a distinct political actor.
Our prediction that majority grievances would negatively affect intergroup attitudes received partial empirical support. We observed the expected effect on anti-Muslim attitudes, but not on other intergroup attitudes. Interestingly, both majority politicization and majority solidarity played mediating, but opposing roles in the effect on anti-Muslim attitudes. Whereas majority politicization seemed to translate majority grievances into stronger anti-Muslim attitudes, majority solidarity seemed to pave the way for a reduction in anti-Muslim attitudes. Although the experimental effect was limited to attitudes toward Muslims, the opposite associations with intergroup polarization of majority politicization and majority solidarity emerged across target groups when we analyzed the correlational relationships above and beyond experimental influences. We will resume the discussion of these opposing forces if the critical associations can be replicated in Experiment 2.
Finally, our obvious failure to experimentally disseminate a cultural injustice frame portraying the majority population as a victim of political correctness needs discussion. This failure clearly points to the limits of possible experimental manipulations, but it simultaneously demonstrates that we did not conduct an experiment in a social vacuum (Tajfel, 1972). With hindsight, it appears that we should have anticipated the resistance against the cultural injustice frame in our sample consisting predominantly of university students. Although, on average, we observed only a very moderate left-wing political orientation in our sample, university students may generally be reluctant to see political correctness as an injustice. They may even endorse it as an ingroup norm, the reference to which invigorates their concerns about prejudice and discrimination against minorities. From the perspective of populist movements, a reference to political correctness could then actually backfire with regard to the enlistment of this kind of audience. In fact, while neither perceived responsibility of political correctness nor that of economic interest groups or established political parties varied across conditions, comparisons of the perceived responsibility scores within the control condition (see Table 1) revealed that political correctness was generally least likely to be blamed for the suffering of the majority population (ps < .05). In this connection it is also noteworthy that, although the control condition never differed significantly from the other conditions, the mean scores for majority politicization and anti-Muslim attitudes in the control condition consistently fell between those in the cultural frame condition and those in the economic and political frame conditions. It would certainly be questionable to regard the scores in the control condition as a marker of zero politicization or neutral intergroup attitude, because even control conditions are not socially vacuous. Still, the pattern of means indicates that the cultural injustice frame and the corresponding lack of perceived majority grievances may even have triggered some depoliticization and deliberate rejection of anti-Muslim attitudes. This would be consistent with research that uncovered depoliticization among nonaggrieved minority members (Simon, Reichert, Schaefer, Bachmann, & Renger, 2015) as well as with our speculation about the acceptance of the political correctness norm among our research participants. We further develop this line of reasoning in Experiment 2.
Experiment 2
In Experiment 1, participants obviously refused to construe the political correctness norm as a majority grievance and consequently reacted with less majority politicization and less intergroup polarization. We reasoned that this refusal reflected participants’ center-left political orientation and we therefore conducted a second, complementary experiment. We used the same experimental variation and dependent measures as in Experiment 1, but recruited participants with a more conservative, right-wing political orientation. More specifically, we recruited sympathizers of the German populist party “Alternative für Deutschland” (AfD, Alternative for Germany). The AfD was founded in 2013 as a Eurosceptic national conservative party and has subsequently especially sharpened its culturally conservative profile advocating the preservation or restoration of the cultural hegemony of ethnic Germans and traditional Christian family values. Accordingly, its website features the slogan: “It’s about us – Our culture – Our home – Our Germany – Our goals” (www.afd.de). In 2017, the AfD claimed 28,000 registered members and received 12.6 percent of the vote in the federal election winning 94 of the 709 seats in the German parliament. In light of the AfD’s emphasis on cultural preservation, if not restoration, we reasoned that, for its sympathizers, the cultural injustice frame would be particularly effective, at least as much as or even more so than the economic or political injustice frames. Thus, unlike the participants in Experiment 1, AfD sympathizers should perceive political correctness as a majority grievance and react with majority politicization and intergroup polarization.
Method
Participants and Design
The sample consisted of 199 German citizens who self-identified as AfD sympathizers (140 men and 55 women; four missing gender specifications). Average age was 46 years (SD = 13). Fifty-five percent of the participants indicated an education level that was below the level necessary for university admission (i.e., below the German “Abitur”), and 22% reported to have a college or university degree. The majority of participants (64%) were employed or self-employed. The remaining participants were unemployed or looking for work (4%), were students (3%), or undertook a vocational training (2%), but a sizeable number provided no pertinent information (28%). On average, participants evinced a center-right political orientation with a mean of 4.76 (SD = 0.90) using the same 7-point left–right self-rating scale as in Experiment 1.
Participants were recruited via an ISO-certified German online panel (www.respondi.com), which ensures high quality data by minimizing participation frequency and conducting continuous controls.v Participants were screened for sympathy with the AfD and then randomly assigned to the four conditions of the experimental variable (frame: cultural, economic, political, control). The number of participants per condition varied between 48 and 51. Because, in the screening questionnaire, participants could indicate more than one political party with which they sympathized, we compared the political orientation of participants who – in addition to their sympathy with the AfD – also sympathized with at least one established political party (i.e., Christian Democratic/Social Union, Social Democratic Party, Green Party, Liberal Democratic Party, Left Party) with the political orientation of the participants who did not express such additional sympathies. The comparison revealed that the subsample with no sympathies with established parties was more right-wing (MEP_sympathy = 4.52, SDEP_sympathy = 0.83; Mno_EP_sympathy = 4.95, SDno_EP_sympathy = 0.91), t(188) = 3.36, p = .001, d = 0.49.vi We therefore included established party sympathy (yes vs. no) as a quasi-experimental variable in our design.
Procedure and Dependent Measures
Data were collected between November 21 and 25, 2016. Except for the fact that Experiment 2 was conducted online, the experimental procedure was identical to that followed in Experiment 1. We also used the same dependent measures: perceived majority grievances (two items, rSB = .79), majority identification (two items, rSB = .89), majority politicization (three items, Cronbach’s α = .94), majority solidarity (two items, rSB = .64), radicalization (two items, rSB = .41), negative attitudes toward homosexuals, Muslims, millionaires, and politicians (each with nine items, Cronbach’s αs ≥ .71), perceived responsibility for majority grievances of political correctness, economic interest groups, and established political parties (each with one item).
Results
Unless otherwise noted, we conducted a 4 x 2 ANOVA with frame (cultural, economic, political, control) and established party sympathy (yes, no) as between-subjects variables for each dependent measure. The means, standard deviations, and main effects of frame (including the results of subsequent post-hoc tests with Bonferroni correction) are presented in Table 3. Significant main effects of established party sympathy are reported in the text below (in all other cases, Fs ≤ 2.58, ps ≥ .110). The interactions were all nonsignificant (Fs ≤ 1.97, ps ≥ .121).
Table 3
Means and Standard Deviations for the Dependent Measures (Experiment 2)
| Measures | Cultural injustice frame
|
Economic injustice frame
|
Political injustice frame
|
Control condition
|
FFrame (3,188–191) | ||||
|---|---|---|---|---|---|---|---|---|---|
| n = 49
|
n = 51
|
n = 51
|
n = 48
|
||||||
| M | SD | M | SD | M | SD | M | SD | ||
| Perceived majority grievance | 6.22 | 1.06 | 6.00 | 1.28 | 5.78 | 1.36 | 5.89 | 1.35 | 1.46 |
| Majority identification | 5.33 | 1.75 | 5.08 | 1.72 | 4.93 | 1.62 | 5.10 | 1.66 | 0.40 |
| Majority politicization | 6.54 | 0.88 | 6.32 | 0.88 | 6.07 | 1.22 | 6.28 | 1.03 | 2.68* |
| Majority solidarity | 5.70 | 1.28 | 5.56 | 1.08 | 5.32 | 1.27 | 5.41 | 1.17 | 0.88 |
| Radicalization | 3.34 | 1.63 | 3.30 | 1.39 | 3.54 | 1.66 | 3.54 | 1.37 | 0.32 |
| Negative intergroup attitudes | |||||||||
| Homosexuals | 3.63 | 1.69 | 3.50 | 1.48 | 3.48 | 1.48 | 4.03 | 1.69 | 1.42 |
| Muslims | 5.67 | 1.01 | 5.58 | 1.04 | 5.29 | 1.20 | 5.58 | 1.01 | 2.14† |
| Millionaires | 5.40a | 1.14 | 5.03ab | 1.08 | 4.97ab | 1.17 | 4.58b | 1.08 | 4.00** |
| Politicians (establ. parties) | 5.71 | 0.90 | 5.30 | 0.98 | 5.56 | 1.02 | 5.22 | 0.86 | 2.29† |
| Perceived responsibility | |||||||||
| Political correctness | 5.47 | 1.80 | 4.73 | 1.63 | 4.90 | 1.75 | 5.06 | 1.69 | 1.83 |
| Economic interest groups | 5.84 | 1.49 | 5.53 | 1.51 | 5.51 | 1.41 | 5.13 | 1.51 | 1.89 |
| Established political parties | 6.18 | 1.15 | 5.98 | 1.14 | 5.96 | 1.15 | 5.90 | 1.17 | 0.62 |
Note. Higher mean scores indicate a higher level of the construct in question. Scores could range from 1 to 7. Means in the same row with completely different subscripts differ significantly from each other (p ≤ .05, post-hoc tests with Bonferroni correction). Where necessary, we conducted tests for unequal variances. For the attitude measures, ns and degrees of freedom may vary owing to missing values.
†p ≤ .100. *p ≤ .050. **p ≤ .010. ***p ≤ .001.
Manipulation Check
As can be seen in Table 3, the perception of majority grievances was slightly increased in the cultural frame condition, but there were no significant differences between conditions. Instead, the level of perceived majority grievance was generally very high, both in absolute terms and relative to the level observed in Experiment 1.
However, it could be expected that participants who, in addition to their sympathy with the AfD, also sympathized with established political parties would have some difficulty in embracing the political injustice frame with its explicit criticism of established political parties. We therefore specifically compared these participants’ perceptions in the political frame condition with their perceptions in the (presumably most effective) cultural frame condition. The contrast was indeed significant (Mpolitical = 5.47, SDpolitical = 1.23; Mcultural = 6.37, SDcultural = 1.08), t(85) = 2.41, p = .018, d = 0.77, while the two other means, as well as the four means for the participants without sympathies with established parties, fell between these benchmarks without producing any other significant contrast. For participants without sympathies with established political parties, the corresponding contrast was nonsignificant (Mpolitical = 5.95, SDpolitical = 1.45; Mcultural = 6.13, SDcultural = 1.06), t(106) = 0.51, p = .608, d = 0.14. Below, we continue to inspect this contrast separately for participants with sympathies with established political parties and participants without such sympathies, while we again acknowledge that the interactions were nonsignificant.
Majority Identification
There were no significant effects on majority identification.
Majority Politicization
Although the significant main effect of frame and the pattern of means point to increased majority politicization in the cultural frame condition relative to the political frame condition (see Table 3), the subsequent post-hoc tests did not yield any significant differences (ps ≥ .125). However, in line with the pattern observed for the perception of majority grievances, the contrast between the political frame condition and the cultural frame condition was significant for participants who also sympathized with established political parties (Mpolitical = 5.46, SDpolitical = 1.57; Mcultural = 6.30, SDcultural = 1.28), t(85) = 2.21, p = .030, d = 0.58. For participants without sympathies with established political parties, the corresponding contrast was only marginally significant (Mpolitical = 6.39, SDpolitical = 0.84; Mcultural = 6.69, SDcultural = 0.46), t(50.71) = 1.75, p = .087, d = 0.44, test for unequal variances. Finally, the main effect of established party sympathy was significant, F(1, 191) = 7.90, p = .005, η2 = .04. Participants who did not additionally sympathize with established political parties generally showed stronger majority politicization (M = 6.47, SD = 0.85) than participants with such additional sympathies (M = 6.10, SD = 1.17).
Majority Solidarity
There were no significant effects on majority solidarity.
Radicalization
There was only a significant main effect of established party sympathy, F(1, 191) = 4.73, p = .031, η2 = .02. Participants who did not additionally sympathize with established political parties generally showed stronger radicalization (M = 3.64, SD = 1.50) than participants with such additional sympathies (M = 3.17, SD = 1.49).
Intergroup Attitudes
As shown in Table 3, there were significant or marginally significant main effects of frame on three of the four intergroup attitudes (i.e., attitudes toward Muslims, millionaires, and politicians). Post-hoc tests revealed that attitudes toward millionaires were significantly more negative (p = .002), and attitudes toward politicians marginally more negative (p = .060), in the cultural frame condition than in the control condition. In addition, the main effect of established party sympathy was significant for attitudes toward Muslims, F(1, 191) = 4.57, p = .034, η2 = .02, and for attitudes toward politicians, F(1, 190) = 6.87, p = .009, η2 = .03. Participants who did not additionally sympathize with established political parties generally showed more negative attitudes toward both groups (MMuslims = 5.66, SDMuslims = 1.05; Mpoliticians = 5.63, SDpoliticians = 0.94) than participants with such additional sympathies (MMuslims = 5.36, SDMuslims = 1.09; Mpoliticians = 5.23, SDpoliticians = 0.93).
Overall, attitudes toward homosexuals and attitudes toward Muslims were similar in that the least negative attitudes seemed to emerge in the political frame condition, whereas attitudes toward millionaires and attitudes toward politicians were similar in that the most negative attitudes seemed to emerge in the cultural frame condition. Principal-components analysis (with subsequent varimax rotation) on the four intergroup attitudes indeed suggested that they fell into two categories: attitudes toward cultural minorities (i.e., homosexuals and Muslims; 27% explained variance; rSB = .47) and attitudes toward power elites (millionaires and politicians; 44% explained variance; rSB = .63). We therefore also conducted a mixed 4 x 2 x 2 x 2 ANOVA adding type of outgroup (cultural minorities vs. power elites) and subtype (homosexuals vs. Muslims; millionaires vs. politicians) as two within-subjects variables to our original design. Theoretically most important, we observed a significant interaction between frame and outgroup type, F(3, 186) = 4.59, p = .004, η2 = .07, which was not qualified by any higher-order interaction (Fs ≤ 1.88, ps ≥ .135).vii While nonsignificant for cultural minorities, F(3, 186) = 1.85, p = .140, η2 = .03, the effect of frame was significant for power elites, F(3, 186) = 4.08, p = .008, η2 = .06. Post-hoc tests with Bonferroni correction confirmed that attitudes toward power elites were significantly more negative in the cultural frame condition (M = 5.56, SD = 0.82) than in the control condition (M = 4.90, SD = 0.81; p = .002). The means for the two other conditions fell in between (see Figure 2).
Figure 2
Negative attitudes toward power elites as a function of injustice frame (Experiment 2). Error bars represent ± 1 SEM.
Although, overall, there were no significant differences between frame conditions for attitudes toward cultural minorities, participants who additionally sympathized with established political parties evinced a pattern of anti-Muslim attitudes that closely followed the patterns they had produced for perceived majority grievances and majority politicization. That is, they showed significantly more negative attitudes toward Muslims in the cultural frame condition than in the political frame condition (Mpolitical = 4.78, SDpolitical = 1.32; Mcultural = 5.67, SDcultural = 1.02), t(85) = 2.57, p = .012, d = 0.76. For participants without sympathies with established political parties, the corresponding contrast was nonsignificant (Mpolitical = 5.56, SDpolitical = 1.06; Mcultural = 5.67, SDcultural = 1.02), t(106) = 0.41, p = .683, d = 0.10. As in Experiment 1, we examined whether majority politicization served as a mediator. Majority solidarity and radicalization were again included in the analysis as possible parallel mediators (Hayes, 2013). The indirect effect via majority politicization was marginally significant (bindirect = 0.347, 90% CI [0.018, 0.978], bias-corrected bootstrapping based on 5,000 re-samples). Majority solidarity (bindirect = 0.018, 90% CI [-0.055, 0.389]) and radicalization (bindirect = -0.004, 90% CI [-0.167, 0.068]) played no roles as additional mediators. The remaining direct effect was nonsignificant (bdirect = 0.530, 90% CI [-0.075, 1.134]).
We also examined the relationships between majority politicization, majority solidarity, and radicalization, on the one hand, and attitudes toward cultural minorities and power elites, on the other hand, in the entire sample using multiple regression analyses. Gender, age, political orientation, experimental conditions, and established party sympathy served as control variables. As presented in Table 4, stronger majority politicization was associated with more negative attitudes toward power elites, but not toward cultural minorities. Majority solidarity was generally associated with more positive and radicalization with more negative intergroup attitudes. Finally, it is noteworthy that, relative to the control condition, both the cultural injustice frame and the political injustice frame increased negative attitudes toward power elites.
Table 4
Multiple Regression Analysis With Negative Attitudes Toward Cultural Minorities and Power Elites as Criteria (Experiment 2)
| Control and predictor variables | Cultural minorities
|
Power elites
|
||||||
|---|---|---|---|---|---|---|---|---|
| b | SE | ß | t(173) | b | SE | ß | t(173) | |
| Gender | 0.342 | .175 | .139 | 1.95† | 0.065 | .143 | .031 | 0.46 |
| Age | 0.016 | .006 | .192 | 2.59* | 0.023 | .005 | .323 | 4.45*** |
| Political orientation | 0.360 | .087 | .298 | 4.15*** | -0.046 | .071 | -.046 | -0.66 |
| Cultural injustice frame | -0.058 | .220 | -.023 | -0.27 | 0.547 | .179 | .258 | 3.06** |
| Economic injustice frame | -0.212 | .213 | -.084 | -0.99 | 0.235 | .174 | .112 | 1.36 |
| Political injustice frame | -0.282 | .213 | -.115 | -1.33 | 0.367 | .173 | .178 | 2.12* |
| Majority politicization | 0.136 | .094 | .130 | 1.44 | 0.185 | .077 | .211 | 2.42* |
| Majority solidarity | -0.187 | .080 | -.204 | -2.35* | -0.127 | .065 | -.165 | -1.95† |
| Radicalization | 0.120 | .049 | .170 | 2.43* | 0.089 | .040 | .150 | 2.21* |
| Established party sympathy | 0.046 | .160 | .021 | 0.28 | -0.189 | .130 | -.103 | -1.45 |
Note. Gender was coded 1 for women and 2 for men. Experimental conditions were dummy coded with the control condition as reference category. Established party sympathy was coded 1 for no and 2 for yes. Cultural minorities: R2 = .21, R2adj = .16, F(10, 173) = 4.52***; power elites: R2 = .25, R2adj = .21, F(10, 173) = 5.78***; all variance inflation factors (VIFs) ≤ 1.8.
†p ≤ .053. *p ≤ .050. **p ≤ .010. ***p ≤ .001.
Perceived Responsibility for Majority Grievances
There were no significant effects on perceived responsibility of political correctness or of economic interest groups, but there was a significant main effect of established party sympathy on perceived responsibility of established political parties, F(1, 191) = 6.82, p = .010, η2 = .03. Participants who did not additionally sympathize with established political parties generally perceived more responsibility of established political parties (M = 6.20, SD = 1.10) than participants with such additional sympathies (M = 5.76, SD = 1.17).
Discussion
In Experiment 2, which was conducted with sympathizers of a populist party as research participants, we observed generally very high levels of perceived majority grievances and majority politicization, even in the control condition. While these observations confirm that we sampled a segment of society with a pertinent mobilization potential (Klandermans, 1997), they also point to the possibility of a ceiling effect, which may have limited the effectiveness of our experimental manipulation. Still, at least for the subsample of participants who had not severed all ties with established political parties we secured consistent experimental effects of medium size. For one, they clearly differentiated between cultural and political injustices perceiving the (alleged) political correctness norm more clearly as a majority grievance than the influence of established political parties, with which, after all, they still had some sympathy. Moreover, in line with these perceptions, they showed stronger majority politicization in response to the cultural grievance than the political grievance, or more precisely, less majority politicization in response to the latter. In light of the nonsignificant interactions, however, these observations need to be interpreted with due caution. In fact, with hindsight, it appears that our study lacked statistical power to detect the relevant interactions. We will return to the issue of statistical power in the general discussion.
The experiment also provided evidence that majority grievances negatively affect intergroup attitudes. For one, participants who still had some sympathy with established political parties and accordingly saw the majority population as more culturally than politically aggrieved showed more negative attitudes toward Muslims in response to the cultural grievance than the political grievance. Like in Experiment 1, majority politicization seemed to play a mediating role in this effect. In addition, the cultural grievance generally increased negative attitudes toward (economic and political) power elites relative to the control condition as did the political grievance (after controlling for simultaneous influences of possible confounding variables; see Table 4).
We also replicated the opposite associations with intergroup polarization of majority politicization and majority solidarity observed in Experiment 1. The association between majority politicization and negative intergroup attitudes was obtained with regard to power elites, the opposite association between majority solidarity and positive (or less negative) intergroup attitudes emerged for both cultural minorities and power elites.
Finally, the relation between cultural and political grievances deserves discussion. As evidenced by the perceived responsibility scores in the control condition (see Table 3), established political parties were more likely to be blamed for the suffering of the majority population than the political correctness norm, and this was true for both participants with and those without sympathies with established political parties (ps ≤ .021). However, the political grievance revolving around the influence of established political parties did not produce more or stronger effects on majority politicization or intergroup polarization than the cultural grievance revolving around the norm of political correctness. If anything, it produced weaker or even opposite effects while the cultural grievance entailed a further deterioration of the attitudes toward politicians of established parties. Taken together, while established political parties (and most likely other power elites as well) obviously are the prime target for adversarial attributions made by sympathizers of populist parties or movements, the cultural grievance, symbolized by the political correctness norm, seems to develop its own momentum as a powerful force driving sympathizers further toward majority politicization and social polarization.
General Discussion
A distinctive feature of the research presented in this article is that it combined the experimental method enabling causal inference with the ecological realism afforded by the employment of highly topical issues and different samples of participants. The samples employed in our two experiments systematically differed in political orientation and consequently in their perceptions of and reactions to potential majority grievances revolving around cultural, economic, or political issues. The most striking differences between experiments were observed for the cultural grievance symbolized by the political correctness norm. Participants in Experiment 1, who evinced a center-left political orientation, clearly refused to construe it as a majority grievance and reacted with less majority politicization and less intergroup polarization. Conversely, participants in Experiment 2, who were more conservative or even right-wing, clearly construed the majority population as culturally aggrieved and showed pronounced majority politicization and, especially in response to the cultural grievance, increased intergroup polarization with regard to power elites.
We also want to highlight that, far from conducting experiments in a social vacuum, we were not able to generate grievances arbitrarily and from scratch. Rather, we effected experimental manipulations that consisted in shifting research participants’ attention to possible majority grievances that they were prepared or willing to see (or not to see) in the first place. Nevertheless, the resulting variation in perceived majority grievances enabled us to test and actually confirm our prediction that the perception of majority grievances facilitates politicization in the name of the majority as well as intergroup polarization. Both experiments also provided evidence that majority politicization mediates the effect of majority grievances on intergroup polarization and that, in contrast to the divisive role of majority politicization, majority solidarity likely fosters social inclusion.
Limitations and Directions for Future Research
A limitation of the present research was its relatively low statistical power. This limitation is largely a consequence of the exploratory aspects of our research, which led to a number of unanticipated statistical comparisons. In fact, post-hoc power analyses revealed that there was only between 40% and 79% statistical power to detect a significant difference for majority politicization in our focal tests (i.e., combined economic/political frame conditions vs. cultural frame condition with d = 0.65 in Study 1 and political frame condition vs. cultural frame condition for participants with sympathies with established political parties with d = 0.58 in Study 2; all two-tailed). In future attempts to replicate the critical effects, researchers would need a sample size of 64 (two-tailed) or 52 (one-tailed) per cell to achieve a more satisfactory statistical power of 90%. It would be particularly important that replication studies focus on the role of cultural grievances. While, at first glance, it seemed that we observed inconsistent effects of such grievances across our two studies, we suggest that the apparent inconsistency can be traced back to differences in the political orientation of the participants in the two studies. Theoretical integration along these lines is another important task for future research.
Similarly, there was some indication in our research that directing people’s attention to the political correctness norm not necessarily promotes their politicization in the name of the majority, but also has the potential to undermine it. We suspect that people’s response to the political correctness norm critically depends on whether they construe political correctness primarily as a symbol of majority misrecognition or of minority recognition, or in Kymlicka’s (1995, p. 35) terms, as an “internal restriction” of the majority or an “external protection” for minorities. In the latter case, people may even politicize in the name of minorities. The present research provided no direct empirical evidence for this line of reasoning. The broader challenge for future work is therefore to investigate systematically when members of the majority population politicize either in support of the recognition (or even the hegemony) of the majority population’s culture or in support of the (equal) recognition of minority cultures (Fraser & Honneth, 2003; Taylor, 1992).
A related issue that deserves attention in future research concerns the specificity of majority versus minority grievances. From a theoretical point of view, there seems to be little reason to expect that certain grievances are the “prerogative” of majorities and others that of minorities. Our selection of cultural, economic, and political grievances clearly resonates with the spectrum of grievances identified in the literature as critical antecedents of minority politicization (Klandermans, 1997; Simon, 2011; Simon & Klandermans, 2001). In principle, we would therefore expect a parallelism between majority and minority grievances such that, as a function of social context, majority and minorities can experience comparable grievances (e.g., cultural misrecognition, economic deprivation, or political underrepresentation), though not necessarily at the same time. There may be one notable exception, however. The fear of losing numerical superiority should qualify as a grievance unique to (numerical) majorities (Craig & Richeson, 2014, 2015). The experience of the indigenous population of a nation-state in times of massive immigration would be a case in point. In fact, many contemporary populist movements, including the one targeted in Experiment 2, seem to capitalize on, if not exploit, this specific majority grievance, which readily lends itself to a mobilization strategy revolving around national culture rather than social class.
Finally, research on majority politicization needs to be complemented by research on majority radicalization. We secured a number of parallel findings which suggest that majority politicization and majority radicalization share antecedents (e.g., perceived majority grievances) as well as consequences (e.g., intergroup polarization). But the findings were not sufficiently stable across experiments, which may be partly due to insufficient statistical power and/or low reliability of our radicalization measure. More systematic research with pertinent methodological improvements is necessary to reliably identify both the commonalities and differences between majority politicization and majority radicalization, as well as the factors and processes responsible for the transition from one form of engagement to the other. Such research will be highly valuable not only from a scientific, but also from an applied or political point of view. In particular, it shall help us to uncover the dangers of majority politicization without prematurely denouncing and delegitimizing majority politicization as majority radicalization or extremism.