Introduction

A substantial amount of research shows that when racial/ethnic threat is salient, whites are more opposing of policies intended to help racial and ethnic minorities. For example, evidence demonstrates that context (e.g. percentage of Hispanics/blacks living in a county) and information (e.g. racialized campaign messages) cause whites to lower their support for racial and immigration policies (Hopkins 2010; Kinder and Kam 2009; Kinder and Sanders 1996; Mendelberg 2001; Merolla et al. 2013; Nelson and Kinder 1996; Rocha et al. 2011; Valentino et al. 2002; White 2007). Ethnocentrism is one explanation offered for whites expressing anti-black and anti-Hispanic policy preferences. For example, Kinder and Kam (2009) show that ethnocentrism is a latent attitude, and its effect on policy opinions grows substantially when activated by group threat. In fact, they find that when Americans (or whites) believe immigrants (or blacks) hurt their group’s interest, ethnocentrism is a strong predictor of policies perceived to benefit the out-group (Kinder and Kam 2009). Thus far, we know that when group threat is salient, ethnocentrism strongly predicts policies considered to help racial and ethnic minorities. But, can ethnocentrism still predict whites’ policy views, even when group threat is absent? My paper turns to research on emotion for an answer.

Research indicates that the political environment is often filled with emotion (Valentino et al. 2011). Passions usually run high when an election approaches or an economic crisis cripples a country. For example, after the U.S. housing market collapsed in 2008 leading to massive layoffs and putting the country in a deep recession, many Americans felt angry about the economic downturn. According to a 2009 ABC News/Washington Post poll, a majority of Americans opted for “angry”—angry about the role banks (70 %), the Bush administration (58 %), and large corporations (68 %) played in the economic recession. But, some whites’ anger over the financial meltdown triggered an “us versus them” mentality—blaming racial and ethnic minorities for the country’s economic woes. For instance, when asked about the economic recession, Ann Coulter, a conservative pundit, angrily stated that the reason for the current financial crisis was “they gave your mortgage to a less qualified minority.” Richard Cohen, President of the Southern Poverty Law Center, responded to such remarks by stating “some commentators and politicians on the right are pinning blame on poor minorities” and causing some to “look for scapegoats in all the wrong places.”Footnote 1 This example raises an important question. Can feelings of anger, not directed at racial and ethnic minorities, trigger whites’ out-group animosity, such as ethnocentrism, thereby causing them to strongly oppose racial and immigration policies? Moreover, does non-racial/ethnic anger have a similar effect on whites that have positive racial and ethnic attitudes? That is, can anger increase their support for policies considered to benefit racial and ethnic minorities?

In this paper, I argue that experiencing anger, unrelated to racial/ethnic groups, should increase opposition to racial and immigration policies among ethnocentric whites. I also contend that non-racial/ethnic anger should increase support for racial and immigration polices among the least ethnocentric (i.e. racial liberals). By knowing if experiencing anger, unrelated to racial/ethnic groups, activates racial and ethnic considerations among whites, we can better understand when these attitudes will shape American public opinion on race and immigration. Anger-inducing events, such as an economic crisis, problems in the health care system, or a crumbling educational system (basically any undeserved outcome within society), may cause certain whites to scapegoat minorities while persuading others to defend the rights and interests of disadvantaged groups. Perhaps political appeals focusing on immigrants or blacks are not the only effective means of causing group attitudes to matter in American politics. Anger—itself—may produce a similar effect. My findings would suggest that thoughts about racial and ethnic groups are so ingrained in American society via anger that group cues do not need to be salient for whites’ racial and ethnic considerations to impact their policy decisions. In the following pages, I propose a series of hypotheses by integrating the theory of ethnocentrism with two prominent theories of emotion: appraisal theories of emotion and mood and memory theory. Each emotion theory provides important insights as to why non-racial/ethnic anger can activate whites’ racial and ethnic attitudes.

Ethnocentrism

People strive for differentiation by adopting an “us versus them” view of the social world. Tajfel (1981) argues that a basic function of human nature is to strive for a positive self-identity, and membership in social groups influences people’s view of themselves and of others. As a result, he created social identity theory (SIT) to explain in-group bias and out-group prejudice. Under this theory, we organize the world into a basic set of categories (e.g. racial background) such that people fall into some categories but not others. Because of this categorization, people enhance the similarities between themselves and their in-group and accentuate how they differ from out-groups. Thus, their identity takes on an “us versus them” mentality. This social differentiation lays the groundwork for ethnocentric thinking.

According to Kinder and Kam (2009), “ethnocentrism is a general outlook on social difference, it is prejudice, broadly conceived” (p. 42). Ethnocentric individuals characterize themselves and members of their in-group in a positive fashion such as hardworking, honest, or friendly, while members of the out-group are considered lazy, untrustworthy, and hostile. As a result, an ethnocentric person develops a value system that characterizes its members positively, while out-groups are not considered to possess these traits. As a consequence, when in-group favoritism and out-group prejudice occur, they are justified by these value differentials (Tajfel 1981). This difference in values preserves the positive image of the in-group and reinforces how out-groups differ. Value preservation produces a strong sense of moral superiority among in-group members over out-groups. These individuals believe they live life by a strict moral code—adhering to the rules and traditions that govern society- while out-groups do not. This perceived lack of subscription to moral values by out-groups leads to blaming the out-group for their own misfortunes. Consequently, when out-groups receive rights and resources from the government, members of the in-group consider them unfair and unjust. This line of thinking often reflects how dominant groups (e.g. whites) justify their advantage over lower status groups (e.g. blacks or Hispanics); they blame them for their failings (Gurin et al. 1980; Miller et al. 1981; Sidanius and Pratto 1999). Out-groups not adhering to the values of the in-group are also likely to spur strong emotions among in-group members. In fact, Kinder and Kam (2009) argue that ethnocentrism has a strong affective component. If so, what emotion might underlie whites’ ethnocentric thinking?

Appraisal Theories of Emotion

Before we can understand which emotion is linked to ethnocentrism, it is important to know the circumstances causing an individual to experience a particular emotion. An appraisal theory framework determines the broad contours of emotional experience (Lazarus 1991; Smith and Ellsworth 1985). In other words, people’s appraisal of their environment determines the type of emotion they experience (Lazarus 1991; Smith and Ellsworth 1985; Smith et al. 1993). This process involves two types of appraisals. Primary appraisals inform us of whether our environment is congruent with our goals. If so, positive emotions result. On the other hand, if the environment is incongruent with our goals, negative emotions occur. Secondary appraisals identify coping strategies and future expectations, which further differentiate emotions. These appraisals are more conscious and include attributions of blame/responsibility, certainty, and control. They help determine when environmental threats produce anger, fear, disgust, or sadness.

For example, a person experiences fear when he/she negatively evaluates an event that leads to threat and is uncertain of what happened and will happen in the future (Lazarus 1991; Lerner and Keltner 2001). Lack of control is another important attribute of fear; people believe the situation is beyond their control (Smith and Ellsworth 1985). Therefore, when one is threatened, not in control, and uncertain of how to cope with the problem, fear is likely to be aroused. On the other hand, a person experiences anger when he/she feels threatened and is certain who is responsible or blameworthy for the offensive action. That is, the individual places the blame outside of him/herself (i.e. external attribution of blame) (Lazarus 1991; Smith and Ellsworth 1985). These individuals also believe they can influence the likely outcome, so coping potential and future expectations are considered bright (Averill 1983; Lerner and Tiedens 2006). Relying on an appraisal theory framework, researchers demonstrate that anger and fear have different effects on risk-seeking choices (Lerner and Keltner 2001), political participation (Valentino et al. 2011) and information seeking (MacKuen et al. 2010; Valentino et al. 2009).

I argue that anger should be strongly linked to ethnocentric attitudes among whites because they blame racial and ethnic minorities for their lower position in society.Footnote 2 For instance, since the civil rights movement, many white Americans believe that racial inequality would be reduced if blacks would adhere to American traditional values, such as the Protestant work ethic. A quintessential example of this characterization is the highly vitriolic welfare debate, in that, many whites believe a lack of motivation on the part of blacks is responsible for their shortcomings in society (Gilens 1999). In the case of immigration, Schildkraut (2011) also finds that many white Americans have strong resentment toward immigrants. This resentment stems from viewing immigrants as not adhering to the norms and values of American identity. The media often reinforces this resentment by focusing on how immigrants are responsible for the country’s immigration problem.Footnote 3 For instance, Merolla et al. (2013) find that 42 % of news coverage on immigration describes immigrants as “illegal”, “unauthorized”, or “undocumented.” This type of frame puts emphasis on immigrants as the culprit of wrongdoing rather than focusing on their motivation for immigrating to the United States.Footnote 4 In fact, Kinder and Kam (2009) find a strong correlation between ethnocentrism and racial resentment—a belief system that blames racial and ethnic minorities for their problems in society (Sears and Kinder 1971; Kinder and Sanders 1996; Schildkraut 2011). As a result, I suspect that anger underlies whites’ ethnocentric thinking.

Emotion and Ethnocentrism

Why might non-racial/ethnic anger trigger ethnocentrism? Bower and Forgas’ (2001) mood and memory theory helps us understand how this process works. They propose a learning model in that information learned in one emotional state is more likely to be retrieved from memory when subjects are returned to that same emotional state. In fact, Kinder and Kam (2009) argue that ethnocentrism originates, in part, from social learning from parents, and this learning process largely takes place at a fairly young age. For example, they state “the ethnocentrism of parents and the ethnocentrism of children is due not only to genetic transmission but to social learning” (p. 33). Perhaps at a young age ethnocentric whites begin to learn about out-groups during angry events, such as parents discussing a news story that focuses on blacks cheating the welfare system or Hispanic immigrants entering the U.S. illegally and unfairly putting a strain on the economy. After repeated exposure to such information, anger becomes tightly linked to ethnocentric thinking among whites. Once this linkage is firmly established, I expect that experiencing anger should make ethnocentric attitudes more accessible in whites’ minds. This link might be so strong that experiencing anger, even unrelated to racial and ethnic groups, sets ethnocentric thinking into motion. In fact, researchers have demonstrated that incidental anger (i.e. feelings that arise from general experiences in a person’s life) causes people to rely on stereotypes (Bodenhausen et al. 1994) and to deny assistance to welfare recipients (Small and Lerner 2008). Moreover, scholars have shown that general anger triggers anti-black attitudes (Banks and Valentino 2012; Banks 2014a). However, it is unclear whether general anger activates whites’ prejudice toward other minority groups, such as Hispanic and Asian Americans. Since ethnocentrism captures prejudice toward racial and ethnic minorities in general, I examine whether non-racial/ethnic anger activates this form of out-group bias. As a result, I test whether the experience of anger, unrelated to racial and ethnic groups, will boost opposition to racial and immigration policies among ethnocentric whites.

Scholars have mainly focused on the conditions that trigger racial and ethnic prejudice among whites. Very little attention is devoted to understanding the circumstances activating the beliefs of whites sympathetic to the plight of racial and ethnic minorities. What impact might emotions have on the beliefs of whites that score low in ethnocentrism—those favoring out-groups over their in-group? According to Kinder and Kam (2009), a person that scores at the low end of the ethnocentrism scale “represents a topsy-turvy world in which out-groups are seen as virtuous and in-groups as utterly without virtue” (p. 56). So how many whites score low in ethnocentrism? Based on data from the American National Election Study (from 1992 to 2008), I find about 11 % of whites rate racial and ethnic out-groups more favorably than their in-group. And who exactly are whites that rate blacks, Hispanics, and Asians more positively than their in-group? I suspect these individuals are racial liberals that believe the inequities between whites and racial and ethnic minorities are due to discrimination, racial oppression, and social injustice. To them, the moral character of racial and ethnic minorities is not the problem; instead, it is whites’ unfair treatment of these out-groups. Whites that score low in ethnocentrism hold their in-group responsible for the problems in the black and immigrant community, which explains why they feel more favorable toward racial and ethnic minorities than their own group. As a result, anger should also be strongly linked to their beliefs because blame appraisals are an important aspect of their group attitudes.Footnote 5 In fact, Banks (2014b) finds that racially liberal whites, induced to feel angry, are more supportive of health care reform (i.e. a policy considered to help minorities) than similar individuals that feel enthusiastic and relaxed. Therefore, I also examine whether anger, unrelated to racial and ethnic groups, will increase support for racial and immigration policies among whites that score low in ethnocentrism.

Perhaps fear, and not anger, causes ethnocentric whites to be more opposing of racial and immigration policies. An individual who looks different from us (e.g. phenotypically), speaks a different language, and follows different cultural norms may evoke a fearful reaction. In fact, several studies have shown that anxiety causes whites to be more hostile toward immigrants (Brader et al. 2008; Valentino et al. 2013) and policies considered to help African Americans (Suthammanont et al. 2010). What is unclear about this work is whether fear itself causes ethnocentric whites to be more opposing of racial and immigration policies. For example, (Brader et al. 2008) find that whites are more opposing of immigration when they view a threatening news story that contains a Latino migrant, rather than a European migrant. They find that anxiety mediates this effect. Their measure of anxiety, however, is a combination of self-reported feelings of anxiety and anger.Footnote 6 As is often the case with survey measures of emotion, people report feeling a mix of fear and anger (Marcus et al. 2000). In fact, Suthammanont et al. (2010) use a similar emotion measure to test if anxiety increases reliance on racial prejudice in shaping racial policy opinions. Since both of these studies combine anger and fear into one measure, it is unclear whether fear drives whites’ opposition to racial and immigration policies. As a result, I examine whether non-racial/ethnic fear increases opposition to racial and immigration policies among ethnocentric whites.

I test these hypotheses using an adult national experiment on whites to independently induce anger and fear using an emotion induction task. This study provides a precise test of the causal mechanism I believe is at work in activating ethnocentrism during racial and immigration policy opinion formation.

Experimental Study

Polimetrix/YouGov, an Internet survey company, ran the experiment from April 21 to April 30, 2008. The company uses a matching technique to draw its adult sample. Respondents are matched to the national population on gender, age, race, education, party identification, and political interest. There was substantial variation on age (23 % were 18–34; 42 % 35–54; 35 % were 55 and over), gender (58 % female) and education (43 % high school degree or less; 24 % some college; 33 % college graduate). The sample was more likely than the nation to live in the South (51 %) and identify as Republican (47 %) and conservative (47 %). I employ weights, since my goal is to estimate the average effect of the treatment for the population. Weights are randomly distributed across cells, and the results are unchanged if I do not use weights. The total sample size for the experiment is 180 whites.Footnote 7

The experiment was conducted in two waves. This design choice is important, although costly. Similar priming studies measure the primed dimension in the post-test because researchers fear the pretest measure may itself activate thoughts about the group – thus eliminating any experimental effects. However, measuring group attitudes in the post-test carries a different risk: that the stimulus itself will lead to changes in the primed attitude dimension. A preferable design is to measure ethnocentrism in a pretest far enough in advance that it is unlikely to remain salient by the time the individual is exposed to the stimulus in the second wave (Mendelberg 2008).Footnote 8 As a result, group attitudes are measured a week prior to exposure to the emotion induction task. The first wave consisted of group and political attitude measures, such as the ethnocentrism scale. Since my interest is in whether emotions are linked to ethnocentrism, I use the measure that captures the emotional aspect of ethnocentrism—the feeling thermometer scale (Kinder and Kam 2009). To measure ethnocentrism, I use the feeling thermometer ratings of blacks, whites, Hispanics, and Asians. Patterned after Kinder and Kam’s (2009) scale, I create a measure of ethnocentrism by taking the average of the three out-group thermometers (i.e. blacks, Hispanics, and Asians) and subtracting them from the in-group thermometer (i.e. whites).Footnote 9 Several days later, Polimetrix/YouGov re-contacted respondents to participate in wave 2. In the second wave, subjects were randomly assigned to three emotional conditions (anger, fear, and relaxed (control group)), which was followed by measures of racial and immigration policy opinions.

Experimental Manipulation

The experimental manipulation utilizes two induction techniques common in psychological studies of emotion (Banks and Valentino 2012; Bower 1981; Ekman 1993; Lerner and Keltner 2001). Subjects are asked to recall and focus on things in general that lead them to experience a given emotion while viewing an image of a person with a facial expression corresponding to that emotion.Footnote 10 The combination of written and visual stimuli ensures respondents experience distinct negative emotions (i.e., anger and fear). Subjects are asked via the computer to respond to the following query.

Here is a picture of someone who is ANGRY/AFRAID. We would like you to describe in general things that make you feel like the person in the picture. It is okay if you don’t remember all the details, just be specific about what exactly it is that makes you ANGRY/AFRAID and what it feels like to be ANGRY. Please describe the events that make you feel the MOST ANGRY/MOST AFRAID, these experiences could have occurred in the past or will happen in the future. If you can, write your description so that someone reading it might even feel ANGRY/AFRAID.

The control condition asks subjects to focus on things that make them feel relaxed.Footnote 11 As you can see, the emotional stimulus does not focus the respondent’s attention on groups in any way. It is, therefore, a precise test of the hypothesis that emotions themselves can activate specific group attitudes. By not asking respondents to focus on racial/ethnic objects that cause them to experience specific emotions, I can test whether the emotion itself is responsible for the changes I observe in policy views. Response length to the emotional prompt was unrestricted, but subjects were told to take a few minutes to write down anything in general that made them feel the intended emotion. After the induction, subjects completed a post-test questionnaire that included a variety of policy-opinion measures.

Manipulation Check

First, I conducted a manipulation check to determine if the induction procedure operated as expected. Open-ended responses to the induction task were double-coded by two trained graduate students unaware of the hypotheses. They identified the intensity of any negative emotions expressed in the responses.Footnote 12 The scale ranged from 0 to 1, 0 = none, .5 = some and 1 = extreme. The reliability of coders was high: Cronbach’s alpha for anger = .85 and fear = .93. As expected, participants in the anger condition expressed significantly more intense anger (β = .51, p ≤ .001) than those in the control (relaxed) condition but did not express more fear than subjects in the control condition. Correspondingly, respondents in the fear condition expressed more fear (β = .48, p ≤ .001) but not more anger relative to the control condition. These results indicate that the emotion induction performed as intended. Altogether, participants discussed events in their personal lives that reflected the intended emotion.

Results

My first hypothesis is that the anger condition should increase opposition to racial policies among ethnocentric whites—relative to comparable individuals in the control condition.Footnote 13 I also expect that the anger condition should increase support for racial policies among the least ethnocentric, in comparison to similar individuals in the control condition. To test these hypotheses, I regressed Racial Policy Opinions on emotion dummies (Anger Condition and Fear Condition), Ethnocentrism and the interaction between the two, controlling for Ideology, Authoritarianism, Education, Income, South, and Age. Column 1 of Table 1 shows that the interaction coefficient between the anger condition and ethnocentrism is positive while the coefficient on the anger condition is negative. Meanwhile, the interactive coefficient between the fear condition and ethnocentrism is negative while the coefficient for fear is zero. Due to the complexity involved in interpreting the statistical significance of interaction effects based on the coefficients, (Braumoeller 2004; Brambor et al. 2006; Kam and Franzese 2007; Ai and Norton 2003), I present the marginal effects of the anger condition and fear condition visually in Fig. 1a, b. The figures illustrate where the effect of the anger condition or fear condition on racial policy opinions is statistically significant across the ethnocentrism scale.

Table 1 2008 experiment: the priming effect of emotion for ethnocentrism on opposition to racial and immigration policies
Full size table
Fig. 1
figure 1

2008 experiment: marginal effects of the anger condition and fear condition on racial policy opinions as ethnocentrism changes for whites. a This figure is based on the results in column 1 of Table 1. The solid black line is the marginal effect of anger and the dotted lines are its 95 % confidence interval. b This figure is based on the results in column 1 of Table 1. The solid black line is the marginal effect of fear and the dotted lines are its 95 % confidence interval

Full size image

Figure 1a displays the marginal effect of the anger condition on racial policy opinions across levels of ethnocentrism. A 95 % confidence interval is displayed, in the dotted lines, around the marginal effect of anger in the solid black line. The vertical axis on the right indicates the magnitude of the marginal effect. The vertical axis on the left is for the histogram, in grey bars, which shows the distribution of observations in the analysis on the ethnocentrism variable. Looking at the figure, I find that the anger condition enhances opposition to racial policies among whites that score at .2 on the ethnocentrism scale and higher- relative to comparable individuals in the control condition. The figure also shows that the effect of the anger condition is statistically different from the control condition among whites that score at the high end of the ethnocentrism scale (.6 of the scale and higher). Thus, the anger condition does not increase opposition to racial policies among all ethnocentric whites but only those that have strong negative attitudes toward racial and ethnic minorities. Figure 1a also shows that the anger condition significantly increases support for racial policies among whites that score at the low end of ethnocentrism scale (staring at about −.05). In other words, anger causes white racial liberals to be more supportive of racial policies—relative to similar individuals in the control condition. Looking at the distribution of observations along the ethnocentrism scale, we see that the anger effect is statistically significant among a small group of respondents (8 %).Footnote 14 Turning to the fear condition in Fig. 1b, I find that its effect on racial policy opinions is statistically insignificant across the ethnocentrism scale.Footnote 15

Another prediction of mine is that the anger condition should enhance the effect of ethnocentrism on immigration policy opinion—relative to similar individuals in the control condition.Footnote 16 Column 2 of Table 1 shows that the coefficient on the interaction between the anger condition and ethnocentrism is positive while the anger coefficient is negative. Figure 2a shows that as whites are more ethnocentric (starting at .2 on the scale), the anger condition increases opposition to immigration. The figure also indicates that the anger condition is statistically different from the control condition among whites at the very high end of the ethnocentrism scale (starting at .7). Turning to whites that score low in ethnocentrism, I find that the anger condition significantly increases their support for immigration policies—relative to comparable individuals in the control condition. Similar to the previous figure, the graph shows that the effect of anger is statistically significant among a minority of respondents.Footnote 17 Meanwhile, Fig. 2b shows that the marginal effect of fear on immigration policy opinion is flat and insignificant across levels of ethnocentrism.Footnote 18

Fig. 2
figure 2

2008 experiment: marginal effects of the anger condition and fear condition on immigration policy opinion as ethnocentrism changes for whites. a This figure is based on the results in column 2 of Table 1. The solid black line is the marginal effect of anger and the dotted lines are its 95 % confidence interval. b This figure is based on the results in column 2 of Table 1. The solid black line is the marginal effect of fear and the dotted lines are its 95 % confidence interval

Full size image

In summary, I find that the anger condition pushes whites at the very high end of the ethnocentrism scale to be more opposing of racial and immigration policies—relative to similar individuals in the control condition. That is, the findings show that general anger only moves a small number of ethnocentric whites. Anger does not cause all ethnocentric whites to be more opposing of racial and immigration policies. This result demonstrates that the link between anger and out-group attitudes is particularly powerful among whites that score high in ethnocentrism. The experimental findings also show that the anger condition causes whites that score low in ethnocentrism to be more supportive of racial and immigration policies- relative to comparable individuals in the control condition. On the other hand, the fear condition has no effect on ethnocentrism (i.e. anywhere across the scale) in the case of racial or immigration policy opinions.

Although the results from the experimental study largely support my expectations, I still need to be cautious in generalizing these effects to the national population. The main threat to external validity in the experiment is the realism of the manipulation. I directly induced emotions in order to maximize the distinctiveness of my respondents’ reactions. As a result, I now turn to a survey-based test. A benefit of the American National Election Study (ANES) is that I can see if my effects hold up on different emotion measures, across time, and a nationally representative sample.

American National Election Study

The ANES cumulative file is composed of pooled cross section studies from 1948 to 2012. My analysis focuses on election years between 1992 and 2004 because these years include measures of ethnocentrism, racial and immigration policy opinions, as well as emotion measures of anger and fear. To measure people’s emotional states, I use the emotion items about the presidential candidates. For example, in 1992, survey respondents are asked if Bill Clinton (or George H.W. Bush) has ever made you feel angry/afraid? Responses are summed to create mean level measures of anger and fear. The strength of this measure comes from taking the average emotional reaction to all targets and not an individual’s preferred candidate. Another benefit of this measure is that racial or ethnic groups are not the objects being evaluated; therefore, I can test the effect of non-racial/ethnic anger on ethnocentrism.Footnote 19 As in the experimental study, I use the feeling thermometer scale to measure ethnocentrism.Footnote 20

I regressed Racial Policy Opinions on Anger and Fear, Ethnocentrism and the interaction between the two, controlling for Ideology, Party Identification, Limited Government, Egalitarianism, Male, Education, Income, Age, South, Assessment of Economic Conditions, and year dummies with 1992 as the baseline. Column 1 of Table 2 shows the findings for racial policy opinions. The interaction coefficient between anger and ethnocentrism is positive while the anger coefficient is negative. The column also shows that the interactive coefficient between fear and ethnocentrism is positive while the standalone coefficient for fear is zero. Figure 3a visually shows the marginal effect of anger and its 95 % confidence interval. The figure indicates that anger increases opposition to racial policies, such as affirmative action, among ethnocentric whites, relative to similar individuals that experience neither anger nor fear. This effect is statistically significant among whites that score at about .4 on the ethnocentrism scale and higher. Turning to whites that score low in ethnocentrism, I find that anger significantly increases their support for racial policies. Similar to the experimental results, the figure shows that the effect of anger is statistically significant among a minority of respondents (17 %) in the analysis.Footnote 21 Figure 3b displays the marginal effect of fear on racial policy opinions across the ethnocentrism scale. The figure shows that the effect of fear on racial policy opinions is insignificant across the scale.Footnote 22

Table 2 ANES: the effect of emotion for ethnocentrism on opposition to racial and immigration policies
Full size table
Fig. 3
figure 3

1992–2004 ANES: marginal effects of anger and fear on racial policy opinions as ethnocentrism changes for whites. a This figure is based on the results in column 1 of Table 2. The solid black line is the marginal effect of anger and the dotted lines are its 95 % confidence interval. b This figure is based on the results in column 1 of Table 2. The solid black line is the marginal effect of fear and the dotted lines are its 95 % confidence interval.

Full size image

Column 2 of Table 2 shows the results for immigration policy opinion. The dependent variable is coded 1 for respondents that want to decrease immigration levels and 0 for those that want to keep it the same or increase immigration levels.Footnote 23 Since the measure is a dummy variable, I run a probit regression model. Column 2 shows that the interaction coefficient between anger and ethnocentrism is positive. Meanwhile, the anger coefficient on opposition to immigration is negative. The column also shows that the interaction coefficient for fear and ethnocentrism is negative while the coefficient for fear is positive. Due to the complexity involved in interpreting interaction effects in non-linear models, an analysis of the predicted probabilities is more informative (Ai and Norton 2003).

Figure 4 is the converted predicted probabilities from the results in column 2 of Table 2. The figure shows the probability for decreasing immigration levels at varying levels of ethnocentrism conditional on one’s emotional experience.Footnote 24 The solid black line represents anger while the long-dashed black line represents fear. The dotted black line represents no anger and no fear (referred to as no emotion). The vertical axis on the right shows the predicted probability of decreasing immigration levels. The vertical axis on the left displays the histogram, which illustrates the percent of observations in the analysis across the ethnocentrism scale. Figure 4 indicates that ethnocentrism has a strong effect on immigration opinions, absent anger and fear. For example, the difference in support for decreasing immigration levels is about 55 % age points between whites that score at the very low and high ends of the ethnocentrism scale. Consistent with my expectation, the figure also shows that anger pushes whites that score high in ethnocentrism (at .6 and higher) to be about 25 % age points more willing to decrease immigration levels than similar individuals that experience no emotion. Turning to whites that score low in ethnocentrism, I find that anger has a strong effect on their willingness to increase immigration levels or keep them the same. Specifically, anger increases their support by about 10 % age points. Moreover, the marginal effect of anger is statistically significant among whites that score high and low in ethnocentrism.Footnote 25

Fig. 4
figure 4

1992–2004 ANES: probability of whites support for decreasing immigration levels across emotions as their level of ethnocentrism changes. These are the predicted probabilities calculated from the results in column 2 of Table 2. I calculate the probabilities by manipulating the emotion variables while holding all the other independent variables at their own values observed in the data and then averaging over all of the cases

Full size image

Moving to fear, I find that it pushes ethnocentric whites to be more willing to decrease immigration levels than similar individuals that experience no emotion. However, this difference is not statistically significant. The figure also shows that anger leads whites high in ethnocentrism to be more willing to decrease immigration levels (by about 10 % points) than similar individuals that experience fear. As we move to whites that score low in ethnocentrism, there is not much of a difference between fear and no emotion. On the other hand, the figure shows that fear pushes whites that don’t have strong out-group attitudes (i.e. whites scoring around 0 on the ethnocentrism scale) to be more willing to decrease immigration levels than those that experience no emotion. This difference is statistically significant (two-tailed p ≤ .05).Footnote 26 Thus, the figure shows that fear increases opposition to immigration among whites that don’t have strong negative or positive attitudes toward racial and ethnic minorities.

In summary, the ANES findings largely corroborate the experimental results. The survey findings demonstrate that anger pushes ethnocentric whites to oppose racial and immigration policies and those that score low in ethnocentrism to support these policies. The results for fear are somewhat mixed. Fear does not have a significant effect on racial policy opinions among whites that score anywhere along the ethnocentrism scale. On other hand, fear pushes whites that don’t have strong out-group attitudes to be more opposing of immigration.

Conclusion

Research shows that racial and ethnic prejudice matters a great deal in American public opinion (Kinder and Kam 2009; Kinder and Sanders 1996). The impact of prejudice, however, waxes and wanes depending on the political environment. According to Kam and Kinder (2012), the activation of ethnocentrism “is more likely, when politics is portrayed as a conflict between ingroups and outgroups—between the virtuous and alien” (p. 335). But, can ethnocentric attitudes be activated even in the absence of group threat? One purpose of this paper was to examine if non-racial/ethnic anger contributed to making ethnocentric attitudes more important in whites’ political decision-making. The paper also intended to show if non-racial/ethnic anger had a similar effect among whites that score low in ethnocentrism.

The experimental findings show that the anger condition increases opposition to racial and immigration policies among whites that score high in ethnocentrism—relative to comparable individuals in the control condition. The results also demonstrate that the anger condition enhances support for racial and immigration policies among whites that score low in ethnocentrism—relative to similar individuals in the control condition. Furthermore, I find that the anger effect is significant among a minority of whites. Even so, we should be paying careful attention to these individuals. Research shows that anger causes people with strong group attitudes to act out against an enemy (Mackie et al. 2000) even in violent ways (Claassen n.d.). By knowing if anger, unrelated to racial and ethnic groups, exacerbates whites’ ethnocentric thinking, we can better predict when these individuals may act out against (or in favor of) racial and ethnic minorities. The experimental findings also show that the effect of anger on racial and immigration policy opinions is statistically different from fear. Meanwhile, the fear condition does not have a significant effect on racial and immigration policy opinions among ethnocentric whites nor those that score low in ethnocentrism.

The survey-based test largely supports the experimental findings. The survey results demonstrate that non-racial/ethnic anger pushes ethnocentric whites to be more opposing of racial and immigration policies, and it moves those low in ethnocentrism to be more supportive of these policies—relative to comparable individuals that feel neither anger nor fear. Moving to non-racial/ethnic fear, I find that it has no effect on racial policy opinions for anyone across the ethnocentrism scale. That is, fear does not increase the effect of group attitudes on racial policy opinions. On the other hand, the survey findings indicate that fear increases opposition to immigration among whites that don’t have strong out-group attitudes—relative to similar individuals that experience neither anger nor fear. Why does fear have an effect on this group? Perhaps they feel uncertain about how the growing number of immigrants will impact American society. That is, their willingness to decrease immigration levels is not being driven by out-group attitudes but the uncertainty this change will pose to the country. I also find that the effect of fear on immigration is not statistically different from anger. In fact, although the anger effect is much larger than the fear effect in all of the survey findings, these differences do not reach the conventional level of significance (two-tailed p ≥ .05). One possible explanation is that having survey respondents recall how the presidential candidates make them feel is a difficult task. When recalling their emotional experiences, some respondents are likely to mix up their feelings of anger with anxiety—making it harder to isolate the independent effect of these emotions on some political variable. Even with these concerns, the survey findings show that anger has a substantively larger effect on ethnocentrism than fear.

The political landscape is often colored with anger. Although this anger may not be directed at racial and ethnic minorities, these results show that it can still have a detrimental effect on these groups. Whites’ anger about a failing economy or a crumbling school system may make their negative attitudes toward racial and ethnic groups more salient. Therefore, the absence of racial and ethnic conflict in the public discourse on politics does not mean that ethnocentric attitudes will be less potent. Simply the experience of non-racial/ethnic anger leads these attitudes to play an important role in whites’ evaluations of racial and immigration policies.