Are emotionally intelligent people less prejudiced? The importance of emotion management skills for outgroup attitudes

People vary in their ability to understand, process, and manage information about one's own and others’ emotions, a construct known as Emotional Intelligence (EI). Past research highlighted the importance of EI in interpersonal relations as well as the key role of emotions underlying outgroup prejudice. Remarkably, hardly any research has investigated the associations between EI and outgroup prejudice. In three studies (total N = 922) conducted in Spain and the United Kingdom, we measured emotional intelligence using self- report and performance tests and prejudice toward a variety of outgroups. Results showed that those with stronger performance- based emotion management skills expressed lower generalized ethnic prejudice (Studies 1– 3), more positive attitudes toward immigrants (Study 2a) and refugees (Study 2b), and less homophobic attitudes (Study 3). This negative association between emotion management and prejudice was found with different performance- based EI measures and held after controlling for self- perceived EI (Study 1) and self- reported abilities to regulate emotions (Study 3). Study 3 further demonstrated that higher empathy partly accounted for the association between emotion management and prejudice. The findings suggest that emotion management abilities play an important, but so far largely neglected role in generalized prejudice.


| INTRODUC TI ON
People from ethnic, religious, and sexual minority groups are often victims of prejudice and discrimination (Carter et al., 2019;House et al., 2011;Francis & McKenna, 2018;Rodriguez et al., 2018;Shavers et al., 2012;Zschirnt & Ruedin, 2016) and such experiences can have a profound and lasting impact. For instance, those who have been targets of prejudice tend to report poorer mental and physical health outcomes (Burton et al., 2013;Dover et al. 2020;Samari et al., 2018;Williams et al., 2019). Therefore, understanding the psychological roots of prejudice is crucial to the development of more effective interventions that aim to reduce prejudice and discrimination.
Individuals differ in how they typically think and feel toward outgroups in general, and some people are particularly predisposed to hold prejudiced views about outgroups (Allport, 1954;. Previous research has demonstrated that individual differences in personality and cognitive ability can shape this generalized disposition to adopt prejudiced attitudes . However, limited research has explored the role of individual differences in emotion-related traits and abilities in explaining prejudice. People vary in their "ability to carry out accurate reasoning about emotions and the ability to use emotions and emotional knowledge to enhance thought" (Mayer et al., 2008, p. 511), a construct known as Emotional Intelligence (EI; Mayer & Salovey, 1997). High EI individuals are typically more socially competent, experience less conflicts in social relationships, and report greater psychological well-being (Brackett et al., ,2011(Brackett et al., , , 2019Cabello & Fernandez-Berrocal, 2015;Mayer et al., 2016;Schutte et al., 2001), while lower levels of EI are related to socially problematic behaviors such as interpersonal aggression (García-Sancho, Salguero, & Fernández-Berrocal, 2014García-Sancho et al., 2017). Given the key role of emotions underlying outgroup prejudice and intergroup bias, the ability to manage one's emotions may allow individuals to process intergroup emotions more effectively (e.g., regulating negative emotions toward outgroups), leading to greater tolerance and positivity toward outgroups and reducing negative attitudes associated with negative intergroup emotions. Remarkably, whereas there is a vast literature demonstrating the importance of EI in interpersonal and intimate relations (Brackett et al., 2011;Lopes et al., 2011;, hardly any research has investigated the associations between EI and prejudice or outgroup attitudes. The current research addresses this gap in the literature. Allport's (1954) classic notion that those who are prejudiced toward one outgroup are more likely to be prejudiced toward other outgroups has been widely supported by empirical research (e.g., Ekehammar & Akrami, 2003;McFarland, 2010;Zick et al., 2008).

| Individual differences and prejudice
Such covariation between different types of prejudice suggests that this pattern of generalized prejudice might be shaped by latent psychological individual differences (Allport, 1954;Duckitt, 1992;. For example, people scoring lower (vs. higher) on the personality dimensions agreeableness or openness tend to hold more negative attitudes toward outgroups (Ekehammar & Akrami, 2003Ekehammar et al., 2004;Sibley & Duckitt, 2008). Furthermore, research examining the specific facets of agreeableness and openness found that particularly tender-mindedness (i.e., the tendency to be guided by one's emotions), and openness to feelings (i.e., receptiveness to one's emotions) showed strong negative relationships with prejudice (Ekehammar & Akrami, 2007). This indicates that personality traits related to the understanding of and receptiveness to positive emotions are implicated in generalized prejudice, and thus, emotion-related personality dimensions are important to consider in the intergroup domain.
Differences in cognitive ability might also partly account for why some people are more prone to prejudice than others (Dhont & Hodson, 2014;. Several cross-sectional and longitudinal studies relying on a range of intelligence tests and prejudice measures reported that those who have higher cognitive abilities are less likely to express negative outgroup attitudes (e.g., Hodson & Busseri, 2012;Meeusen et al., 2013; for a meta-analysis, see Onraet et al., 2015). These studies tend to use measures that capture broad cognitive abilities including memory, knowledge and comprehension, and reading and writing skills. However, the question of whether cognitive abilities related to the processing and managing of emotions are also related to prejudice has only received scant research attention.
People can experience a range of emotions during intergroup encounters or simply by thinking of a specific outgroup. Whereas some outgroups elicit feelings of warmth and admiration, others make people feel anxious, angry, or disgusted (Cottrell & Neuberg, 2005;Seger et al., 2017). Processing, expressing, and managing these diverse emotional experiences can be challenging and may have implications for how people react toward outgroups. For example, perceiving outgroup members as threatening is typically associated with increased intolerance and derogation, whereas empathizing with outgroups seems critical in establishing positive and sustainable intergroup relations (Stephan & Finlay, 1999;Stephan & Stephan, 2000;Swart et al., 2011). Given that people´s emotions in intergroup contexts are associated with intergroup attitudes, it is also likely that their general ability to manage emotions plays an important role in shaping relatively stable intergroup emotions and attitudes. Indeed, in keeping with the personality approach to prejudice, people differ in how they typically feel toward outgroups, with, for instance, some being more inclined to feel anxious, threatened or disgusted by outgroups than others (Britt et al.,1996;Choma et al., 2012;Hodson et al., 2013;Stephan, 2014). These consistent emotional reactions, specifically toward outgroups, may partly be rooted in aspects of people's emotional intelligence.

| Emotional intelligence
Two main theoretical approaches of EI have been proposed in the literature. The first approach conceptualizes EI as cognitive ability and uses performance-based tests that measure people's ability to solve problems related to emotions (Mayer et al., 2001;Mayer, Salovey, et al., 2008). The second approach conceptualizes EI as a set of emotionrelated self-perceptions and dispositions and uses self-reported tests that measure traits or characteristics associated with emotionally intelligent behavior (Petrides, 2011;Petrides & Furham, 2000).
According to the well-established ability model proposed by Mayer and Salovey (1997), EI is composed of four branches of distinct emotional abilities (Mayer & Salovey, 1997). This set of abilities include (a) the perception of emotions, referring to the ability to identify emotions accurately, (b) understanding emotions, referring to the capacity to reflect upon and analyze the experience of emotions, (c) using emotions to facilitate thought, referring to the skills of employing emotions for problem solving, decision making, and interpersonal communication, and (d) emotion management, referring to the ability to lessen, enhance or adjust an emotional response in oneself or others (Mayer et al., 2016;Mayer & Salovey, 1997;Mayer, Salovey, et al., 2008).
To measure ability EI, Mayer and colleagues developed performance-based tests that aim to measure effortful information processing about emotions. For instance, the Mayer-Salovey-Caruso Emotional Intelligence Test (MSCEIT; Mayer et al., 2002) is a widely used test measuring the capacity to solve emotion-related problems in each of the four branches of the Mayer and Salovey model of EI (1997), scores can also be combined to give a total score of EI. Other scholars have also developed scales to measure specific branches such as the Situational Tests of Emotional Management and Understanding (STEM and STEU, MacCann & Roberts, 2008).
As for all cognitive abilities, performance-based tests are considered the most appropriate way to measure ability EI (Mayer et al., 1999(Mayer et al., , 2002Mayer, Salovey, et al., 2008. However, due to the subjective nature of emotions, individuals can express and manage their emotions in any given situation in several different, relatively adaptive ways. This makes it difficult to evaluate definitive correct or incorrect responses unlike other tests of cognitive abilities (Izard, 2001;Matthews, et al., 2007;Roberts et al., 2001). As such, responses are scored with a degree of correctness, as determined by consensus or expert scoring Mayer et al., 2003).
An alternative conceptualization of EI has been proposed by Petrides and Furnham (2000) who proposed a trait model of EI (also referred to as trait emotional self-efficacy). This model focuses on emotion-related self-perceptions and dispositions, not cognitive abilities or levels of competence, and therefore is theoretically and psychometrically distinct from the ability model (Petrides, 2009(Petrides, , 2010Petrides, Mikolajczak, et al., 2016). More specifically, evidence suggests that trait EI is connected to existing personality dimensions but is significantly discriminant from other models of personality (e.g., Big Five), demonstrating the utility of trait EI over and above existing personality models (Andrei et al., 2016;Gardner & Qualter, 2010;Petrides et al., 2007). By placing EI within the domain of personality rather than intelligence, the trait EI approach emphasizes that there are many different, yet adaptive, ways a person can respond to a situation and some emotional "profiles" might be more useful in some circumstances but not others (Petrides, 2010(Petrides, , 2011. According to Petrides (2010) the trait EI approach better accounts for the subjective nature of emotions compared to the ability EI approach, by recognizing that the appropriateness of the emotional experience might differ based on the person and context. Unlike ability EI which is assessed through maximum performance, trait EI is usually assessed using self-report instruments which measure individuals' subjective perceptions of their own emotional abilities (Petrides & Furnham, 2001;. Although self-report EI measures vary in the number and types of EI components they assess (for an overview see Pérez et al., 2005), these measures typically cover the perceived ability to attend to one's own emotions, understand different emotional states, and regulate feelings. As such, self-report measures are potentially biased due to self-report biases such as socially desirable response patterns. A frequently used self-report EI measure is the Trait Meta-Mood Scale (TMMS, Salovey et al., 1995), which measures individual differences in the tendency to pay attention to emotional experiences, the level of clarity in understanding one's feelings, and the capacity to repair emotions through regulation. Performance-based and self-report EI tend to be weakly correlated and can be considered distinct, complimentary aspects of EI that can show differential associations with other constructs (Brackett & Mayer, 2003;O'Connor & Little, 2003;Van Rooy et al., 2005). Self-reported EI tends to be more strongly correlated with personality traits than with cognitive ability, whereas performance-based EI tends to be more strongly correlated with cognitive ability than with personality traits Van Rooy et al., 2005).
Furthermore, when examined separately, both self-reported and performance-based EI are related to greater social competence (e.g., Brackett et al., 2006;Mavroveli et al., 2007). Yet, when examined simultaneously, scores on performance-based measures seem to play a more prominent role, particularly for men (Brackett et al., 2006). More specifically, Brackett and colleagues (2006) found that performancebased EI, but not self-reported EI, was associated with greater levels of social competence exhibited in both pre-existing social relationships and in the initiation of new social relations. Such findings indicate that the association between EI and greater social competence can be largely attributed to the mental ability to process and use emotional information (i.e., performance-based EI), rather than self-perceived EI.
Other research further highlighted that not all four branches of EI are equally implicated in positive social relations. For instance, Lopes and colleagues found that out of the four EI branches, only emotion management was significantly correlated to both selfreported and peer-reported friendship quality (Lopes et al., 2004).
Moreover, emotion management abilities predicted friendship quality more strongly than self-reported emotion regulation, highlighting the specific importance of performance-based emotion management in positive social interactions (Lopes et al., 2004).
Indeed, the ability to manage emotions is also uniquely related to a range of socially adaptive characteristics such as positive outcomes in daily life, greater interpersonal sensitivity and higher quality social interactions, and better social and intimate relations (Cabello & Fernandez-Berrocal, 2015;Lopes et al., 2003;Lopes et al., 2005, for an overview see Lopes et al., 2011). Furthermore, Extremera and Fernández-Berrocal (2004) explored the relationships between selfreported EI, performance-based emotion management, and quality of interpersonal relationships. They found that stronger emotion management abilities predicted greater levels of intimacy and affect in their social relations, yet higher self-perceived attention to emotions predicted higher antagonism levels.
Taken together, not only does this previous research highlight the role of emotion management in positive social relations, it demonstrates the distinction between performance-based emotion management and self-perceived emotional competence. Based on these previous findings, we propose that individual differences in emotion management skills are relevant to the understanding of not only interpersonal relations but also intergroup relations and are expected to be meaningfully associated with outgroup attitudes.

| Emotional intelligence and outgroup attitudes
Theoretically, it can be expected that those who have greater emotional skills, and especially emotion management skills, not only show greater interpersonal sensitivity in the domains of work and intimate relations but are also cognitively well equipped to better empathize with members of other groups, which in turn is related to lower prejudice. Indeed, the emotion management branch of EI shows the strongest relationship with empathy (Ciarrochi et al., 2000;Mayer et al., 1999), while empathy, in turn, is vital in establishing harmonious intergroup relations (Batson et al., 1997;Stephan & Finlay, 1999;Swart et al., 2011) and shows a robust negative association with generalized prejudice (McFarland, 2010;Levin et al., 2016; see also Onraet et al., 2017). In line with this rationale, the first study by Onraet and colleagues demonstrated that those who perceive themselves as being less emotionally competent show greater prejudice toward immigrants (Onraet et al., 2017). Furthermore, a significant part of the negative association between self-reported EI and anti-immigrant prejudice was explained by greater levels of empathy among those higher in EI. However, by focusing on the relations between self-reported EI and anti-immigrant prejudice, it remains unclear if (a) similar relations exist for performance-based EI after accounting for self-reported EI, (b) whether different emotional abilities are equally predictive of outgroup prejudice, or alternatively, whether emotion management, in particular, plays a more important role (similar to the findings obtained for interpersonal relations) and (c) whether the relations between EI and prejudice generalize to different types of prejudice.

| Overview of current research
The current research investigates the associations between EI, particularly emotion management, and prejudice in two different counties (Spain and the United Kingdom) using both student and community samples. We hypothesize that people with a greater ability to process and manage their emotions hold less prejudiced attitudes toward outgroups. In Study 1, we measured EI using performance-based and self-report tests to investigate their associations with generalized ethnic prejudice and to determine the relative importance of each EI component for prejudice. Next, Studies 2a and 2b aim to replicate the findings using different performance-based measures of emotion management and emotional understanding. Finally, in Study 3, we aim to replicate and extend the findings by examining the unique relationship between performance-based emotion management and prejudice, while also accounting for self-reported emotion regulation.
Furthermore, we investigate the relations for different types of prejudice by focusing on attitudes toward ethnic groups and toward sexual minority groups (i.e., homophobia). Importantly, Study 3 also investigates the role of empathy in explaining (i.e., mediating) the association between emotion management and different types of prejudice. For each study, we aimed to recruit over 200 participants, providing sufficient power to detect associations of average ("typical," Gignac & Szodorai, 2016) effect size as observed in personality and social psychological research (with an alpha of .05 and a power of .80).

| S TUDY 1
The aim of Study 1 was to examine the relations between different components of emotional intelligence and generalized ethnic prejudice. More specifically, we assessed the four branches of EI, using a performance-based measure (i.e., MSCEIT v2.0), as well as different self-reported facets of EI using the TMMS. This allowed us to first investigate the zero-order correlations between all facets of the performance-based and self-report measures of EI and prejudice, and then simultaneously test the predictive value of the significant EI correlates of prejudice.

| Participants and procedure
The participants for this study were 233 Spanish undergraduate students (70% females) aged between 19 and 48 (M = 22.01, SD = 3.61) studying psychology at a Spanish university. Participants were informed that the study aimed to investigate how people feel toward groups and received course credits for participating. Participants first completed the MSCEIT v2.0 in group sessions in a classroom.
The TMMS and the prejudice measure were completed individually as part of an electronic survey.

| Self-reported emotional intelligence
We assessed self-reported EI using the validated, Spanish version

| Generalized ethnic prejudice
A five-item scale measured attitudes toward different ethnic outgroups using 10-point feeling thermometers ranging from 0 to 10° (extremely unfavorable) to 91-100° (extremely favorable). Specifically, participants were asked to indicate how they generally feel toward immigrants, ethnic minorities, Arabic people, Black people, and Asian people (see also Bobo & Zubrinsky, 1996;Dhont et al., 2016;Sears, 1988). The scores were reversed such that high scores indicated higher levels of prejudice and were then averaged in a single score of generalized ethnic prejudice. This scale demonstrated high internal consistency (α = .90).

| Results
Means, standard deviations, and zero-order correlations are presented in Table 1. Some missing data were evident in the sample. All analyses were conducted in MPlus (version 8, Muthén & Muthén, 1998, using FIML to deal with missingness. Although most EI scales were negatively related to prejudice, emotion management was the only performance-based EI branch that showed a significant negative correlation with prejudice (r = −.17, p = .019), corroborating the importance of emotion management for intergroup attitudes. Furthermore, for the self-report EI scores measured with the TMMS, both the attention and repair of emotions subscales were significantly negatively correlated with prejudice (r = −.18, p = .010; r = −.14, p = .040, respectively).
Having established the associations of emotion management, self-reported attention to emotion, and repair of emotion with prejudice, we then simultaneously tested these associations while controlling for gender and age. We used path analysis with the robust maximum likelihood estimator in Mplus (version 8, Muthén andMuthén (1998-2017)). The results confirmed that emotion management was still significantly negatively related to prejudice (β = −.134, p = .041). Attention to emotion was also still significantly associated with generalized prejudice (β = −.163, p = .037), but repair of emotions was not (β = −.126, p = .126).

| S TUDY 2
The first aim of Study 2 was to replicate the association between emotion management and prejudice using a student sample (Study 2a) and an adult sample (Study 2b) recruited from a different cultural context (i.e., the United Kingdom). Furthermore, this study aimed to establish this relation using a different, widely used, and wellvalidated performance-based measure of emotion management, the Situational Test of Emotion Management (STEM, MacCann & Roberts, 2008). Demonstrating the relations between emotion management and prejudice with a different measure of emotion management would provide convergent evidence for the hypothesized association, increasing confidence in the robustness of the findings. The second aim was to test the associations with different outgroup attitude measures in order to assess if emotional management is also related to prejudice toward a wider range of outgroups.
Specifically, in addition to the generalized ethnic prejudice, we included a measure of negative attitudes toward immigrants in Study 2a and a measure of negative attitudes toward refugees in Study 2b.

| Study 2a: Participants and procedure
The sample for Study 2a consisted of 246 Psychology undergraduate students recruited from a UK university (85% female participants).
Participants were informed that the study focused on the possible associations between personality, personal experiences, and attitudes toward several social groups and issues. The measures were completed on a computer, in supervised group sessions, as part of a demonstration on psychological measurements.

| Performance-based emotion management
We used the short 18-item Situational Test of Emotional Management (STEM-B) to measure the ability to manage emotions (Allen et al., 2015). For each item, participants are asked to choose the most effective response to manage an emotional situation. To calculate the scores, a partial scoring procedure is used in which participants are given scores depending on the appropriateness of the answer, with higher scores for better answers as determined by expert ratings. These are summed to give a total score (M = 10.88, SD = 2.26), the maximum score that can be achieved is 15, and the minimum is 0. Allen and colleagues (2015) reported high internal consistency (α = .83), the Cronbach's alpha for the current study was.67.

| Performance-based emotional understanding
To measure participants' ability to understand emotions, we admin-

| Generalised ethnic prejudice
To measure generalized ethnic prejudice, we used similar attitude thermometers as in Study 1, but adapted them to the UK context.
Specifically, we asked participants to indicate their general feelings toward ethnic minorities, Muslims, Hispanic people, Black people, South Asian people, immigrants, and refugees. Scores were reversed scored so that higher scores indicated higher levels of prejudice and were averaged into a single score of generalized ethnic prejudice (M = 2.90, SD = 1.65, α = .94).

| Attitudes toward immigrants
In addition to the attitude thermometers, we also presented four items asking respondents to indicate how they generally feel to-

| Results
The zero-order correlations showed that, as expected, emotion management was negatively and significantly correlated with gen- Furthermore, using path analysis in Mplus, we simultaneously tested the associations of the STEM scores (i.e., emotion management) and the STEU scores (i.e., emotional understanding) with generalized prejudice and attitudes toward immigrants, while controlling for age and gender. The results confirmed that emotion management was significantly negatively related to both generalized prejudice and attitudes toward immigrants, β = −.257, p < .001 (R 2 = .076) and β = −.225, p = .004 (R 2 = .066), respectively.
Emotional understanding was not significantly related to either prejudice measures β = −.021, p = .747 and β = .054, p = .456, respectively. Consistent with the findings of Study 1, these findings provide further evidence for the association between emotion management and prejudice. Study 2b aimed to replicate this association in a heterogeneous community sample of adults from the United Kingdom.

| Measures and results
Emotion management was measured using the same STEM scale as

| S TUDY 3
Studies 1 and 2 examined the relationship between emotion management and prejudice based on ethnicity and immigration status. In Study 3, we extended the research scope and focused on attitudes toward sexual minority groups (i.e., homophobia) to provide a more comprehensive test of the relationships between emotion management and different types of prejudice.
Second, we also examined how emotion management is related to outgroup prejudice and expected empathy to be an important factor in this relationship. Dispositional empathy has been consistently linked to various forms of prejudice including generalized prejudice, and homophobia (Johnson et al., 1997;McFarland, 2010) and is also associated with both performance-based and self-reported EI (Brackett et al., 2006;Ciarrochi et al., 2000;Mayer et al., 1999;Schutte et al., 2001). When considering the associations between specific ability EI branches and empathy, management of emotions is most closely related to empathy (Mayer et al., 1999). Furthermore, other research has demonstrated that empathy mediates the relationship between emotion management (using both performancebased and self-reported tests) and a range of interpersonal outcomes including effective conflict management, forgiveness of others, and higher quality of person-centered care in nursing (Hodgson & Wertheim, 2007;Pérez-Fuentes et al., 2020;Rizkalla et al., 2008). As such, those who are better able to manage emotions likely tend to be more empathetic, which in turn would relate to lower prejudice.
Finally, the findings from Study 1 suggested that emotion management measured with a performance-based EI test is related to outgroup prejudice after controlling for self-reported abilities to regulate negative emotions (the repair facet of TMMS). In order to provide further support that emotion management ability is related to prejudice after controlling for self-reported difficulties in regulating distressing emotions, we included a different measure of self-reported emotion regulation, namely the Difficulties in Emotion Regulation Scale (DERS). The DERS is a validated self-report measure used to assess emotion dysregulation and taps into people's non-acceptance of negative emotions, difficulties engaging in goaldirected behavior, difficulties with impulse control, lack of emotional awareness, inability to use emotion regulation strategies, and lack of emotional clarity (Gratz, & Roemer, 2004;Kaufman et al., 2016).
The scale is typically used in clinical settings and has previously been associated with the self-reported tendency to avoid negative experiences, self-harm, anxiety, depression, and behavioral problems (Gratz, & Roemer, 2004;Kaufman et al., 2016).

| Participants and procedure
A total of 224 undergraduate students from a UK university took part in this study, with 16% male and 84% female participants (M age = 19.29, SD age = 2.54). Regarding ethnicity, 144 participants selfidentified as white, 27 as Black, 22 as Asian, 3 as Arabic, 1 as Latin/ South American, and 26 as "Other." One person did not disclose their ethnicity. We also asked participants to disclose their sexual orientation; 92.4% (207 participants) self-identified as heterosexual, 11 as bisexual, two as homosexual, one as queer, and three as "other." Participants were provided with an anonymous link to a larger survey which included the measures used in the current study. They were informed that the study aimed to collect people's opinions about several social topics. Scales measuring emotion management and emotion regulation were presented first, followed by empathy and the prejudice measures (i.e., attitudes toward lesbians and gay men and generalized ethnic prejudice).

| Performance-based emotion management
As in Study 2, we used the STEM to measure emotion management.

| Self-reported difficulties in emotion management
We administered the shortened 18-item self-report scale measuring difficulties in emotion regulation (DERS, Kaufman et al., 2016), derived from the full version developed by Gratz and Roemer (2004).

| Empathy
We measured dispositional empathy using the 14 items of the Interpersonal Reactivity Index (Davis, 1980(Davis, , 1983 to assess the people's general capacity for compassion and concern for others and to take on another's point of view. Sample items are "I often have tender, concerned feelings for people less fortunate than me" and "Before criticizing somebody, I try to imagine how I would feel if I were in their place." Participants responded to each item using a 7-point scale from 1 (strongly disagree) to 7 (strongly agree). Reversescored items were recoded and mean scores were calculated such that high scores indicated higher levels of empathy (α = .86).

| Homophobia
This construct was measured using the Attitudes toward Lesbians and Gays scale (ATLG-Short Form; Herek, 1988). The short 10-item version of this scale was derived from the full 20-item scale devised by Herek (1984) and aims to measure negative attitudes toward lesbians and gay men (i.e., "Lesbians just can't fit into our society" and "I think male homosexuals are disgusting"). Statements were scored on a 7-point scale anchored by 1 (strongly disagree) and 7 (strongly agree), reverse-scored items were recoded and averages were calculated such that high scores indicate more negative attitudes. Caviola which is comparable to the current study (α = .85).

| Generalized ethnic prejudice
The same measure of generalized prejudice was used as the one used in Study 2a. The internal consistency of this scale was comparable to the previous studies (α = .92). TA B L E 2 Zero-order correlations, means, standard deviations, and reliability scores for variables in Study 3 via empathy using Mplus. The results of this model, presented

| Results
in Figure 1, showed that the direct path between STEM and homophobia was significant, whereas the direct path between STEM and generalized prejudice was not. Furthermore, corroborating our hypotheses, the indirect associations from STEM, via empathy, to both homophobia and generalized prejudice were

| G ENER AL D ISCUSS I ON
Research on the affective factors related to prejudice has typically focused on specific emotions elicited in intergroup contexts as well as on the social nature of these emotions (Cottrell & Neuberg, 2005;Mackie et al., 2008;Tapias et al., 2007 .328 *** develop positive outgroup attitudes and harmonious intergroup relations (e.g., Batson et al., 1997;McFarland, 2010). This finding is in line with previous research that has highlighted the association between emotion management and empathy (Mayer et al., 1999), as well as with recent work of Onraet et al. (2017) who demonstrated that dispositional empathy is key in understanding why those who score higher on self-report measures of EI tend to be less prejudiced toward immigrants.
Moreover, the present findings move beyond this recent work on the associations between subjectively perceived EI skills and anti-immigrant prejudice (Onraet et al., 2017) by revealing that EI, measured with performance-based tests, is significantly related to prejudice toward a variety of outgroups, even after controlling for self-report EI scores. This way, our findings add to the wider debate regarding the differentiation between performance-based and self-report EI measures (Petrides & Furnham, 2001;Petrides, Mikolajczak, et al., 2016) and show that facets of both types of measures are likely uniquely related to outgroup attitudes (e.g., Study 1).
The current work also adds to the literature that highlights the importance of cognitive abilities and styles for prejudicial attitudes.
Specifically, higher levels of cognitive ability tend to be related to lower levels of prejudice (Dhont & Hodson, 2014;Onraet et al., 2015), while cognitive motivational styles reflecting preferences for order and predictability, and aversion toward ambiguity (i.e., need for closure) are positively related to prejudice (Dhont et al., 2013;Makwana et al., 2018;Roets & Van Hiel, 2011). By establishing the relationship between the ability to manage emotions and outgroup attitudes, the current findings demonstrate that other mental abilities are also relevant to intergroup relations in ways that have previously been unaccounted for. Although previous research has demonstrated that links between EI and social relations exist even after controlling for cognitive ability (Brackett et al., 2006;Ciarrochi et al., 2000;Lopes et al., 2005), future research would benefit from investigating the simultaneous associations between cognitive abilities, motivated cognitive styles, and the ability to manage emotions. Examining the incremental validity of EI over and above these other variables in the prediction of prejudice could shed further light on the unique importance of EI in intergroup relations. Furthermore, our findings provide new avenues for the development of prejudice reduction interventions. For instance, future research could test whether existing EI training packages, especially the ones that focus on improving emotion management (Castillo-Gualda et al., 2018;Hodzic et al., 2018;Nelis et al., 2009), have unexplored benefits for intergroup relations and decrease prejudice. Especially promising may be the RULER program (Brackett et al., 2019;Nathanson et al., 2016), which is an evidence-based learning program that seeks to help teachers and students build their emotional abilities with the aim to foster a more supportive climate in the classroom. Research showed that schools that implemented the RULER approach reported a more positive emotional climate compared to control groups. More specifically, after participating in the RULER program, both students and teachers were more supportive and showed greater cooperation, which are key factors in both interpersonal and intergroup relations (e.g., Allport, 1954;Pettigrew et al. 2011).

| Limitations and future research
Before closing, it is important to highlight some limitations of the current research. First, we relied on self-reported measures of prejudice which might be susceptible to social desirability bias. It could be argued that those with higher EI may be more prone to respond to attitudinal measures in a socially desirable manner because of their higher interpersonal sensitivity and an enhanced understanding and appreciation of socially appropriate responses. However, Lopes and colleagues (2004) reported nonsignificant associations between social desirability and ability EI, using the MSCEIT (but see also Rode et al., 2008). Furthermore, participants in all three studies completed the measures anonymously and independently, which should reduce inclinations to complete the attitude measures in a socially desirable way. That said, we cannot completely rule out this alternative interpretation and future studies should replicate the findings while controlling for social desirability.
Additionally, we only relied on cross-sectional data to demonstrate the relations between emotion management and prejudice.
This limitation prevents us to make any claims about the causal directions of these associations. Future studies relying on longitudinal design could shed further light on the long-term changes and effects of emotion management on outgroup prejudice. Furthermore, experimental designs could further investigate the specific mechanisms employed to manage and regulate emotions and the use of empathy in intergroup interactions (i.e., with emotion regulation strategies such as emotion reappraisal and suppression; Gross & John, 2003). More specifically, people high in EI can process and regulate their emotions in ways that suit the social situation (Mayer et al., 2016;Mayer, Roberts, et al., 2008) and they may purposefully up-or down-regulate empathy depending on the circumstances (Zaki, 2014). More research is, therefore, needed to reveal when and how people manage empathy and other emotions in intergroup relations.

| CON CLUS ION
While there is substantial literature examining to what extent differences in beliefs, personality, and cognitive factors are implicated in prejudice (see , for a review), the possible impact of EI on outgroup attitudes has been largely ignored. The current research demonstrated that an individual's capacity to manage emotions is negatively related to prejudice and this relationship is underpinned by higher empathy levels. These findings have implications for research lines on the roles of emotions and cognitions in prejudice and offer promising directions for prejudice reduction interventions based on EI training.

CO N FLI C T O F I NTE R E S T
The authors declare that there is no conflict of interest.

E TH I C A L A PPROVA L
All studies received ethics approval from the ethics committee at the researchers' University; informed consent was obtained from all participants who took part.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.