Big Five Personality Traits and Attitudes Towards Immigrants: The role of general categorization and context
Master’s Thesis
Research Masters Programme Communication Science Graduate School of Communication
Dr. Rachid Azrout
Natalie Jones 11357665
nataliepsjones@gmail.com
Abstract
This study examines addresses the dearth of research surrounding personality and attitudes towards immigrants by testing a conditional, context-dependent argument about their relationship in two studies. Specifically, this article tested the effects of the Big Five personality traits on attitudes towards immigrants both directly and, more importantly, in conjunction with general categorization mediation and contextual factors such as immigrant threat and the timing of the European Migration Crisis of 2015. Through both a survey experiment with U.S.
participants (N = 213) and latent growth curve modeling with a Dutch representative panel (N = 11,312), Openness and Agreeableness were found to have direct effects on attitudes towards immigrants. Specifically, higher Agreeableness and higher Openness both led to more positive attitudes towards immigrants. Moreover, the contingent nature of Openness was discovered in both studies. Specifically, in the experimental study, the main effect of Openness to attitudes towards immigrants was found to be party mediated by general categorization: individuals low on Openness related positively to the categorization of others, which subsequently led to more negative attitudes towards immigrants. Similarly, in the latent growth curve model Openness was found to significantly predict differences in the respondents’ slope, with those low in Openness becoming more negative in their attitudes towards immigrants more rapidly over time from December 2013 to January 2017. The results imply that personality is important for attitudes towards immigrants, and serves as an impetus for rigorous future research on the effects of personality on immigrant attitude formation and even political behavior as a whole.
Keywords: Big Five; personality; attitudes toward Immigrants; Social Identity Theory, general categorization
The Big Five Personality Traits and Attitudes Towards Immigrants: The Role of General Categorization and Context
International migration is a worldwide phenomenon that has only continued to increase in scope, complexity, and impact in recent decades. Modern developments in technology and transportation have enabled widespread geographical mobility to become a reality, empowering large numbers of people to travel further, cheaper, easier, and faster (UN, 2017). In 2017, the UN estimate of international migrants worldwide reached 258 million, up from 220 million in 2010 and 173 million in 2000 (UN, 2017).
Instigated by the rapid growth in immigration to Western societies, coupled with the intensification of media attention on the issue, migration has become one of the most prominent political issues affecting almost every aspect of political life. However, there are few topics more contentious than immigration (Health & Richards, 2016). In spite of the extensive literature on immigration as a formidable engine for growth, such as through alleviating crucial labor shortages, stimulating job development, and the enrichment of cultural diversity, its perceived costs are often those targeted in the political and public spheres (Card, Dustmann, & Preston, 2005; West, 2011; Münz, Straubhaar, Vadean, & Vadean, 2006). Engendered by these negative conceptualizations, anti-immigrant sentiment has continued to rise in many Western societies, epitomized by the rise in support for radical-right parties (Heath & Richards, 2016). Specifically, prejudice towards immigrants robustly predicted support for the 2016 Brexit referendum, while in the United States the Trump administration’s plans to end the Deferred for Childhood Arrivals (DACA) program and enact travel restrictions highly depends upon public sentiment regarding these measures (Meleady, Seger, & Vermue, 2017; Schaffner, MacWilliams, & Nteta, 2017). Thus, public perceptions about immigrants form the foundation for how international migration
is addressed in the global policy agenda, with widespread consequences for the structure of the social fabric (West, 2011).
Given its implications for national identity, citizenship, and the development of essential immigration policies, the dynamics of public opinion about attitudes towards immigrants is necessitated. Numerous scholars have investigated the determining influences of individual motivational and dispositional factors on attitudes towards immigrants (Espenshade & Calhoun, 1993; Lahav & Courtemanche, 2012; Knoll, Redlawsk, & Sanborn, 2011) However, while extensive literature exists on the socio-economic, ideological, and interpersonal contact factors underlying attitudes towards immigrants, and even on how personality affects a variety of other political domains, only recently have scholars started to explore the relationship between personality and prejudice towards immigrants (Vecchione, Caprara, Schoen, Castro, & Schwartz, 2012; Gallego & Pardos-Prado, 2014; Dinesen, Klemmensen, & Nørgaard, 2016).
In order to address the dearth of literature on personality traits and attitudes towards immigrants, this study will make two major contributions. First, this study will test the effects of personality on attitudes towards immigrants both directly and mediated by general categorization. To the author’s knowledge, this study will be the first to investigate the extent to which the impact of personality traits on attitudes towards immigrants is mediated by the general tendency to classify others into groups. Second, this study will also examine the contextual role of personality traits on attitudes towards immigrants through both experimental situational triggers and latent growth curve modeling over time. To the author’s knowledge, this study will also be the first to test the long-term relationship between personality traits and attitudes towards immigrants. These research goals are ambitious, but are vitally important in the aim to serve as an impetus for rigorous scholarly attention on how personality influences attitudes towards
immigrants, and even political behavior as a whole.
Theoretical Framework The Big Five Model of Personality
A multifaceted concept, personality is conceptualized by Caprara and Vecchione (2013) as, “a dynamic system of psychological structures and processes that mediates the relationship between the individual and the environment and accounts for what a person is and may become (p. 24).” Even with this general formulation, personality is often summarized into the Big Five traits, which most personality psychologists agree validly captures the bulk of differences in individual personality structures (John, Naumann, & Soto, 2008; Winter, 2003). Derived through lexical analysis, the Big Five personality model categorizes individual variation in temperament and behavior into five abstract trait dimensions: Extraversion, Conscientiousness, Agreeableness, Openness (to Experience), and Neuroticism (Digman, 1990). Those who score high in Agreeableness are characterized as warm, trustful, and cooperative, while those scoring low in this trait often competitive and manipulative, with little interest in others. A high score in Openness refers to people who are curious, imaginative, and overall open-minded, while those low in this trait more traditional and resistant of change. Conscientious people are dependable, organized, with a preference for dependability, while in contrast those low in this trait dislike rigid structure and are apathetic about responsibility. High Extraversion indicates a person that is bold and social, with an energetic approach to life, while low extraverted (introverted) people are more reserved and prefer solitude. Finally, Neuroticism is often associated with tenser, anxious people with a high sensitivity to negative emotions, while those low in this trait are often more relaxed and emotionally resilient.
As the stability of these five factors has repeatedly been evidenced across a wide array of cultures, languages, and nations, they are often considered to embody a human universal (Heine & Butchel, 2009; Allik & McCrae, 2004). Moreover, personality traits are considered consistent; while the variation in personality can be attributed to genetic components and childhood environment, personality stabilizes early and becomes increasingly stable throughout adulthood (Roberts & Mroczek, 2008; Bazana & Stelmack, 2004). Personality traits are consequently largely positioned as exogenous, distinguished as causally prior to attitudes and behaviors (Mondak, Hibbing, Canache, Seligson, & Anderson, 2010). As such, general beliefs and tendencies are expected to mediate the link between personality traits and political attitude formation.
Formation of Attitudes Towards Immigrants
Concern over immigration is often rooted in ethnocentrism and group animus, as Social Identity Theory [SIT] postulates that citizens are motivated to divide themselves into groups and express prejudice towards other groups classified outside their homogenous social identities (Sniderman, Peri, De Figueiredo, & Piazza, 2000). Through an experiment comparing attitudes towards varying groups of immigrants, Sniderman and his colleagues (2000) demonstrated that respondents who expressed negative evaluations about immigrants held these views regardless of particular immigrant group. Hence, prejudice is not inherent to a certain immigrant group, but rather the consequence of the tendency of people to categorize immigrants as ‘other’ that induces intergroup hostility irrespective of context (Sniderman et al., 2000). Also, scholarly attention on the topic has illustrated that not only do citizens display prejudice through this general classification tendency, but citizens also exhibit a favorable bias towards members of their own
group at the expense of the others (De Vreese & Boomgaarden, 2005; Brown, 2000).
However, these outcomes are not only general; substantial evidence exists differentiating the general tendency to categorize and issue-specific categorization based on context. Race, cultural proximity, perceived skill level, and country of origin are all socio-demographic factors that have consistently shown to influence attitudes towards particular immigrant groups (Ford, 2011; Ford, Morrell, & Heath, 2012; Lee & Fiske, 2006). Similarly, societal stereotypes and particular events exacerbated by the mass media have been found to also play a role in public opinion of specific immigrant groups (Hainmueller & Hangartner, 2013; Valentino, Brader, & Jardina, 2013; Hartman, Newman, & Bell, 2013; Hopkins, 2015). Branton, Cassese, Jones, and Westerland (2011) demonstrated that both media exposure and affect towards Latinos significantly predicted changes in immigration attitudes among non-Hispanic whites after the September 11, 2011 terrorist attacks in the United States.
Azrout, Spanje, and de Vreese (2010) condense this theory into the three-fold argument that (1) citizens with prejudice have a general tendency to categorize, such that when encountering a specific issue, these people will (2) apply this general classification mechanism to frame issue-specific members in terms of out-groups (Azrout et al., 2010). This framing then (3) leads to a negative bias towards these issue-specific “others,” based upon out-group rejection and/or in-group favoritism. To test this theory, the authors evidenced that when confronted with the issue of potential Turkish membership to the EU, participants with negative attitudes towards immigrants framed the issue in terms of out-groups, which consequently led to less support for Turkish membership. As such, the next section will explore how personality relates to attitudes towards immigrants, reviewing the literature to date and proposing theoretical links worthy of investigation.
Direct Effects of Personality on Attitudes Towards Immigrants
Despite recent work on the impact of personality on political outcomes, the influence of the Big Five personality traits on attitudes towards immigrants has only received sporadic research attention. The author finds it imperative that the conditional nature of this relationship is understood, and to begin explores the hypothesized direct links between personality traits and attitudes towards immigrants. As people who score low in Agreeableness are overall less trusting towards others, they should also be less tolerant of difference and hold more negative attitudes towards immigrants (Gallego & Pargos-Prado, 2014; Ackermann & Ackermann, 2015). Similarly, a low score in Openness refers to people who eschew new perspectives and people, and as such should be more negative towards immigrants (Ackermann & Ackermann, 2015; Sibley & Duckitt, 2008). Finally, the effects of the last three traits, Conscientiousness, Extroversion and Neuroticism, on attitudes towards immigrants are more ambiguous: all of these three traits have shown weak or inconsistent links to general prejudice in the past (Sibley & Duckitt, 2008). As a low score in Conscientiousness refers to impulsive people with avoidance for order, they may perhaps welcome the change immigrants have on the existing social order (Sibley & Duckitt, 2008). However, low Conscientious people are also more often those with lower levels of education and income, which are both negative socio-economic predictors of attitudes towards immigrants (Vecchione et al., 2012; Espenshade & Calhoun, 1993). Low Extraversion, a trait characterized by an anti-social nature towards others, should affect social activities such as activism, but its connection to attitudes such as towards immigration is unclear (Dinesen et al., 2016; Sibley & Duckitt, 2008). Lastly, relaxed people with low scores for Neuroticism may apply this relaxed nature towards immigrants as fellow citizens, but may also be more sensitive to the negative associations of immigrant threat (Sibley & Duckitt, 2008).
In regards to the current literature on the influence of personality on attitudes towards immigrants, a small number of scholarly work stand at the forefront of this domain. The first to test these relationships comprehensively, Gallego and Pargos-Prado (2014) utilize regression analyses to find Agreeableness and Openness to Experience to be positively correlated and unconditionally predict positive attitudes towards immigrants in the Netherlands. Building upon their study, Ackermann and Ackermann (2015) also find direct effects of Agreeableness and Openness on attitudes towards immigrants, but also a moderating effect of perceived ethnic diversity in Switzerland, illustrating the contingent relations between personality and attitudes towards immigrants. The third, a study by Vecchione et al. (2012) investigated this conditional relationship further by illustrating that the associations between Openness and Agreeableness on perceptions of immigration were fully mediated by basic values of security and universalism in Italy, Spain, and Germany. Finally, Dinesen et al. (2016) find that people react differently when encountering the same situational triggers based upon their personality traits through a Danish survey experiment; hence, Agreeableness and Openness serve as moderators of economic and cultural threat on attitudes towards immigrants. While all four studies found consistent results for the main effects of Agreeableness and Openness on attitudes towards immigrants, it should be noted that Conscientious was also evidenced to negatively correlate and impact positive attitudes towards immigrants in the Netherlands, Switzerland, and Denmark, and Neuroticism to prejudice towards immigrants in the Netherlands; however, these relationships were negligible, conditional or moderated by other factors such as perceived neighborhood diversity (Gallego & Pardos-Prado, 2014; Ackermann & Ackermann, 2015; Dinesen et al., 2016; Vecchione et al., 2012).
In sum, scholarly research has confirmed the expected differing effects of personality on attitudes towards immigrants from theory: people who score low in Agreeableness and Openness
repeatedly tend to view immigrants more negatively, while Conscientiousness, Extraversion, and Neuroticism have conflicting or negligible results. As personality traits are causally prior, this study expects the following direct effects from this theoretical synthesis of literature.
H1a: People with lower scores in Agreeableness will have more negative attitudes towards immigrants
H1b: People with lower scores in Openness will have more negative attitudes towards immigrants
The Mediating Role of Group Categorization
As aforementioned, this study seeks to build upon these four studies to not only explore these contradicting results and confirm previous findings of the direct effects, thus furthering our knowledge in this new domain of personality and attitudes towards immigrants, but also most critically explore the potential mediating effect of group classification. Personality traits are expected to differentially affect attitudes towards immigrants; however, attitudes towards immigrants can be viewed more broadly as a general mechanism of the categorization of others. The relationships between political attitudes and personality traits, as stable psychological tendencies manifesting in consistent attitudinal and behavioral patterns independent of situational factors, have been researched by a myriad of previous studies. Mondak and his colleagues (2010) propose that the variation in individual psychological predispositions results in differing responses to exposure of environmental stimuli and, accordingly, that personality trait expression will vary by situational triggers. Scholars have tested this assumption across a variety of political outcomes, behaviors, and opinions, finding the Big Five personality traits to be predictive of political ideology, vote choice, candidate preference, party affiliation, political interest, civic
duty, and voter turnout, among many others (Caprara, Barbaranelli, & Zimbardo, 1999; Caprara, Vecchione, Barbaranelli, & Fraley, 2007; Gerber, Huber, Doherty, Dowling, & Ha, 2010; Gerber, Huber, Doherty, & Dowling, 2012; Mondak & Halperin, 2008). Notably, an important strand of research emphasizes the links between authoritarian personality and ethnocentrism, prompted by Adorno, Frenkel-Brunswick, Levin, and Sanford’s seminal book in 1950. However, authoritarianism can be understood as an expression of social attitudes and values, while this study seeks to test the generalized dispositions of personality traits that in fact underlie authoritarianism (Sibley & Duckitt, 2008).
Thus, personality traits are expected to be causally prior to political attitudes and tendencies, and differentially affect general categorization of prejudice when encountering the same situational triggers. Those who score low on Openness embrace traditional norms about the world and resist change, and are therefore likely to employ conventional negative stereotyping of minority groups and intergroup attitudes (Flynn, 2005; Jost, Glaser, Kruglanski, & Sulloway, 2003; Sibley & Duckitt, 2008). These motivations are consistent with group categorization, with those low in Openness expectedly more likely to position those different from themselves as out-groups threatening their in-group norms and prescripts. This research further hypothesizes that those low on Agreeableness should also express higher preference for group categorization. People who score low in Agreeableness have little interest or sympathy in the problems of others, and will focus on pursuing their own self-interests without concern of the consequences for others (Sibley & Duckitt, 2008). Furthermore, those low in Agreeableness prefer positions of power, with a socially competitive view of the world that should result in heightened motivation for group-based dominance and higher levels of group classification.
personality traits, and conduct an experimental design to test how general categorization mediates the relationship between personality and attitudes towards immigrants. Individuals low on Openness and Agreeableness will relate positively to the categorization of others, and as such are more likely to classify groups into in-groups and out-groups and express subsequent favoritism or rejection, respectively. In turn, the categorization of others, through in-group favoritism and out-group rejection, will result in negative attitudes towards immigrants. The following hypotheses summarize the expectations from this theoretical review of literature.
H2a: General group categorization will mediate the effects of Agreeableness on attitudes towards immigrants.
H2b: General group categorization will mediate the effects of Openness on attitudes towards immigrants.
The Crucial Role of Context
In order to causally assess the mediating effect of general group categorization on the relationship between personality and attitudes towards immigrants, this study will assess the critical role of contextual factors. Previous scholars have illustrated the conjunctional relationship between personality traits and various situational factors when affecting political outcomes (Gerber et al., 2010; Mondak, 2010; Mondak et al., 2010). Specifically, conflict theory explains that conflict occurs when members of a dominant social group perceive an out-group, such as immigrants, as an economic, cultural, or political threat (Blumer, 1958; Billiet, Meuleman, & de Witte, 2014).
When considering threat context when relating to attitudes towards immigrants, one traditional exemplifier involves social threat, or the perception of immigrants as a threat to social
identity or the overall social order. The personality traits most likely to be stimulated by this social threat are Agreeableness and Openness. First, due to their socially competitive nature, those low on Agreeableness are expected to be affected more by immigrant social threat than those scoring high in this trait: they are likely to see the influx of immigrants as threats to the social system providing resources (Sibley & Duckitt, 2008). Next, those low in Openness seek stability and social conformity, and should be more sensitive to the social threat of immigrants compared to their peers who score higher on this threat (Sibley & Duckitt, 2008).
In this study, the role of context is examined in two manners: the situational trigger in a survey experiment, and the passage of time through Latent Growth Curve [LGC] modeling. The context in which people are in results in the issue salience of threat, with the hypothesis being that the effect will be stronger for people low on Openness and those low in Agreeableness. While the directly manipulated threat context for the survey experiment will allow for analysis on personality traits and attitudes towards immigrants, the threat context for the LGM model relies upon the event of the European migration crisis beginning in 2015 and the subsequent media salience that should expectantly have caused changes in attitudes towards immigrants in the Netherlands. Thus, the second study will investigate through LGC modeling the long-term effects of personality on attitudes towards immigrants from December 2013 to January 2017, expectantly driven by the contextual threat factor of the European migration crisis. The second study will draw upon data from the representative Longitudinal Internet Studies for the Social Sciences [LISS] panel by CentERdata. To the author’s knowledge, this study will be the first to examine the long-term effects of personality on attitudes towards immigrants, providing a unique opportunity for a reliable and valid understand the contextual causes and consequences in this relationship through time.
H3a: The effect of Agreeableness on attitudes towards immigrants that is mediated by group categorization will also be moderated by the threat context. In specific, the effect will be stronger in high immigrant social threat context compared to a low threat context.
H3b: The effect of Openness on attitudes towards immigrants that is mediated by group categorization will also be moderated by the threat context. In specific, the effect will be stronger in high immigrant social threat context compared to a low threat context.
Study 1: The causal mechanisms underlying the relationship between personality and attitudes towards immigrants
Methods
The first study assesses the direct effects of Agreeableness and Openness on attitudes towards immigrants, as well as the mediating impact of generalized group categorization. To investigate the causal effects, an online survey experiment was conducted in which participants were asked several background and personality questions, asked to read a short story, and then report their attitudes towards immigrants and general categorization measures. Such an experiment allows for the testing of the causal effects of various components in survey experiment treatments, specifically allowing us to randomly assign participants to the two treatment conditions: low threat context (n=113) vs. high threat context (n=100).
Sample
281 participants were recruited through the Amazon Mechanical Turk (MTurk) Platform, but participants with incorrect data for age above 200 years (n = 2) and with illogical answers demonstrating non-investment (n=66) (i.e. “nice” as an answer to the general categorization
open question) were excluded to result in a final sample of 213 participants. Only adults aged 18 and over located in the United States were allowed to participate, and all participants gave their informed consent to the survey. The relevant summary statistics can be found in Table 1.
Stimulus materials
The stimulus material consisted of three paragraphs per each of the two conditions. The story covered the increased influx of immigrants worldwide, and contained identical opening paragraphs on the situation. In the high threat condition, potential threats of immigration to the social order were stressed due to its aforementioned hypothesized ability to activate low Agreeableness and low Openness to negatively view immigrants. In the low threat condition, the benefits of immigration are emphasized to outweigh its costs, implying little to no threat on the existing social order. The researcher constructed the articles to prevent the respondents from having previous knowledge about the specific issues, and wrote the articles to appear as if they could be a short news story. A pilot study of 36 participants successfully found the level of threat-context to be appropriate in the articles, as the chi-square test of independence (p=0.016) comparing the feelings of threat in the low threat context (M=1.22, SD=1.06) to high threat context (M=2.82, SD=2.30) was significant. To view the full stimulus materials, see Appendix B.
Attitudes towards immigrants: The dependent variable is comprised of level of agreement responses to a series of five statements from (1) “fully disagree” to (5) “fully agree.” The statements are identical to that in the LISS Panel, in order to make stronger comparisons between the two studies.
- It is good if society consists of people from different cultures - It should be made easier to obtain asylum in the Netherlands
- Legally residing foreigners should be entitled to the same social security as Dutch citizens
- There are too many people of foreign origin or descent in the Netherlands
- It does not help a neighborhood if many people of foreign origin or descent move in After reverse-coding the last two items, a principal component analysis was performed that indicated one component has an eigenvalue clearly above 1, which is also above the elbow in the scree plot, suggesting the extraction of one factor (component) for the three items. Factor loadings can be found in Appendix D. A reliability analysis of the five items also indicated high
reliability based on the associated Cronbach’s alpha (α=0.82). As such, the five items were combined into an immigration attitude subscale.
Personality traits: Personality traits were measured with BFI-S, a short version of the “Big Five Inventory” [BFI] with each of the five personality dimensions measured by three survey items. Several scholars have evidenced the robustness of this short instrument to validly operationalize the Big Five personality traits (Gerlitz & Schupp, 2005; Lang, Weiss, Stocker, & Rosenbladt, 2007). The present study confirms the number of dimensions through a principal component analysis for each of the personality traits, all indicating one component for each of the three trait items with an eigenvalue above 1 and also above the elbow in the scree plot; hence, the extraction of one factor (component) for each trait was performed with three items. Appendix D contains the relevant factor loadings. Reliability analyses illustrated high reliability for four of the five subscales with the three items, Openness (α=0.82), Conscientiousness (α=0.73), Extraversion (α=0.82), and Neuroticism (α=0.85), while Agreeableness was fairly reliable (α=0.60). Each of the five traits was then rescaled from (0) “not expressed” to (1) “expressed” for subsequent analyses.
General (group) categorization: This mediating variable was measured in two distinct manners. First, a broad-bandwidth factor was created from a series of statements assessing attitudes towards three types of social groups: dangerous, derogated, and dissident groups. Similar to generalized prejudice scales by previous scholars, Asbrock, Sibley, & Duckitt (2010) developed this version of the scale in which participants rated the warmth or coldness of their feelings towards 21 different groups and social categories varying in level of specificity from 1 (very cold) to 7 (very warm) (Duckitt & Sibley, 2007; Ekehammer, Akrami, Gylje, & Zakrisson, 2004; McFarland & Mathews, 2005). The scores were reversed such that higher scores
represented more negative (cold) feelings in order to index prejudice. A principal component analysis confirmed the three expected factors with eigenvalues above 1, so a hierarchical factor analysis for the three factors was performed to find one factor (component) with an eigenvalue above 1 and above the elbow in the scree plot. Although the dangerous groups dimension did not load strongly on the factor (0.47), all three dimensions are necessary as general categorization should encompass not just groups that are the targets of prejudice (i.e. minorities, women), but any groups seen as competitive or threatening (Asbrock, Sibley, & Duckitt, 2010). Cronbach’s α’s for the seven-item dangerous (α=0.94), seven-item derogated (α=0.86), and seven-item dissident groups (α=0.87) also all indicate high internal reliability. The Cronbach’s alpha value for the full 21-item scale was also high (α=0.90), with the reliability between the three factors adequate (α=0.61). Hence, the 21 items were averaged and combined into a General Categorization scale. Appendix D contains the relevant factor loadings.
Next, the second operationalization of this categorization pertained to an open question that asked participants to imagine a scenario in which a family of immigrants is seeking asylum in their home country, based upon a similar measure developed by Azrout, Spanje, & de Vreese (2010). Participants wrote a few sentences on their position, and the degree to which respondents framed the issue in terms of out-group was examined. Following the operationalization in Azrout, Spanje, & de Vreese (2010), the first item coded was the explicit reference of immigrants as an out-group (i.e. “They will not assimilate” or “They don’t have a right to be here with us”) from 0 (absent) to 1 (present). Next, as Sniderman et al. (2000) argue that categorization of immigrants depends upon their perceived threats (i.e. “They are weakening our traditions” or “They are putting us in danger”), explicit mentions of threat resulting from immigrants was recorded from 0 (absent) to 1 (present). In contrast, respondents could also
frame immigrants as not an out-group (i.e. “We are all human beings with rights” or “Our country of immigrants”), and this third item was measured from 0 (absent) to 1 (present) then reverse-coded. Finally, the last item measured explicit references to positive consequences of immigration (i.e. “They will enrich our culture and social experiences” and “They are a positive force”) from 0 (absent) to 1 (present) then reverse-coded, as Azrout et al. (2010) argue perceived positive results should signify less categorization. The four items formed a strong, reliable Mokken Scale (H = 0.58) (Mokken, 1971). The four items were then added to construct a measure of the degree to which the issue is framed in terms of out-group (M= 2.10, SD= 0.83) from 0 (explicit mention of immigrants as an in-group with positive consequences) to 4 (explicit mentions that immigrants are an out-group with negative consequences).Finally, a second coder also scored the items for 58 respondents. The inter-coder reliability was sufficient, with Krippendorff’s alpha values ranging from between 0.68 to 0.85 for the four scored items and a Krippendorff’s alpha value of 0.74 for the framing scales. Full descriptives can be found in Appendix D.
Control Variables: Age, gender, education, political orientation, and immigrant identity were all measured to control for their potential confounding influence.
Results Randomization and Manipulation
First, after Levene’s tests confirmed the equal variance assumption for each (p>0.05), four independent mean t-tests and one chi-square test were conducted that determined participants in the two conditions did not significantly differ with respect to education (p=0.421), political orientation (p=0.664), immigrant identity (p=0.433), age (p=0.754), and gender (χ2(1) =
0.73, p=0.392), indicating successful randomization between the conditions. Next, after Levene’s tests also confirmed the equal variance assumption for each (p>0.05) of the Big Five personality traits, five independent mean t-tests were performed that also identified successful randomization between the conditions as participants in the two conditions did not significantly differ with respect to Openness (p=0.618), Neuroticism (p=0.673), Agreeableness (p=0.179), Extraversion (p=0.506), and Conscientiousness (p=0.471). Participants were also asked to what extent they found the story on immigrants threatening, setting aside their personal feelings. A significant chi-square test of independence (p=0.022) comparing the feelings of threat in the low threat context (M=1.58, SD=1.80) to high threat context (M=2.68, SD=2.27) indicates a successful manipulation as intended in terms of immigrant threat context.
The Role of Agreeableness and Openness
First, to describe model fit, Maximum Likelihood estimation was utilized to fit the path models depicted in Figure 1 and Figure 2. The model for Agreeableness, mediated by general categorization as an open-ended question, with a direct effect to attitudes towards immigrants resulted in an adequately fitting model (χ2(3) = 2.71, p = 0.439; CFI = 1.00; RMSEA = 0.00, 90%CI = [0.00,0.11]); however, removing the direct effect to test the full mediation model was not satisfactory (χ2(4) = 9.25, p = 0.055; CFI = 1.00; RMSEA = 0.08, 90%CI = [0.00,0.15]). Similarly, the model for Agreeableness, mediated by general categorization as a scale, with the direct effect resulted in an adequately fitting model (χ2(3) = 2.48, p = 0.479; CFI = 1.00; RMSEA = 0.00, 90%CI = [0.00,0.11]), while the removal of the direct effect was also not satisfactory (χ2(4) = 14.19, p = 0.007; CFI = 0.99; RMSEA = 0.11, 90%CI = [0.05,0.17]). The model fit for Openness was similar, as the model without the direct effect with general categorization as a
scale did not adequately fit (χ2(4) = 16.09, p = 0.003; CFI = 0.98; RMSEA = 0.12, 90%CI = [0.06,0.18]), but the inclusion of its direct effect to attitudes towards immigrants resulted in a suitable model (χ2(3) = 2.47, p = 0.481; CFI = 1.00; RMSEA = 0.00, 90%CI = [0.00,0.11]). Finally, the model fit for Openness with general categorization as an open question also did not fit (χ2(4) = 14.28, p = 0.006; CFI = 0.99; RMSEA = 0.11, 90%CI = [0.05,0.17]), but the inclusion of the direct effect to attitudes towards immigrants resulted in a satisfactory model (χ2(3) = 3.18, p = 0.365; CFI = 1.00; RMSEA = 0.02, 90%CI = [0.00,0.12]). The results of these satisfactory models are reported in Table 2 and Table 3.
For Agreeableness, its direct effect to attitudes towards immigrants is significant in both general categorization models, meaning that scoring one point higher from the minimum to the maximum value of Agreeableness results in 1.04 points more positive attitudes towards immigrants in the model where general categorization is a scale and 0.73 points more positive attitudes when general categorization is an open-ended coded variable, confirming H1a. However, in regards to the mediation itself, it was found that all effects of the study variables to the general categorization variables, both as a scale and as an open question, were non-significant. Thus, H2a of the mediating role of general categorization from Agreeableness to attitudes towards immigrants is not confirmed. Also, H3a is not confirmed, as the effects of condition and the interaction term to general categorization are non-significant, meaning that the threat context did not significantly affect general categorization or interact with Agreeableness to have an effect on attitudes towards immigrants.
Figure 2. Structural Equation (Path) Model for the relationships between personality traits, immigrant threat context (i.e. Condition), and attitudes towards immigrants, as mediated by general
categorization. Agreeableness was modeled separately in the same manner as Openness, and general categorization separately as both scale and open-ended question variables to result in four separate interaction models. Age, gender, and immigrant identity were included as control variables.
Figure 1. Structural Equation (Path) Model for the relationships between personality traits, immigrant threat context (i.e. Condition), and attitudes towards immigrants, as mediated by general
categorization. Agreeableness was modeled separately in the same manner as Openness, and general categorization separately as both scale and open-ended question variables to result in four separate models for the main effects. Age, gender, and immigrant identity were included as control variables.
Similarly, the direct effect of Openness to attitudes towards immigrants is significant in both general categorization models, meaning that scoring one point higher from the minimum to the maximum value of Openness results in 0.88 points more positive attitudes towards immigrants in the model where general categorization is a scale and 0.74 points more positive attitudes when general categorization is an open-ended coded variable, corresponding to H1b. In contrast, the effect of Openness to general categorization is significant in both operationalizations of general categorization; as such, the main effects are examined in a model without the interaction term (Table 4). Specifically, for the open-ended question model, one point higher from the minimum to the maximum value of Openness leads to 0.53 points less general categorization, which in turn leads to more positive attitudes towards immigrants as one point higher in general categorization leads to 0.24 points less in positive attitudes towards immigrants. Similarly, for the general categorization scale model, one point higher from the minimum to the maximum value of Openness leads to 0.58 points less in general categorization, which in turn leads to more positive attitudes towards immigrants as one point higher in general categorization leads to 0.47 points less in positive attitudes towards immigrants. Thus, H2b is confirmed of the mediating role of group categorization between Openness and attitudes towards immigrants. However, H3b is not confirmed, as the effects of condition and the interaction term to general categorization are non-significant, meaning threat context did not significantly affect general categorization or interact with Openness to have an effect.
Study 2: A latent growth curve model examining the long-term effects of personality on attitudes towards immigrants
To examine the research purpose of the contextual (longitudinal) effect of Agreeableness and Openness on attitudes towards immigrants, Latent Growth Curve modeling [LGC] was employed to a three-wave LISS panel dataset from December 2013 to January 2017. LGC modeling is a structural equation modeling technique particularly suited to longitudinal data analysis through modeling change in the dependent variable as random effects: hence, estimation is possible of the influence of stable individual characteristics between respondents (i.e. demographics or personality) on the slope and intercept of the dependent variable (i.e. attitudes towards immigrants) over time (i.e. the context) (Preacher, Wichman, MacCallum, & Briggs, 2008). Thus, by estimating the intercept and slope for the entire sample, the LGC technique optimally models not only individual changes over time for attitudes towards immigrants during the period of study, but also the antecedents and consequences of such change (Kim & Kim, 2012). First, all relevant LGC models will be built and the appropriateness of their fit discussed, and afterwards their results will be interpreted.
Data
The present study utilizes longstanding panel data collected by the Longitudinal Internet Studies for the Social Sciences [LISS], administered by CentERdata in Tilburg. The LISS panel
is a representative sample of Dutch individuals 16 years and older from about 5,000 households, randomly selected from the population register by Statistics Netherlands. Although the data is gathered online, selected households were provided with the necessary infrastructure (i.e. Internet and computer access) to participate. Further information about the LISS panel is available at www.lissdata.nl.
For this study, data from the three waves collected in the period December 2013 to January 2017 is analyzed, with details presented in Table 1. As the given research focuses on understanding how personality affects attitudes towards immigrants, only relevant variables were chosen accompanied by potential covariates. As a result, the finalized dataset is a three-wave panel from December 2013 to January 2017 with 11,312 observations.
Attitudes towards immigrants: The dependent variable is measured equivalently to the operationalization in Study 1 with five items, with the identical question formulation between the three waves. A principal component analyses were performed that indicated one component for the five items in each wave with an eigenvalue above 1, which was also above the elbow in the scree plot, suggesting the extraction of one factor (component) in each wave. Relevant factor loadings can be found in Appendix D. The associated Cronbach’s alphas for the five items in wave 1 (α=0.77), wave 2 (α=0.77), and wave 3 (α=0.80) all indicate high reliability. As such, the immigration attitude subscale for each wave contained all five items.
Personality traits: Personality traits were measured by the 50-item International Personality Item Pool - Five Factor Model (IPIP-FFM), such that ten items measure each personality domain (Ehrhart, Roesch, Ehrhart, & Killian, 2008). To manage the many missing values due to the large battery of items, the subscale for each trait was created from the ten items per wave, combined across four waves from December 2013 to January 2018. The relative stability of personality traits in adulthood enables this treatment (Roberts & Mroczek, 2008; Bazana & Stelmack, 2004). Principal component analyses for each wave indicated more than one component for the ten items per trait with an eigenvalue above 1. After one item for Agreeableness, five items for Openness, three items for Conscientiousness, one item for Extraversion, and one item for Neuroticism were dropped, principal component analyses suitably found only one component for each wave with an eigenvalue above 1, which was also above the elbow in the scree plot, leading to the extraction of one factor (component) for the five traits in each of the four waves. All reliability analyses across the four waves for the resulting five-item Openness (α>0.64), seven-item Conscientousness (α>0.73), nine-seven-item Extraversion (α>0.86), nine-seven-item Neuroticism (α>0.88), and nine-item Agreeableness (α>0.78) indicated adequate reliability. Finally, the
subscale for each trait was rescaled from 0 (not expressed) to 1 (expressed) and utilized for further analyses. All reliability results and factor loadings are found in Appendix D.
Control Variables: Analyses are adjusted for the potential confounding influences of age, gender, education, economic ideology, and political orientation. All control variables are only measured in Wave 1 as they are considered time-invariant demographics.
Analysis and Model Specification
To estimate the intercept and slope (linear change) of attitudes towards immigrants from December 2013 to January 2017, a Latent Growth Curve [LGC] model using maximum likelihood estimation was specified. As the LGC model did not yield satisfactory fit (χ2(3) = 34.68, p = 0.000; CFI = 1.00; RMSEA = 0.03, 90%CI = [0.02,0.04]), modification indices were examined to result in the construction of error correlations between the successive three waves. This yielded satisfactory model fit (χ2(1) = 0.04, p = 0.846; CFI = 1.00; RMSEA = 0.00, 90%CI = [0.00,0.01]), suggesting that the growth factors do not completely capture the similarities between subsequent observations.
Next, the LGC model was extended by including age, education, political orientation, economic attitude, and gender as predictors of the slope and intercept of attitudes towards immigrants. Thereby, the intercept and slope factors became endogenous variables, with a necessary error correlation due to the significant covariance between the two factors (p<0.000). All direct effects of the control variables on the slope and intercept factors found to be insignificant (p > 0.05) were removed to create an optimal functioning model, due to parsimony and the avoidance of multi-collinearity: insignificant effects indicate relationships unlikely to exist in the sample, and hence unnecessary to control for significantly (Kline, 2011). This model
not only resulted in satisfactory fit (χ2(9) = 12.19, p = 0.203; CFI = 1.00; RMSEA = 0.01, 90%CI = [0.00,0.01]), but also was confirmed through a comparative chi-square difference test to fit the data significantly better than the initial LGC model without predictors (p=0.000).
Finally, all Big Five personality traits were inserted as additional exogenous variables predicting the slope and intercept of attitudes towards immigrants. This approach corresponds with the aforementioned notion of the relative stability of personality traits (Bazana & Stelmack, 2004; Mondak et al., 2010). While the final model (χ2(14) = 27.37, p = 0.017; CFI = 1.00; RMSEA = 0.01, 90%CI = [0.00,0.01]) has a significant chi-square value, it can be argued that this is mainly a consequence of the large sample size in this study, and may still be considered appropriate based upon the RMSEA and CFI values (Thompson, 1992). Figure 1 depicts this final model.
As this sample contains missing values, FIML is employed to obtain the Maximum Likelihood parameter estimates for the LGC models. After a Levene’s test confirmed the equal variance assumption (p>0.05), an independent mean t-test illustrated that there exist no significant differences in attitudes towards immigrants for participants in only the first wave to those in all three waves (p=0.64). Thus, panel attrition does not appear to be dependent on attitudes towards the EU, and full information maximum likelihood (FIML) estimation is thus more efficient and less biased than other methods of data imputation or deletion of partial data (Preacher et al., 2008). Specifically, when compared to listwise and pairwise deletion and imputation methods, FIML has been evidenced to have greater efficiency, decreased bias, more accurate model rejection rates, and lower occurrence of convergence failure (Enders & Bandalos, 2001). However, results are robust when a balanced panel is artificially created through eliminating individuals with missing observations in any wave (n= 2,141).
Figure 3. Latent Growth Curve model for the relationships between the Big Five personality traits and attitudes towards immigrants
Results Attitudes Towards Immigrants: Intercept and Slope
The initial latent growth curve [LGC] model without predictors estimated Dutch respondents’ average initial attitudes towards immigrants in Wave 1 (December 2013 to January 2014) to be about 2.87, illustrating overall more positive initial attitudes towards immigrants. As the variance of the intercept is significant, substantial variation exists among individuals’ initial levels of attitudes towards immigrants, potentially explainable by the independent variables. Also, the slope of attitudes towards immigrants differed significantly from zero, illustrating that attitudes towards immigrants became 0.04 points more negative on average between December 2013 and January 2017, which can be explained by the study’s context and timing. The significant variance in the slope demonstrates that substantial difference exists among individual growth trajectories for attitudes towards immigrants over the three waves, variation worth investigating with the independent variables. Finally, the correlation between the intercept and slope factors for attitudes towards immigrants (r = 0.08, p<0.000) illustrate that the growth trajectories for attitudes towards immigrants are dependent upon initial attitude levels; as such, the final extended models included a covariance. Including the personality traits and
demographics fairly affected the initial level and change of attitudes towards immigrants, as in the final model the intercept was 3.02 (SE=0.093, p < 0.001) with a significant positive slope of 0.300 (SE=0.073, p < 0.001). Thus, while initial attitudes are still overall more positive, attitudes towards immigrants became 0.30 points more positive on average between December 2013 and January 2017 when including the demographics and personality traits.
The Role of Agreeableness and Openness
As expected, Openness and Agreeableness were found to both have significant effects on the intercept of attitudes towards immigrants, with higher Agreeableness and higher Openness related to more positive initial attitudes towards immigrants. Stated differently, lower Agreeableness and lower Openness is associated with more negative attitudes towards immigrants in Wave 1, December 2013 to January 2014, also confirming H1a and H1b and corresponding to the significant direct effects of both personality traits on attitudes towards immigrants found in Study 1.
Contrary to H3a, the effect Agreeableness on the slope factor of attitudes towards immigrants was not significant, meaning that individual changes in attitudes towards immigrants over time did not depend upon the personality traits. However, as predicted, Openness was found to significantly predict differences in the respondents’ slope. For every additional point increase
from the min value to the max value in Openness, positive attitudes towards immigrants increased by 0.14 points more strongly between waves. Hence, while the average respondent in the final model increased in their positive attitudes towards immigrants 0.30 points from December 2013 to January 2017, this became 46.7% more (i.e. 0.14 points) for every additional point increase from the min value to the max value in Openness. As such, high Openness positively affects increases in positive attitudes towards immigrants over time from December 2013 to January 2017, confirming H3b of the moderating role of threat context for the effect of Openness on attitudes towards immigrants. Finally, it should be noted that results were robust when comparing the final LGC model for personality traits with and without the control variables, except that the effect of Extraversion on the slope was also significant (p=0.002) for the LGC model without control variables.
Discussion
This study addressed the dearth of research surrounding personality and attitudes towards immigrants by positing a conditional, context-dependent argument about their relationship. Specifically, this study tested the causal effects of personality on attitudes towards immigrants both directly and, more importantly, in conjunction with general categorization mediation and contextual factors of immigrant threat. Through both a survey experiment with U.S. participants and latent growth curve modeling with a Dutch representative panel, Openness and Agreeableness were found to have both direct effects on attitudes towards immigrants. As expected, higher Agreeableness and higher Openness led to more positive attitudes towards immigrants, confirming previous findings (Dinesen et al., 2016; Gallego & Pardos-Prado, 2014). However, in Study 1 only the main effect from Openness to attitudes towards immigrants was
found to be partly mediated by general categorization: as predicted, individuals low on Openness related positively to the categorization of others, which subsequently led to more negative attitudes towards immigrants. Similarly, in Study 2 only Openness was found to significantly predict differences in the respondents’ slope, with those low in Openness becoming more negative in their attitudes towards immigrants more rapidly over time from December 2013 to January 2017. However, neither Openness nor Agreeableness interacted with immigrant threat context in Study 1 to affect attitudes towards immigrants. In sum, from the two studies both the direct and contextual nature of Openness on attitudes towards immigrants through situational trigger and mediation by general categorization were found, while only the direct effect of Agreeableness on attitudes towards immigrants was observed. Collectively, the results illustrate the vital importance of scholarly attention to the contextual manner in which personality influences attitudes towards immigrants, and even political behavior as a whole.
The results raise two theoretical enigmas worthy of investigation. First, why is Openness, and not Agreeableness, the personality trait that was mediated by general categorization and influenced by the context of the time period chosen? Second, how can it be understood that immigrant threat context had no effect on either Openness or Agreeableness and attitudes towards immigrants, but context only in terms of the time period around the European Migration Crisis influenced Openness?
To address the former, perhaps the differing dimensions of Openness and Agreeableness that determine generalized prejudice provide one such explanation. Specifically, the Dual-Process Motivational [DPM] approach to prejudice illustrates that personality affects prejudice through mediation by social attitude dimensions Right-Wing Authoritarianism [RWA] and Social Dominance Orientation [SDO]. Low Openness has been repeatedly evidenced to be
associated to the RWA values of social cohesion and collective security, with a strong motivation to protect the existing social order and stability (Duckitt, 2001; Sibley & Duckitt, 2008). In contrast, the SDO values of group-based dominance and superiority correspond more to low Agreeableness, with those low in this trait more likely to view the world as competitive in which only the strong succeed (Duckitt, 2001; Sibley & Duckitt, 2008). Thus, it is possible that the general categorization measures in this study, a scale of generalized prejudice towards stereotyped groups and an open-question dependent on stereotypes of an asylum-seeking family, appeal more to the RWA dimension than SDO; low Openness is the proximal dimension of personality underlying stereotype bias, an aspect of RWA (Flynn, 2005; Jost et al., 2003). Specifically, those low in Openness were more likely to adhere to adhere to conventional negative stereotypes of minorities, reflecting their epistemic motivation to utilize the most readily accessible (i.e. normative) information when assessing others (Flynn, 2005; Jost et al., 2003). Future research should examine measures of prejudice by bandwidth and specificity; effects may differ, for example, with more generalized forms of prejudice, as these are more abstracted and less likely to be influenced by prior experiences with specific social groups (Sibley & Duckitt, 2008; Zick et al., 2008).
In regards to the latter theoretical puzzle, perhaps the extent of threat provoked may provide an explanation. Threat can be distinguished into two separate types: personality and group (collective) motivations (Schwartz, 1992). While personality security serves individual motivations, group security encompasses wider national and societal group interests (Schwartz & Boehnke, 2004). Vecchione et al. (2012) found personal security values to have no significant effects on perceptions of immigrants in Italy, Germany, and Spain, while group security values in particular affected the relationship of Openness on attitudes towards immigrants and not
Agreeableness. Thus, perhaps the immigrant threat context focusing on social order threat through personalization such as fabricated direct quotes from individuals did not stimulate anxiety at a group-level for participants reading online in the United States, which experiencing the situation of the European Migration Crisis certainly brought forth in the Netherlands. Future research should examine further situational triggers with varying degrees of threat to examine how the separate types of threat interact, as well as other aspects of threat such as cultural threat, economic threat, and so forth.
The current study is also not without its limitations. First, the samples utilized in Study 1 and Study 2 both are characterized by their own unique benefits and drawbacks. As crowd-sourced labor, the sample obtained through Amazon’s Mechanical Turk [MTurk] program in Study 1 must be carefully screened for any noticeable signs of incompetence or non-investment (Oppenheimer, Meyvis, & Davidenko, 2009). However, as aforementioned the author exhaustively examined all responses from the MTurk participants for any indicators of incompetence or non-investment and eliminated 66 respondents, as it was easily noticeable in the open-ended question on general categorization and in the levels of warmth battery towards various objectionable ungroups whether or not participants understood the questions and invested adequate time in the survey. Also, MTurk users have often been found to be more representative of the general population and diverse than student and convenience samples, respond consistently to experimental stimuli in prior research, and provide high quality data for powerful insights (Berinsky, Huber, & Lenz, 2012; Casler, Bickel, & Hackett, 2013). Conversely, the panel dataset in Study 2 may also contain biases such as panel attrition, as de Vos (2010) found the LISS panel response rates to be lower among young and old people, those single in relationship status, people living in highly urbanized neighborhoods, and so forth.
However, in order to examine the study’s purpose of the contextual factors behind personality and attitudes towards immigrants or even more specifically the role of time, panel data is uniquely qualified for this research. Future research should explore whether these relationships also hold amongst other samples to more adequately generalize these findings to the general population.
Also, the analysis techniques utilized in both studies also allow for further, conflicting interpretations. Specifically, Structural Equation Modeling [SEM] techniques acknowledge the possibility that alternative models may also suit the data (Eveland Jr., Hayes, Shah, & Kwak, 2005). In the SEM mediation models, interactions are symmetrical and in such case the ability to empirically distinguish between the independent variable (i.e. personality) and the moderator of immigrant threat context not possible. However, from previous literature the hypothesized relations in which context should activate certain personality traits has strong theoretical basis as Mondal (2010) argues, “variation in people’s psychological dispositions leads them to respond differently when exposed to common environmental stimuli, and, correspondingly, […] the expression of personality traits will vary by situation” (p. 90). Future research should examine this relationship further and explore the vital interplay between personality and context in explaining political behavior, such as through economic threat, contextual diversity, and so forth. Similarly, in the conditional LGC model, there may exist other predictors of intercept and slope beyond those hypothesized in the research. However, the LGC model is an ideal method to appropriately analyze longitudinal data, allowing for the study of inter and intra individual variability in attitudes towards immigrants over time as related to the various Big Five personality traits (Preacher et al., 2008). Future research may include additional potential predictors of the slope and intercept of attitudes towards immigrants such as employment status.
Also, it should be noted that the effects of Agreeableness and Openness on attitudes towards immigrants couldn’t be separated into their component parts. Specifically, are the effects of Agreeableness and Openness on attitudes towards immigrants a reflection of in-group favoritism or out-group rejection? Thus, future research should assess more possible mediation and moderation relationships between personality and attitudes towards immigrants to determine the exact mechanisms underscoring political attitude formation. Finally, as a multifaceted concept, personality and its effects on political behavior should be examined in a more nuanced approach that allows for interactive combinations of the Big Five, as well as interactions with other individualized determinants such as those outlined in the Supernumerary Personality Inventory (SPI) framework (Hong, Koh, & Paunonen, 2012).
Despite its limitations, this study is of theoretical significance through illustrating the important role of personality on attitudes towards immigrants, serving as an impetus for necessitated scholarly research on the role of individual predispositions in political outcomes. The results support the notion of the contextual, contingent influence of personality on attitudes towards immigrants, with the effect of Openness mediated by general categorization and moderated by context over time. As such, this study not only advances the study of attitudes towards immigrants, but also more broadly serves as a vehicle for direct policy implications for addressing public opinion on immigration. As personality characteristics are associated to prejudice towards immigrants, policy makers should focus on communicating information on immigration that reduces stereotypes for general categorization and group security threat, reducing citizen concerns of the effects of immigration.
References
Ackermann, K. & Ackermann, M. (2015). The Big Five in Context: Personality, Diversity and Attitudes toward Equal Opportunities for Immigrants. Swiss Political Science
Review, 21(3), 396–418. doi:10.1111/spsr.12170.
Adorno, T., Frenkel-Brunswik, E., Levinson, D., & Sanford, R. (1950). The Authoritarian Personality. New York, NY: Harper & Row.
Allik, J., & McCrae, R. R. (2004). Toward a Geography of Personality Traits: Patterns of Profiles across 36 Cultures. Journal of Cross-Cultural Psychology, 35(4), 13-28. doi:10.1177/0022022 103260382.
Asbrock, F., Sibley, C., & Duckitt, J. (2010). Right-Wing Authoritarianism and Social
Dominance Orientation and the Dimensions of Generalized Prejudice: A Longitudinal Test. European Journal of Personality, 24(7), 583-601. doi:10.1002/per.772
Azrout, R., van Spanje, J., & de Vreese, C. H. (2011). Talking Turkey: Anti-immigrant attitudes and their effect on support for Turkish membership of the EU. European Union Politics, 12(1), 3–19.
Bazana, P. G., & Stelmack, R. M. (2004). Chapter 8 - Stability of personality across the life span: A meta-analysis. In M. Zuckerman (Ed.). On the Psychobiology of Personality (pp. 113–144). New York, NY, US: Cambridge University Press.
Berinsky, A., Huber, G., & Lenz, G. (2012). Evaluating Online Labor Markets for Experimental Research: Amazon.com's Mechanical Turk. Political Analysis, 20(3), 351-368.
Billiet, J., Meuleman, B., & de Witte, H. (2014). The relationship between ethnic threat and economic insecurity in times of economic crisis: Analysis of European Social Survey data. Migration Studies, 2(2), 135–161.
Blumer, H. (1958). Race Prejudice as a Sense of Group Position. Pacific Sociological Review, 1, 3-7.
Branton R., Cassese, E. C., Jones B. S., & Westerland, C. (2011). All along the watchtower: acculturation fear, antiLatino affect, and immigration. Journal of Politics, 73(3), 664-679. Brown, R. J. (2000). Social identity theory: Past achievements, current problems and future
challenges. European Journal of Social Psychology, 30(6), 745–778.
Caprara, G., Barbaranelli, C., & Zimbardo, P. (1999). Personality Profiles and Political Parties. Political Psychology, 20(1), 175–197.
Caprara, G. V., Vecchione, M., Barbaranelli, C., & Fraley, R. C. (2007). When likeness goes with liking: The case of political preference. Political Psychology, 28(5), 609-632. https://doi.org/10.1111/j.1467-9221.2007.00592.x
Caprara, G. & Vecchione, M. (2013). Personality Approaches to Political Behavior. In Huddy, L., D. Sears & J. Levy (Eds.), Oxford Handbook of Political Psychology (pp. 23–58). Oxford: Oxford University Press.
Card, D., Dustmann, C., & Preston, I. (2005) Understanding attitudes to immigration: The migration and minority module of the first European Social Survey. CReAM Discussion Paper Series CDP No 03/05.
Casler, K., Bickel, L., & Hackett, E. (2013). Separate but equal? A comparison of participants and data gathered via Amazon's MTurk, social media, and face-to-face behavioral testing. Computers in Human Behavior, 29, 2156-2160.
De Vreese, C. H., & Boomgaarden, H. G. (2005). Projecting EU referendums: Fear of
immigration and support for European integration. European Union Politics, 6(1), 59–82. de Vos, K. (2010). Representativeness of the LISS-panel 2008, 2009, 2010. CentERdata, Tilburg Digman, J. M. (1990). Personality structure: Emergence of the five factors model. Annual
Review of Psychology, 41, 417–440.
Dinesen, P. T., Klemmensen, R., & Nørgaard, A. S. (2016). Attitudes toward immigration: The role of personal predispositions. Political Psychology, 37(1), 55-72.
Duckitt, J. (2001). A dual-process cognitive-motivational theory of ideology and prejudice. In M. P. Zanna, (Ed.), Advances in experimental social psychology (Vol. 33, pp. 41-113). New York: Academic Press
Duckitt, J., & Sibley, C. G. (2007). Right wing authoritarianism, social dominance orientation and the dimensions of generalized prejudice. European Journal of Personality, 21, 113– 130.
Ehrhart, K.H, Roesch, S. C., Ehrhart, M. G., & Killian, B. (2008). A test of the factor structure equivalence of the 50-item IPIP ve-factor model measure across gender and ethnic groups. Journal of Personality Assessment, 90(5), 507–516.
Ekehammar, B., Akrami, N., Gylje, M., & Zakrisson, I. (2004). What matters most to prejudice: BigFive personality, Social Dominance Orientation, or Right-Wing Authoritarianism? European Journal of Social Psychology, 18, 463–482.
Enders, C. K., & Bandalos, D. L. (2001). The relative performance of full information maxi-mum likelihood estimation for missing data in structural equation models. Structural Equation Modeling, 8, 430–457.
