Changing “us” and hostility towards “them”—Implicit theories of national identity determine prejudice and participation rates in an anti‐immigrant petition

A rise in anti-immigrant hostility can currently be observed in Europe. The anti-immigrant movement of the “Patriotic Europeans against the Islamization of the West” (short PEGIDA) has attracted thousands of participants to its demonstrations. Founded 2014 in Germany, PEGIDA is now spreading across Europe (Wolf & Alexe, 2016). The rising anti-immigrant sentiment has also found its way off the streets and into politics. Populist parties, like the above quoted “Sverigedemokraterna” and the “Alternative für Deutschland”, are gaining more and more popularity and seats in parliaments on both the national and the European level. The above quotations demonstrate one commonality that can be observed in anti-immigrant movements across Europe: a rhetoric about national identity (e.g., “us Swedish/German people”) which excludes immigrants (e.g., Non-Whites and Muslims not being considered true Swedes/Germans) on the basis of fixed and descent-focused characteristics (e.g., longstanding ethnic or religious roots). Received: 17 June 2019 | Accepted: 24 January 2020 DOI: 10.1002/ejsp.2666


| 811
CHANGING "US" AND HOSTILITY TOWARDS "THEM" Research on intergroup processes indicates that this rhetoric may not just be a mere byproduct of anti-immigrant hostility. Instead, this rhetoric, which portrays immigrants as distinct outgroups, may itself actively contribute to increased hostility against immigrants.
In this article, we introduce the concept of implicit theories regarding the fixedness versus malleability of the ingroup's national identity definition (in short: implicit theories of national identity). We hypothesize that the hostility-inducing perceived divide between people who are considered "true compatriots" and those who are not can be exacerbated or blurred by portraying national identity definitions (i.e., the definition of what makes someone a "true compatriot") as fixed or malleable. Viewing the ingroup's national identity definition as fixed (i.e., thinking that a "true German" was always and will always be defined by ethnicity, for example) should exacerbate the perception of a rigid and clear-cut divide. At any given time, it is very clear who belongs to the national in-or outgroup and who will do so in the future. This intensified perception of a national ingroup-outgroup is expected to increase hostility toward immigrants.
In contrast, a malleable perception of national identity definitions implies that what it means to be a "true compatriot" can change over time. For example, a person who acknowledges the malleability of national identity definitions could think that "true Germans", who Our hypotheses are based on two lines of research: (a) Social Identity Theory (SIT; Tajfel & Turner, 1979), and (b) the concept of malleable versus fixed implicit theories (Dweck, 2008;Dweck & Leggett, 1988).

| Social identity theory and its explanation of outgroup-hostility
According to SIT (Brown, 2000;Tajfel & Turner, 1979), the mere activation of group memberships can increase hostility towards outgroupsa process called outgroup bias (for a review, see Hewstone, Rubin, & Willis, 2002). The theory rests on the assumption that (a) people generally strive for a positive self-image and (b) the ingroup (e.g., "us Germans") is seen as a part of the self. By devaluing outgroups (e.g., immigrants), individuals' ingroup and by extension their overall selfimage can be improved on a relative basis. The resulting outgroup bias could be shown using as diverse range of group categorizations and even random assignment to fictitious groups (Hewstone et al., 2002;Mullen, Brown, & Smith, 1992). Among the many categories humans can use to sort others, national identity-together with other demographics like race and sex-constitutes one of the most commonly used organizing schemes. This may lead to especially pervasive effects of national identity-based outgroup bias.
Previous research has indeed used SIT to explain covariations between how individuals define their national ingroup and their anti-immigrant hostility. Work in this area has focused on the contents and, by implication, the width of national ingroup definitions.
It was shown that people who exclude immigrants with a narrow definition of the national "us" (e.g., Germans defined by contents like ethnicity or inherited culture) exhibit more hostility toward immigrants than individuals who define national identity more inclusively (e.g., Germans defined by contents such as shared values). This relation has been shown in both cross-sectional as well as longitudinal studies including diverse populations from over 30 countries and all four populated continents (e.g., Billiet, Maddens, & Beerten, 2003;Ha & Jang, 2015;Jones, 1997Jones, , 2000Maddens, Billiet, & Beerten, 2000;Meeus, Duriez, Vanbeselaere, & Boen, 2010;Pehrson, Brown, & Zagefka, 2009;Pehrson, Vignoles, & Brown, 2009). Wakefield et al. (2011) have tested the link between the inclusiveness of the criteria that define national identity (narrow ethnic vs. broader formality-based criteria) and intergroup relations in an experimental setting. This study focused on a specific form of intergroup relationships, with positive attitudes and supportive behaviors towards Asian Scottish individuals as outcome variables.
They found the induction of an inclusive national identity definition to improve Scottish participants' supportive behaviors, but not participants' liking of an Asian Scottish confederate. It is unclear, however, whether effects found in this study regarding positive and supportive attitudes toward Asian minorities, which are often considered a particularly successful minority group, can be applied to attitudes toward other immigrant groups (i.e., hostile attitudes regarding the larger group of immigrants in general, or less favorably viewed immigrant subgroups). Also, the exact mechanisms of the link between narrow national identity definitions and anti-immigrant hostility are still under investigation. While Pehrson and Green (2010) found evidence for a moderation model in which more exclusive definitions of national identity strengthened the negative effect of a strong sense of national attachment on anti-immigrant hostility, findings by Meeus et al. (2010) supported a mediation model in which more narrow national identity definitions mediated the relationship between attachment and anti-immigrant hostility.
With our research, we aim to explore a new dimension of national identity definitions. While previous research has investigated which kinds of contents are used to define national identity, our research on perceived malleability describes whether people think that these contents (be it, e.g., narrow ethnic, or broader value-based definitions) change over time. Thus, our research takes on a meta-theoretical perspective: Instead of investigating individuals' content theories ("what makes a true compatriot?"), we investigate the perceived malleability of individuals' content theories ("Is the definition of 'what makes a true compatriot' malleable?").

| Implicit theories about malleability versus fixedness
People differ in their implicit theories about the malleability versus fixedness of diverse matters like attributes of the world, conflict, emotions, and abilities (Cohen-Chen, Halperin, Crisp, & Gross, 2014;Tamir, John, Srivastava, & Gross, 2007;Yang & Hong, 2010)-with multiple implications for motivation, wellbeing, and interpersonal relationships (e.g., Dweck, Chiu, & Hong, 1995). A major part of the research on implicit theories has focused on the perceived malleability of human attributes.
The first studies on implicit theories regarding human attributes specifically focused on attributes of the self like a person's own intelligence or skills (Dweck, 2008;Dweck & Leggett, 1988).
It was shown that viewing those self-attributes as malleable rather than fixed resulted in more adaptive judgements and behaviors: Failures, for example, were more likely to be seen as an opportunity to learn instead of seeing them as evidence for lack of intellectual potential (Dweck et al., 1995). As a result, individuals who believe that attributes like intelligence are malleable show improved academic results (e.g., Blackwell, Trzesniewski, & Dweck, 2007).
More recently, research on the malleability of human attributes has turned to the perceived malleability of attributes of other people and groups and the resulting judgements of those others. By aiming to investigate the usefulness of the malleability framework for the improvement of interpersonal and intergroup relationships, it could, for instance, be shown that the perceived malleability of people was linked to more adaptive attribution patterns: Individuals who were led to believe that people can change regarding their personality and moral character were less likely to attribute other people's negative behaviors to unchangeable, fixed traits, and more likely to see wrongdoing as a result of other factors like people's temporary thoughts and feelings or situational factors (Chiu, Hong, & Dweck, 1997;Levy, Plaks, Hong, Chiu, & Dweck, 2001). As a result, a malleable rather than a fixed view of other people also improved individuals' attitudes in cases of conflict. For instance, it was shown that both Israeli and Palestinian participants were more ready to compromise for peace when they were led to view characteristics of groups as inherently malleable rather than fixed (Halperin, Russell, Trzesniewski, Gross, & Dweck, 2011). While a fixed view of groups who may be considered aggressive and evil implies that these groups keep their (detested) attributes, a malleable view implies that groups could become better in the future. Together, these findings suggest that a malleable rather than a fixed view of other people results in more adaptive attribution patterns, attitudes, and behaviors in situations of conflict.
In our research, we use the concept of implicit theories to show that a fixed view of the ingroup's national identity definition increases outgroup hostility, whereas a malleable view decreases these negative attitudes as it implies that the boundaries separating in-and outgroup can change. While previous research on implicit theories of human attributes has highlighted the benefits of a malleable view regarding individuals, we investigate the effects of implicit theories of a social category, namely of the national ingroup category "true compatriots". Note that if the definition of the national ingroup category changes over time (e.g., from an ethnic to a value focus), this does not necessarily imply that the individuals themselves change regarding their personalities and behaviors; rather, a malleable national identity implies a change in the defining "entry criteria" that determine who may cross the line and be considered part of the national ingroup category. Instead of focusing on the changeability of individuals' personal characteristics, we thus focus on the malleability of the categorizing, group-dividing boundaries. We specifically hypothesized that, while a fixed view on these categorizing boundaries exacerbates the perceived divide between "us" and "them" and thus increases anti-immigrant hostility, a malleable view blurs this divide and thus reduces anti-immigrant hostility.

| Conceptually related constructs
To investigate our new constructs' convergent and discriminant validity, we included measures of other related constructs in our studies. In particular, we expected implicit theories of national identity to be moderately correlated with measures of perceived group malleability (as discussed above), essentialism, and historical tolerance. As our outcome variables measuring anti-immigrant hostility are expected to be strongly correlated with right-wing political orientation (Prezza, Zampatti, Pacilli, & Paoliello, 2008) and a strong sense of national attachment (e.g., Ditlmann & Kopf-Beck, 2019;Leyens et al., 2003), we additionally included measures of both variables as controls. 1

| Essentialism
According to Bastian and Haslam (2008), essentialism describes the belief in a biological essence within humans that determines their behavior. More specifically, essentialism consists of three components: biological basis, discreteness, and informativeness. Individuals with an essentialist belief system tend to think that (a) people's behaviors are determined by their biological make-up (biological basis), (b) there are clear-cut boundaries between different "types" of people (discreteness), and (c) learning about individuals' traits enables accurate predictions about their future behavior (informativeness). Like narrow national identity definitions (based on, for instance, shared ethnicity), essentialism has been associated with worse intergroup-relations (e.g., Bastian & Haslam, 2008;Roets & Van Hiel, 2011). 1 As can be seen in Table S3 in the Appendix S1, demographic variables (age, education, gender) would not have added consistent predictive value to our analyses. This supports our reasoning that our control variables are more closely related to the dependent variables and the predictive power of demographics is already summarized in our controls.
It seems plausible that-like narrow national identity definitions-essentialism may also heighten the degree to which individuals perceive ingroup-outgroup divides, which may in turn spark outgroup hostility. In contrast to essentialism, our implicit theory concept focuses on the perceived malleability or fixedness of national identity. While perceived fixedness of national identity may be empirically related to essentialist beliefs, our concept does not imply a belief in a biological essence to national identity, in the discreteness of national group boundaries, or in the predictability of individuals' behavior based on their national identity.

| Historical tolerance
Some research outside the SIT-framework has investigated whether hostility can also be decreased by making pro-immigrant values like tolerance a defining feature of ingroup membership (irrespective of the width of national ingroup definitions). Specifically, Smeekes, Verkuyten, and Poppe (2012) showed that individuals who were led to see religious tolerance as an important part of their national identity exhibited less hostile (and more tolerant) attitudes toward Muslims. These findings can best be explained by assimilation processes, as described by self-categorization theory (SCT, Turner, 2010), rather than by SIT-based ingroup-outgroup processes: the description of one's own national group as tolerant leads individuals to self-stereotype and assimilate to be more tolerant, too. Despite these differences in targeted processes, implicit theories of national identity and historical tolerance may both affect similar outcomes like anti-immigrant hostility.

| National attachment and political orientation
As mentioned, we included measures of political orientation and national attachment as controls. Two major forms of individuals' sense of national attachment are national identification and nationalism. National identification constitutes a rather general operationalization of national attachment (which we used as a control in our Prestudy). In contrast, nationalism (which we used in all other studies) is a form of national attachment that specifically "reflects apperception of national superiority" (Kosterman & Feshbach, 1989, p. 271) and is even more predictive of anti-immigrant hostility than general national attachment measures (Blank & Schmidt, 2003;Kosterman & Feshbach, 1989;Wagner, Becker, Christ, Pettigrew, & Schmidt, 2010).

| The present research
As outlined, we introduce the concept of implicit theory of national identity, aiming to extend research both on SIT-based outgroup hostility and on malleable versus fixed implicit theories.
We chose Germany and England to conduct our studies for two reasons: Firstly, both Germany and England have witnessed a recent rise of right-wing populism. Our research topic is thus of great societal importance in both countries. Secondly, as in other nations with a long and ethnically rather homogeneous history, national identity is primarily defined over narrow culture-based contents in both Germany and England (Ditlmann, Purdie-Vaughns, & Eibach, 2011;Stokes, 2017). Germany and England thus constitute good exemplars with which to investigate effects of national identity perceptions in Europe.
To test our hypotheses, we conducted five studies with independent samples. In our Prestudy, we created a scale to measure individuals' views about the malleability versus fixedness of their own national identity. In Studies 1 and 2, we tested if our scale predicted self-reported prejudice and behavioral measures of anti-immigrant hostility (participation rates in a hypothetical anti-immigrant referendum in Study 1 and a real anti-immigrant petition in Study 2) in Germany and England-taking related variables into account (essentialism, historical tolerance, perceived group malleability, sense of national attachment, and political orientation). In Studies 3 and 4, we obtained experimental evidence for the proposed relationships. In Study 3, we tested the effect of an experimental manipulation of a malleable versus a fixed view of national identity. Study 4 served as a replication and additionally included a control condition.
To capture anti-immigrant hostility comprehensively, we chose to assess (a) prejudice as a mental and affective form of hostility as well as

| PR E S TU DY
The aim of the Prestudy is to create and validate a self-report measure of individuals' implicit theories of national identity. To test the 2 According to our institution's guidelines and national regulations, no formal ethics approval was required, since our surveys were anonymous and no risks were anticipated.
All surveys were conducted with the informed consent of each participant. scale's convergent validity via correlational analyses, we included political orientation and a measure of national attachment as two related variables. We expected a fixed view of national identity to be connected to a more right-leaning political self-description and stronger national attachment.

| Participants and procedure
Participants were recruited through German-speaking groups on the social media platform Facebook. Facebook users were told that researchers were looking for German participants to investigate opinions on German national identity. No incentives were given. Fifty-eight participants (30 male, 23 female, 2 other, 3 nonresponders) completed our online questionnaire. The mean age was M = 25.3, SD = 7.3. Six participants (10%) reported having a migration background (i.e., they agreed to the question "Did your family immigrate to Germany after 1949?"-a question commonly used in Germany as a replacement for questions about race and ethnicity; e.g., German Federal Statistical Office, 2005). Sixty-seven percent of participants were students while 24% had a job. Fifty-two percent of participants held a university degree.

| Implicit theory of national identity
Using previous measures of perceived malleability (Blackwell et al., 2007;, we generated a pool of 14 items describing malleability beliefs of national identity (e.g., "True Germans are characterized by certain cultural attributes that will always remain as they are", "Different generations may do things differently, but there are certain cultural characteristics that true Germans will always share"). Participants rated all items on a scale from 1 ("totally disagree") to 6 ("totally agree").

| Political orientation
Political orientation was assessed with the following item adapted from Green (2009): "How would you describe your overall political orientation?" with responses ranging from 1 to 6 ("extremely left", "left", "slightly left", "slightly right", "right", and "extremely right").

| Results and discussion
In selecting our final items from the item pool, we considered results from factor analyses, item-total correlations, and internal reliability. A detailed description of the item selection can be found in the Supplemental Material. The resulting Implicit Theory of National Identity Scale consists of 6 items (α = 0.95). Items and factor loadings are shown in Table 1. In line with previous implicit theory scales (e.g., implicit theories of intelligence, Hong, Chiu, Dweck, Lin, & Wan, 1999; implicit theories of personality, Chiu et al., 1997; implicit theories of groups, Halperin et al., 2011), our scale only includes items that endorse a fixed view of individuals' ingroup national identity. As has been previously discussed (e.g., Chiu et al., 1997;Levy, Stroessner, & Dweck, 1998), malleability items are rather universally endorsed and thus less helpful in distinguishing individuals with malleable and fixed implicit theories. Furthermore, evidence was obtained indicating that disagreement with the fixedness items can be interpreted as endorsement of malleability views (Levy & Dweck, 1996).
Items were reversed, so that higher scores represent a greater perceived malleability of national identity. A factor analysis revealed one factor that explained 77% of the variance. For each participant, we calculated a mean score across all items. Descriptive statistics (M = 2.37, SD = 1.27, range = 1.00-6.00) indicated that participants overall tended to lean towards a fixed view of national identity, while there was considerable variance between individuals. As expected, individuals' implicit theory of national identity was significantly correlated with both political orientation (r = −.52, p < .001) and national attachment (r = −.53, p < .001), indicating good convergent validity. Descriptive statistics of both variables are summarized in Table 2.
TA B L E 1 Items and factor loadings of the "Perceived Implicit Theory of National Identity Scale" Item Factor loading 1 True Germans are characterized by certain cultural attributes that will always remain as they are.
0.89 2 Different generations may do things differently, but there are certain cultural characteristics that true Germans will always share.

0.91
3 True Germans share certain cultural characteristics that will always stay the same.

4
There is something about the culture and heritage true Germans share that you just can't really change.
0.73 5 There are certain cultural characteristics that define a true German and that will always remain the same. 0.92 6 When I think of true Germans, I think of people with cultural characteristics that will remain the same even over generations.

| S TUDY 1
Study 1 aimed at testing the relationship of our Implicit Theory of National Identity Scale with anti-immigrant hostility. Besides a measure of prejudice, we also used a behavior-related measure of antiimmigrant hostility: whether or not individuals would participate in a hypothetical anti-immigrant referendum. We hypothesized that our scale would predict both measures of anti-immigrant hostility.

| Participants and procedure
Recruiting procedures were equivalent to the ones used in our Prestudy.
From the results of previous studies in related fields (e.g., Halperin et al., 2011), we expected a small to medium effect size of partial R 2 = .05 and OR = 0.3 for our analyses. With a target of 80% statistical power, we aimed to recruit 152 participants, based on power analyses using G*Power (Faul, Erdfelder, Buchner, & Lang, 2009). In the end, 154 participants (67 male, 83 female, 1 other, 3 non-responders) completed the online questionnaire. The mean age was M = 31.29, SD = 11.38.
Fifty-five percent of the participants had a job while 32% were students.
Forty-nine percent of our sample had obtained a university degree.

| Implicit theory of national identity and political orientation
These variables were measured respectively by the same 6 items (α = 0.96) and the same single item as in our Prestudy.

| National attachment
We used the following two items adapted from Wagner et al. (2010), who have demonstrated the scale's validity in a German population: "I am proud about Germany's history", and "I am proud to be German" (1 = "strongly disagree", 6 = "strongly agree"; α = 0.71).

| Participation in a hypothetical antiimmigrant referendum
We asked individuals about their willingness to participate in an antiimmigrant referendum that argued for "the tightening of immigration restrictions which is expected to cause a decrease in the number of migrants". Participants could choose to vote for or against the proposal ("I would vote for [/against] the proposal") or to abstain from the vote ("I would not vote"). Since our SIT-based hypothesis only allows for predictions regarding negative attitudes toward immigrants, 3 responses were dummy-coded, so that supporting the anti-immigrant referendum was counted as 1 (indicating anti-immigrant hostility), and not supporting the proposal (by supporting the pro-immigrant version or abstaining from participation) was counted as 0.

| Results and discussion
To test if individuals' implicit theory of national identity predicts levels of prejudice and individuals' hypothetical participation in the anti-immigrant petition, we conducted a linear and a logistic regression, respectively, controlling for national attachment and political orientation. As is evident from the regression information in Table 3, individuals' implicit theory of national identity significantly predicted levels of prejudice, β = −0.47, p < .001. The more participants thought that national identity was malleable rather than fixed, the less prejudice they reported. 3 Our research on implicit theories of national identity focusses on the manipulation of negative outgroup bias by exacerbating versus blurring the lines between in-and outgroup. As shown by   Similar relationships were obtained for the behavior-related measure of prejudice. Overall, 39.6% of participants indicated they would vote for tighter immigration restrictions in a referendum (while 46.1% wanted to vote against it and 14.3% would not vote). As can be seen in Table 4, a binary logistic regression revealed that individuals' implicit theory of national identity significantly predicted hypothetical referendum participation, with an odds ratio coefficient for perceived malleability of OR = 0.45, p = <.001. That is, the more participants thought that national identity was fixed, the more likely they would be to vote for tighter immigration restrictions, and the more they thought national identity was malleable, the less likely they would be to vote for a tightening of immigration restrictions. Above and beyond national attachment and political orientation, individuals' implicit theory of national identity thus seems to be a relevant predictor for prejudiced attitudes as well as behavioral intentions hostile toward immigrants. A mediation analysis following Preacher and Hayes (2004)

| S TUDY 2
This study aimed to (a) test the generalizability of our findings in another cultural context (England) and (b)

| Participants and procedure
With a conservative estimate of our effect size based on Study

| Measures
If not otherwise noted, measures were rated on a scale from 1 ("strongly disagree") to 6 ("strongly agree"). Political orientation, implicit theory of national identity (α = 0.96) and prejudice (α = 0.85) were measured exactly as in Study 1.

| New control variables
To measure national attachment in the British context, we adapted a nationalism scale from Kosterman and Feshbach (1989) containing 8 items (e.g., "Generally, the more influence England has on other nations, the better off they are"). Historical tolerance was assessed with two items (e.g., "England has a long history of religious tolerance") adapted from Smeekes et al. (2012). Perceived group malleability was assessed with a 4-item scale ("Groups can do things differently, but the important parts of who they are can't really be changed") developed by Halperin et al. (2011). Essentialism was assessed with a 15-item scale of Bastian and Haslam (2008) containing the three factors discreteness (e.g., "The boundaries that define the differences between people are clear-cut"), informativeness ("Generally speaking, once you know someone in one or two contexts it is possible to predict how they will behave in most other contexts"), and biological basis (e.g., "The kind of person someone is can be largely attributed to their genetic inheritance").

| Participation in the anti-immigrant petition
Participants had the option to vote in favor of a real anti-immigrant petition arguing for the deportation of recent refugee migrants in England.
To prevent individuals from viewing our survey as politically biased, we also offered the option to vote for a pro-immigrant petition, arguing that refugees should not be deported. As a third option, participants could choose to support none of the petitions. Participants who chose a petition were directed to the webpage of the respective petition where they could finalize their vote by entering their name and email address. Responses were dummy-coded. Support of the deportation petition was counted as 1 and not participating in this petition (by supporting the no-deportation petition or no petition at all) as 0.

| Correlations and factor analyses
To test the hypothesis that individuals' implicit theory of national identity is distinct from other related constructs, we conducted correlational analyses and confirmatory factor analyses. As evident in Table 5, correlations between our scale and other tested constructs ranged between |r| = .28 and |r| = .60. This can be seen as a first indication in support of our assumption that our scale is related to,

| Regression analyses predicting antiimmigrant hostility
To test if individuals' implicit theory of national identity predicts levels of prejudice and individuals' petition participation above and beyond the additional variables we accounted for (nationalism, political orientation, perceived group malleability, historical tolerance, essentialism),

| S TUDY 3
Building on the findings of Study 1 and 2, we conducted an experiment to test the causal relationship between individuals' implicit theory of national identity and anti-immigrant hostility in Study 3.

| Participants
From the effects of our correlational studies as well as previous experimental studies in the field (e.g., Halperin et al., 2011), we expected an effect size of partial R 2 = .05 and OR = 0.40. With the goal of reaching 80% statistical power, we recruited 225 participants, as planned. Participants were recruited online through Germanspeaking groups on Facebook and a student mailing list in Berlin.
Facebook users and subscribers to the mailing list were told that researchers were looking for German participants to investigate Germans' opinions about current political topics. As an incentive, participants could win one out of 5 Amazon vouchers worth €5.
In addition, students could receive course credit. At the end of the survey, participants were directed to a separate page to enter their email address for the lottery and a personal identification code for

| Procedure and experimental manipulation
After giving informed consent, participants were randomly assigned to one of two conditions in a between-subjects design: a malleable and a fixed condition (a control condition was to be included later, After reading the article, participants were asked to take an active part in the manipulation by finding explanations for the findings presented in the research article. This so-called saying-is-believing task (Higgins & Rholes, 1978) is a commonly used technique that leads participants to repeat the manipulation in their own words and link it to their own experiences whereby the manipulation message can be processed more deeply. Following this task, participants filled in our questionnaire including our dependent variables, demographic information, and our Implicit Theory of National Identity Scale.

| Participation in the anti-immigrant petition
Participants had the option to vote for a German anti-immigrant petition similar to the one used in Study 2. The petition specifically argued for the deportation of Syrian refugees. Alternatively, participants could vote for a pro-immigrant petition calling on the federal government to suspend deportations or abstain from both petitions. Participants who chose a petition were directed to the webpage of the respective petition where they could finalize their vote.
Responses were again dummy-coded.

| Implicit theory of national identity
To test if our manipulation successfully changed individuals' implicit theory of national identity, we conducted an ANCOVA on our Implicit Theory of National Identity Scale, controlling for national attachment and political orientation. We found a significant difference between malleable and fixed condition, F(1, 221) = 5.70, p = .018, η 2 = 0.03, showing that the malleable condition has led participants to view national identity as more malleable (M adj = 4.08) than the fixed condition (M adj = 3.74).

| Prejudice
To test if our manipulation was successful in changing participants' levels of prejudice, we conducted an ANCOVA, controlling for national attachment and political orientation. We found a significant difference between malleable and fixed condition, F(1, 221) = 5.66, p = .018, η 2 = 0.03. As expected, participants in the malleable condition reported lower levels of prejudice (M adj = −0.12) than participants in the fixed condition (M adj = 0.12). Results confirm that leading participants to view national identity as malleable resulted in significantly lower levels of prejudice compared to the fixed condition.

| Participation in an anti-immigrant petition
In line with the sample's left-bias in political orientation with only 30% describing themselves as politically right (see Table 2), the overall endorsement rate for the anti-immigrant petition was rather low.
Only 10% of participants voted for the deportation of refugees (28% voted for the pro-refugee petition and 62% participated in neither petition). While 7% of participants advocated deportation in the malleable condition, 13% did so in the fixed condition. This difference amounts to an 86% increase of endorsement rates in the predicted direction. To test the condition effect, a binary logistic regression was conducted, controlling for national attachment and political orientation. Results are depicted in Table 8. Contrary to our expectations, the effect of the condition was not significant, OR = 0.55, p = .27. Despite the 86% difference between conditions, the effect did not reach a significant level.

| Mediation analyses
Mediation analyses using the SPSS macro following Preacher and Hayes (2004)  Leading participants to view national identity as malleable rather than fixed seemed to result in lower levels of prejudice and thus prevented individuals from taking part in an anti-immigrant petition.

| Discussion
In Study 3, we implemented an experimental manipulation that successfully changed participants' view of the malleability versus fixedness of national identity. As a result, significant differences in levels of prejudice were found. Participants who were led to believe that national identity is fixed showed more prejudice than participants who were led to believe that national identity was malleable.
A similar but not consistently significant pattern emerged for the behavioral measure of anti-immigrant hostility. Only about half as many individuals voted in favor of the anti-immigrant petition in the malleable than the fixed condition-thus replicating the pattern of behavioral measures in Studies 1 and 2)and experimental condition had a significant indirect effect on petition participation through prejudice.
In contrast, logistic regression analyses did not reach statistical significance. Compared to the overall endorsement of the referendum for tighter immigration laws in Study 1 (39.6%), pro-votes for the petition on the deportation of refugees in Study 2 (9.7%) and Study 3 (10%) were very low. This indicates that we were successful in capturing a rather extreme form of anti-immigrant hostility with the petitions of Studies 2 and 3. However, on a statistical level, low baseline rates are known to lead to overly conservative significance tests and thus limit the sensitivity of logistic regressions (Peduzzi, Concato, Feinstein, & Holford, 1995). It is thus likely that the low baseline participation rates stemming from a politically left-leaning sample contributed to the non-significant and overall inconclusive results.
In sum, Study 3 provided first experimental evidence for a causal relationship between individuals' implicit theory of national identity and self-reported anti-immigrant hostility. The inconsistent results regarding the behavioral measure of anti-immigrant hostility were further investigated in Study 4.

| S TUDY 4
The goal of Study 4 was threefold: We aimed to (a) replicate the findings of Study 3, (b) add a control group, and (c) clarify the rather inconclusive results of the manipulation effect on petition participation.
For the latter goal, we aimed to recruit a politically more right-leaning sample with a higher baseline participation rate in the anti-immigrant petition that is more adequate for logistic regression analysis.

| Participants
Assuming that by recruiting a less politically left-leaning sample, our effect size would be slightly higher than in Study 3, we expected an effect size of partial R 2 = .04 and OR = 3.5. With the goal of reaching 80% statistical power, we aimed to recruit 235 participants.
Participants were again recruited through German-speaking groups on Facebook and, in addition, via mailing lists of a right-leaning party (Christian Democratic Union, CDU) to increase the share of rightwing participants. Facebook users and subscribers to the mailing lists were told that researchers were looking for German participants to investigate their opinions about current political topics. Participants could win one out of 8 Amazon vouchers worth 5 Euros. At the end of the survey, participants were directed to a separate page to enter their email-address for the lottery, so that participants' survey data could not be linked to their email-address. Two hundred and twentyfive participants (121 male, 99 female, 3 other, 2 non-responders) completed our online questionnaire. Nineteen individuals (8%) reported having a migration background. Forty-one percent of participants had a job while 45% were students. Twenty-nine percent of participants had obtained a university degree.

| Procedure and experimental manipulation
Procedure and experimental manipulation were the same as in Study 3, with the exception that a control group was added. For the control condition, another ostensible research article was created. Instead of the malleability or fixedness of national identity definitions, the control article discussed German farming methods for strawberries. Overall, the article was written in a similar manner to the two articles about national identity. For example, all three articles presented results of surveys that were repeatedly conducted since 1951 by a German research team.

| Measures
National attachment (

| Results
As shown in Table 2, we were able to increase the percentage of people who self-described as politically right-leaning ("rather right", "right", or "extremely right") to 39% (compared to 30% in Study 4).
This increase in right-leaning participants is also evident in a higher overall endorsement rate in the anti-immigrant petition. Compared to 9.7 and 10% in Studies 2 and 3, 15% of the participants voted for the deportation of refugees (while 27% voted for the pro-refugee petition, and 58% voted for neither petition).

| Implicit theory of national identity
We conducted an ANCOVA on the Implicit Theory of National Identity Scale, controlling for national attachment and political orientation. We found a significant overall effect of condition, F(2, 220) = 7.80, p = .001, η 2 = 0.07. To examine the differences between the respective groups, we conducted pairwise post-hoc tests using Bonferroni-corrected alpha-values and computed 95% confidence intervals for the mean differences (M diff ). Post-hoc tests revealed a significant difference between fixed and malleable condition,

| Prejudice
We conducted an ANCOVA on prejudice, again controlling for national attachment and political orientation. Analyses revealed a significant overall effect of experimental condition, F(2, 220) = 3.13, p = .046, η 2 = 0.03. Pairwise post-hoc tests using

| Participation in anti-immigrant petition
We conducted a binary logistic regression, again controlling for national attachment and political orientation. As can be seen in Table 9, results showed that the fixed condition significantly increased deportation-endorsement rates compared to the malleable condition, unstandardized b = 1.16, OR = 3.12, p = .046, and to the control condition, unstandardized b = 1.64, OR = 5.16, p = .011.
There was no significant difference between the malleable and the control condition, unstandardized b = 0.48, OR = 1.61, p = .468. The odds ratios showed that chances to endorse deportation were 5.16 and 3.12 times higher for participants who were led to believe that national identity was fixed than for participants in the control and malleable condition, respectively. While in the control and malleable condition, respectively 11% and 13% of participants voted for the deportation referendum, endorsement rates in the fixed condition rose to 21%. Compared to the control and malleable condition, respectively 85% and 66% more individuals were in favor of the deportation of refugees when they were led to view national identity as fixed.

| Mediation analyses
Mediation analyses using the SPSS macro following Preacher and Hayes (2004) were conducted to test whether our manipulation effects were explained by prejudice.

| Discussion
The experimental manipulation with three conditions led to a significant change in both self-reported prejudice and endorsement rates in an anti-immigrant petition. Compared to the malleable condition, fixed views of national identity increased levels of prejudice and the likelihood that participants voted for the deportation of refugees.
While the political left was clearly overrepresented in the sample of Study 2 and 3, the share of participants with more right-wing views was larger in Study 4. This increase in right-leaning participants corresponded to a 50% increase of the overall participation rate in the anti-immigrant petition which enabled more sensitive logistic regression analyses (Peduzzi et al., 1995) that confirmed our hypotheses. Leading participants to view the ingroup's national identity as fixed rather than malleable approximately led to a 66% increase in endorsement rates in the anti-immigrant petition. This effect is in line with the significant predictive effect of individuals' implicit theory of national identity on referendum votes and petition participation we found in Study 1 and 2. The fact that we found consistently significant effects in both studies with high petition/ referendum participation rates (Studies 1 and 4), but mixed effects in the studies with low participation rates (Studies 2 and 3), is in line with research showing that low participation rates decrease the sensitivity of logistic regressions (Peduzzi et al., 1995). It seems likely that low baseline participation rates contributed to the nonsignificant result of Study 3.
As expected, all results of the control condition lie roughly between the levels of both national identity conditions. This is important, because it rules out the possibility that both national identity manipulations may inadvertently have increased anti-immigrant hostility simply by activating a sense of shared national identity which unites all "true Germans" and distinguishes them from other outgroups. The control condition clearly shows that this was not the case and only the fixed national identity condition (and not the malleable condition) increased anti-immigrant hostility. The fact that the control condition differed significantly from the fixed, but not from the malleable, condition regarding the implicit theory scale and main outcomes indicates that our fixed condition was more successful in altering individuals' implicit theory than the malleable condition.

| G ENER AL D ISCUSS I ON
National identity definitions draw a dividing line between national in-(e.g., "us Germans") and "foreign" outgroups, with the latter being prone to hostile outgroup bias. In five studies with a total sample of over 1,000 participants, we found evidence that viewing the ingroup's national identity as fixed exacerbates the perceived divide between in-and outgroup and thus increases anti-immigrant hostility, while a malleable view may blur the divide and reduce antiimmigrant hostility.
After developing the Implicit Theory of National Identity Scale in a Prestudy, Study 1 and 2 showed that, both in Germany and England, individuals' implicit theory of national identity was a reliable predictor of prejudiced attitudes and behavior (i.e., participation rates in an anti-immigrant referendum/petition)-even when taking several related constructs into account. In Studies 3 and 4, we extended these findings with experimental evidence. We specifically showed that leading participants to view their ingroup's national identity as fixed resulted in higher levels of prejudice towards immigrants, as well as hostile behaviors towards refugees, than leading participants to view national identity as malleable.
Compared to the malleable condition, participants in the fixed condition were 66% more likely to vote for the deportation of refugees in a real petition. While the threshold to really participate in a petition is high (as evidenced by the fact that the majority of participants did not participate in either petition), we could show that implicit theories can change these rather strong forms of anti-immigrant hostility, too.
We think that this research makes important contributions to both theory and practice in several ways. Firstly, we were able to identify and highlight the importance of a previously overlooked dimension of national identity: its perceived malleability or fixedness.
We show that it is not only important to look at how people define national identities at a given time, but also whether or not they think these definitions can change over time. This perspective can help us reach a more complete picture of the complex way individuals think about national identities.
Since results from our second study showed implicit theories to predict anti-immigrant hostility beyond essentialism, historical tolerance and group malleability, we could further show that these variables and our scale seem to be related, but distinct, constructs.
Merely believing in the fixedness of the ingroup's national iden-

| Limitations and future research
While we think that our research makes important contributions, it is not without limitations. The size of the effect our experimental manipulation had on individuals' implicit theory of national identity was rather small, suggesting that our stimulus material was limited in its persuasiveness. The small effect translated into similarly small effects on prejudice and a larger effect on the participation rates in the anti-immigrant petitions. Hence, while our manipulation has been rather weak, the change it did cause in participants' implicit theory of national identity has thus still transferred quite well into a change in prejudiced attitudes and behavior. To this end, future research should develop more persuasive manipulations, for example, by increasing the dosage and by implementing more engaging materials (e.g., multimedia videos, which have successfully been used in experiments by Crum, Salovey, & Achor, 2013).
Furthermore, our malleable condition manipulation in Study 4 was not successful in changing implicit theories of national identity as compared to the control condition (while the differences between fixed vs. control and fixed vs. malleable were significant). This may at least in part be due to our sample in Study 4, which already tended to lean toward a malleable view of national identity at baseline. Our manipulation was not strong enough to move participants further in the direction of a malleable view. As a result, while we could show that a more fixed view of national identity increases baseline levels of anti-immigrant hostility, we could not test whether an increase in individuals' malleable view of national identity would decrease baseline levels of anti-immigrant hostility. This remains to be verified in future research. In doing so, future research should improve on the manipulation content (as discussed above), and recruit a more politically conservative and fixed-leaning sample.
Another limitation concerns the generalizability of our findings.
While our samples were quite diverse in terms of occupation, political orientation, and gender, our participants were quite young on average (means varied between 30 and 36 years), quite highly educated (most participants either held a university degree or were currently enrolled in higher education programs) and from two European countries (Germany and England). Future research should replicate our manipulation with more diverse samples and in even more countries. shows the highest endorsement rates on exclusive contents).
Future research needs to empirically investigate the psychological processes underlying the effectiveness of our manipulation. We reasoned that our manipulation could change hostile outgroup bias by making the lines between in-and outgroup more or less clear and thus altering the extent to which individuals perceive others as distinct inand outgroup members. Since self-report measures are unable to capture such categorization processes, we did not include any potentially mediating process variables in our questionnaire studies. Still, non-reactive methods like the "Who said what?"-paradigm (Taylor, Fiske, Etcoff, & Ruderman, 1978) or the Implicit Association Test (Greenwald, McGhee, & Schwartz, 1998) could help researchers to investigate the underlying processes. These methods can be used to assess the extent to which participants implicitly categorize people as national in-or foreign outgroup members without invoking reactive responses (since participants are not aware of the assessment goal). Specifically, it could be tested whether participants who are led to view national identity as malleable or fixed would change the degree to which they see immigrants as foreign outgroup members and whether this effect would mediate effects on anti-immigrant hostility measures.

| Conclusion
The International Organization for Migration (2018) estimates that there are about 244 million migrants worldwide. An increasing share of those migrants is coming to Europe. In 2016 alone, 4.3 million migrants emigrated to one of the EU member countries (Eurostat, 2018). Although these migrants have long crossed geographical borders, mental borders of rigid national identity definitions still work to maintain a divide between residents and immigrants. Our research invites us to rethink the lines that are being drawn between the national in-and outgroup. Embracing a dynamic way to think about these lines may be a promising new way to overcome rigid divides between "us" and "them" and improve residents' relationships with newly arrived immigrants.

ACK N OWLED G EM ENT
This work was supported by a grant of the "Studienstiftung des Deutschen Volkes" allocated to the first author. We thank Carol Dweck and Gregory Walton for their feedback.

CO N FLI C T O F I NTE R E S T
The authors confirm they have no conflict of interest to declare.
Authors also confirm that this article adheres to ethical guidelines specified in the APA Code of Conduct as well as the authors' national ethics guidelines.