The personalization model revisited: An experimental investigation of the role of five personalization-based strategies on prejudice reduction

Abstract

In this paper, we report findings from a series of experimental studies investigating the role of five personalization-based strategies (individuation, self-disclosure, decategorization, self–other comparison, and empathy) on prejudice reduction. As anticipated, participants who were in the self-disclosure or individuation conditions developed more favorable attitudes towards other out-group members (Study 1). In Study 2, decategorization or self–other comparison allowed group members to discover similarities between the self and members of the out-group, thereby reducing negative out-group memory. Interestingly, when we explored the processes underlying the generalization of prejudice reduction to the out-group (Study 3), we found that the affective component of personalization (i.e., empathy) emerged as a more robust predictor of prejudice reduction than self–other comparison and decategorization. In conclusion, these studies provide the first empirical evidence of the effects of the five components of personalization in reducing prejudice across different contexts and different experimental paradigms.

Keywords

personalization decategorization empathy intergroup relations bias reduction

Casual contact with members of an out-group is part of our daily lives. It may occur at a bus stop, on an airplane, or standing in line at the bank. Conversation with an out-group member might have a number of distinct conceptual components. The conversation might serve to individuate that stranger (i.e., it might differentiate him/her from other members of his/her social category), it might induce self–other comparisons, it might arouse feelings of empathy, and it might also elicit feelings of trustworthiness in response to the out-group member’s disclosure of personal and unique information. These processes—individuation, decategorization, self–other comparison, empathy, and self-disclosure—are the ingredients of personalization. Any, or all, of their effects are likely to lead us to be more accepting towards this out-group member. More critical, however, is whether the processes affect our attitudes toward other members of that out-group category as well as to the individual.

Herein, we examine the effects of personalization in reducing prejudice toward members of the out-group category of the person with whom personalized interaction occurs. Of particular interest are the independent effects of its five enumerated components. We aim to clarify their conceptual distinctions to gain a better understanding of their prejudice-reducing effects onto both new out-group members, and the out-group as a category.

The personalization model

The personalization model, based on the contact hypothesis (Allport, 1954), suggests that personalized contact with an out-group member will facilitate perceptions of his/her unique characteristics, and will enhance familiarity between the members of the groups (Brewer & Miller, 1984, 1988). As a result, quality as well as quantity of contact with out-group members will reduce prejudice towards out-group members within the “contact situation” (Bettencourt, Brewer, Croak, & Miller, 1992; Binder et al., 2009; Brewer & Miller, 1984; Ensari & Miller, 2001, 2002). The reduction of prejudice produced by personalization has been attributed to the independent effects of its five components: specifically to individuation (Galinsky & Moskowitz, 2000), decategorization (Ensari & Miller, 2001), self–other comparison (Bettencourt et al., 1992), empathy (Batson, Chang, Orr, & Rowland, 2002; Wilder, 1978), and to the induction of trust that results from self-disclosure (Ensari & Miller, 2002).

The five components of personalization

Personalization sometimes involves direct experience with an out-group member that produces more acceptance of, or more positive attitudes toward that person (Fazio & Zanna, 1978), and this can potentially generalize to the social group too. In other cases, personalization involves indirect or partial interaction which can also reduce inter-group bias (Gaertner et al., 1999). Personalization necessarily involves individuation, and also contains other conceptually independent components which are separately discussed in what follows: Decategorization, self–other comparison, self-disclosure, and empathy. In everyday life these components typically co-occur. From an analytical perspective, however, they are independent.

Individuation

Acquisition of individuating information is a necessary step in any personalized interaction. Individuation creates recognition of a person’s distinct identity by providing information about his/her unique traits and/or attributes. It does not require direct interaction in that such information can be acquired from a third party (Miller, 2002). Moreover, individuation need not be accompanied by any of the other four components of personalized interaction (Miller, 2002). Both direct and indirect forms of individuation promote an awareness of the distinctiveness of individual members of a social category. Individuation reduces the better-than-average effect (Alicke, Klotz, Breitenbecher, Yurak, & Vredenburg, 1995), increases the perception of intracategory differences and intracategory variability (Brewer, Dull, & Lui, 1981), allows the out-group to be seen less as a unit, and thus diminishes intergroup bias (Wilder, 1978). It results in an assessment and evaluation of each person on his/her own merits, not according to stereotypes.

At the same time, when an individual’s attributes are perceived as distinct from those of other group members, that person can be subtyped, and thereby be viewed as an exception to the rule. Hence, Hewstone and Brown (1986) have argued that although positive attitudes toward a subtyped individual may emerge, there is no reason to expect that they will have a bearing on attitudes towards others who belong to that person’s social category, precisely because the subtyped individual is seen as atypical and unlike other out-group members. Hence, individuation may not necessarily alter stereotypes or perceptions of the out-group.

Decategorization

Decategorization is a type of top–down processing that produces differentiated perceptions of group members by inducing thinking that differentiates them from the stereotype of that group. During decategorization, individuals focus on the “big picture” of the out-group, retrieving the stereotypical traits that support it and altering these category-based cognitions, or groupings that have emerged from classifications based on our past experiences and expectations (Brewer, 1988). It can be achieved either by noting subgroups and differentiating them from the out-group stereotype, or by providing individuating information that differentiates individual group members from the out-group stereotype. Consequently, it necessarily reduces perceptions of out-group homogeneity (Miller, 2002), and develops more favorable attitudes towards out-group members (Ensari & Miller, 2001).

An important conceptual distinction between individuation and decategorization is that a group member can be individuated without being compared to his/her group. For instance, providing a list of the personality traits that characterize an out-group member can be sufficient to create an individuated perception of that person. Because decategorization requires that out-group member be distinguished from other out-group members, however, it requires thought about the mapping of that out-group member’s attributes onto those implicit in the stereotype of that out-group. Although reading a list of the personality traits of an out-group member starts this process via individuation, the individual’s out-group category (or the stereotypes about that category) still needs to be salient for that individual to be perceived as distinct from the group’s stereotype.

Self–other comparison

While decategorization is a form of top–down processing, self–other comparison is a bottom–up processing style (Brewer, 1988). During self–other comparison, bottom–up perceptions of another person’s attributes are evaluated in comparison to the self (Myers, 2001). When individual attributes of an out-group member are presented in relationship to self, impression formation begins mainly with recognizing and decoding the attributes of that out-group member, and proceeds from individual attributes to higher levels of comprehension. Consequently, similarities and differences between self and another are compared; as such, similar and dissimilar others are differentiated, and perceptions of group variability increases. Self–other comparison and individuation are conceptually independent of each other, in that self–other comparison requires individuation of the other, but individuation does not require self–other comparison (Miller, 2001). In addition, however, group members might also step into each others’ perspective during this process, and in turn, this may promote empathetic understanding (Batson, Early, & Salvarini, 1997).

Self-disclosure

Self-disclosure, or “the voluntary provision of information to another that is intimate or personal” involves freely providing unique, intimate information about oneself that is not ordinarily known and not readily available (Ensari & Miller, 2001, p. 196). It also involves risk and vulnerability on the part of the person self-disclosing it, and subsequently promotes trust and familiarity, reduces anxiety implicit in intergroup encounters, and creates perceptions of the out-group member as friendlier and more desirable to have “as a friend” (Ensari & Miller, 2001; Rothbart & John, 1985; Stephan & Stephan, 1985). Simultaneously, these processes provide an opportunity to disconfirm the negative stereotypes of disliked out-group members and break down the monolithic perception of the out-group as a homogeneous unit (Cook, 1978; Wilder, 1978). Consequently, self-disclosure by an out-group member increases interpersonal liking, and reduces prejudice towards the social group (Bettencourt et al., 1992; Ensari & Miller, 2002; Worchel, 1986).

Empathy

Empathy is an emotional response that is linked to, and elicited, by another’s circumstances (Batson, Klein, Highberger, & Shaw, 1995). Empathizing does not cause the other to be seen as resembling of the self, but in addition to individualizing the other (Batson, Early, et al., 1997), it motivates people to help others because of the guilt and shame that comes from not helping, and the social praise and self-rewards that arise from helping the target individual (Batson, 1991). At a general level, empathy for out-group members facilitates conflict resolution between opposing groups, and leads to less out-group bias (Stephan & Finlay, 1999).

There are two components of empathy. Emotional empathy refers to experiencing the same emotions as another person. It arouses feelings of injustice, which counteract prejudice, in turn leading to attitude change, and evoking compassion-related emotions that arise from a feeling of concern (Finlay & Stephan, 2000; Stephan & Finlay, 1999). Cognitive empathy (i.e., perspective taking) refers to understanding the perspectives of others (Batson, Batson, et al., 1995). Perspective taking not only elicits emotional empathy, but also extends self-descriptive traits to a target/the target’s group (Galinsky & Moskowitz, 2000), and fosters enhanced closeness with the other person in the form of self–other merging (Vorauer, Martens, & Sasaki, 2009). Moreover, perspective taking leads to favorable attributions at the interpersonal level (i.e., internal attributions for others’ success and external attributions for others’ failure), and increases helpful and supportive behavior (Toi & Batson, 1982). Also at the intergroup level, it increases merging of self and out-group members, enhances perceptions of similarity between self and out-group members, and decreases both stereotypic biases and intergroup bias (Aron, Aron, Tudor, & Nelson, 1991; Galinsky & Moskowitz, 2000; Vescio, Sechrist, & Paolucci, 2003).

The present studies

Our purpose herein is to examine the independent effects of personalization components on prejudice. Although past studies examined some of them in an interpersonal or intergroup context (e.g., self-disclosure in Ensari & Miller, 2002; decategorization in Ensari & Miller, 2001), none manipulated them within a single study. Here we aim to extend our understanding of personalization and its components by (a) providing experimental evidence of their independent contributions in reducing prejudice; (b) showing how their beneficial effects can be generalized to new out-group members in an entirely different context; (c) examining their effects on distinct outcome measures such as prejudice toward new out-group members, prejudice toward the entire out-group, and intergroup memory bias; (d) demonstrating the comparatively stronger effects of empathy by comparison with other components; and (e) replicating personalization effects in different experimental paradigms across three studies.

We expected each component of personalization to reduce prejudice. More specifically, in comparison to the baseline control condition, we hypothesized that individuation and self-disclosure will augment the favorability of evaluations of new members of the out-group category (Study 1), decategorization and self–other comparison will reduce intergroup memory bias (Study 2), and prejudice toward the entire out-group (Study 3). We did not make a prediction with respect to the differential effects of decategorization, individuation, self–other comparison, and self-disclosure, because there is no strong theoretical basis to predict so. On the other hand, we predicted that empathy will further reduce prejudice beyond the level induced by these components (Study 3), as further explained in what follows.

Study 1

Participants

Eighty-eight university students (63 women, 25 men) voluntarily participated for course credit. The participants were older than 18 years, and were from diverse racial backgrounds. All were liberal in their political orientation. In all three studies reported herein, the participants were randomly assigned to one of the experimental conditions.

Procedure

After the participant arrived in the experimental room, the confederate, who acted as another participant, appeared in the same room. The experimenter welcomed both of them, and gave them the “Informed Consent Form” to read and sign. Next, she asked them to sit across from each other, and told them that they were participating in a large-scale study designed to investigate how people from different backgrounds work together in a cooperative task. Before telling them that there were three consecutive tasks, they filled out a questionnaire in which they indicated their political orientation (conservative or liberal), and the party they vote for (Republican or Democrat). At that point, if the real participant indicated that s/he was conservative and/or voted for the Republican party, the experimenter stopped the experiment and released the participant. Otherwise, the experimenter collected the psycho-demographic questionnaires from the participants, leaving the confederate’s form on the top and visible to the real participant. The confederate always categorized herself as a “conservative.”

The constant features of the experiment

According to Ensari and Miller (2002), category salience and typicality are necessary conditions for personalization to be generalized to the out-group. Therefore, to maintain high salience of the participant’s out-group category membership and high out-group typicality as constant features of the experiment, the participants were told that we were especially interested in their political orientation; and they were given badges indicating their political orientation, which were to be worn throughout the experiment. Furthermore, the confederate emphasized her political identity (as the out-group member) to participants during the disclosure task (see, Ensari & Miller, 2002, the Procedure section on pages 316–317).

Manipulation of personalization

The self-disclosure condition

The manipulation of self-disclosure was taken directly from Ensari and Miller’s (2002, 2005) procedures. Therefore, we only briefly describe it here. After initial introduction, the experimenter presented the communication task, which consisted of an interview of one participant by the other. The experimenter then gave real participants a list of five interview questions that were personal, with instructions on how to use them as an interviewing guide, and further urged them to engage in as natural a conversation as possible for 10 minutes. The key interview question was the following: “Tell me 2 good and 2 bad things about yourself that most people who already know you don’t know” (from Relationship Closeness Induction Task [RCIT]; Sedikides, Campbell, Reeder, & Elliot, 1998). To encourage the participant to view the act of disclosure as voluntary and hence, feel entrusted with personal information, the confederate was publicly told that “anything you wish to tell about yourself is okay – It is completely up to you.” The experimenter then left the room. At that point, the participant, who always was the interviewer, started to ask the listed questions one by one, and listened to the confederate (i.e., the interviewee) who answered them in a natural conversation. The responses provided by the confederate were previously carefully selected to include two moderately positive and two moderately negative pieces of information that were of moderately high intimacy (Ensari & Miller, 2002, 2005).

In the individuation and control conditions, the confederate responded to the interview questions with neutral impersonal statements (e.g., aspects of campus life) that included no additional individuating information beyond what had previously been provided during the introduction. As a manipulation check, directly after the communication task, the participants indicated the extent to which the information revealed by the confederate was personal and intimate, and rated the confederate’s friendliness using 7-point scales. Higher scores indicated greater self-disclosure and friendliness.

The individuation condition

After the completion of the communication task, the participants were separated into two different rooms to complete the second task: the personality profile task. After this point, the participant remained alone in the room. S/he was told that the purpose of this part was to investigate how university students interpret the outputs of personality tests. They were told that some participants filled out these tests as part of their job application packet, and that the results had been interpreted and summarized by a clinical psychologist.

In the individuation condition, the participant was told that the other participant (viz. the confederate who had been introduced as a Republican out-group student) had actually filled out one of these personality tests as part of a job application procedure, and we had the results summarized as a brief report which contained five moderately positive and five moderately negative personality characteristics preselected from Anderson (1968). There were two different versions of the report that were used in alternating order. The order of the positive and negative traits was balanced. In the self-disclosure and control conditions, the instructions were identical with the exception that the personality inventory belonged to an unknown student whose political orientation was not disclosed, instead of the Republican confederate.

After reading the report, the participant used 7-point scales to complete a questionnaire that requested evaluations of the profiled student’s friendliness and trust and also included a manipulation check of individuation (“To what degree was the information on the personality inventory individuating?” “To what degree was the information on the personality inventory unique?”).

The main dependent measures

The participants’ final task was to examine how people like themselves, who are not from a personnel department, evaluate student applications for employment. The participant was given a folder containing materials of two applicants who had applied for employment the previous year. We gave the participant job applications of a liberal (in-group) and a conservative (out-group) applicant (both female) and told them to examine the application materials carefully before evaluating them. The information indicating the political orientation of the applicants appeared under the “Skills and Extracurricular Activities” section of the form. To manipulate in-group/out-group membership of the applicants, one was depicted as having served 5 years at the Student Liberals Community whereas the other had similarly volunteered for the Organization of Young Republicans Conference. We constrained all other information to be sparse and to provide little basis for subtyping the individual applicants. We counterbalanced the order of the in-group and out-group application forms and the descriptions of all other information about the applicants. This latter constraint meant that (between-subjects) there were actually two in-group and two out-group applicants. In the dependent measure packet, as a prejudice measure, the participants used 7-point scales (with higher scores indicating more of the attribute) to rate the degree to which each applicant was qualified, competent, friendly, and trusting. This measure allowed us to examine generalized prejudice toward out-group members who were not involved in the personalized interaction with the participants.

Check of typicality, in-group identity, and category salience

To assess whether out-group typicality, category salience, and high in-group identity were maintained, the participants were asked to answer the following questions on a 7-point scale (1 = “not at all,” 7 = “extremely”): “What is the extent to which you feel your political orientation influenced your evaluations of the other participant?”; “At this moment, how much do you feel that you identify with Liberals?”; “At this moment, how much do you distinguish yourself from Conservatives?”; “What is the extent to which the other participant was a typical Conservative?”; “What is the extent to which the other participant is representative of Conservatives?” and “What is the extent to which the other participant behaved like a typical Conservative?”.

Assessment of suspicion

Upon completion, we assessed suspicion by asking several questions that focused on whether the procedures seemed odd or uncomfortable, or whether there were any suspicions regarding the confederate in connection to the communication and problem-solving tasks. Based on the answers, only one participant was later removed from analyses.

Results

Check of contextual levels of typicality, in-group identity and salience

Although typicality and salience were intentionally made constant across the conditions, we also sought to induce high levels of each. When averaged (alpha = .87), the mean typicality (M = 5.21) exceeded the scale midpoint (4.0), t(87) = 5.66, p < .001, indicating that the out-group confederate was indeed seen as a typical conservative across the conditions. Additionally, participants indicated that they felt their political orientation influenced their evaluations of the other participant (M = 5.16), that they identified with Liberals (M = 4.50), and that they distinguished themselves from Conservatives (M = 4.77), t(55) = 5.17, p < .001; t(55) = 2.38, p < .05; t(55) = 3.96, p < .001 respectively. A one-way ANOVA revealed no main effect, indicating that the confederate’s political identity was highly salient, and that the participants had identified with their in-group across conditions.

Manipulation checks of self-disclosure and individuation

After the communication task that involved personal (self-disclosure), or impersonal questions (individuation and control conditions), the participants rated the extent to which the information revealed by the confederate was personal and intimate on a 7-point scale. When we combined the two measures of self-disclosure (i.e., personal and intimate, r = .71, p < .00), a one-way ANOVA revealed a main effect, F(2, 87) = 26.46, p < .001. Tukey’s post hoc analyses indicated that the revealed information (during the communication task) was seen as more personal and intimate in the self-disclosure condition (M = 3.97) by comparison with the individuation (M = 2.27) and control conditions (M = 1.65, p < .05). As expected, the means for the individuation and control conditions did not reliably differ (p > .05). These results provide evidence for a successful manipulation of self-disclosure.

The measures of individuation manipulation were taken after completion of the personality profile task which summarized the personality traits of either the Republican out-group confederate (individuation condition), or a student whose group identity is unknown (self-disclosure and control conditions). On 7-point scales, the participants rated the degree to which the information on the personality inventory was individuating and unique; and evaluated how friendly and trustworthy the profiled student was. Two manipulation check measures of individuation (i.e., individuating and unique) were combined (r = .82, p < .01) and their mean (M = 4.48) was tested against the midpoint of the scale (i.e., 4). A one-sample t test showed that the information presented on the personality profile was seen as individuating and unique, t(87) = 3.48, p < .001. Additionally, we analyzed the two evaluative measures of the student on the personality profile (friendliness and trustworthy). A one-way ANOVA revealed a significant effect only for the friendliness measure, F(2,85) = 3.75, p < .05. When the student applicant on the personality profile was presented as the out-group confederate that they previously met during the communication task (i.e., the individuation condition), she was seen as friendlier (M = 4.48) than when the group membership and the identity of the student applicant on the personality profile were not revealed (i.e., in the self-disclosure [M = 3.56] and the control conditions [M = 3.28]). Positive evaluations of the out-group confederate after the personality profile task not only show the anticipated positive effect of individuation on interpersonal perceptions, but also further confirm a successful manipulation of individuation.

Prejudice towards new out-group members

To measure prejudice, the participants were asked to evaluate the job application forms of an in-group and an out-group applicant both of whom had applied for university employment. Because analysis of the counterbalancing of the irrelevant information on the application forms revealed no effects, we collapsed over this variable in subsequent analyses. Four 7-point scales had assessed the degree to which each applicant was viewed as qualified, competent, friendly, and trusting, with higher scores reflecting more of each attribute. Based on Abele and Wojciszke’s conceptualization (2007; see Fiske, Cuddy, & Glick, 2006), the two items reflecting the competence measure (qualified and competent) were combined (r = .71, p < .05), and the two items reflecting warmth measure (trusting and friendly) were combined (r = .54, p < .05). The results of MANOVA on the competence and warmth measures revealed a main effect for competence measure (F(2, 85) = 18.55, p < .001, η² = .30), and for warmth measure (F(2, 85) = 12.72, p < .001, η² = .23). Tukey’s post hoc analyses revealed that both competence and warmth measures were higher in the self-disclosure and individuation conditions than in the control condition (p = .001) (see Table 1). However, the self-disclosure and individuation conditions did not differ (p > .05). As shown by planned contrast analyses using weights of 1, 1, −1 to self-disclosure, individuation, and control conditions, confirming the post hoc results, the out-group evaluations were more favorable in the self-disclosure and individuation conditions by contrast to the control condition (F(1, 85) = 43.93, p < .05 for competence measures, and F(1, 85) = 23.38, p < .05 for warmth measures). Evaluations of the in-group applicants, however, were equally favorable in all three conditions (p > .05). This result is consistent with previous findings (Ensari & Miller, 2002), showing that the effects of self-disclosure and individuation are attributable to more favorable evaluations of the out-group applicants.

Table 1.

Means and standard deviations of the experimental conditions (self-disclosure, individuation, and control) on warmth (trusting and friendly combined), and competence measures (competent and qualified combined) in Study 1 (N = 88)

		Competence measure	Warmth measure
Self-disclosure (n = 30)	Mean	5.15_a	5.13_a
Self-disclosure (n = 30)	SD	.70	.85
Individuation (n = 28)	Mean	5.39_a	4.84_a
Individuation (n = 28)	SD	.94	.80
Control (n = 30)	Mean	4.13_b	4.10_b
Control (n = 30)	SD	.88	.80

Note: Seven-point rating scales were used. Higher values indicate more favorable rating for the out-group member. Means that do not share common subscripts differ reliably from each other (p < .05).

Study 2

Study 2 differed from Study 1 in that we: (a) manipulated decategorization and self–other comparison: two components of personalization that had not been examined in Study 1; (b) utilized a different experimental paradigm; (c) focused on a different outcome measure, intergroup memory bias; and (d) explored the mediational role of perceived similarity to the out-group. Thus, whereas Study 1 examined intergroup attitudes to new out-group members as a means of assessing prejudice, Study 2 examined intergroup memory bias, a related, but conceptually distinct aspect of prejudice. Memory bias refers to a recall advantage for unfavorable information associated with the out-group compared to that associated with the in-group (Howard & Rothbart, 1980). In the present study, we specifically focused on negative out-group memory (i.e., recalling negative traits of out-group) because past research found evidence for recall bias on negative, but not positive traits (Crisp & Hewstone, 2001; Howard & Rothbart, 1980). Negative information is more salient, unusual, or informative than positive information. Therefore, it is the natural dimension of representational differentiation between the in-group and out-group (Crisp & Hewstone, 2001; Fiske, 1981; Kanouse & Hanson, 1971). As a result, this positive–negative asymmetry in intergroup memory bias leads to a preference for encoding more negative information in association with out-group than in-group members.

We expected that the positive effects of personalization should extend to intergroup memory bias such that decategorization and self–other comparison conditions should reduce out-group memory bias, resulting in a weaker memory of negative out-group attributes. Furthermore, we expected perceived similarity to the out-group to mediate this effect. Personalization allows group members to discover similarities between themselves and the out-group, and consequently to focus on out-group traits and attributes with a more positive point of view. The resulting decrease in the tendency to focus on negative traits should reduce their availability in memory. Self–other comparison, for instance, allows individuals to discover similarities, as well as differences, between one’s self and the out-group member. Consequently, as more self–other similarities are perceived, the negative impact of prejudice on memory should diminish. Likewise, as decategorization reduces reliance on the out-group stereotype and substitutes the processing of new information about the focal out-group member, similarities between the self and that focal out-group members can emerge. Thus, we predicted that decategorization or self–other comparison will increase perceptions of similarity of self and out-group members, in turn, reduce negative memories of out-group.

Methods

Participants

Twenty-seven Liberal undergraduate students from USC (University of Southern California) (18 women, 9 men) participated in this study in exchange of a course credit. All participants were 18 years or older, and their ethnicity and race were diverse.

Procedure

Selection of only Liberal participants was identical to Study 1. The participants were told that this study was designed to investigate the relationship between personality characteristics and people’s political choices, and they would engage in several tasks consecutively. The experimental manipulations involved sequentially completing a series of worksheets in the order that they were positioned in an envelope. The experimenter had previously prepared three versions of the questionnaire packet, corresponding to the three experimental conditions, and placed them in envelopes, which were then randomly ordered. To avoid experimenter’s bias, and keep the experimenter blind to the conditions, there was no identifying information on these identical envelopes. The total time allotted for completion was constant across conditions. The first page of the packet provided general information that indicated that the participant would receive information about the personality of another person, and be asked to complete various evaluative tasks. The following pages involved the experimental manipulations as follows.

The decategorization condition

The second page in the envelope presented a personality profile of a Republican student whose identity was unknown. The participants were told that the information on this profile was obtained from an analysis of questions that this Republican student, who was randomly chosen from others, answered in a previous experiment ran last semester at the same college. The personality profile listed 20 traits (10 positive and 10 negative, randomly ordered): 15 traits were selected randomly from amongst a larger set of traits from Dumas’ personality trait list (Dumas, Johnson, & Lynch, 2002), and five traits were selected as most commonly stereotypical traits of Republicans which were selected from amongst the pretested characteristics used in Ensari and Miller (2002) (see Appendix A). This profile rated the Republican person on each trait either “high,” “medium,” or “low”. In order to provide a basis of comparison, the ratings were presented as compared to the average student of USC. For instance, “high” score on “mature” indicates that the person is more mature than most students at USC. To maintain high typicality, this page included ratings of “average Republican students” that represent typical Republicans. This allowed us to show that the Republican student resembles “average Republican students” on traits that are stereotypical of Republicans (e.g., both scored high on “conservative”). We created two different versions of this personality profile (keeping the five stereotypical traits constant), and randomly assigned each participant to one of the two versions. The overall valence of the out-group member’s profile was made neutral by balancing valence across traits. Sex of person represented by the profile was not indicated.

The self–other comparison condition

In the self–other comparison condition, the profile of the Republican student was presented next to the participant’s own trait profile, using the same traits (see Appendix B). The participant was told that we obtained his/her personality information through the Introduction to Psychology questionnaire packet that s/he completed at the beginning of the semester. This cover story did not evoke any suspicion because all students in this class had actually completed a number of questionnaires presented in a packet at the beginning of the semester.

The control condition

The participants in the control condition were given a personality profile, including the same traits that described a student whose political orientation was unknown (See Appendix C). Thus, they reviewed a profile of a student whose personality traits were compared with average USC students.

The dependent measures

Perceived similarity to the out-group

After reviewing the personality profile, the participants were asked to indicate how similar they felt (a) to the Republican student on the profile, (b) to the average Conservative (the key measure), and (c) to the average Liberal on 7-point scales (7 = extremely, 1 = not at all).

Out-group memory

On the following page, the participants were asked to remember the personality profile that they had examined, and write down as many of the personality traits of the student on the profile as they could. They were also told to indicate whether the person being evaluated was low, moderate, or high on each trait. Later, the experimenter counted the number of negative (the key dependent measure) and positive traits that were correctly remembered.

Finally, we asked the participants to report whether the average Republican (in the decategorization condition), they/themselves (in the self–other comparison condition), or the student depicted in the profile (in the control condition) were low, medium, or high on each trait. The experimenter later counted the number of correct responses, which were used as a direct manipulation check of the personalization components, thus ensuring that the participants carefully reviewed and processed the information presented on the profile. At the end, a debriefing was used to simultaneously probe the subject for suspicions and to reveal the details and deceptions used. None of the participants were found to be suspicious.

Results

Tests of correctly remembered traits

The number of traits correctly remembered (M = 11.41 out of 20 traits) exceeded the midpoint of 10 (t(26) = 2.40, p < .05), indicating that the participants carefully reviewed and then remembered the traits of the student on the personality profile. A one-way ANOVA showed that the number of traits correctly remembered did not differ as a function of the conditions (M = 11.33, 10.75, and 12.00 for the decategorization, self–other comparison, and control conditions respectively, F[2, 24] = .36, p > .05).

A 3 (decategorization, self–other comparison, vs. control) x 2 (negative vs. positive traits) ANOVA with repeated measures on the second factor revealed an interaction, F (2, 24) = 3.66, p < .05 (see Table 2). The experimental conditions differed in negative memory (F(2, 24) = 5.08, p < .05), but not in positive memory (F(2, 24) = 1.18, p > .05). Tukey’s post hoc tests showed that participants remembered fewer negative traits in the decategorization (M = 4.00) than in the control condition (M = 6.40), p < .05, and similarly fewer negative traits in the self–other comparison (M = 3.13) than in the control condition (p < .05)^1,2.

Table 2.

Means and standard deviations of the experimental conditions (decategorization, self–other comparison, and control) on positive and negative traits correctly recalled in Study 2 (N = 27)

		Positive Traits	Negative Traits
Decategorization (n = 9)	Mean	7.33_a	4.00_a
Decategorization (n = 9)	SD	2.67	2.00
Self–other comparison (n = 8)	Mean	7.62_a	3.13_a
Self–other comparison (n = 8)	SD	2.82	1.88
Control (n = 10)	Mean	5.60_a	6.40_b
Control (n = 10)	SD	2.67	1.78

Note: Higher values indicate better memory. Means that do not share common subscripts within a column differ reliably from each other (p < .05).

Negative memory of the out-group

To test the hypothesis that perceived similarity to the out-group mediates the relationship between personalization and negative memory (of the out-group), we have first combined the two personalization conditions as an independent variable, and used the number of negative traits correctly remembered as a dependent measure in the mediational analyses. We have used bootstrapping (Preachers & Hayes, 2004), which is considered to be superior than other techniques in terms of power, control over Type 1 error, and intuitive appeal (MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002; Preacher & Hayes, 2004). The results of the bootstrapping method (based on 1,000 bootstrap resamples) showed that the bootstrapped estimate of the indirect effect fell between 0.18 and 1.30 with a 95% confidence interval which excluded zero, indicating that the indirect effect from personalization to negative memory was significantly different than zero at p < 0.05. Thus, altogether, these analyses lead us to conclude that similarity to the out-group partially mediated the effect of personalization on negative memory^3,4.

In sum, these results showed that decategorization and self–other comparison mechanisms allow group members to discover similarities between self and members of the out-group, and by implication reduce negative out-group memory. The results support previous research which offers further evidence for the presence of a recall bias on negative, but not positive traits (Crisp & Hewstone, 2001; Howard & Rothbart, 1980).

Study 3

Study 3 aimed at expanding on the previous studies by (a) adding another important component of personalization, empathy, to the design; (b) examining a different experimental paradigm in manipulating the components of personalization; and (c) using a general measure of prejudice that examined attitudes towards the entire out-group, rather than focusing on a single out-group member, thus increasing the validity of our findings.

Generalization to the entire out-group

In previous research (e.g., Desforges, Lord, & Ramsey, 1991; Ensari & Miller, 2001; Hewstone & Brown, 1986; van Oudenhoven, Willemsma, & Prins, 1996; Wilder, 1984), intergroup contact led to improved attitudes towards the out-group. Pettigrew (1997) raised the critical question of whether these altered attitudes generalize beyond the immediate situation—to new situations, to the entire out-group, or even to out-groups not involved in the contact. For example, students who studied cooperatively with students of other races increased the number of their cross-racial friendships, yet unfortunately they did not change their racial attitudes in general (Weigel, Wiser, & Cook, 1975). If personalization only improves attitudes to the out-group members in the contact situation, then its application is very limited, and the value of contact theory is constrained.

Generalization was examined in previous studies. In an experimental study, Rhodes, Halberstadt, and Brajkovich (2001) showed that the increases in liking that derive from exposure can generalize to greater liking for both related and unknown targets. Pettigrew and Tropp (2006) meta-analytically examined generalization beyond the contact situation and found that attitudes toward the immediate participants usually became more favorable, but more importantly attitudes toward (a) out-group members not involved in the contact; (b) out-group members in other situations; and (c) the entire out-group were improved. Corresponding to this first type of generalization, in Study 1 and 2, we have focused on attitudes toward an out-group member not involved in the contact (a Republican job applicant in Study 1, and a Republican student on the personality profile in Study 2). Corresponding to the third type of generalization, Study 3 assessed attitudes toward the entire out-group, Republicans as a group.

Additionally, Ensari and Miller (2001, 2002) suggested an integration of prior theoretical models that emphasize the importance of personalization (Brewer & Miller, 1984) versus category salience (Hewstone & Brown, 1986) in producing generalization toward the out-group category during cooperative contact. This integrative approach was empirically tested in a number of studies that illustrated that the positive effects of personalization will only be generalized to the out-group if both typicality and salience are maintained (Ensari & Miller, 2001, 2002). Consistent with this approach, all three studies reported herein maintained high typicality and salience throughout the experiments.

Empathy

In this study we predicted that, consistent with Study 1 and 2, all personalization components will reduce prejudice toward the new out-group members as well as toward the entire out-group. We also predicted that, by comparison with the other personalization components, empathy will more strongly reduce prejudice. The rationale for this prediction is that decategorization and self–other comparison (as well as individuation) only involves cognitive mechanisms that increase recognition of an out-group member’s distinct identity, whereas empathy involves both cognitive and affective mechanisms and so it should be a more powerful predictor given that it cross-cuts both domains.

The cognitive component of empathy, perspective taking, is an ability to cognitively understand another’s internal state (e.g., Hogan, 1969; Underwood & Moore, 1982). Although Batson (1987) showed that experimental inductions of perspective taking reliably elicit both the tender emotions such as sympathy, compassion, and warmth, as well as emotions reflecting personally felt distress, like worried, alarmed, and troubled, nonetheless, this does not alter the fact that in some fundamental sense, perspective-taking manipulations necessarily contain a cognitive component. Galinsky and Moskowitz (2000) showed that perspective taking extends self-descriptive traits both to a target and the target’s group, and in turn, decreases both stereotypic and inter-group bias. It also reduced stereotypical perceptions of an individual who spoke English as a second language (Weyant, 2007), promoted more favorable attitudes toward racial groups (Vescio, Sechrist, & Paolucci, 2003), and mediated the relationship between contact and intergroup anxiety (Aberson & Haag, 2007).

The affective component of empathy is viewed by some as tapping the experiencing of an emotion similar to that felt by another person (e.g., Eisenberg & Miller, 1987; Feshbach & Roe, 1968; Stephan & Finlay, 1999; Stotland, 1969; Worchel, 1986). Batson (1987) has presented convincing evidence showing that the emotions elicited by another’s plight have a two-factor structure consisting of the tender emotions and personal distress, and that it is the former set that more strongly contributes to prosocial behavior. Intergroup friendships lead to an increased emotional empathy for the out-group to which a friend belongs, and these emotional connections lead to more favorable intergroup attitudes (Pettigrew, 1997). When the out-group member is in need of help, or is in danger, personalization of that out-group member will increase feelings of a close connection with that person, and elicit sympathetic reactions that act to increase helping behavior towards others and reduce prejudice.

In a meta-analytical study, Pettigrew and Tropp (2008) found that empathy and perspective-taking are strong mediators of the positive effect of contact on prejudice. Although their results were limited (owing to small number of studies available for analysis n = 9), the fact that the effects of empathy were not detached from those of perspective-taking, suggests the results were consistent with the robust findings reported in the previous literature (e.g., Batson, Ahmad, & Stocks, 2004). Pettigrew and Tropp (2008) concluded that “Future research should distinguish more sharply between the effects of empathy and perspective taking – something we could not attempt with the paucity of research to date. But the patterns of effects observed in the present analysis suggest that greater emphasis on processes involved in empathy and perspective taking would be a fruitful direction for future research.” (2008, pp. 929–930). In sum, empathy differs from decategorization and self–other comparison in that it involves not only a cognitive component, but also an affective component, and hence, it is more likely to elicit a stronger reduction of prejudice.

Methods

Participants

Participants were 94 Liberal undergraduate psychology students from USC (University of Southern California) who were older than 18, ethnically diverse, (76 women, 18 men) who received extra course credit for participation.

Procedure

The experimental procedure consisted of two parts. First, decategorization and self–other comparison were manipulated using the same procedure as in Study 2 (i.e., reviewing a personality profile). In the empathy and control conditions, similar to the control condition in Study 2, the participants were instructed to review a personality profile of a student whose political identity was unknown, and whose traits were compared to the average USC student.

In the second part of the experiment, the participants were given a list of problems that an out-group member (a Republican student) faced during the last 6 months. In the empathy condition, they were asked to read this list, and to think for 5 minutes about the way this out-group member felt when the problem occurred. The rationale for this is that a person’s prior experience of an unpleasant event increases their empathy for another person who is currently experiencing that incident. Imagining how another would feel produces empathy, whereas imagining how you would feel produces distress as well as empathy (Batson, Early, & Salvarini, 1997). Therefore, it was important that the participants were told to imagine “how the other person feels” in the situation and not “how they feel,” to insure that empathy, and not distress, was being induced.

In the decategorization and self–other comparison conditions, the participants were first asked to read the same list, and then to think for 5 minutes about the way this person solved each problem, in contrast to the way that most “Republican college students” would solve it. Thus, in these conditions, the participants were encouraged to engage in a cognitive process that required them to compare the Republican out-group student to other Republicans, instead of engaging in an affective process. In the control condition, participants were told that these problems were written by a randomly chosen participant from a previous experiment (whose group identity was unspecified), and asked to think for 5 minutes about each of the four problems that the person faced. These problems were previously selected to be neutral or positive in valence. In sum, there were two key distinctions differentiating the conditions: First, decategorization and self–other comparison conditions involved reviewing the list of traits that belong to an out-group member, whereas empathy and control conditions did not. Second, the participants were engaged in an affective, empathic process with an out-group member in the empathy condition, whereas those in the other conditions did not.

The manipulation check measures and the dependent measures

Study 3 aimed to examine the role of personalization on attitudes toward the entire out-group, not just attitudes towards a single out-group member. As such, we assessed general attitudes toward Republicans. At the end of the experiment, the participants completed the manipulation check measures (Batson’s [1991] empathy adjectives: sympathetic, warm, softhearted, compassionate, tender, moved), and the generalization measures. These latter measures asked them to indicate the likelihood that Republicans will engage in the following behaviors: “volunteer to do community service,” “become a successful person,” “save a person’s life,” “contributes positively to the community,” and “fail at college (reversed)” on a 9-point Likert-type scale. The target behaviors were carefully piloted prior to the experiment, and considered to be positive and socially acceptable activities. Upon completion, a funnel debriefing was used to simultaneously probe the participants for suspicions about the experiment and to reveal the details and deceptions used. None indicated suspicion in Study 3.

Results

Manipulation check of empathy

To assess the effectiveness of the manipulation of empathy, Batson’s (1991) empathy adjectives (sympathetic, warm, softhearted, compassionate, tender, moved) were combined (alpha = .89). A one-way ANOVA showed that participants in the empathy condition scored higher on the combined empathy measure (M = 4.47) than those in the decategorization (M = 2.84), self–other comparison (M = 2.84), and control conditions (M = 2.89), F(3, 56) = 8.28, p < .001. This shows that the empathy manipulation was successful.

General attitudes toward the entire out-group

We combined the five behavioral items which are considered to be positive, as indicated above (alpha = 0.68). A one-way ANOVA revealed that the participants indicated that Republicans are more likely to engage in positive actions in the experimental conditions (M = 4.80 in the decategorization condition, 4.71 in the self–other comparison condition, 5.19 in the empathy condition) than did those in the control condition (M = 4.29), F(3, 56) = 3.59, p < .05.

Differential impact of empathy

We hypothesized that there would be a further reduction in prejudice in the empathy condition in comparison to the decategorization and self–other comparison conditions. We conducted a planned contrast analysis wherein we assigned the following contrast weights to the empathy, decategorization, self–other comparison, and control conditions respectively: 1, ½, ½, −2. The results of the contrast analyses confirmed this hypothesis, F(1, 56) = 9.00, p < .05, with no residual variance (p > .05)⁵. In sum, empathy was shown to be the most effective component of personalization in reducing prejudice toward the entire out-group.

Discussion

Taken together, the three studies provide strong convergent evidence suggesting that people can draw on a number of personalization strategies, ranging from decategorization to empathy, to reduce prejudice toward an out-group and its members. Among the five personalization components, empathy outperformed the other components of personalization in its ability to diminish prejudice associated with the out-group. Although we interpreted this stronger effect as reflecting both components of the empathy induction, namely, the elicitation of both cognitive and affective responses, there are at least two alternative explanations for this outcome. First, it is possible that affect is a more robust component of personalization and therefore diminished the influences of more cognitively focused manipulations such as, individuation, self–other comparison, and decategorization. Alternatively, it might be that the more powerful prejudice-reducing effect linked to empathy should be a consequence of the strength of implementation issue, and therefore if a different implementation were employed to induce self–other comparison and decategorization, our induction of empathy may not have shown as strongly, and these cognitive components of personalization might have accounted for more of the predictive variance. Hence, it unwittingly may simply have been as stronger induction and therefore produced a stronger effect.

Study 1 showed the beneficial effects of self-disclosure and individuation, and supported our previous findings (Ensari & Miller, 2002, 2005). It also extends them by showing that individuation of an out-group member can be as powerful as an out-group member’s self-disclosure in improving attitudes toward other members of that out-group. Of course, what is not known is whether the experimental inductions of individuation and self-disclosure were implemented with equivalent strength. Hence, apart from the ordinary caveats about the meaninglessness of a failure to detect a difference, whatever manifest meaning is taken from the similar means of the two experimental conditions, it is compromised by the fact that their comparative strength of implementation remains unknown.

In Study 2, we address important questions about reducing negative out-group memory and diminishing prejudice by the induction of decategorization and self–other comparison. This finding supports our hypothesis, and is consistent with previous research, which shows that recall bias exists on negative, but somewhat surprisingly not on positive traits (Crisp & Hewstone, 2001; Howard & Rothbart, 1980). The caveat, however, is that these results may diverge from previous findings that have found that intergroup bias was evidenced when using positive, yet not negative stimuli items (e.g., Crisp & Hewstone, 2000; Gaertner & Dovidio, 2000; Hewstone, Fincham, & Jaspars, 1981; Mummendey & Otten, 1998; Wenzel & Mummendey, 1996). This positive/negative asymmetry effect in intergroup bias could be explained as a normative effect, such that differentiation of one’s group from another on negative dimensions compared to positive ones is perceived as less socially acceptable than if the contrast were on positive to negative characteristics (Hewstone et al., 1981; Wenzel & Mummendey, 1996). In these studies (Wenzel & Mummendey, 1996), the researchers asked participants to rate the groups on either positive or negative attributes, which evoked the norm of “no derogation” in negative terms. On the other hand, however, when participants were presented mixed positive/negative lists, differentiation on negative attributes was clearly counter the norm of “no derogation,” and as a result the positive/negative asymmetry effect disappeared (see, Crisp & Hewstone, 2001). In our experiments, similar to those reported by Crisp and Hewstone’s, the list of personality traits included both positive and negative characteristics of the target person. Consistent with the normative hypothesis, the positive/negative asymmetry effect was not observed.

Yet, in spite of their important contributions to the literature, the findings were limited by some shortcomings. Such issues included a small sample size (especially in Study 2) which can be seen as restricting the generalizability of our results. Also, while we used randomization in our experiments, it is always possible to argue that sampling error may have affected the emerging pattern. But, as far as we are ample to determine employing a post hoc examination of the power in each of the three studies, using the G*Power software (Faul, Erdfelder, Buchner, & Lang, 2009) the power ranged from .87 (Study 2) to .99 (Study 1) for various analyses we have conducted. Thus, the results our analysis suggests, that despite the small sample sizes, the findings generated high power across the studies, and therefore the effects are reliable.

Turning our attention to the more practical implications, one of the more important applied features of the work is that personalization effects (via empathy) can be extended to the entire out-group, which it can be argued carries benefits for reducing bias, as well as helping to enhance important group functions such as productivity and effectiveness. Focusing on employee–organization relations for example, the context for this research, one of the points highlighted by the series of studies is the importance of maintaining an awareness of respective similarities and differences—in an effort to view diversity between in-group and out-group as a perceived strength. In terms of implementations, utilizing opportunities to engage in personalized interactions could be facilities through forums centered on work information or conflict resolution sessions; and personalizing tasks could be structured to increase one’s own vulnerability to the other group. Generally speaking, however, if “workgroup” is not the salient reason underlying the stereotype in question, the link between “known people of a different employee groups” will be broken, and positive contact will not lead to generalized positive attitudes (towards the out-group employees). Thus, in applying such a strategy, it is essential to mediate possible wider interaction dynamics that are likely to unsettle the intervention—otherwise it would be possible to argue that the intergroup relationships might not become strong enough for empathy to enhance group attitudes.

This echoes findings reported in other domains of contact research, particularly research that has centered on promoting stronger relationships between the elderly people and younger adults (see, Holt, Christian, & Larkin, in press). In a recent review of intergeneration contact interventions, Turner, Christian, Holt, and Larkin (2010) report that empathy not only had a direct effect on positive out-group attitudes, but it also played a critical underlying role mediating the relationship between self-disclosure and positive attitudes towards the out-group. It is possible to argue that this mediation occurs, because in such intergroup situations, empathy lowers levels of intergroup anxiety (Batson, Sager, et al., 1997). That is, the more someone empathizes with an out-group member, the more likely he/she is to understand and predict the discloser’s behavior (Berger & Bradac, 1982), and the more control he/she is likely to feel over the situation, and subsequently the less anxious the participant will report feeling. In turn, the less anxiety people perceive, the more positive their attitudes towards the out-group are (Islam & Hewstone, 1993).

In closing, this paper provides experimental evidence that each component of personalization is effective in diminishing negative perceptions of out-groups and their members. Although each component was effective in reducing prejudice, there was no evidence of their differential or interactive effects. Future studies are needed to investigate whether there is any theoretical basis for expecting distinct effects of these components, or whether they might interact with each other. Our research also highlights some outstanding issues that still need redressing, primarily centering on when people are likely to adopt various personalization mechanisms, empathy over self–other comparison for example, and the situations and social pressures that are likely to lead them to do so.

Footnotes

Appendix A

Appendix B

Appendix C

Notes

References

Abele

A. E.

Wojciszke

(2007). Agency and communion from the perspective of self versus others. Journal of Personality and Social Psychology, 93, 751–763.

Aberson

C. L.

Haag

S. C.

(2007). Contact, perspective taking, and anxiety as predictors of stereotype endorsement, explicit attitudes, and implicit attitudes. Group Processes and Intergroup Relations, 10, 179–201.

Alicke

M. D.

Klotz

M. L.

Breitenbecher

D. L.

Yurak

T. J.

Vredenburg

D. S.

(1995). Personal contact, individuation, and the better-than-average effect. Journal of Personality and Social Psychology, 68, 804–825.

Allport

G. W.

(1954). The nature of prejudice. Cambridge, MA: Addison-Wesley.

Anderson

(1968). Likableness ratings of 555 personality-trait words. Journal of Personality and Social Psychology, 9, 272–279.

Aron

E. N.

Tudor

Nelson

(1991). Close relationships as including the other in the self. Journal of Personality and Social Psychology, 60, 241–253.

Batson

C. D.

(1987). Prosocial motivation: Is it ever truly altruistic? Advances in Experimental Social Psychology, 20, 65–122.

Batson

C. D.

(1991). The altruism question: Toward a social-psychological answer. Hillsdale, NJ: Lawrence Erlbaum.

Batson

C. D.

Ahmad

Stocks

E. L.

(2004). Benefits and liabilities of empathy-induced altruism. In Miller

A. G.

(Ed.), The social psychology of good and evil (pp. 359–385). New York, NY: Guilford Press.

10.

Batson

C. D.

Batson

J. G.

Todd

R. M.

Brummett

B. H.

Shaw

L. L.

Aldeguer

C. M. R.

(1995). Empathy and the collective good: Caring for one of the others in a social dilemma. Journal of Personality and Social Psychology, 68, 619–631.

11.

Batson

D. C.

Chang

Orr

Rowland

(2002). Empathy, attitudes, and action: Can feeling for a member of a stigmatized group motivate one to help the group? Personality and Social Psychology Bulletin, 28, 1656–1666.

12.

Batson

C. D.

Early

Salvarini

(1997). Perspective taking: Imagining how another feels versus imaging how you would feel. Personality and Social Psychology Bulletin, 23, 751–758.

13.

Batson

C. D.

Klein

T. R.

Highberger

Shaw

L. L.

(1995). Immorality from empathy-induced altruism: When compassion and justice conflict. Journal of Personality and Social Psychology, 68, 1042–1054.

14.

Batson

C. D.

Sager

Garst

Kang

Rubchinsky

Dawson

(1997). Is empathy-induced helping due to self–other merging? Journal of Personality and Social Psychology, 73, 495–509.

15.

Berger

C. R

Bradac

J. J.

(1982). Language and social knowledge: Uncertainty in interpersonal relationships. London, UK: Edward Arnold.

16.

Bettencourt

Brewer

Croak

Miller

(1992). Cooperation and the reduction of intergroup bias: The role of reward structure and social orientation. Journal of Experimental Psychology, 28, 301–319.

17.

Binder

Zagefka

Brown

Funke

Kessler

Mummendey

… Leyens

J. P.

(2009). Does contact reduce prejudice or does prejudice reduce contact? A longitudinal test of the contact hypothesis in three European countries. Journal of Personality and Social Psychology, 96, 843–856.

18.

Brewer

M. B.

(1988). A dual process model of impression formation. In Srull

T. K.

Wyer

R. S.

Jr. (Eds.), Advances in social cognition (pp. 1–36). Hillsdale, NJ: Erlbaum.

19.

Brewer

M. B.

Dull

Lui

(1981). Perceptions of the elderly: Stereotypes as prototypes. Journal of Personality and Social Psychology, 41, 656–670.

20.

Brewer

M. B.

Miller

(1984). Beyond the contact hypothesis: Theoretical perspectives on desegregation. In Miller

Brewer

(Eds.), Groups in contact: The psychology of desegregation (pp. 281–302). New York, NY: Academic Press.

21.

Brewer

M. B.

Miller

(1988). Contact and cooperation: When do they work? In Katz

P. A.

Taylor

D. A.

(Eds.), Eliminating racism: Profiles in controversy (pp. 315–326). New York, NY: Plenum.

22.

Cook

S. W.

(1978). Interpersonal and attitudinal outcome in cooperating interracial groups. Journal of Research and Development in Education, 12, 97–113.

23.

Crisp

R. J.

Hewstone

(2000). Crossed categorization and intergroup bias: The moderating roles of intergroup and affective context. Journal of Experimental Social Psychology, 36, 357–383.

24.

Crisp

R. J.

Hewstone

(2001). Multiple categorization and implicit intergroup bias: Differential category dominance and the positive–negative asymmetry effect. European Journal of Social Psychology, 31, 45–62.

25.

Desforges

D. M.

Lord

C. G.

Ramsey

S. L.

(1991). Effects of structured cooperative contact on changing negative attitudes toward stigmatized social groups. Journal of Personality and Social Psychology, 60, 531–544.

26.

Dumas

J. E.

Johnson

Lynch

A. M.

(2002). Likableness, familiarity, and frequency of 844 person-description words. Personality and Individual Differences, 32, 523–531.

27.

Eisenberg

Miller

P. A.

(1987). The relation of empathy to prosocial and related behaviors. Psychological Bulletin, 101, 91–119.

28.

Ensari

Miller

(2001). Decategorization and the reduction of bias in the crossed categorization paradigm. European Journal of Social Psychology, 31, 193–216.

29.

Ensari

Miller

(2002). Out-group must not be so bad after all: The effects of disclosure, typicality and salience on intergroup bias. Journal of Personality and Social Psychology, 83, 313–329.

30.

Ensari

Miller

(2005). Prejudice and intergroup attributions: The role of personalization and performance feedback. Group Processes and Intergroup Relations, 8, 391–410.

31.

Faul

Erdfelder

Buchner

Lang

A.-G.

(2009). Statistical power analyses using G*Power 3.1: Tests for correlation and regression analyses. Behavior Research Methods, 41, 1149–1160.

32.

Fazio

R. H.

Zanna

M. P.

(1978). Attitudinal qualities relating to the strength of the attitude–behavior relationship. Journal of Experimental Social Psychology, 14, 398–408.

33.

Feshbach

N. D.

Roe

(1968). Empathy in six- and seven-year-olds. Child Development, 39, 133–145.

34.

Finlay

K. A.

Stephan

W. G.

(2000). Improving intergroup relations: The effects of empathy on racial attitudes. Journal of Applied Social Psychology, 30(8), 1720–1737.

35.

Fiske

S. T.

(1981). Social cognition and affect. In Harvey

(Ed.), Cognition, social behavior, and the environment (pp. 227–264). Hillsdale, NJ: Erlbaum.

36.

Fiske

S. T.

Cuddy

Glick

(2006). Universal dimensions of social cognition: Warmth and competence. Trends in Cognitive Sciences, 11, 77–83.

37.

Gaertner

S. L.

Dovidio

J. F.

(2000). Reducing intergroup bias: The common ingroup identity model. New York, NY: Psychology Press.

38.

Gaertner

S. L.

Rust

M. C.

Mottola

G. R.

Dovidio

J. F.

Nier

J. A.

Banker

B. S.

… Houlette

(1999). Reducing intergroup bias: Elements of intergroup cooperation. Journal of Personality and Social Psychology, 76, 388–402.

39.

Galinsky

A. D.

Moskowitz

G. B.

(2000). Perspective-taking: Decreasing stereotype expression, stereotype accessibility, and in-group favoritism. Journal of Personality and Social Psychology, 78, 708–724.

40.

Hewstone

Brown

(1986). Contact is not enough: An intergroup perspective on the “contact hypothesis.” In Hewstone

Brown

(Eds.), Contact and conflict in intergroup encounters (pp. 1–44). New York, NY: Blackwell.

41.

Hewstone

Fincham

Jaspars

(1981). Social categorization and similarity in intergroup behavior: A replication with penalties. European Journal of Social Psychology, 12, 241–269.

42.

Hogan

(1969). Development of an empathy scale. Journal of Consulting and Clinical Psychology, 33, 307–316.

43.

Holt

Christian

Larkin

(in press). Maintaining connectedness: Exploring the experiences of older homeless men living in hostel accommodation. Journal of Community and Applied Social Psychology.

44.

Howard

J. W.

Rothbart

(1980). Social categorization and memory for in-group and out-group behavior. Journal of Personality and. Social Psychology, 38, 301–310.

45.

Islam

Hewstone

(1993). Intergroup attributions and affective consequences in majority and minority groups. Journal of Personality and Social Psychology, 64, 936–950.

46.

Kanouse

D. E.

Hanson

L. R.

Jr. (1971). Negativity in evaluations. In Jones

E. E.

Kanouse

D. E.

Kelley

H. H.

Nisbett

R. E.

Valins

Weiner

(Eds.), Attribution: Perceiving the causes of behavior (pp. 47–61). Morristown, NJ: General Learning Press.

47.

MacKinnon

D. P.

Lockwood

C. M.

Hoffman

J. M.

West

S. G.

Sheets

(2002). A comparison of methods to test mediation and other intervening variable effects. Psychological Methods, 7, 83–104.

48.

Miller

(2002). Personalization and the promise of contact theory. Journal of Social Issues, 58, 387–410.

49.

Mummendey

Otten

(1998). Positive–negative asymmetry in social discrimination. European Review of Social Psychology, 9, 107–143.

50.

Myers

D. G.

(2001). Psychology. Holland, MI: Worth.

51.

Pettigrew

T. F.

(1997). Generalized intergroup contact effects on prejudice. Personality and Social Psychology Bulletin, 23, 173–185.

52.

Pettigrew

T. F.

Tropp

L. R.

(2006). A meta-analytic test of intergroup contact theory. Journal of Personality and Social Psychology, 90, 751–783.

53.

Pettigrew

T. F.

Tropp

L. R.

(2008). How does intergroup contact reduce prejudice? Meta-analytic tests of three mediators. European Journal of Social Psychology, 38, 922–934.

54.

Preachers

K. J.

Hayes

A. F.

(2004). SPSS and SAS procedures for estimating indirect effects in simple mediation models. Behavior Research Methods, Instruments and Computer, 36, 717–731.

55.

Rhodes

Halberstadt

Brajkovich

(2001). Generalization of mere exposure effects to averaged composite faces. Social Cognition, 19, 57–70.

56.

Rothbart

John

(1985). Social categorization and behavioral episodes: A cognitive analysis of the effects of intergroup contact. Journal of Social Issues, 41, 81–104.

57.

Sedikides

Campbell

W. K.

Reeder

G. D.

Elliot

A. J.

(1998). The self-serving bias in relational context. Journal of Personality and Social Psychology, 74, 378–386.

58.

Sobel

(1982). Asymptotic confidence intervals for indirect effects in structural equations models. In Leinhart

(Ed.), Sociological methodology (pp. 290–312). San Francisco, CA: Jossey-Bass.

59.

Stephan

W. G.

Finlay

K. A.

(1999). The role of empathy in improving intergroup relations. Journal of Social Issues, 55(4), 729–743.

60.

Stephan

C. W.

Stephan

W. G.

(1985). Two social psychologies: An integrative approach. Homewood, IL: Dorsey.

61.

Stotland

(1969). The psychology of hope: An integration of experimental, clinical, and social approaches. San Francisco, CA: Jossey-Bass.

62.

Toi

Batson

C. D.

(1982). More evidence that empathy is a source of altruistic motivation. Journal of Personality and Social Psychology, 43, 281–292.

63.

Turner

R. N.

Christian

Holt

Larkin

(2010). Does intergenerational contact reduce ageism? Working paper, University Of Birmingham.

64.

Underwood

Moore

(1982). Perspective taking and altruism. Psychological Bulletin, 91, 143–173.

65.

Van Oudenhoven

J. P.

Willemsma

Prins

K. S.

(1996). Integratie en assimilatie van Marokkanen, Surinamers en Turken in Nederland [Integration and assimilation of Moroccan, Surinamese and Turkish people in the Netherlands]. De Psycholoog, 31, 468–471.

66.

Vescio

T. K.

Sechrist

Paolucci

M. P.

(2003). Perspective taking and prejudice reduction: The mediational role of empathy arousal and situational attributions. European Journal of Social Psychology, 33, 455–472.

67.

Vorauer

J. D.

Martens

Sasaki

S. J.

(2009). When trying to understand detracts from trying to behave: Effects of perspective-taking in intergroup interaction. Journal of Personality and Social Psychology, 96, 811–827.

68.

Weigel

R. H.

Wiser

P. L.

Cook

S. W.

(1975). The impact of cooperative learning experiences on cross-ethnic relations and attitudes. Journal of Social Issues, 31, 219–244.

69.

Wenzel

Mummendey

(1996). Positive–negative asymmetry of social discrimination: A normative analysis of differential evaluations of in-group and out-group on positive and negative attributes. British Journal of Social Psychology, 35, 493–507.

70.

Weyant

(2007). Perspective taking as a means of reducing negative stereotyping of individuals who speak English as a second language. Journal of Applied Social Psychology, 37, 703–716.

71.

Wilder

D. A.

(1978). Reduction of intergroup discrimination through individuation of the out-group. Journal of Personality and Social Psychology, 36, 1361–1374.

72.

Wilder

D. A.

(1984). Intergroup contact: The typical member and the exception to the rule. Journal of Experimental Social Psychology, 20, 177–194.

73.

Worchel

(1986). The role of co-operation in intergroup conflict. In Worchel

Austin

(Eds.), Psychology and intergroup relations (pp. 288–304). Chicago, IL: Nelson-Hall.