Group Processes & Intergroup Relations http://gpi.sagepub.com/ A meta-analytic test of the imagined contact hypothesis Eleanor Miles and Richard J. Crisp Group Processes Intergroup Relations 2014 17: 3 DOI: 10.1177/1368430213510573 The online version of this article can be found at: http://gpi.sagepub.com/content/17/1/3 Published by: http://www.sagepublications.com Additional services and information for Group Processes & Intergroup Relations can be found at: Email Alerts: http://gpi.sagepub.com/cgi/alerts Subscriptions: http://gpi.sagepub.com/subscriptions Reprints: http://www.sagepub.com/journalsReprints.nav Permissions: http://www.sagepub.com/journalsPermissions.nav Citations: http://gpi.sagepub.com/content/17/1/3.refs.html >> Version of Record - Dec 25, 2013 What is This? Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 510573 2013 17110.1177/1368430213510573Group Processes & Intergroup RelationsMiles and Crisp G Group Processes & P Intergroup Relations I Article R Group Processes & Intergroup Relations A meta-analytic test of the 2014, Vol 17(1) 3­–26 © The Author(s) 2013 Reprints and permissions: imagined contact hypothesis sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/1368430213510573 gpir.sagepub.com Eleanor Miles1 and Richard J. Crisp2 Abstract Imagined intergroup contact (Crisp & Turner, 2009) is a new indirect contact strategy for promoting tolerance and more positive intergroup relations. Despite its relatively recent inception, there have now been over 70 studies showing that imagining a positive interaction with an outgroup member can reduce prejudice and encourage positive intergroup behavior. With this meta-analysis, we provide the first quantitative review of imagined contact effects on four key measures of intergroup bias: attitudes, emotions, intentions, and behavior. We also test for moderators arising from both group and study design characteristics. The analysis found that imagined contact resulted in significantly reduced intergroup bias across all four dependent variables (overall d+ = 0.35). The effect was significant for both published and unpublished studies, and emerged across a broad range of target outgroups and contexts. The effect was equally strong for explicit and implicit attitude measures, but was stronger on behavioral intentions than on attitudes, supporting the direct link between imagery and action proposedly underlying mental simulation effects. Most design characteristics had no significant impact, including valence of the imagined interaction, type of control condition, and time spent imagining contact. However, the more participants were instructed to elaborate on the context within which the imagined interaction took place, the stronger the effect. The imagined contact effect was also stronger for children than for adults, supporting the proposition that imagined contact is a potentially key component of educational strategies aiming to promote positive social change. Keywords imagined contact, intergroup contact, mental simulation, prejudice Paper received 8 August 2013; revised version accepted 4 October 2013. When you visualised a man or a woman 1 carefully … when you saw the lines at the University of Sussex, UK 2 corners of the eyes, the shape of the mouth, University of Sheffield, UK how the hair grew, it was impossible to hate. Corresponding author: Hate was just a failure of imagination. Eleanor Miles, School of Psychology, University of Sussex, Graham Greene, The Power and the Glory. Brighton BN1 9QH, UK. Email: e.miles@sussex.ac.uk Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 4 Group Processes & Intergroup Relations 17(1) Researchers developing prejudice-reduction inter- Imagined contact may, for instance, make people ventions have a powerful tool at their disposal in more likely to seek out and seize opportunities Gordon Allport’s contact hypothesis (1954). This for contact. It may also improve the quality of hypothesis proposes that contact between groups direct contact by preparing them to engage in reduces prejudice, a prediction which has been these interactions with a positive and open mind. confirmed across more than 500 studies (Petti- In turn, this may make it more likely that future grew & Tropp, 2006). Intergroup contact works direct contact will meet the conditions Allport via a range of cognitive and affective processes believed were necessary for prejudice reduction including reduced anxiety, increased self-disclosure, (equal status, common goals, a co-operative envi- and increased trust, and it is unquestionably inte- ronment, and support from authority). The idea gral to efforts aimed at improving intergroup that imagined contact prepares people for future relations (see Hodson & Hewstone, 2013). But contact can be seen as a logical extension of what if contact is prevented, either by physical or Allport’s original theorizing. Indeed, Crisp and psychological barriers? In many of the situations Turner (2012) note that in The Nature of Prejudice where intergroup conflict is most pervasive, geo- Allport discusses the potential of “fantasy level” graphical or social segregation means that contact contact as an effective first step to improving between groups is either impossible or unlikely, intergroup relations, particularly when “realistic and there are many reasons why people may not discussion” between groups could constitute a seize opportunities for contact even when they threat (1954, p. 453). Where contact encounters become available. In recent years, however, it has are possible, but where groups do not act upon become apparent that the contact hypothesis has these opportunities, imagined contact may there- an application that transcends face-to-face inter- fore be a critical first step towards establishing action. The imagined intergroup contact hypoth- direct contact. esis (Crisp & Turner, 2009) proposes that the very concept of contact, mentally articulated in the form of an imagined interaction, can have Processes and Outcomes a positive impact on intergroup perception and Since its inception, research has shown positive behavior. There have now been over 70 studies effects of imagined contact on intergroup atti- of imagined contact, testing its impact on a range tudes, emotions, behavioral intentions, and of measures related to the reduction of prejudice behavior. There is substantial evidence showing and intergroup bias. In this meta-analysis, we pro- that imagined contact, especially when positively vide the first quantitative review of these imag- toned, has beneficial effects on intergroup rela- ined contact effects. tions. For instance, early enquiries demonstrated that it improves intergroup attitudes (Turner, Crisp, & Lambert, 2007), perceptions of outgroup variability Imagined Intergroup Contact (Turner et al., 2007), and enhances projection of Imagined intergroup contact is defined as “the positive traits to the outgroup (Stathi & Crisp, 2008). mental simulation of a social interaction with a Subsequent studies found that it fosters more member or members of an outgroup category” positive intentions to engage in outgroup contact (Husnu (Crisp & Turner, 2009, p. 234). Crisp and Turner & Crisp, 2010a, 2010b), increases self-efficacy con- argued that imagined contact may be valuable as cerning future outgroup contact (Stathi, Crisp, & Hogg, an application of the contact hypothesis where 2011), and facilitates outgroup trust (Pagotto, actual contact is impossible or unlikely; for exam- Visintin, De Iorio, & Voci, 2012; Vezzali, ple, where there are a lack of opportunities to Capozza, Stathi, Giovannini, 2012). Other studies meet people from other groups. However, they have shown it reduces negative aspects of out- also noted that imagined contact may be valuable group evaluation such as intergroup anxiety (Birtel as a means of preparing people for future contact. & Crisp, 2012b; Husnu & Crisp, 2010a; Turner Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 5 et al., 2007; West, Holmes, & Hewstone, 2011), on four key dependent measures: attitudes, emo- infrahumanization of the outgroup (Vezzali, Capozza, tions, intentions, and behavior. We also examine Stathi, et al., 2012), negative stereotypes (Brambilla, moderating conditions that may limit or enhance Ravenna, & Hewstone, 2012; Cameron, Rutland, its effectiveness. To identify these moderators, we Turner, Holman-Nicolas, & Powell, 2011; Stathi, draw upon key theoretical predictions, as well as Tsantila, & Crisp, 2012) and stereotype threat the contributions of other researchers. We were (Abrams et al., 2008). Research has also revealed able to assess the impact of multiple moderators that imagined contact can combat subtle forms relating both to group-based characteristics (i.e., of bias such as implicit prejudice (Turner & Crisp, variability relating to participants or to the out- 2010; Vezzali, Capozza, Giovannini, & Stathi, group being imagined) and design-based charac- 2012) and negative nonverbal behaviors (Birtel & teristics (i.e., variability relating to the imagined Crisp, 2012a; Turner & West, 2012). contact manipulation). We discuss how we identi- As well as a large number of supportive stud- fied and assessed these characteristics next. ies, imagined contact has stimulated a great deal of debate. With this meta-analysis, we address some of these critiques. For example, the “real Group Characteristics world” significance of the effect has been ques- Is imagined contact more effective for some outgroups than tioned (Lee & Jussim, 2010), and others have for others?  Pettigrew and Tropp (2006) found that questioned whether the effect is present for non- the beneficial effect of direct contact “applies self-report measures, suggesting that it is subject beyond racial and ethnic groups to embrace other to demand characteristics (Bigler & Hughes, types of groups as well” (p. 768). In order to 2010). While these issues have been debated from determine whether the same is true for imagined a conceptual standpoint (Crisp, Birtel, & Meleady, contact, we coded the type of outgroup partici- 2011; Crisp & Turner, 2010), with this meta-anal- pants were asked to imagine contact with. All but ysis we provide a quantitative riposte. For five studies could be categorized into the out- instance, if imagined contact has significant groups of ethnic group, nationality, mental illness, dis- effects when its outcomes are measured implic- ability, age, sexual orientation, religion, or weight. itly, this would suggest the effect cannot be attrib- utable to demand characteristics. If it has Do age, gender, and nationality of participants predict the significant effects on actual behavior towards effectiveness of imagined contact?  We coded the par- outgroup members, then the implications for ticipant characteristics of age, gender, and nation- real-world interaction are clear. ality as potential moderators. Our decision to examine these factors as moderators was guided Aims of the Present by the diversity of participants represented in our sample, and our aim was to discover whether Meta-Analysis imagined contact was effective across all partici- Previous theoretical and narrative reviews have pant groups. Of particular note, recent evidence established key principles of imagined contact shows that imagined contact improves intergroup (Crisp & Turner, 2009), documented its impact relations not only among adults but also among on behavioral intentions (Crisp, Husnu, Meleady, children (Cameron et al., 2011; Vezzali, Capozza, Stathi, & Turner, 2010), proposed both cognitive Giovannini, et al., 2012; Vezzali, Capozza, Stathi, and affective pathways through which it has a et al., 2012). Given that the formative years of positive impact (Crisp & Turner, 2012; Crisp prejudice development are school years (Cam- et al., 2010), and elucidated links with Allport’s eron & Rutland, 2006; Cameron, Rutland, Brown, original theorizing (Crisp & Turner, 2012). With & Douch, 2006) and the clear appeal of imagery this meta-analysis, we provide the first quantitative interventions for children, this is an obvious place review of the effectiveness of imagined contact to intervene with imagined contact. Thus, it is Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 6 Group Processes & Intergroup Relations 17(1) important to establish its effectiveness within this nonoutgroup member, these explanations are group. We therefore tested for differential effec- effectively controlled for, representing a more tiveness of imagined contact on adult partici- stringent test of the hypothesis (and, potentially, a pants as compared to child participants, as well as smaller effect size). We examined the control con- coding age as a continuous variable to test for any ditions in our included studies, and found that changes in the effectiveness of imagined contact they could be categorized into four types: (a) across the lifespan. imagine contact with a nonoutgroup member, (b) imagine a neutral scene, (c) think about the out- group, and (d) no task. Design Characteristics Is positive imagined contact more effective than neutral Does elaboration improve the effectiveness of imagined imagined contact?  Based on findings that quality of contact?  Husnu and Crisp (2010a) found that an contact is particularly important in order for it to elaborated version of the imagined contact task benefit intergroup relations (e.g., Eller & Abrams, (where participants thought about specifically 2004; Stathi & Crisp, 2010; Voci & Hewstone, when and where the imagined interaction took 2003), studies of imagined contact often specify place) enhanced the effect of imagined contact that participants should imagine a positive inter- on anxiety, attitudes, and intentions towards the action. The importance of this has been argued outgroup. The reason why elaborated imagined theoretically (Crisp & Turner, 2009), and empiri- contact should enhance intentions was derived cal studies comparing positive contact with neu- from research showing that when we make more tral contact also support this idea (Stathi & Crisp, detailed plans, this provides an available behavio- 2008; West et al., 2011). Specifying a positive ral script that can provide the cognitive roadmap interaction may be important because it guards for future behaviors (Gollwitzer, 1993). The elab- against a possible negative tone, which might orated instructions should therefore have a emerge if participants were given no direction greater impact on intentions because they help and relied upon negative stereotypes as a basis for participants to create a more cue-rich simulation the imagined interaction (cf. West et al., 2011). that makes the imagined behavior subsequently Thus, we coded whether participants were asked more available at the judgmental phase. More to imagine a positive interaction, or whether this general research on mental simulation supports was not specified. this hypothesized relationship between elabora- tion and future behavior. For many situations we Does it matter what participants in the control condition have a behavioral script, and the more elaborate do?  When calculating effect sizes on the basis of a and detailed the script, the stronger its impact on comparison between two groups, it is important subsequent attitudes and behavior (Anderson, to consider the quality of the control condition. If 1983; Ross, Lepper, & Hubbard, 1975). There- an intervention is very effective when compared fore, it seems likely that studies which include to one type of control group, but less effective greater elaboration will obtain larger effects. We when compared to another, this may suggest pos- found that studies varied in a number of ways in sible alternative explanations. For example, where terms of the level of elaboration specified by one group is asked to imagine a positive interac- their imagined contact instructions, and coded tion with an outgroup member and is then com- four variables accordingly: the amount of detail pared with another group which does nothing, provided about the situation or context of the any differences in intergroup bias between the imagined interaction, the amount of detail speci- two groups could be attributable to the general fied about the outgroup target, the amount of time effect of imagined social interaction, or to posi- participants spent imagining contact, and whether tive affect. However, if the second group is asked participants described what they had imagined after to imagine a positive interaction with a the manipulation. Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 7 Publication Bias, or the File Drawer intergroup bias. We employed the following inclu- Problem sion criteria in order to select these studies. Finally, as with all reviews, our overall effect sizes Criterion 1.  The study included an experimental need to be considered in light of the “file drawer manipulation of imagined contact, as defined by problem” (Rosenthal, 1979)—the likelihood that Crisp and Turner (“the mental simulation of a additional studies have been conducted on imag- social interaction with a member or members of ined contact, but neither been published nor an outgroup category”; 2009, p. 234). In order to made available to us. To quantify the impact of be eligible, the manipulation had to include both the file-drawer problem on our findings, we the mental simulation and the interaction compo- sought both published and unpublished studies nents. Therefore, studies where contact was not (with the final sample containing 50% of each real, but where participants believed it was, were type), and coded publication status as a potential excluded because they did not include the simula- moderator of effect size. We also conducted a tion component (e.g., Finchilescu, 2010; Vorauer, number of different analyses to assess the poten- Hunter, Main, & Roy, 2000, Study 2; e.g., interact- tial size and impact of publication bias on our set ing with an outgroup member via an Internet of studies, including fail-safe N, Egger’s regres- chat room, where the outgroup members’ sion, and trim and fill analyses. responses were programmed via computer). Fur- thermore, studies in which participants mentally simulated an outgroup member but did not imag- Summary of Aims ine interacting with that outgroup member were In this meta-analysis, we draw upon 71 published excluded because they did not include the interac- and unpublished studies to present a comprehen- tion component (e.g., taking the perspective of an sive assessment of the size and variability of outgroup member, Todd et al., 2011; imagining imagined contact effects, their relative size across being in the presence of an outgroup member, different measures of intergroup bias, their uni- Desforges et al., 1997; imagining a counterstereo- versality across participant groups and outgroups, typic outgroup member, Blair, Ma, & Lenton, and whether there are any necessary or facilitating 2001). We also further specified that the interac- conditions in order for them to occur. tion must occur between the participant and the outgroup member; therefore, studies in which participants mentally simulated social interactions Method that did not involve themselves were excluded (e.g., watching or reading about ingroup–out- Inclusion Criteria group interactions; Cameron et al., 2006; Mazzi- We adopted an inclusive approach when searching otta, Mummendey, & Wright, 2011). Thus, we for studies, both to ensure that our effect sizes limited our analysis to studies which met the strict represented a comprehensive assessment of the definition of imagined intergroup contact, and effectiveness of imagined contact, and to address excluded studies on various related constructs concerns over whether the imagined contact such as counterstereotypic mental imagery (Blair effect is only apparent for certain groups of peo- et al., 2001), perspective taking (Todd, Hanko, ple or under certain circumstances (Bigler & Galinsky, & Mussweiler, 2011), vicarious contact Hughes, 2010). Thus, we included any study, pub- (Mazziotta et al., 2011) and experimental extended lished or unpublished, which randomly assigned contact (Cameron et al., 2006). one group of participants to imagine positive or neutral contact with an outgroup, assigned another Criterion 2.  As there is a theoretical and empirical group of participants to complete an alternative basis to suggest that negative imagined contact is task, and included a subsequent measure of ineffective or even harmful (Crisp & Turner, Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 8 Group Processes & Intergroup Relations 17(1) 2009; see also Harwood, Paolini, Joyce, Rubin, & We also conducted an extensive search for Arroyo, 2011; West et al., 2011), we did not unpublished work. To obtain unpublished include any studies or conditions in which the research, we (a) searched ProQuest Dissertations interaction was explicitly negative. and Theses using the search terms described ear- lier; (b) contacted the authors of relevant confer- Criterion 3.  The study had to include a compari- ence papers; and (c) contacted the email listservs son condition, in which participants completed of major social psychological societies (Social for any task which did not involve imagined contact Personality and Social Psychology, Society of with the same outgroup. This criterion was Experimental Social Psychology, Society for the employed in order to ensure that all our effect Psychological Study of Social Issues) to request sizes were comparable, in that they represented unpublished or in press work. the effectiveness of imagined contact on inter- Finally, we employed ancestry and descendancy group bias, rather than the relative effectiveness approaches (Johnson & Eagly, 2000) to ensure that of one type of imagined contact versus another. our search included all relevant studies. The refer- Studies in which all participants imagined contact ence lists of all papers included in the meta-analysis with the same outgroup (e.g., Babbitt & Som- were examined, as were later citations of each mers, 2011; Husnu & Crisp, 2011, Experiment 2; paper (retrieved using the Social Sciences Citation Kuchenbrandt, Eyssel, & Seidel, 2013) were Index and Google Scholar), and later citations of therefore excluded, as they did not allow us to key theoretical imagined contact papers. calculate an effect size representing the effective- Our search identified 6,490 papers, theses, and ness of imagined contact. unpublished works. Titles and abstracts were reviewed independently by two coders in order to Criterion 4.  The study had to include at least one identify potentially relevant studies, and full text measure indicative of intergroup bias, taken after articles were then reviewed independently by two the imagined contact manipulation. Eligible coders in order to determine eligibility. In total, dependent variables were measures of emotion, we were able to compute effect sizes for 71 inde- attitude, behavior, or intended behavior towards pendent tests of imagined contact versus a con- the outgroup (see the following for specific infor- trol condition which met our inclusion criteria. mation concerning each category). Of these tests, 34 were taken from 24 published papers, and 37 were taken from unpublished studies. We were able to calculate precise effect Search Strategy sizes based on means and standard deviations for Our primary strategy was to search social scien- the vast majority of these (67 studies, or 94% of tific databases (PsycInfo, Web of Knowledge) for the total sample), either from information pro- studies published before the 6th of June 2013 and vided in the report or from correspondence with containing any term related to intergroup contact the authors. For the remainder (6%), effect sizes (e.g., contact, interaction, intergroup, outgroup), were estimated using summary statistics, such as t as well as any term related to mental simulation values. This resulted in k = 71 separate tests of (e.g., imagine, mental simulation, mental imagery). imagined contact effects, which were included in To ensure that we did not omit any study which the meta-analysis. Table 1 presents the effect sizes used nonstandard terms to describe an imagined and characteristics for each of these tests. contact manipulation, we conducted additional searches including more general terms which Selection of Comparisons Within might be used to describe imagined contact (e.g., hypothetical, simulated, vicarious) in conjunction Studies with terms relating to both contact and outgroups The basis of our effect sizes was a comparison (e.g., outgroup, ingroup, intercultural, prejudice, between an imagined contact condition and a con- disability, ethnic, nationality, schizophrenia). trol condition. Where studies included more than Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 9 one experimental or more than one control condi- about the outgroup (e.g., imagining a neutral tion, we adopted a systematic approach in either scene). We then empirically assessed the influence selecting or averaging those conditions to com- of type of control condition on effect sizes pute a single effect size. This was because includ- through moderator analyses. ing two separate effect sizes would violate the assumptions of independence in meta-analysis. Coding of Dependent Measures and For studies that included more than one manipulation of imagined contact, we adopted Calculation of Effect Sizes the recommended approach of combining Again, we adopted an inclusive approach when groups in order to create one pair-wise compari- selecting eligible dependent variables, to provide son (Higgins & Green, 2009). In other words, we the broadest possible test of the imagined con- allowed all manipulations of imagined contact to tact hypothesis. Thus, any measure related to contribute to the final effect size, whether or not intergroup bias was included in our analysis. As they were hypothesized to be effective (e.g., West imagined contact may not have the same effect & Bruckmüller, 2013, presented imagined con- across different components of intergroup bias, tact instructions in either an easy-to-read font or we coded dependent variables into four catego- a hard-to-read font, proposing that imagined ries, to enable us to assess whether imagined con- contact would be ineffective in the latter condi- tact was equally effective across these categories. tion; we included both of these conditions when These categories were attitudes, emotions, inten- computing our effect size). In order to compute tions, and behavior towards the outgroup. combined effect sizes, individual meta-analyses The category attitudes towards the outgroup were performed on data from each study, to included explicit measures of attitudes towards ensure that summary effect sizes were correctly the outgroup (both cognitive and affective), weighted by the number of participants in each measures of implicit attitudes (e.g., implicit asso- condition. ciation tasks), measures indicative of general out- However, where studies included multiple group evaluation (e.g., feeling thermometers), control conditions, we did not simply average ratings of specific outgroup characteristics (e.g., these, as the most stringent test of the imagined perceived variability, warmth, competence, dan- contact hypothesis was provided by a more selec- gerousness), measures assessing the relationship tive approach. Theoretical predictions suggest between the participant and the outgroup (e.g., that certain control conditions are likely to inflate perceived commonality, inclusion of other in self, effect sizes; for example, simply thinking about social distance), projection of positive traits to the target group (Turner et al., 2007, Study 2) may the outgroup, and measures of stereotyping. actually have a negative effect on intergroup bias The category emotions towards the outgroup (by priming outgroup stereotypes), thus increas- included measures of intergroup anxiety, ratings ing the observed effect size for imagined contact. of other intergroup emotions (e.g., trust, anger, Therefore, where studies included multiple con- fear), and ratings of general positive or negative trol conditions, we selected the condition that affect towards the outgroup. most closely resembled the imagined contact The category intended behavior towards the out- condition, and used this as the basis for our com- group included measures of future contact inten- parison. Where possible, we selected a control tions, intentions to help an outgroup member, condition in which participants imagined contact contact self-efficacy, perceived importance of with a nonoutgroup member (e.g., for Chen & contact, self-disclosure, perceived tolerance, Mackie, 2013, we chose “imagine contact with a motivation to respond without prejudice, stranger” rather than “think about Muslims”); approach and avoidance tendencies, and meas- otherwise, where possible, we selected another ures concerning future interactions (e.g., antici- control condition which did not involve thinking pated enjoyment, uncertainty). Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 10 Group Processes & Intergroup Relations 17(1) Table 1.  Characteristics and effect sizes for studies included in the meta-analysis. Study Experiment Country Outgroup N E NC Effect sizes included Att Emot Bhvr Int Asbrock (2012a) 1 - Other 34 34 0.09 0.34 Asbrock (2012b) 1 - Weight 51 61 0.16 −0.16 Asbrock (2012c) 1 - Mental illness 29 31 −0.11 0.3 Asbrock et al. (2013) 1 Germany Nationality 62 58 0.34   Asbrock et al. (2013) 2 Germany Ethnic group 29 29 0.36 Bajrektarevic et al. 1 - Nationality 27 29 0.24   (n.d.) Bergeron (2012) 1 USA Religion 38 37 0.68 0.87   Birtel & Crisp (2011) 1 UK Religion 17 15 −0.37   Birtel & Crisp (2011) 5 UK Sexual 30 31 −0.13 −0.29 0.1 orientation Birtel & Crisp (2011) 7 UK Religion 39 39 −0.56 0.57 Birtel & Crisp (2011) 9 UK Religion 22 24 0.2 0.21 0.1 Birtel & Crisp (2011) 10 UK Disability 21 19 0.09 0.48 Birtel & Crisp (2011) 11 UK Religion 32 32 0.27 0.46 0.13 Birtel & Crisp (2011) 12 UK Nationality 18 18 −0.07 −0.03 Birtel & Crisp (2011) 13 UK Disability 30 31 −0.07 0.33 Brambilla et al. (2012) 1 Italy Nationality 65 58 0.49   Broad (2011) 1 UK Mental illness 40 40 1.03 −0.45   Cameron et al. (2011) 1 UK Disability 63 60 0.39 0.41 Capozza et al. (2013) 1 Italy Disability 92 88 0.03   Chen, Cook, et al. 1 USA Weight 72 26 0.07   (2013) Chen & Mackie 2 USA Religion 60 61 0.02   (2013) Chen, Richards, et al. 3 USA Other 75 36 −0.03   (2013a) Chen, Richards, et al. 4 USA Other 100 53 −0.01   (2013b) Crisp & Husnu 1 UK Age 30 30 0.64 0.79 (2011) Frye et al. (2012) 1 USA Sexual 25 19 1.02   orientation Giacobbe et al. 1 Australia Mental illness 28 27 0.19 0.69   (2013) Harwood et al. (2011) 1 USA Nationality 42 48 0.4   Hughes et al. (2013) 1 - Ethnic group 45 44 0.12   Hughes et al. (2013) 1 - Age 52 52 0.59   Husnu & Crisp 1 UK Religion 16 17 0.84 (2010a) Husnu & Crisp 1 Cyprus Ethnic group 60 30 0.51 (2010b) Japhet (2010) 1 UK Mental illness 26 53 0.11 0.24 0.05 Jaworska et al. (2012) 1 Poland Religion 42 44 −0.12 −0.06 Jaworska et al. (2012) 2 Poland Ethnic group 39 42 −0.25   Jaworska et al. (2013) 1 Poland Religion 63 62 0.12 0.11   Jaworska et al. (2013) 2 Poland Ethnic group 36 40 0.01   Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 11 Table 1. (Continued) Study Experiment Country Outgroup N E NC Effect sizes included Att Emot Bhvr Int Kuchenbrandt & 1 Germany Other 26 20 0.69 1.09   Eyssel (2012) Lai et al. (2013) 1 USA Ethnic group 272 216 0.05 Menkinoska (2011) 1 Australia Mental illness 44 22 0.41   Miller et al. (2013) 1 USA Sexual 37 39 0.57 0.08 0.05 orientation Pagotto et al. (2012) 1 Italian Religion 40 19 0.35 0.18 0.3   Slater (2011) 1 UK Weight 21 23 0.25 0.86   Stathi et al. (in press) 1 UK Ethnic group 64 65 0.42 0.38 Stathi & Crisp (2008) 2 UK Nationality 31 28 −0.13   Stathi & Crisp (2008) 3 UK Nationality 49 49 0.08   Stathi et al. (2011) 1 UK Religion 16 16 0.8 Stathi et al. (2012) 1 UK Mental illness 23 24 0.8 0.53 0.69 Turner & Crisp 1 UK Age 13 12 0.99   (2010) Turner et al. (2007) 1 UK Age 14 14 0.05   Turner et al. (2007) 2 UK Age 12 12 0.94   Turner et al. (2007) 3 UK Sexual 14 13 0.82 1.43   orientation Turner & West (2012) 1 UK Weight 25 25 0.58   Turner & West (2012) 2 UK Religion 20 21 0.77 0.77   Turner et al. (2013) 1 UK Nationality 18 18 1.2 0.97 0.97 Turner et al. (2013) 2 - Sexual 20 21 0.89 1.07 0.92 orientation Vezzali, Capozza, 1 Italy Nationality 22 22 0.68 0.94 Giovannini, et al. (2012) Vezzali, Capozza, 1 Italy Nationality 17 17 0.31 1.68 1.28 Stathi, et al. (2012) Vezzali, Crisp, et al. 1 Italy Nationality 15 22 −0.02 0.68 −0.08   (2013) Vezzali, Crisp, et al. 2 - Nationality 19 21 0.4 0.55 (2013) Vezzali, Stathi, Crisp, 1 Italy Disability 43 38 1.36 1.27 1.3 & Capozza (2013) Vezzali, Stathi, Crisp, 2 Italy Nationality 29 31 1.14 0.46 0.72 & Capozza (2013) Vezzali, Stathi, Crisp, 1 Italy Other 10 13 1.11 0.98 Giovanni, & Capozza (2013) Vezzali, Stathi, Crisp, 2 Italy Nationality 53 52 0.58 0.83 Giovanni, Capozza, & Gaertner (2013) Vezzali, Stathi, Crisp, 1 Italy Nationality 50 22 0.54 Giovanni, Capozza, & Gaertner (2013) Wallace (2010) 1 UK Mental illness 30 30 1.03 0.79   (Continued) Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 12 Group Processes & Intergroup Relations 17(1) Table 1. (Continued) Study Experiment Country Outgroup N E NC Effect sizes included Att Emot Bhvr Int West & Bruckmüller 1 UK Mental illness 376 132 0.02   (2013) West & Bruckmüller 2 Germany Religion 50 51 −0.05   (2013) West et al. (2011) 1 UK Mental illness 44 43 −0.28 −0.69   West et al. (2011) 2 UK Mental illness 24 25 0.73 0.77   West et al. (2011) 3 UK Mental illness 19 19 0.94 0.77   West et al. (2011) 4 UK Mental illness 23 24 0.65 0.77   Note. Country = nationality of participants, or country where study took place; Outgroup = group with which participants imagined contact; NE = number of participants in the experimental condition; NC = number of participants in the control condition; Att = effect size of imagined contact on attitudes; Emot = effect size of imagined contact on emotions; Bhvr = effect size of imagined contact on behavior; Int = effect size of imagined contact on intended behavior. Finally, the category behavior towards the outgroup This method would also be problematic from a included measures of actual behavior towards statistical point of view, as meta-analysis assumes outgroup members (whether self-reported or that each effect size is independent, whereas two observed), such as seating distance from an out- measures taken from the same study will clearly group member, making the decision to work with share variance with one another. The options, as an outgroup member, self-disclosure to an out- with multiple conditions, are either to select one group member, time spent with outgroup mem- dependent variable or to compute an average. bers, and number of outgroup friends formed Consistent with our inclusive approach, we chose after the intervention. to include all eligible dependent variables and to Decisions about which measure belonged in compute summary effect sizes. We first com- which category were made independently by two puted up to four summary effect sizes for each postgraduate coders, who agreed with high relia- study, representing the average effect of imagined bility (Cohen’s κ = 0.94). Effect sizes for each contact on attitudes, emotion, intentions, and measure were also calculated independently by behavior, as applicable. Then, we averaged these both coders (their effect size calculations differed effect sizes into a single overall effect size for by less than d = 0.02). In both cases, disagree- each study, representing an estimate of the effect ments or discrepancies were resolved by the first of imagined contact across all measures of inter- author. For all dependent variables, effect sizes group bias, with effects on attitude, emotion, were coded such that a positive effect represented intention, and behavior contributing equally. reduced intergroup bias (i.e., more positive atti- tudes, fewer negative emotions, or increased Coding and Moderators intentions for contact). The majority of the studies included in the We coded a total of 11 group and design char- meta-analysis (77%) assessed intergroup bias acteristics which might moderate the effective- using more than one dependent measure. If we ness of imagined contact. First, we coded the were to include a separate effect size in our meta- group characteristics of the outgroup with analysis for each of these dependent measures, which participants imagined contact (ethnic studies with more measures would have a dispro- group, nationality, mental illness, disability, age, portionate influence on the overall effect size. sexual orientation, religion, weight, other); Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 13 participant gender (recorded as percentage of (Harris et al., 2008), which implements the ran- women in the sample); participant nationality dom effects model specified by DerSimonian and (where this was not reported, we substituted Laird (1986). The effect sizes in this analysis were nationality with the country where the study computed using Cohen’s d (d+), and the standard took place, if known); and participant age. Age errors used to weight each effect size were calcu- was coded both as a continuous variable and as lated according to the formula specified by Lipsey a dichotomous one (children, adults; i.e., and Wilson (2001). We interpreted these effect whether participants were under or over 18). sizes using standard convention (Cohen, 1992), in Second, we coded a number of design charac- which values of 0.2, 0.5, and 0.8 represent small, teristics assessing the methodological decisions medium, and large effect sizes, respectively; these made in each study. These were the valence of roughly correspond to Pearson’s r values of 0.1, the imagined contact (positive, neutral/unspeci- 0.25, and 0.4. We examined our effect sizes for fied); the type of control condition (imagine con- outliers, and as no effect sizes exceeded 2.5 stand- tact with a nonoutgroup member; imagine a ard deviations from the mean (either within each situation that does not involve contact; think of the four categories of dependent measure, or about the outgroup member; no task); and four overall), we made no adjustments. However, we variables relating to the level of elaboration. observed that two studies had a sample size over 6 These variables were the level of detail partici- times the average (Chen & Mackie, 2013; Lai et al., pants were given about the context of the imag- 2013). To ensure that these studies did not contrib­ ined scenario (coded on a 5-point scale from ute disproportionately to the summary effect size, minimal to very detailed); the level of detail partici- we capped their sample size at 180 (the size of the pants were given about the outgroup target in the next largest study) when computing the standard imagined scenario (coded on a 5-point scale from error variable used to weight each effect size. minimal to very detailed); the amount of time partici- Heterogeneity was evaluated using Cochran’s pants spent imagining contact (in minutes); and homogeneity Q statistic and the I2 statistic. Where whether participants wrote about or described the the Q statistic is significant, this indicates that the imagined contact afterwards (yes, no). effect of imagined contact across the relevant set Coding was performed independently by two of studies is moderated by factors other than postgraduate students, and a subset of the sampling error. The I2 statistic estimates the per- papers was also coded by the first author. centage of variability in the effect size estimate Interrater reliability was assessed using Pearson’s that can be attributed to these moderating fac- r for the continuous variables and Cohen’s kappa tors, rather than to sampling error. As a general for the categorical variables. Overall, there was a guideline, an I2 statistic of 30% to 60% indicates high level of agreement between coders (mean moderate variability, and over 75% indicates con- r = 0.89, range 0.78–0.97; mean κ = 0.86, range siderable variability (Higgins & Green, 2009). 0.74–1.00), and disagreements were resolved Moderator analyses were conducted using two through discussion. approaches. Our main approach was to employ metaregression (Thompson & Sharp, 1999), which can be used to assess the effect of both Meta-Analytic Strategy continuous and categorical moderators, in order Calculations were performed using STATA to assess whether each moderator was associated Version 12, and were based on random effects with significant variation in the effect size (the models. These assume that the true effect size of beta and p values in metaregression indicate the imagined contact in each study varies as a function strength and significance of this association, of differences in study characteristics as well as respectively). These analyses were performed sampling error. We computed weighted average using the STATA command metareg (Hardbord & effect sizes using the STATA command metan Higgins, 2008). However, where it was most Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 14 Group Processes & Intergroup Relations 17(1) Table 2.  Sample-weighted average imagined contact effect as a function of measure of intergroup bias. Dependent measure d k n 95% CI χ2 I2 Attitudes 0.346*** 57 4935 0.24, 0.45 159.14*** 64.8%   Explicit attitudes 0.364*** 52 4021 0.25, 0.48 154.85*** 67.1%   Implicit attitudes 0.307* 10 1686 0.05, 0.58 32.94*** 72.7% Emotion 0.410*** 28 1697 0.22, 0.61 102.75*** 73.7% Behavior 0.459** 10 530 0.16, 0.76 24.74** 63.6% Intentions 0.459*** 32 2076 0.32, 0.59 68.77*** 54.9%  Overall 0.351*** 71 5770 0.26, 0.44 158.37*** 55.8% Note. CI = confidence interval. *p < .05. **p < .01. ***p < .001. informative to examine the absolute size of the size of 0.31 on measures of implicit attitudes, imagined contact effect for each level of a mod- (95% CI [0.05, 0.58]), and an average effect size erator rather than to assess whether moderation of 0.36 on measures of explicit attitudes (95% CI was significant, we used meta-analysis (as [0.25, 0.48]). These effect sizes did not differ described in the previous section) to give an esti- from one another (Q[1] = 0.95, p = .329). mate of the magnitude of the size of the imag- The average effect size of imagined contact ined contact effect. on emotions towards the outgroup was d+ = 0.41, with a 95% confidence interval from 0.22 to 0.61, based on 28 comparisons and a total sample size Results of 1,697. Imagined contact had a medium effect Effectiveness of Imagined Contact in on both intentions (d+ = 0.46, 95% CI [0.32, 0.59]) and actual behavior towards the outgroup Reducing Intergroup Bias (d+ = 0.46, 95% CI [0.16, 0.76]), although far Meta-analysis showed that imagined contact had more studies included a measure of intentions a reliable small-to-medium effect across all meas- (32 studies, 2,076 participants) than included a ures of intergroup bias (see Table 2). The overall measure of actual behavior (10 studies, 530 sample-weighted effect of imagined contact on participants). intergroup bias was d+ = 0.35 (95% CI [0.26, We also compared the relative effectiveness of 0.44]), based on 71 studies and 5,770 participants. imagined contact between these four types of There was significant variation in the effect of intergroup bias. These analyses suggested that imagined contact across studies (Q[70] = 158.37, imagined contact had a larger effect on intentions p < .001), with a moderate-to-high level of het- than on attitudes (Q[1] = 4.04, p = .033), but no erogeneity across studies (I2 = 55.8%). Therefore, other comparisons approached significance (Qs moderator analyses are justified in order to deter- < 1.49, ps > .22). mine the sources of this variability. We next examined the effect of imagined con- tact on different types of intergroup bias. The Moderators of the Effectiveness sample-weighted average effect of imagined con- of Imagined Contact: Group tact on attitudes to the outgroup was d+ = 0.35, with a 95% confidence interval from 0.24 to 0.45, Characteristics based on 57 comparisons and a total sample size We evaluated five group characteristics as mod- of 4,935. We also computed separate effect sizes erators of the effectiveness of imagined contact based only on explicit or implicit measures of (see Table 3). Across the studies included in our attitudes; imagined contact had an average effect analysis, dozens of different outgroups were Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 15 Table 3.  Moderators of the imagined contact effect. Moderator Regression Standard k n 95% CI I2 Adj R2 coefficient error Publication status .244** .089 37, 34 3635; 2135 0.068, 0.420 52.5% 11.3% (unpublished, published) Group characteristics Percentage of female –.003 .002 65 4831 –0.008, 0.002 56.6% 1.28% participants (range 0–100) Participant age (range –.025* .010 54 4050 –0.044, –0.005 56.1% 17.24% 5–31) Adult or child .495** .160 48, 7 3638; 456 0.174, 0.815 55.0% 24.4% participants (adults, children) Design characteristics Valence of imagined .010 .119 14, 52 948; 3908 –0.227, 0.248 56.0% –3.14% contact (neutral, positive) Control condition   (absent, present)  Imagine contact with a –.078 .094 43, 28 3331; 2439 –0.266, 0.109 55.7% 0.08% non-outgroup member  Imagine a neutral .035 .094 40, 31 3719; 2051 –0.151, 0.222 56.1% –1.85% scene  Think about the –.035 .178 66, 5 5391; 379 –0.389, 0.320 56.4% –2.97% outgroup   No task .137 .153 64, 7 4869; 901 –0.167, 0.442 56.2% –1.61% Level of elaboration (range 1–5, from minimal -to very detailed)  Context .133** .046 71 5770 0.042, 0.225 51.0% 17.9%  Target .034 .040 71 5770 –0.046, 0.114 56.1% –1.51% Did participants –.102 .134 10, 56 730; 4564 –0.369, 0.165 56.9% –0.36% describe what they imagined? (no, yes) Time spent imagining .158 .114 30, 20 1860; 1379 –0.071, 0.388 55.6% 0.98% (1 or 2 minutes, over 2 minutes) Note. Columns k and n represent number of studies and number of participants, respectively. Where applicable, these are reported separately for each level of the moderator variable (indicated in parentheses at the end of each moderator name). Where a variable is coded as “absent, present,” absent was coded as 0 and present was coded as 1; thus, a positive regression coefficient indicates that studies in which the variable was present had larger effect sizes, and a negative regression coefficient indicates that studies where that variable was present had smaller effect sizes. *p < .05; **p < .01; ***p < .001. represented, including some which were unique these categories (see Table 4). Imagined contact to their specific study. We classified the majority had a positive effect on intergroup bias across all of the studies (k = 66) into eight outgroup cate- outgroups, with particularly robust effects for gories, and calculated effect sizes for each of outgroups based on nationality and age. Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 16 Group Processes & Intergroup Relations 17(1) Table 4.  Sample-weighted average imagined contact effect in different participant groups and outgroups. Group d k n 95% CI χ2 I2 Nationality of participants  UK 0.361*** 32 1,835 [0.220, 0.501] 66.3*** 53.2%  USA 0.225* 9 1,643 [0.022, 0.428] 18.3* 56.3%  European 0.412*** 20 1,601 [0.231, 0.592] 57.4*** 66.9%   Other or unknown 0.300*** 10 691 [0.135, 0.465] 10.5 14.0% Outgroup     Ethnic group 0.165 7 1,011 [−0.023, 0.353] 9.09 34.0%  Nationality 0.439*** 15 1,010 [0.274, 0.604] 22.5 37.7%   Mental illness 0.352** 12 789 [0.115, 0.590] 29.2** 62.4%  Disability 0.420 5 485 [−0.018, 0.858] 21.0*** 80.9%  Age 0.612*** 5 241 [0.353, 0.872] 3.8 0.0%   Sexual orientation 0.592* 5 249 [0.104, 1.081] 13.7** 70.8%  Religion 0.224* 13 1,280 [0.038, 0.409] 22.0* 45.4%  Weight 0.234 4 304 [−0.062, 0.530] 4.5 33.6%  Other 0.310 5 401 [−0.058, 0.679] 11.2* 64.2% Note. CI = confidence interval. *p < .05; **p < .01; ***p < .001. The confidence intervals for some outgroups metaregression confirmed that whether a study indicated that there was insufficient evidence for was conducted in one of these geographical areas a significant effect of imagined contact on that did not predict significant variation in its effect specific outgroup (ethnic group, disability, size, compared to the rest of the sample (βs < weight); however, all of the nonsignificant effect .138, ps > .268). Likewise, the gender of partici- sizes were based on seven or fewer studies, and pants did not moderate the observed effect size metaregression found that none of the outgroup (β = –.003, p = .197). categories were associated with significantly The effect of imagined contact was reliable larger or smaller effect sizes than the rest of the for both children (d+ = 0.81, 95% CI [0.53, sample as a whole (βs < .303, ps > .130). Thus, 1.09]) and adults (d+ = 0.32, 95% CI [0.21, there seem to be no substantive differences in the 0.43]), as neither confidence interval included effectiveness of imagined contact across out- zero. Furthermore, within adult participants groups, but there is a need for more studies to only, age did not influence the effectiveness of provide evidence for the effectiveness of imag- imagined contact (β < –.002, p = .866). Age was, ined contact with some specific outgroups. however, a significant moderator when consid- There was also considerable variation in group ered as a continuous variable (β = −.025, p = factors relating to the participants themselves. .013), which appeared to be due to the fact that The studies in our analysis were conducted with the effect of imagined contact was larger in chil- samples from nearly a dozen different countries, dren than in adults, resulting in a significant and we found little evidence that imagined contact moderation of effect size (β = .495, p = .003). was more or less effective across these different However, we also observed differences in design samples. Categorizing studies according to characteristics between studies with young par- whether they were performed in the UK, USA, ticipants and studies with adult participants, elsewhere in Europe, or elsewhere in the world, which could partly account for this effect. For we found that imagined contact had a significant example, studies with children tended to be effect in all these samples (see Table 4), and delivered in multiple sessions, and provided Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 17 their participants with significantly more detail effect in unpublished studies (d+ = 0.24, 95% CI about the imagined interaction than did studies [0.13, 0.35]), and a medium effect in published with adults (t[53] = 2.58, p = .013; see the fol- studies (d+ = 0.49, 95% CI [0.36, 0.62]). Thus, lowing section on elaboration). while the effect was significantly larger in pub- lished studies, the effect of imagined contact was reliable in both published and unpublished stud- Moderators of the Effectiveness ies, as neither confidence interval includes zero. of Imagined Contact: Design We also computed further analyses to assess Characteristics the likelihood that yet more unpublished studies exist, and to quantify the possible implications We evaluated six design characteristics as moder- for the size of our observed imagined contact ators of the effectiveness of imagined contact. effect. We assessed bias in our effect sizes using Whether participants were asked to imagine con- three methods, which all had converging results: a tact for longer than 1 or 2 minutes (β = .158, p = significant correlation between effect size and .172), explicitly told they should imagine a posi- sample size (r[69] = −.32, p = .006), significant tive interaction (β = .010, p = .931), or asked to bias on an Egger’s regression (β = 3.06, p < .001), describe what they had imagined afterwards (β = and inspection of the funnel plot (see Figure 1). −.102, p = .447) did not influence the effective- As a whole, these results suggest that small stud- ness of imagined contact. Similarly, the level of ies with small positive or negative effect sizes are detail participants were given about their imag- underrepresented in our sample, and that the ined interaction partner had no influence on smaller studies in our sample were more likely to effect size (β = .034, p = .395). Overall, the con- display positive effect sizes. trol condition with which imagined contact was The relevance of these results to our findings compared did not significantly influence effect is that if additional unpublished studies have sizes (βs < .137, ps > .372; see Table 3). been conducted but did not show up in our litera- However, the amount of detail participants ture search, then the true effect of imagined con- were given about the context of the imagined tact could be smaller than we believe it to be interaction significantly moderated how effec- (although it should be noted that publication bias tive it was at reducing intergroup bias (β = .133, is not the only possible explanation for small study p = .005). In the 30 studies which provided par- effects). However, it is also possible to allay these ticipants with no information or minimal infor- concerns by performing additional analyses to mation about the situation they should imagine, determine whether our results are indeed resist- the average effect size was 0.21 (95% CI [0.09, ant to publication bias. Accordingly, using the 0.33], p = .001); in the 41 studies which provided trim and fill procedure (Duval & Tweedie, 2000), more detailed information, effect size was 0.46 we found that the effect of imagined contact (95% CI [0.34, 0.58], p < .001). remained significant when 16 small studies with negative effect sizes were imputed (resulting in an overall effect size of d+ = 0.22, 95% CI [0.13, Publication Bias 0.32]). In addition, fail safe N (Rosenthal, 1979) Given that our analysis included a large number indicated that 3,481 studies with null effects of both unpublished and published studies, we would need to exist in order to overturn the con- were able to assess the impact of publication sta- clusion that imagined contact has a significant tus on effect size and to quantify the effect size effect on intergroup bias, which greatly exceeds within both types of study. Publication status was the recommended value of 5n + 10 (which a significant moderator of the observed effect size equates to 365 for our meta-analysis). Therefore, (β = .244, p = .007). Imagined contact had a small the overall effect of imagined contact is reliable Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 18 Group Processes & Intergroup Relations 17(1) Figure 1.  Funnel plot of effect sizes (d) for imagined contact, across all measures of intergroup bias. despite the small study effects, and appears to be effective even when our significant moderators resistant to any publication bias. We conclude were absent. For example, while giving partici- that, while unpublished studies in our sample do pants more detail about the to-be-imagined inter- have smaller effects than published ones, and action resulted in a larger impact on intergroup while it is likely that yet more unpublished studies bias, significant reductions in bias were also exist in file drawers, the effect of imagined con- observed in studies which gave participants little tact on intergroup bias is robust. or no detail. Similarly, while the effect of imag- ined contact was larger in children than in adults, the effect was still robust and significant in adults. General Discussion This meta-analytic review provides the first quan- Moderators of the Imagined Contact titative tests of the effectiveness of imagined contact on four key measures of intergroup bias: Effect attitudes, emotions, intentions, and behavior. Type of dependent measure.  Our analysis revealed an Overall, the effects of imagined contact appear to overall significant impact of imagined contact on parallel those of direct contact: there is a clear all dependent variables, and for both published and robust effect on all dependent measures, and and unpublished studies. In particular, we while some group or design characteristics facili- observed no significant difference between the tate the effect, none of them appear capable of effects on implicit and explicit attitudes, a finding eliminating it. Furthermore, just as Pettigrew and that has significant theoretical implications, par- Tropp (2006) observed that contact situations ticularly for the debate over the role of demand which met Allport’s desirable criteria were more characteristics in imagined contact effects (cf. effective, but that even situations that did not Bigler & Hughes, 2010). In combination with the meet those criteria were associated with reduced highly significant effect of imagined contact on prejudice, so we found that imagined contact was actual behavior, this provides the most convincing Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 19 evidence to date that demand characteristics can- ethnicity. This could be because prejudices against not account for imagined contact effects. It is these groups are stronger and more embedded in highly unlikely that participants would have been the culture, meaning that elaborated variants of able to modify their responses on such implicit imagined contact are needed to overcome bias in tasks even if they were able to guess the hypothe- these settings. On the other hand, there were few sis. Furthermore, the observation of a clear imag- studies with this type of outgroup (N = 7), so the ined contact effect in studies employing measures absence of an effect here could simply be due to of subtle nonverbal behavior, assessed by inde- low power. Either way, the clear implication is pendent coders (Birtel & Crisp, 2012a; Turner & that more research is needed into the effects of West, 2012) further strengthens this assertion. imagining contact with other ethnic groups. Thus, our results appear to dispel the demand Another important observation arising from characteristics critique. our analysis is that the imagined contact effect is Effect sizes were similar across most depend- stronger for children than adults. This makes ent variables. However, we did observe that sense: at school age, children are at a formative imagined contact has a stronger effect on behav- stage where imagery is a key component of how ioral intentions than on attitudes (the effect of they learn about the world (Cameron & Rutland, imagined contact on actual behavior was identi- 2006; Cameron et al., 2006). This finding that cal to the effect on intentions, but due to the imagined contact is not only effective in chil- smaller number of studies, the comparison with dren, but is actually more powerful, may be attitudes was not significant). This is consistent related to the typical features of the interven- with the wider literature on mental simulation, tions used with school-age children, as well as to in particular with evidence that mental simula- their age. Such interventions are typically more tion taps directly into the neurological architec- involved than those used with adults, often ture involved in action initiation (e.g., Kosslyn, occurring over multiple sessions (e.g., three ses- Ganis, & Thompson, 2001). It is also consistent sions in Vezzali, Capozza, Giovannini, et al., with research on the “perception–behavior 2012) with highly elaborated instructions. expressway” (Dijksterhuis, Bargh, & Mark, Although no study has directly compared multi- 2001), which has demonstrated that activating ple sessions of imagined contact with the rela- representations in memory can automatically tively brief instructions typically given to adults activate the associated behaviors. Through these in the laboratory, this chimes with our meta- processes, we suggest that imagined contact may analytic finding that elaborated instructions are operate at a different psychological stage to more powerful. While this type of extended and other attitude change interventions. Whereas detailed task is partly a consequence of the attitude change interventions first change atti- requirements and constraints of the educational tudes, and then exert their impact on behavior setting, it is possible that an extended program through intentions (Ajzen, 1991), imagined con- of imagined contact may also reinforce and sus- tact arguably intervenes at a point more proxi- tain the effect in adults, and it is also reassuring mal to actual behavior. Thus, it may eventually that these studies find results extended over be found to have an advantage over prejudice- weeks, rather than the typical single session in reduction interventions that focus instead on adult studies (e.g., Brambilla et al., 2012; Crisp & changing precursors of behavioral intention. Husnu, 2011; Turner et al., 2007). The confirmation that imagined contact works Group characteristics. We observed little variation well in school contexts has important practical in the imagined contact effect across a broad implications for extending the application and range of participant groups and target outgroups, impact of imagined contact. As discussed by although nonsignificant effects were observed Crisp and Turner (2012, 2013), the majority of for some groups, in particular those based on programs used to reduce prejudice in educational Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 20 Group Processes & Intergroup Relations 17(1) settings, such as the multicultural curricula be positive, consistent with theoretical recommen- approach (Appl, 1996) and the antiracist approach dations (Crisp & Turner, 2009), yet those studies (Dei, 1996), are not developed from evidence- that did not specify the valence of the effect still based theory (Aboud & Levy, 2000). While edu- obtained reliable imagined contact results, which cational psychologists advocate active thought were not significantly smaller. Pettigrew and Tropp over more passive approaches (Randi & Corno, (2006) found that contact reduced prejudice even 2000), existing programs often rely on outdated in nonideal circumstances, proposing a “mere assumptions that children are passive recipients exposure effect” of contact on prejudice. It is pos- of information. Thus, they may fail to reduce sible that there may also be a mere imagined exposure prejudice in children for the same reason that effect on prejudice, whereby thinking about any passive programs often fail to change attitudes in type of imagined interaction has a beneficial effect. adults; because the attitude-incongruent informa- However, there is also evidence that negative imag- tion is forgotten, distorted, or ignored (Rothbart ined contact may actually increase rather than & John, 1985). Imagined contact presents an decrease intergroup bias (e.g., Harwood et al., active, evidence-based approach which may offer 2011; West et al., 2011), which stands against this the means of effectively implementing contact hypothesis. Given this evidence, it is possible that theory in an educational setting. effects of valence appear only when interactions have a strong emotional tone (i.e., a mildly positive Design characteristics.  While few design factors influ- interaction works just as well as a neutral one; most enced the effectiveness of imagined contact, we “positive” imagined contact studies provide only found that the effect was stronger when partici- minimal specification, e.g., by including the word pants were instructed to elaborate on the context “positive” in their instructions). Alternatively, par- within which the imagined interaction took place ticipants may tend to imagine a positive interaction (exemplified by the instructional set developed by even when they are not asked to do so, rather than Husnu & Crisp, 2010a). This finding has practical relying on negative stereotypes to populate their implications for the implementation of imagined imagination (as originally suggested by Crisp & contact to reduce prejudice, and is also consistent Turner, 2009). Perhaps negative stereotypes define with the wider literature on mental simulation. In imagined interactions only in the most intractable particular, researchers in the mental simulation lit- intergroup conflicts, or when imagining interac- erature have proposed that effects of mental tions with the most feared or hated outgroups. imagery on behavior occur through the availability Future research may wish to take a more controlled of mental scripts or cognitive representations of and systematic approach to the valence of imag- sequences of behaviors (Schank & Abelson, 1977). ined contact, by comparing the effects of more As discussed earlier, there is evidence that the nuanced instructions (e.g., mildly positive vs. very more elaborate and detailed the script, the stronger positive), and by including postmanipulation the impact on subsequent attitudes and behavior checks to determine whether the valence of the (Anderson, 1983; Ross et al., 1975). Thus, our find- actual imagined interaction was consistent with the ings are consistent both with previous research on instructions. simulation and with the proposition that imagined contact may work by forming a mental script. Fur- Future Directions ther investigating this and other mechanisms underlying the imagined contact effect is an impor- In addition to the moderators assessed in our tant focus for future work. analysis, there are other factors which may influ- Contrary to theoretical predictions, positive ence the effectiveness of imagined contact, but imagined contact was no more effective than neu- which we were unable to assess due to lack of tral imagined contact. The majority of our included variability across studies. For example, we know studies specified that the imagined contact should little about the duration of imagined contact Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 21 effects over time, which is an important direction behavior more strongly when the person has for future research. To date, few studies have direct experience with the attitude object introduced a delay between performing imagined (Glasman & Albarracín, 2006) leads to a related contact and measuring intergroup bias (but see prediction: that imagined contact may be most Husnu & Crisp, 2010b; Vezzali, Capozza, effective either when participants have already Giovannini, et al., 2012; Vezzali, Capozza, Stathi, had past contact with the outgroup, or in combi- et al., 2012). Our findings and recommendations nation with direct contact (for initial findings in parallel those of Pettigrew and Troop (2006), this regard, see Husnu & Crisp, 2010a). As who called for more research on the duration of research combining elements of mental simula- direct contact effects, citing the relative lack of tion with perspective taking has also shown longitudinal studies. promising results with respect to prejudice reduc- Additionally, many other moderators have tion (Hodson, Choma, & Costello, 2009), we also been suggested by previous theoretical work, but suggest this as a promising focus for future have not been investigated in sufficient numbers research into augmenting and strengthening the to permit moderator analyses; for example, the imagined contact effect. effect of third- versus first-person perspective has been investigated in only one study (Crisp & Conclusion Husnu, 2011). Indeed, many of the studies included in this meta-analysis aimed to compare In Graham Greene’s The Power and the Glory, set in the effect of different types of imagined contact, Mexico during a time of religious persecution, in order to test hypotheses about moderators of the protagonist concluded that “Hate was just a the effect. So far, researchers have identified a failure of imagination” (1940, p. 131). In this number of design characteristics that appear to meta-analysis, we demonstrate that through imag- influence the size of the effect, but for which ination, meaningful reductions in prejudice can there are not yet enough studies for meta-analysis be obtained. Our key finding of a clear, overall (e.g., providing participants with information moderate effect of imagined contact on all about the typicality of the outgroup target; Stathi dependent variables parallels Pettigrew and et al., 2011). Researchers have also identified par- Tropp’s (2006) meta-analysis of direct contact ticipant characteristics which influence the size effects. Across diverse participant groups, of the imagined contact effect, such as majority dependent measures, and experimental designs, versus minority status (Stathi & Crisp, 2008) and imagined contact leads to reduced intergroup authoritarianism (Asbrock et al., 2013). We rec- bias; and, like direct contact, imagined contact ommend, consistent with the conclusions of “applies beyond racial and ethnic groups to Pettigrew and Tropp (2006), that the investiga- embrace other types of groups as well” (2006, p. tion of moderating variables continues to be a 768), with significant effects across diverse direction for future research. outgroups. Finally, the conception of imagined contact as Our finding that imagining an intergroup not only a direct method of reducing prejudice, encounter has reliable effects not only on atti- but also a method of reducing barriers to future tudes and emotions towards that group, but also contact and improving the likelihood that contact on intentions and behavior, means there is great will go well, leads to a number of predictions for potential for imagined contact as a tool to future tests. For instance, imagined contact improve intergroup relations. Echoing Pettigrew should increase the likelihood of engaging in and Tropp, we recommend that researchers con- direct contact in the “real world,” improve the tinue to explore the influence of “individual, quality of that contact, and make that contact structural and normative antecedents of the con- more effective in reducing prejudice. Additionally, tact” (2006, p. 768; see also Brown & Hewstone, the previous finding that attitudes predict 2005), which are largely uninvestigated at the Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 22 Group Processes & Intergroup Relations 17(1) present time. Overall, we recommend that future with psychiatric patients]. Unpublished raw data. research move beyond the debate about whether *Asbrock, F., Gutenbrunner, L., & Wagner, U. (2013). imagined contact works, or whether it is a “real” Unwilling, but not unaffected—Imagined contact effect, to focus on what prevents it from working, effects for authoritarians and social dominators. European Journal of Social Psychology, 43, 404–412. and what facilitates its effectiveness, in different doi:10.1002/ejsp.1956 contexts and with different groups. Babbitt, L. G., & Sommers, S. R. (2011). Fram- Acknowledgements ing matters: Contextual influences on interra- cial interaction outcomes. Personality and Social We gratefully acknowledge the assistance of Carla Psychology Bulletin, 37, 1233–1244. doi:10.1177/ Chivers and Laura Di Bella in coding studies. We would 0146167211410070 also like to thank the authors of the papers included *Bajrektarevic, D., Gulzad, N., & Shahhosseini, N. in this review for providing us with unpublished data (2012). Vorgestellter Kontakt für Minderheiten in and responding to requests for additional information. Deutschland [Imagined contact for minority members in Germany] (Unpublished undergraduate thesis). Funding Philipps University of Marburg, Germany. This research received no specific grant from any fund- *Bergeron, C. J. (2012). Imagine a better world: Two studies ing agency in the public, commercial, or not-for-profit of imagined intergroup contact (Unpublished doctoral sectors. dissertation). University of Maine, Orono. Bigler, R. S., & Hughes, J. M. (2010). Reasons for References skepticism about the efficacy of simulated contact interventions. American Psychologist, 65, 131–132. References marked with an asterisk indicate stud- doi:10.1037/a0018097 ies included in the meta-analysis. *Birtel, M. D., & Crisp, R. J. (2011). [Unpublished im- Aboud, F. E., & Levy, S. R. (2000). Interventions to agined contact studies, University of Kent]. Un- reduce prejudice and discrimination in children and published raw data. adolescents. In S. Oskamp (Ed.), Reducing prejudice Birtel, M. D., & Crisp, R. J. (2012a). Imagining in- and discrimination (pp. 269–293). Mahwah, NJ: Law- tergroup contact is more cognitively difficult rence Erlbaum Associates. for people higher in intergroup anxiety but this Abrams, D., Crisp, R. J., Marques, S., Fagg, E., Bedford, does not detract from its effectiveness. Group L., & Provias, D. (2008). Threat inoculation: Ex- Processes & Intergroup Relations, 15, 744–761. perienced and imagined intergenerational contact doi:10.1177/1368430212443867 prevents stereotype threat effects on older people’s Birtel, M. D., & Crisp, R. J. (2012b). “Treating” math performance. Psychology and Aging, 23, 934– prejudice: An exposure-therapy approach to 939. doi:10.1037/a0014293 reducing negative reactions toward stigma- Ajzen, I. (1991). The theory of planned behavior. Or- tized groups. Psychological Science, 23, 1379–1386. ganizational Behavior and Human Decision Processes, 50, doi:10.1177/0956797612443838 179–211. doi:10.1016/0749-5978(91)90020-T Blair, I. V., Ma, J. E., & Lenton, A. P. (2001). Imag- Allport, G. W. (1954). The nature of prejudice. Cambridge, ining stereotypes away: The moderation of im- MA: Perseus Books. plicit stereotypes through mental imagery. Jour- Anderson, C. A. (1983). Imagination and expectation: nal of Personality and Social Psychology, 81, 828–841. The effect of imagining behavioral scripts on per- doi:10.1037//0022-3514.81.5.828 sonal intentions. Journal of Personality and Social Psychol- *Brambilla, M., Ravenna, M., & Hewstone, M. (2012). ogy, 45, 293–305. doi:10.1037/0022-3514.45.2.293 Changing stereotype content through mental im- Appl, D. J. (1996). Recognising diversity in the early agery: Imagining intergroup contact promotes childhood classroom: Getting started. Teaching Ex- stereotype change. Group Processes & Intergroup Rela- ceptional Children, 28, 22–25. tions, 15, 305–315. doi:10.1177/1368430211427574 *Asbrock, F. (2012a). [Imagined intergroup contact *Broad, J. (2011). Imagined contact and perspective taking: A with unemployed people]. Unpublished raw data. potential stepping stone to reducing prejudice towards people *Asbrock, F. (2012b). [Imagined intergroup contact with schizophrenia (Unpublished undergraduate the- with overweight people]. Unpublished raw data. sis). University of Sussex, UK. *Asbrock, F. (2012c). [Imagined intergroup contact Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 23 Brown, R., & Hewstone, H. (2005). An integrative Review of Social Psychology, 21, 188–236. doi:10.1080/ theory of intergroup contact. In M. P. Zanna (Ed.), 10463283.2010.543312 Advances in Experimental Social Psychology (Vol. 37, pp. Crisp, R. J., & Turner, R. N. (2009). Can imagined in- 255–343). San Diego, CA: Academic Press. teractions produce positive perceptions? Reducing Cameron, L. & Rutland, A. (2006). Extended con- prejudice through simulated social contact. Ameri- tact through story reading in school: Reducing can Psychologist, 64, 231–240. doi:10.1037/a0014718 children’s prejudice towards the disabled. Journal Crisp, R. J. & Turner, R. N. (2010). Have confidence of Social Issues, 62, 489–510. doi:10.1111/j.1540- in contact. American Psychologist, 65, 133–134. 4560.2006.00469.x doi:10.1037/a0018437 Cameron, L., Rutland, A., Brown, R. J. & Douch, R. Crisp, R. J., & Turner, R. N. (2012). The imagined (2006). Changing children’s intergroup attitudes contact hypothesis. Advances in Experimental Social towards refugees: Testing different models of ex- Psychology, 46, 125–182. doi:10.1016/B978-0-12- tended contact. Child Development, 77, 1208–1219. 394281-4.00003-9 doi:10.1111/j.1467-8624.2006.00929.x Crisp, R. J., & Turner, R. N. (2013). Imagined inter- *Cameron, L., Rutland, A., Turner, R., Holman- group contact: Refinements, debates and clarifica- Nicolas, R., & Powell, C. (2011). “Changing atti- tions. In G. Hodson & M. Hewstone (Eds.), Ad- tudes with a little imagination”: Imagined contact vances in intergroup contact (pp. 135–151). Hove, UK: effects on young children’s intergroup bias. Anales Psychology Press (Taylor & Francis). de Psicología, 27, 708–717. Retrieved from http:// Dei, G. J. S. (1996). Anti-racism education: Theory and prac- kar.kent.ac.uk/27770/ tice. Halifax, Canada: Fernwood. *Capozza, D., Visintin, E. P., Favara, I., Falvo, R., DerSimonian, R., & Laird, N. (1986). Meta-analysis in Trifiletti, E., & Di Bernardo, G. A. (2013, January). clinical trials. Controlled Clinical Trials, 7, 177–188. Mental imagery as a moderator of the effects of imagined doi:10.1016/0197-2456(86)90046-2 contact. Paper presented at the 14th Annual Meeting Desforges, D. M. Lord, C. G., Pugh, M. A., Sia, T. of the Society for Personality and Social Psychol- L., Scarberry, N. C., & Ratcliff, C. D. (1997). Role ogy, New Orleans, LA. of group representativeness in the generaliza- *Chen, J. M., Cook, M., & Mackie, D. M. (2013). [Imag- tion part of the contact hypothesis. Basic and Ap- ined contact and weight stigma]. Unpublished raw plied Social Psychology, 19, 183–204. doi:10.1207/ data. s15324834basp1902_3 *Chen, J. M., & Mackie, D. M. (2013). [Imagined con- Dijksterhuis, A., Bargh, J. A., & Mark, P. Z. (2001). tact and religious prejudice]. Unpublished raw data. The perception–behavior expressway: Automatic *Chen, J. M., Richards, B., & Mackie, D. M. (2013a). effects of social perception on social behavior. [Imagined contact and political prejudice I]. Un- Advances in Experimental Social Psychology, 33, 1–40. published raw data. doi:10.1016/S0065-2601(01)80003-4 *Chen, J. M., Richards, B., & Mackie, D. M. (2013b). Duval, S., & Tweedie, R. (2000). Trim and fill: A simple [Imagined contact and political prejudice II]. Un- funnel-plot-based method of testing and adjusting published raw data. for publication bias in meta-analysis. Biometrics, 56, Cohen, J. (1992). A power primer. Psychological Bulletin, 455–463. doi:10.1111/j.0006-341X.2000.00455.x 112, 155–159. doi:10.1037/0033-2909.112.1.155 Eller, A., & Abrams, D. (2004). Come together: Lon- Crisp, R. J., Birtel, M. D., & Meleady, R. (2011). Men- gitudinal comparisons of Pettigrew’s reformulated tal simulations of social thought and action: Trivial intergroup contact model and the common in- tasks or tools for transforming social policy? Cur- group identity model in Anglo-French and Mexi- rent Directions in Psychological Science, 20, 261–264. can-American contexts. European Journal of Social doi:10.1177/0963721411413762 Psychology, 34, 229–256. doi:10.1002/ejsp.194 *Crisp, R. J., & Husnu, S. (2011). Attributional pro- Finchilescu, G. (2010). Intergroup anxiety in interra- cesses underlying imagined contact effects. Group cial interaction: The role of prejudice and metas- Processes and Intergroup Relations, 14, 275–287. tereotypes. Journal of Social Issues, 66, 334–351. doi:10.1177/1368430210390721 doi:10.1111/j.1540-4560.2010.01648.x Crisp, R. J., Husnu, S., Meleady, R., Stathi, S., & Turner, *Frye, G. D. J., Lord, C. G., & Brady, S. E. (2012). Atti- R. N. (2010). From imagery to intention: A dual tude change following imagined positive actions to- route model of imagined contact effects. European ward a social group: Do memories change attitudes, Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 24 Group Processes & Intergroup Relations 17(1) or attitudes change memories? Social Cognition, 30, group contact: A new technique for encourag- 307–322. doi:10.1521/soco.2012.30.3.307 ing greater inter-ethnic contact in Cyprus. Peace *Giacobbe, M. R., Stukas, A. A., & Farhall, J. (2013). and Conflict: Journal of Peace Psychology, 16, 97–108. The effects of imagined versus actual contact with doi:10.1080/10781910903484776 a person with a diagnosis of schizophrenia. Basic Husnu, S. & Crisp, R. J. (2011). Enhancing the imag- and Applied Social Psychology, 35, 265–271. doi:10.108 ined contact effect. Journal of Social Psychology, 151, 0/01973533.2013.785403 113–116. doi:10.1080/00224541003599043 Glasman, L. R., & Albarracín, D. (2006). Forming *Japhet, S. (2010). Boosted imagined contact to reduce preju- attitudes that predict future behavior: A meta- dices against schizophrenia (Unpublished master’s the- analysis of the attitude–behavior relation. Psycho- sis). University of Sussex, UK. logical Bulletin, 132(5), 778–822. doi:10.1037/0033- *Jaworska, M., Górska, P., & Bilewicz, M. (2012). Inter- 2909.132.5.778 group anxiety as a limitation of imagined contact interven- Gollwitzer, P. M. (1993). Goal achievement: The role tions. Paper presented at the 13th Annual Meeting of intentions. European Review of Social Psychology, 4, of the Society for Personality and Social Psychol- 141–185. doi:10.1080/14792779343000059 ogy, San Diego, CA. Greene, G. (1940). The power and the glory. London, UK: *Jaworska, M., Górska, P., & Bilewicz, M. (2013). [Un- Heinemann. published imagined contact data]. Unpublished raw Harbord, R. M., & Higgins, J. P. T. (2008). Meta- data. regression in Stata. Stata Journal, 8, 493–519. Re- Johnson, B. T., & Eagly, A. H. (2000). Quantitative syn- trieved from http://www.stata-journal.com/article. thesis of social psychological research. In H. T. Reis html?article=sbe23_1 & C. M. Judd (Eds.), Handbook of research methods Harris, R., Bradburn, M., Deeks, J., Harbord, R., in social psychology (pp. 496–528). Cambridge, UK: Altman, D., & Sterne, J. (2008). Metan: Fixed- and Cambridge University Press. random-effects meta-analysis. Stata Journal, 8, 3–28. Kosslyn, S. M., Ganis, G., & Thompson, W. L. (2001). Retrieved from http://www.stata-journal.com/arti- Neural foundations of imagery. Nature Reviews Neu- cle.html?article=sbe24_2 roscience, 2, 635–642. doi:10.1038/35090055 *Harwood, J., Paolini, S., Joyce, N., Rubin, M., & *Kuchenbrandt, D., & Eyssel, F. (2012). The mental Arroyo, A. (2011). Secondary transfer effects from simulation of a human–robot interaction: Positive imagined contact: Group similarity affects the gen- effects on attitudes and anxiety toward robots. In eralization gradient. British Journal of Social Psychol- Proceedings of the 21th IEEE International Symposium ogy, 50, 180–189. doi:10.1348/014466610X524263 in Robot and Human Interactive Communication (RO- Higgins, J. P. T., & Green, S. (Eds.). (2009). Cochrane MAN, pp. 463–468). Institute of Electrical and handbook for systematic reviews of interventions, Version Electronics Engineers (IEEE): Paris, France. 5.1.0. Retrieved from www.cochrane-handbook.org Kuchenbrandt, D., Eyssel, F., & Seidel, S. K. (2013). Hodson, G., Choma, B. L., & Costello, K. (2009). Ex- Cooperation makes it happen: Imagined intergroup periencing alien-nation: Effects of a simulation cooperation enhances the positive effects of imag- intervention on attitudes toward homosexuals. ined contact. Group Processes & Intergroup Relations, Journal of Experimental Social Psychology, 45, 974–978. 16, 635–647. doi:10.1177/1368430212470172 doi:10.1016/j.jesp.2009.02.010 *Lai, C. K., Marini, M., Lehr, S. A., Cerruti, C., Shin, J. Hodson, G., & Hewstone, M. (Eds.). (2013). Advances L., Joy-Gaba, J. A.,... Nosek, B. A. (2013). Reducing in intergroup contact. London, UK: Psychology Press implicit racial preferences: I. A comparative inves- (Taylor & Francis). tigation of 18 interventions. Unpublished manu- *Hughes, J. M., Holcombe, C., & Wight, K. (2013). Ef- script. fects of simulated contact on implicit and explicit prejudice Lee, Y.-T., & Jussim, L. (2010). Back in the real world. toward older adults and African Americans. Unpublished American Psychologist, 65, 130–131. doi:10.1037/ manuscript. a0018195 *Husnu, S., & Crisp, R. J. (2010a). Elaboration enhanc- Lipsey, M. W., & Wilson, D. B. (2001). Practical meta- es the imagined contact effect. Journal of Experi- analysis. Thousand Oaks, CA: Sage. mental Social Psychology, 46, 943–950. doi:10.1016/j. Mazziotta, A., Mummendey, A., & Wright, S. C. (2011). jesp.2010.05.014 Vicarious intergroup contact effects: Applying so- *Husnu, S., & Crisp, R. J. (2010b). Imagined inter- cial-cognitive theory to intergroup contact research. Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 Miles and Crisp 25 Group Processes & Intergroup Relations, 14, 255–274. *Stathi, S., & Crisp, R. J. (2008). Imagining intergroup doi:10.1177/1368430210390533 contact promotes projection to outgroups. Jour- *Menkinoska, E. (2011). Imagined intergroup contact and nal of Experimental Social Psychology, 44, 943–957. attitudes toward individuals with schizophrenia (Unpub- doi:10.1016/j.jesp.2008.02.003 lished undergraduate thesis). La Trobe University, Stathi, S., & Crisp, R. J. (2010). Intergroup contact and Melbourne, Australia. the projection of positivity. International Journal of *Miller, A. K., Markman, K. D., Wagner, M. M., Intercultural Relations, 34, 580–591. doi:10.1016/j. & Hunt, A. N. (2013). Mental simulation and ijintrel.2010.04.002 sexual prejudice reduction: The debiasing role *Stathi, S., Crisp, R. J., & Hogg, M. A. (2011). Imagin- of counterfactual thinking. Journal of Applied So- ing intergroup contact enables member-to-group cial Psychology, 43, 190–194. doi:10.1111/j.1559- generalization. Group Dynamics-Theory Research and 1816.2012.00992.x Practice, 15, 275–284. doi:10.1037/a0023752 *Pagotto, L., Visintin, E. P., De Iorio, G. D., & Voci, *Stathi, S., Tsantila, K., & Crisp, R. J. (2012). Imagin- A. (2012). Imagined intergroup contact pro- ing intergroup contact can combat mental health motes cooperation through outgroup trust. stigma by reducing anxiety, avoidance and negative Group Processes & Intergroup Relations, 16, 209–216. stereotyping. Journal of Social Psychology, 152, 746– doi:10.1177/1368430212450057 757. doi:10.1080/00224545.2012.697080 Pettigrew, T. F., & Tropp, L. R. (2006). A meta- Thompson, S. G., & Sharp, S. J. (1999). Explaining het- analytic test of intergroup contact theory. Jour- erogeneity in meta-analysis: A comparison of meth- nal of Personality and Social Psychology, 90, 751–783. ods. Statistics in Medicine, 18, 2693–2708. doi:10.1002/ doi:10.1037/0022-3514.90.5.751 (SICI)1097-0258(19991030)18:20<2693::AID- Randi, J., & Corno, L. (2000). Teacher innovations in SIM235>3.0.CO;2-V self-regulated learning. In M. Boekaerts, P. Pintrich, Todd, A., Hanko, K., Galinsky, A., & Mussweiler, T. & M. Zeidner (Eds.), Handbook of self-regulation (pp. (2011). When focusing on differences leads to simi- 651–685). New York, NY: Academic Press. lar perspectives. Psychological Science, 22, 134–141. Rosenthal, R. (1979). The “file drawer problem” and doi:10.1177/0956797610392929 tolerance for null results. Psychological Bulletin, 86, *Turner, R. N., & Crisp, R. J. (2010). Imagining inter- 185–193. doi:10.1037//0033-2909.86.3.638 group contact reduces implicit prejudice. British Ross, L., Lepper, M. R. & Hubbard, M. (1975). Per- Journal of Social Psychology, 49, 129–142. doi:10.134 severance in self-perception and social perception: 8/014466609X419901 Biased attributional processes in the debriefing par- *Turner, R. N., Crisp, R. J., & Lambert, E. (2007). Im- adigm. Journal of Personality and Social Psychology, 32, agining intergroup contact can improve intergroup 880–892. doi:10.1037/0022-3514.32.5.880 attitudes. Group Processes & Intergroup Relations, 10, Rothbart, M., & John, O. P. (1985). Social categoriza- 427–441. doi:10.1177/1368430207081533 tion and behavioral episodes: A cognitive analysis *Turner, R. N., & West, K. (2012). Behavioural of the effects of intergroup contact. Journal of Social consequences of imagining intergroup contact Issues, 41, 81–104. doi:10.1111/j.1540-4560.1985. with stigmatized outgroups. Group Processes & tb01130.x Intergroup Relations, 15, 193–202. doi:10.1177/ Schank, R. C., & Abelson, R. P. (1977). Scripts, plans, 1368430211418699 goals, and understanding. Hillsdale, NJ: Erlbaum. *Turner, R. N., West, K., & Christie, Z. (2013). Out- *Slater, C. (2011). Shifting the weight of prejudice: Imagining group trust, intergroup anxiety, and out-group contact and inducing empathy to reduce prejudice against the attitude as mediators of the effect of imagined stigmatised group, the overweight (Unpublished under- intergroup contact on intergroup behavioral graduate thesis). University of Sussex, UK. tendencies. Journal of Applied Social Psychology, 43, Stata Statistical Software (Release 12) [Computer soft- E196-E205. ware]. College Station, TX: StataCorp LP. *Vezzali, L., Capozza, D., Giovannini, D., & Stathi, S. *Stathi, S., Cameron, L., Hartley, B., & Bradford, S. (in (2012). Improving implicit and explicit intergroup press). Imagined contact as a prejudice-reduction attitudes using imagined contact: An experimen- intervention in schools: The underlying role of tal intervention with elementary school children. similarity and attitudes. Journal of Applied Social Psy- Group Processes & Intergroup Relations, 15, 203–212. chology. doi:10.1177/1368430211424920 Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014 26 Group Processes & Intergroup Relations 17(1) *Vezzali, L., Capozza, D., Stathi, S., & Giovannini, mediational role of anxiety and the moderational role D. (2012). Increasing outgroup trust, reducing in- of group salience. Group Processes & Intergroup Rela- frahumanization, and enhancing future contact tions, 6, 37–54. doi:10.1177/1368430203006001011 intentions via imagined intergroup contact. Jour- Vorauer, J. D., Hunter, A. J., Main, K. J., & Roy, S. nal of Experimental Social Psychology, 48, 437–440. (2000). Metastereotype activation: Evidence from doi:10.1016/j.jesp.2011.09.008 indirect measures for specific evaluative concerns *Vezzali, L., Crisp, R. J., Stathi, S., & Giovannini, D. experienced by members of dominant groups in (2013). Mental simulation facilitates intercultural commu- intergroup interaction. Journal of Personality and nication for college students on academic exchange programs. Social Psychology, 78, 690–707. doi:10.1037/0022- Unpublished manuscript. 3514.78.4.690 *Vezzali, L., Stathi, S., Crisp, R. J., & Capozza, D. *Wallace, R. (2010). Imagined contact and the role of anxi- (2013). Drawing and imagined contact reduces prejudice in ety and empathy (Unpublished undergraduate thesis). preschool children. Unpublished manuscript. University of Sussex, UK. *Vezzali, L., Stathi, S., Crisp, R. J., Giovannini, D., & *West, K., & Bruckmüller, S. (2013). Nice and easy Capozza, D. (2013). Actual or imagined common ingroup does it: How perceptual fluency moderates the ef- identity, what’s the most powerful? An experimental intervention fectiveness of imagined contact. Journal of Experi- with elementary school children. Unpublished manuscript. mental Social Psychology, 49, 254–262. doi:10.1016/j. *Vezzali, L., Stathi, S., Crisp, R. J., Giovannini, D., jesp.2012.11.007 Capozza, D., & Gaertner, S. L. (2013). Imagined inter- *West, K., Holmes, E., & Hewstone, M. (2011). En- group contact and common ingroup identity: An integrative hancing imagined contact to reduce prejudice approach. Unpublished manuscript. against people with schizophrenia. Group Processes Voci, A., & Hewstone, M. (2003). Intergroup contact & Intergroup Relations, 14, 407–428. doi:10.1177/ and prejudice towards immigrants in Italy: The 1368430210387805 Downloaded from gpi.sagepub.com at University of Sussex Library on March 7, 2014