Abstract
This article has been officially retracted from the Journal.
Introduction
Intergroup Contact and Intergroup Relations
The effect of intergroup contact on intergroup relations was first studied by Gordon Allport. He proposed that prejudices, that is, negative evaluations, can be reduced through outgroup contact, provided that the groups have an equal status, work toward a shared goal, and receive institutional support. 4 Including over 50 years of research, 6 a meta-analysis 7 concluded that direct intergroup contact reduced prejudices also when these conditions were not fulfilled and when ingroup members were merely exposed to the outgroup, for example, in spontaneous encounters.
In addition, direct intergroup contact can foster mutual acceptance of in- and outgroup members, if “personal and collective narratives (…) and suffering,”5(p387) related to the individual's experiences of the intergroup relation, are shared. This strategy is based on the narrative story-telling approach to conflict resolution, which is used to encourage conflict parties to reflect on their unresolved anger, and, by hearing each other's stories, to re-humanize the outgroup members and establish intergroup trust and empathy. 5
It is important to note that contact can only lead to reduced prejudices and increased mutual acceptance, if it takes place in an intergroup as compared to an interpersonal context. 8 This distinction illustrates that individuals' understanding of themselves is compiled of aspects of personal and different social identities, 3 group memberships that are fluid 9 in the way that different facets are salient depending on the social context. Consequently, individuals define themselves and others in different settings either as belonging to different groups (intergroup context) or as being idiosyncratically different (interpersonal context). 8 For example, at a party, two guests might interact as Paul and Peter, while the same individuals would meet as supporters of the respective national teams at an international soccer match, and during a political debate as members of two parties. Only contact in the latter two settings, an intergroup context, can influence intergroup relations, because as the interaction partner's outgroup membership is salient, positive experiences with individual outgroup members can be generalized to the whole outgroup. 10
Computer-Mediated Intergroup Contact
Computer-mediated intergroup contact is an alternative to face-to-face encounters. CMC can facilitate intergroup contact by connecting users across long distances quickly and at relatively low costs. 11 The Internet provides further a context where physical harm from outgroup members does not have to be feared, which can result in lower intergroup anxiety, that is, expecting less negative experiences when meeting the outgroup, and increased openness toward intergroup contact. 12 Direct computer-mediated intergroup contact, text-, image-, or video-based, can be easily observed by in- and outgroup members, making the Internet a source of vicarious intergroup contact. 6 Previous research indicated that observing a representative of the ingroup engage in intergroup contact served as a learning event, showing that intergroup contact is possible, which led to more positive attitudes toward the outgroup and increased participants' willingness to engage in intergroup meetings themselves.6,13
Despite these benefits, few empirical studies assessed the effects of computer-mediated intergroup contact. Stock et al. 14 applied the narrative story-telling approach to encounters between Israeli and Palestinian teenagers, who used one computer interface to create a joint story about their conflict. The acceptance of the outgroup's needs increased significantly during the interactive phase. Evaluations of facilitated online initiatives indicated as well positive effects of computer-mediated intergroup contact on intergroup relations.15–17 Further, Tynes et al. 18 showed that 81 percent of Latino, European, and African-American participants in an informal chat room had spent at least 1 to 5 hours interacting with ethnic outgroup members, and more interactions predicted greater outgroup orientation and learning about the outgroup.
Computer-Mediated Intergroup Contact on Facebook
Extending this research, we investigated computer-mediated intergroup contact and its effects in a particular online setting: Facebook. Facebook is an SNS with more than 900 million users (April 2012). 19 Boyd and Ellison 20 defined SNS “as web-based services that allow individuals to (…) construct a public or semi-public profile (…) (and) (…) articulate a list of other users with whom they share a connection.” The connections, that is, Facebook friends, are mostly users that one met offline or with whom one shares offline elements, such as the educational background or social interests.21,22 While these findings provide insights into interpersonal contacts on Facebook, they allow no conclusions about the frequency and effects of intergroup contact on the site.
We argue that Facebook-groups are a feature where intergroup contact can develop and affect intergroup relations. Intergroup contact requires, as noted earlier, that individuals perceive themselves and others as members of certain groups—in- and outgroup members. Facebook-groups provide a setting to induce this perspective. They serve as virtual discussion tables and an infrastructure for collective interactions around a common interest or need, 23 which defines the Facebook-group and is indicated in the title or description. Joining and engaging in the Facebook-group emphasizes one's belonging to it, making comments or “likes” expressions of in- or outgroup members. 24 It is through these comments that interactions can develop between Facebook users that are not Facebook friends, that possibly do not have similar backgrounds and interests, and that are in- and outgroup members. Indeed, an analysis 25 of one political Facebook-group showed that members did exchange rather diverse opinions in the discussions, even though the Facebook-group itself had a clear group norm.
Based on this argument, the goal of the present study was to assess the extent to which intergroup contact takes place in Facebook-groups, and to test the following hypotheses:
Method
Material for analyses
We examined the research question and hypotheses by analyzing comments from Facebook-groups. The material was collected in January 2011. To highlight the practical relevancy of our analysis, we selected Facebook-groups that referred to social identities of the categories nationality, religion, or gender. Using these increased the possibility that Facebook-group members had unambiguous understanding of being an ingroup or outgroup member, based on criteria that could not be changed without strong commitment. To select the Facebook-groups, a wide range of keywords of the categories gender, nationality, and religion were searched on Facebook (Table 1). We chose those Facebook-groups that were open, had a minimum of 1,000 members, and the title and the majority of comments in English (Table 1).
For each category, three Facebook-groups were sampled. Two had titles, which expressed an ingroup norm and either derogated the outgroup and its values by expressing critique or negative emotions toward them, or not. A third Facebook group had a title that did not explicitly refer to any ingroup norm. We included this distinction as titles that derogate the outgroup might attract more outgroup members who want to defend their group and its status, to uphold their self-esteem. 3 A total of nine Facebook-groups were analyzed (Table 1).
Coding scales
To evaluate the comments, we used the destructive and constructive conflict scale 26 with which utterances are rated on two independent dimensions. The destructive dimension includes four items, referring to a self-centered approach of the speaker, who derogates the interaction partner. The mean score of this scale represented the degree of expressed prejudices (α=0.93). The constructive dimension assesses with four items the speaker's aim to achieve its goals while enhancing relations with the interaction partner by discussing in a way that allows the integration of both parties' perspectives. An exploratory factor analysis showed that all items of the destructive scale loaded on one factor (Factor 1), whereas the items of the constructive scale loaded on two. One (Factor 3) referred to the personal involvement of the speaker (α=0.52) and the other (Factor 2) to a style of argumentation that accepts the interaction partner and integrates both parties' perspectives in joint argument structures (α=0.78) (Table 2). We included only the mean ratings on Factor 2 as an indicator of expressed acceptance.
Shading is a visual aid to identify the factor loadings that are attributed to factor 1, 2, and 3, respectively.
Procedure
The 100 latest comments were extracted from all nine Facebook-groups, pictures, or links not included. Each comment was treated as one utterance. First, the group norm of the Facebook-group was identified through the title and the mission statement. All comments were coded as stemming from the ingroup or outgroup, considering whether support for the ingroup norm and membership in the respective group was indicated or not. Related outgroup and ingroup comments were identified by taking the content, the mentioning of names, and the use of the comment function into account. If a minimum of one outgroup and one ingroup member took turns, it was considered as direct computer-mediated intergroup contact. These comments, outgroup comments that did not relate to a particular ingroup comment and ingroup comments to the outgroup that were not responded to, were coded with the destructive and constructive conflict scale on a 5-point Likert scale. 26 An independent second coder coded 50 percent of the N=223 utterances. The inter-rater reliability yielded a weighted Kappa of κ=0.87 and was acceptable.
Results
Across all Facebook-groups, 13.1 percent of utterances were made by outgroup members (Table 3), and this proportion did not differ significantly between the Facebook-groups taking the social category and the type of title into account (Table 4). On average, 77.86 percent of the outgroup comments were part of intergroup interactions (Table 5). Across all nine Facebook-groups, M=4.1 (SD=3.25) intergroup interactions took place, including one or more turns between in- and outgroup members (Table 3). The overall amount of interactions did not differ significantly between Facebook-groups (Table 4).
p<.05.
The residual and total degrees of freedom appear in parentheses behind the F-values.
Values refer to a total of 100 comments per Facebook-group.
The mean length of interactions indicates the average number of turns taken by in- and outgroup members.
Values refer to a total of 100 comments per Facebook-group.
The mean length of intergroup interactions was computed, representing the average number of turns that were taken by in- and outgroup members (Table 3). This value is a more precise indicator of the amount of intergroup contact than the absolute number of interactions. Facebook-groups related to religion and nationality did not yield a different mean length of intergroup interactions, whereas Facebook-groups referring to gender had significantly shorter intergroup interactions [t(6)=3.09, p<0.05] (Table 4). This was however due to one gender-related Facebook-group not having any outgroup comments or intergroup interactions. It was omitted from all further analyses.
To examine whether the level of expressed prejudices and mutual acceptance in Facebook-groups changed over time, pre- and postscores of destructiveness and constructiveness were computed. The prescores referred to the ratings of destructiveness and constructiveness of the first half of the analyzed utterances in each Facebook-group and the postscores to the ratings of the second half. Findings showed that the expression of prejudices decreased marginal significantly [t(109)=1.79, p=0.07] and that of mutual acceptance increased significantly [t(109)=−2.32, p<0.05] (Table 6). A change score was computed for each scale by subtracting the post score of destructiveness from its prescore and by subtracting the pre score of constructiveness from its postscore. Introducing the users' group membership as the between-subject factor, the level of change did not differ between in- and outgroup members [prejudice: F(1, 106)=2.51, p>0.05; acceptance: F(1,106)=1.84, p>0.05].
Standard deviations appear in parentheses behind means.
To test whether changes in expressed prejudices and mutual acceptance were driven by intergroup contact, we computed two multiple stepwise regressions (Table 7). They included the postscores of destructiveness and constructiveness as dependent variables and the mean length of intergroup interactions as an independent variable, controlling for the number of overall outgroup comments, and the prescore of constructiveness and destructiveness. Results showed that longer intergroup interactions predicted the expression of fewer prejudices [F(4, 105)=8.01, β=−0.26, R2=0.35). Increased expressions of acceptance were not predicted by intergroup contact, supporting only Hypothesis 1.
*p<0.05.
Discussion
The goal of this study was to assess the extent to which intergroup contact unfolds in Facebook-groups and whether it contributes to improved intergroup relations. Our results confirmed that outgroup members were present in Facebook-groups at a low, but yet noticeable level, challenging the idea that Facebook is fragmented in homogenous clusters of like-minded users. 25 Individuals might actually sometimes seek content that is dissonant with their beliefs, to defend the status of their ingroup and self-esteem. 3 With the exception of one, the number of outgroup comments and the mean length of intergroup interactions did not differ between Facebook-groups, which indicates that irrespective of the thematic focus and the way the Facebook-group is presented in its title, direct and vicarious intergroup contact can develop in this online setting.
We demonstrated further that the level of mutual acceptance expressed in the comments increased and that of prejudices decreased over time. The latter was predicted by more intergroup contact in Facebook-groups, which emphasizes that asynchronous and text-based computer-mediated intergroup contact can have the same effects on negative outgroup evaluations as face to face, imagined, heard-about, and observed forms of contact. 6 This finding contradicts the claim that CMC lacks social presence and could therefore not affect prejudices. 27 CMC might, on the contrary, be even beneficial as in- and outgroup members experience not just each others' physical presence, but also opinionated contributions, constructed elaborately without time constraint that might include additional arguments, which can influence levels of prejudices. 27 Open Facebook-groups can also elicit a context of equal status, as all Facebook users can join and participate in the discussions, a setting that should enhance the effect of intergroup contact on reduced prejudices. 4 Ultimately, if communication norms in Facebook-groups encourage the exchange of different opinions by in- and outgroup members, levels of prejudices can be reduced.
However, the expression of mutual acceptance was not affected by intergroup contact in Facebook-groups. Acceptance might indeed depend more on the quality of the shared narratives than on the quantity of interactions. Notably, Maoz argued that stories should be personal, but still relating to the intergroup conflict, and authentic, but not hurting the outgroup, to contribute to mutual acceptance. 28 Our analysis did not include these criteria, and future research should incorporate the categories when assessing the development of mutual acceptance through interactions.
A limitation of our findings stems from the demographics of Facebook users. They represent currently more than ¼ of the population in North America, Europe, Latin America (largely driven by Brazil), and Australia, but in Asia, Africa, the Middle East, and the Caribbean, only 3.6–15 percent of the population use Facebook. 29 This restricts intergroup contact in Facebook-groups, and effects on reduced prejudices, to specific nationalities and cultures.
In addition, while the analysis of comments in Facebook-groups provides valid insights to the actual amount of intergroup contact, it is not possible to conclude whether expressed levels of prejudices and acceptance are in fact representative of privately held attitudes, and hence an indicator of behavior intentions. Attitudes shared in Facebook-groups can be also influenced by a need for positive self-presentation, 22 in particular as content expressed in open Facebook-groups is visible to the authors' Facebook friends, with whom one interacts mostly also offline. In addition, norms of the salient ingroup can drive comments as users are motivated to be accepted and positively valued by their ingroup.3,30 Our findings could hence be also due to norms of intergroup relations that were established in the course of the interactions. Additional self-report attitude measures should be administered in future studies to clarify this issue.
As we analyzed only comments of users that made a reference to the in- or outgroup, or that were involved in direct intergroup contact, we cannot derive whether intergroup contact affected all Facebook-group members in the same way. The frequency with which members visit the Facebook-group could moderate the impact and might be influenced by the degree to which users identify with the ingroup, and view the group membership as central to their self-concept. Group members strive for positive distinctiveness, that is, they aim for their ingroup being perceived as distinct from, and positively discrepant in comparison to, the outgroup. 3 Positive distinctiveness can be under threat when engaging in intergroup contact. Members that are highly identified with the ingroup might be inclined to avoid such situations to maintain a high self-esteem. 3 As a result, they might not join or post a comment to the outgroup, which means that only a selective group of outgroup members being less identified and more open to diverse perspectives might have been included in our analysis.
To sum up our findings and these reflections, Steele and Brown's model of media practice 31 provides a suitable framework. According to the authors, media selection and consumption, that is, the choice to join and to be active in a particular Facebook-group, are influenced by the users' motivations, as just discussed. The engagement and interaction with media, that is, intergroup contact in Facebook-groups, are in return integrated in the users' understanding of her/himself: the user defines her/himself as being less prejudiced toward outgroup members with diverse opinions. This incorporation of media effects will affect future selections of media, for example, potentially more frequent visits of heterogeneous Facebook-groups or reading newspapers that discuss topics from diverse angels and represent not just opinions similar to the users', which might drive the reduction of prejudices further. This emphasizes a relevant and sustainable contribution of intergroup contact in Facebook-groups to improved intergroup relations, making it a considerable alternative to face-to-face contact. 32
Footnotes
Acknowledgments
The research for this article was supported by the Belgian F.R.S.-FNRS aspirant fellowship that was awarded to the first author between 2010 and 2012.
Author Disclosure Statement
For none of the authors, competing financial interests exist.
*
The term Facebook-group is used to emphasize the difference between Facebook-groups as features of the social network site Facebook, and groups as social categories. The data of this study were collected in Facebook-groups, but based on the high similarity with Facebook pages, we include the latter also in the argument, even though they are not further mentioned in the article.
