A social cognitive theory approach to the effects of mediated intergroup contact on intergroup attitudes.

Intergroup contact is an effective approach for the reduction of prejudice, negative stereotyping, and discrimination. In order to produce positive outcomes, Allport (1954) argued that certain conditions within the contact situation have to be met: equal status among the individuals; individuals share common goals; individuals work together to achieve such goals; and, contact has the support of authorities (i.e., social norms favor intergroup cooperation and interaction) (Gaertner & Dovidio, 2000; Pettigrew, 1998). Almost 50 years after Allport's original work, Pettigrew and Tropp's (2000) meta-analysis showed that contact meeting Allport's conditions resulted in decreased intergroup bias. Optimal intergroup contact, however, can be difficult to achieve given the anxiety and hostility that sometimes pervade intergroup relations (Stephan & Stephan, 1985). This anxiety and hostility carries the threat of creating negative rather than positive outcomes. This study examines whether vicariously experiencing optimal intergroup contact in the media provides similar effects to real world intergroup contact, without the risk of accompanying anxiety. Below is an examination of the intergroup contact literature and social cognitive theory as the bases for the specific hypotheses in this study.

Intergroup Contact Theory

One central area of concern in contact theory has been the extent to which a specific positive intergroup experience generalizes to broader attitudes. Can a single conversation with an older adult, for instance, change a young person's more general attitudes about older people? Following Allport's (1954) initial formulation of the contact hypothesis, Hewstone and Brown (1986) argued that group membership typicality or representativeness in intergroup encounters facilitates generalization from a specific experience to more general attitudes. If an outgroup member is not seen as representative of his/her group, then contact is considered interpersonal and the effects will not generalize--the outgroup member may be treated as an exception. When the person is viewed as representative of the group, then treating them as an exception, or ignoring group memberships becomes more difficult and the specific encounter is more likely to be generalized. Evidence for the effects of group typicality in facilitating generalization from individual encounters to intergroup attitudes has emerged in a variety of contexts (e.g., attitudes toward immigrants: Voci & Hewstone, 2003; attitudes toward older adults: Harwood, Hewstone, Paolini, & Voci, 2005). However, maintaining group typicality while also meeting Allport's conditions for optimal intergroup contact is challenging, both because individuals inevitably learn individuating information during interactions (which renders the encounters more interpersonal), and because group-based information activates negative stereotypes and emotions, encouraging negative rather than positive outcomes (Hewstone, 1996).

Negative emotions, particularly anxiety, are common in intergroup contact (Greenland & Brown, 1999; Stephan & Stephan, 1985), and high anxiety suppresses positive effects of contact (Paolini, Hewstone, Cairns, & Voci, 2004). Anxiety also arises at the mere anticipation of future intergroup interaction, as individuals anticipate negative consequences associated with their behavior during such interactions. Anxious people rely more on stereotypes when making judgments and may even avoid intergroup interaction altogether. Prior levels of intergroup contact affect anxiety, such that individuals with low levels of prior contact are more likely to experience anxiety. Anxiety can be reduced by establishing clear expectations for behavior during intergroup contact (Stephan & Stephan, 1985).

Although the majority of previous research has focused on the experience of direct contact with the outgroup, recent work has begun to examine various types of indirect contact, including knowledge that a friend has positive intergroup relations (Wright, Aron, McLaughlin-Volpe, & Ropp, 1997), and contact via the media (Schiappa, Gregg, & Hewes, 2005). Indirect contact has the advantage of being less subject to the anxiety that occurs in direct interaction, and thus less subject to the negative consequences of that anxiety. Schiappa et al. demonstrate that exposure to media portrayals of homosexuals results in reduced prejudice toward gay men. Interpretation of their effects as analogous to a contact effect is strengthened by findings that the effects are strongest among straight people who have relatively little interpersonal contact with gay people. Schiappa et al. frame their findings in terms of a parasocial contact hypothesis. Specifically, they hypothesize that contact with the (mediated) outgroup member results in increased knowledge about the outgroup, and a feeling of increased trust or respect for the outgroup. However, they note the difficulty in pinpointing the precise process by which such change occurs. The current paper considers television's potential to influence intergroup attitudes from the perspective of social cognitive theory (Bandura, 2002), a framework that specifies the process by which attitude change might be occurring, as well as providing some unique hypotheses relating to these effects.

Social Cognitive Theory

According to social cognitive theory (Bandura, 2002), humans are endowed with the capacity to learn from observation. Through observation, individuals can internalize cognitive, affective, and behavioral responses to situations that they do not experience directly. Once learned, individuals can emulate these responses in similar situations (Bandura, 2002). Vicarious experiences can be gained both in one's direct environment and through models observed in the media (Bandura, 2002). This contention has implications for portrayals of intergroup interactions on television. That is, audience members can learn positive intergroup behaviors from observing televised portrayals of characters engaging in positive intergroup contact.

Beyond learning positive behaviors, individuals can also learn positive attitudes concerning intergroup contact and outgroups via abstract modeling, the process by which individuals adopt rules learned through vicarious experience and apply them to different contexts (Bandura, 2002). Through observing media models, people can extract rules governing judgments and behaviors in the observed context, and apply those rules to guide behavior in different situations (Bandura, 1986). These rules also influence people's attitudes and their probability of expressing a particular attitude in a given context (Eyal & Rubin, 2003). When exposed to TV images of positive intergroup contact, for example, viewers may extract a rule that such interaction is open and friendly. They may then extrapolate this rule and use it to guide their behaviors and judgments in future situations where the rule might be applicable (i.e., other intergroup interactions).

Viewers also learn affective responses from symbolic interaction (Bandura, 1999). Social cognitive theory holds that observing characters ("models") display affective expressions creates affective arousal in the viewer. Viewers come to associate targets with emotions based on models' affective responses when encountering the target (Bandura, 1999), and thus develop the same emotional reaction regarding the target. Viewers may emulate ingroup characters' emotional responses to outgroup members. Thus, individuals model positive emotional reactions to outgroup members and develop positive attitudes regarding the outgroup. This leads to the first hypotheses:

[H.sub.1a]:Television exposure to positive intergroup interactions involving an ingroup member will be associated with lower levels of intergroup anxiety.

[H.sub.1b]:Television exposure to positive intergroup interactions involving an ingroup member will be associated with more positive attitudes toward the outgroup.

Social cognitive theory would not predict this association to be similar for all television characters; that is, while viewers are confronted with numerous models from which to gain vicarious experience, they will emulate some characters more than others (Bandura, 1977). Vicarious learning, a central tenet of social cognitive theory, involves immersion into certain characters' perspective (Bandura, 2002). Thus, the theory suggests that viewers' identification with a character influences their modeling of the character (Eyal & Rubin, 2003). Identification occurs when viewers perceive themselves as similar to a character and vicariously participate in the character's experiences (Hoffner, 1996). Identification occurs when individuals view themselves as the character within the program; adopt the character's perspective; experience and understand the character's emotions; and understand how and why the character acts the way he or she does (Cohen, 2001). Thus, in situations of exposure to portrayals of positive intergroup interaction, this leads to the following hypotheses.

[H.sub.2a:] Higher levels of identification with the ingroup character will be associated with lower levels of intergroup anxiety.

[H.sub.2b]: Higher levels of identification with the ingroup character will be associated with more positive attitudes toward the outgroup.

In addition to this main effect of identification, it is predicted that identification will interact with level of viewing specific shows featuring positive intergroup interactions. Specifically, the associations of viewing with intergroup anxiety and attitudes should be greater for those who identify more strongly with the ingroup character.

[H.sub.3]: The associations hypothesized in [H.sub.1a] and [H.sub.1b] will be moderated by ingroup character identification: Stronger associations between viewing and the outcome variables will be observed with higher levels of ingroup character identification.

Finally, the concept of group typicality discussed earlier must be incorporated. Hewstone and Brown's (1986) model says that generalization from a specific intergroup interaction to outgroup attitudes overall is moderated by group typicality. That is, the more the outgroup character is perceived as representative of his or her group, the more likely exposure to positive intergroup contact will translate into positive intergroup attitudes. The Hewstone and Brown effect is consistent with the present theoretical perspective. Social cognitive theory states that viewers categorize events (including other individuals) on shared attributes (Bandura, 1986). The more shared attributes between situations, the greater the likelihood that a learned experience such as positive intergroup contact will generalize to other similar situations (stimulus generalization: Segrin, Taylor, & Altman, 2005). Extending this argument to the intergroup context, exposure to media characters perceived as representative of a specific group should be more influential in future intergroup situations than exposure to contact featuring atypical outgroup members. Exposure to the typical outgroup member results in more shared attributes between the media exposure and the subsequent intergroup situation than exposure to the atypical member, and hence greater generalizability of the media encounter to more general attitudes. Therefore, typicality of an outgroup member in a mediated portrayal of positive interaction should be associated with more positive attitudes. Similarly, perceived typicality should moderate the association between exposure and the outcome variables described in [H.sub.1]--that association should be more intense among those who see the character as typical of the outgroup.

[H.sub.4a]: As perceptions of outgroup members' group typicality increase, levels of intergroup anxiety are predicted to decrease.

[H.sub.4b]: As perceptions of outgroup members' group typicality increase, more positive attitudes toward the outgroup will be reported.

[H.sub.5]: The associations from [H.sub.1a] and [H.sub.1b] will be moderated by outgroup character typicality: Stronger associations will be observed with higher typicality levels.

Thus, the goal of the current study was to extend Schiappa et al.'s (2005) parasocial contact hypothesis by (a) integrating the effect into social cognitive theory and testing links derived from that perspective (the identification hypothesis), (b) examining effects of observing intergroup interaction rather than merely outgroup representations, and (c) incorporating additional hypotheses derived from contact theory (Hewstone and Brown's group typicality effect).

Method

Pilot Study

A pilot study was conducted to discover which televised intergroup interactions were seen most often by the subject population. Participants for the pilot study (N = 25) were drawn from the same population as the main study (none participated in both studies). They were asked to think of a current TV show involving a friendship or romance between a Black and a White character, and to write the name of the show and the relevant characters (or a description of the characters if they could not think of their names). This procedure was repeated for Latino-White, gay-straight, and young-old interactions. The most popular relationships reported were those between Will and Grace from NBC's Will & Grace (N = 17: gay-straight) and between Nehemiah and Wes from MTV's Real World: Austin (N = 4: Black-White). These relationships were used for the study.

Main Study: Participants and Procedure

Undergraduate student volunteers (N = 253, 61.30% women; age M = 21.24, SD = 1.76) were recruited from communication courses at a large southwestern university. Participants received extra credit. Only responses from subjects who identified themselves straight (N = 248; 61.30% women; age M = 21.17, SD = 1.54) or White (N = 210; 61.90% women; age M = 21.17, SD= 1.78) were used for the analyses involving Will & Grace and Real World: Austin, respectively. For hypotheses [H.sub.1a] and [H.sub.1b], responses from all such participants were used. All other analyses require familiarity with the show (e.g., they involve character identification), and so only responses from straight/White participants who were familiar with the show in question were used (Will & Grace: N = 187; 71.10% women; age M = 21.01, SD = 1.16; Real World: Austin: N = 161; 62.10% women; age M = 20.93, SD = 1.07).

Independent Variables

TV Exposure. Five items assessed respondents' total TV exposure (Mastro, Behm-Morawitz, & Ortiz, 2007). Participants indicated the amount of hours they spent watching TV the previous night, afternoon, and morning, and, on average, how many hours of TV they watch in a day. Responses to these questions were summed and divided by 2 to get a measure of average daily TV exposure. Two items assessed exposure to Will & Grace and Real World: Austin on 6-point scales, 1 (never watched it); 6 (watch it multiple times a week).

Ingroup Character Identification. Identification while viewing was measured for the ingroup characters in each intergroup dyad using items from Eyal and Rubin's (2003) character identification scale. Participants indicated how much they agreed with six statements (e.g., "At key moments in the show, I feel like I know exactly what Grace is going through,"), on a 1-5 scale (strongly disagree to strongly agree). (1) Responses were averaged (Cronbach's [alpha] = .93 for Grace, .91 for Wes). Respondents who had never seen the program featuring the character skipped this section.

Group Typicality. This was measured by averaging four items from Harwood et al. (2005). Items assessed perceptions of each outgroup character (Will and Nehemiah) as representative members of their respective group, and how similar they are to others within their respective group (e.g., "How similar is Will to other gay people?"). Responses ranged from 1 (very little)to 7 (a great deal) ([alpha] = .91 for Will and .94 for Nehemiah).

Dependent Variables

Intergroup Anxiety. A short version of Stephan and Stephan's (1985) intergroup anxiety scale was used. The scale measures how much participants would feel six different anxiety-related states interacting with an outgroup stranger (i.e., a Black or gay person). The terms used in this study were: relaxed, awkward, comfortable, threatened, at ease, and tense. Response options ranged from 1 (not at all) to 5 (extremely) with high scores indicating more anxiety. Items were reverse coded where necessary and responses averaged (anxiety with gay contact, [alpha] = .89; Black, [alpha] = .89).

Attitudes Regarding the Outgroup. Participants completed a modified version of Esses and Dovidio's (2002) social distance scale, which asked participants their willingness to engage in a series of behaviors with outgroup members (e.g., confiding in an outgroup member; having an outgroup member as a boss). Social distance is one of the earliest measurements used as an indicator of attitudes toward others (Aiken, 2002; Bogardus, 1925). Response options ranged from 1 (extremely willing) to 5 (not at all willing). Scores on the items were averaged to create composite scores for attitudes toward Black people ([alpha] = .82) and gay people ([alpha] = .94).

Control Variable

Quality of Intergroup Contact. The quality of non-mediated contact with outgroup members was assessed with 12 items, 6 focused on contact with Black people and 6 for contact with gay people (Mastro et al., 2007). First, respondents were asked to think of the outgroup member with whom they have the closest relationship. On a scale ranging from 1-5, respondents indicated how close they felt to this person, how much they value the time they have spent with this person, how much they valued their relationship with this person, and the quality of their relationship. Additionally, respondents were asked to indicate how pleasant and friendly, in general, their contact has been with members of the outgroup. (2) Responses were averaged (Black, [alpha] = .88; gay, [alpha] = .92). Descriptive statistics for all variables are in Table 1.

Results

Hypotheses la and lb postulated that exposure to media interactions involving positive intergroup contact would be associated with lower levels of intergroup anxiety and positive outgroup attitudes. This was tested using regression analyses in which exposure to the respective shows were predictor variables; intergroup anxiety and attitudes toward the outgroup were criterion variables; while quality of intergroup contact and daily television viewing were controlled. (3) Support was not found for [H.sub.1a]. Straight respondents' exposure to Will & Grace did not significantly predict levels of intergroup anxiety ([beta] = -.08, t(209) =-1.34, p > .05), and White respondents' exposure to Real World: Austin did not significantly predict intergroup anxiety ([beta] = .11, t(183) = 1.53, p > .05). However, supporting [H.sub.1b], exposure to Will & Grace significantly predicted low social distance toward gay people ([beta] =-.16, t(236) = -3.30, p = .001, [r.sub.p.sup.2] = .04), and exposure to Real World: Austin was associated with less social distance towards Blacks ([beta] = -.13, t(203) = -1.98, p < .05, [r.sub.p.sup.2] = .02). Across all analyses the control measure of quality of intergroup contact significantly predicted the attitudes measures (more contact associated with less social distance and anxiety, all p < .001). None of the analyses yielded significant effects for overall TV viewing. Full statistics for these analyses are available from the authors.

Multiple regression analyses were performed to test [H.sub.2]-[H.sub.5] using Aiken and West's (1991) procedures for dealing with moderator effects. In each analysis, viewing of the show was entered along with either ingroup character identification or perceived outgroup character typicality. The mean-centered interaction term was entered next (either Viewing x Identification, or Viewing x Typicality). If significant, the interaction effect was decomposed by regressing the criterion variable on show viewing at low (-1 SD), moderate (0 SD), and high (+1 SD) levels of identification. Tables 1 and 2 display the results of these regression analyses.

Hypotheses 2a and 2b postulated that identification with the ingroup character would be related to intergroup anxiety and social distance. These hypotheses were only supported for Will & Grace. Controlling for level of viewing of the show, identification with Grace significantly predicted both intergroup anxiety and social distance (see Table 2). Those who identified more with Grace had less intergroup anxiety and less social distance with regard to gay people. For Wes (the White character in Real World), the effect on anxiety was not significant (see Table 2). However, counter to predictions, identification with Wes was associated with more social distance from Blacks.

The third hypothesis predicted that identification with the ingroup character would moderate the relationship between exposure to the specific program (Will & Grace or Real World: Austin) and the criterion variables. This was partially supported--the interaction between viewing Will & Grace and identification with Grace was marginally significant for anxiety and significant for social distance (see Table 2). For low identification with Grace (-1 SD), increased exposure to Will & Grace is associated with reduced intergroup anxiety ([beta] = -.30, p < .05) and social distance ([beta] = -.38, p < .001). At moderate levels of identification (0 SD) this relationship did not emerge for anxiety ([beta] = -.14, p > .05), but it still emerged for social distance ([beta] = -.25, p < .01). For high identification with Grace (+1 SD), no relationship was found between exposure and anxiety ([beta] = .02, p > .05) or between exposure and social distance ([beta] = -.01, p > .05). No moderator effects occurred with regard to Real World.

Hypotheses 4a and 4b postulated that perceived typicality of the outgroup member would be associated with anxiety and social distance. Partial support only was found for hypothesis 4b. As shown in Table 3, those who perceive Will as more typical of gay men have lower levels of reported social distance (but not anxiety) with regard to gay people. No other effects of typicality emerged, and none of the interaction effects predicted by H5 were significant (see Table 3).

Discussion

This study tested hypotheses derived from a social cognitive theory approach to mediated intergroup contact. Specifically, it was hypothesized that exposure to intergroup interactions involving an ingroup member would reduce intergroup anxiety and improve attitudes toward the outgroup. Limited support was found for this hypothesis in examination of two settings of mediated intergroup contact (Will & Grace, a sitcom featuring a straight-gay friendship, and Real World: Austin, a reality show featuring contact between White and Black cast members). In each setting, exposure to the show was associated with less social distance concerning the pertinent outgroups among, respectively, straight and White viewers.

While causality cannot be inferred due to the cross-sectional nature of the research, it is plausible that viewing of these TV shows may have a unique contribution to attitudes toward the outgroup. One powerful alternative explanation of these findings is that people with more positive attitudes selectively seek out media images of these groups (or conversely, that those with negative attitudes avoid such exposure). Certain control variables in the current research (e.g., level of real world contact with the outgroup) provide partial control for such explanations. Selective seeking of intergroup contact in the real world and on television are likely to be somewhat related, and hence selective seeking on television is partially accounted for by the real world measure. Nonetheless, contact on television and in the real world are clearly distinct, and in neither case is all contact voluntary and actively sought out, so the authors do not claim to have ruled out the selective seeking explanation for the findings. The causal direction preferred is supported by research demonstrating effects similar to these in experimental designs (e.g., Schiappa et al., 2005).

The results provide some support for the application of intergroup contact theory to mediated settings, and are consistent with other similar studies (e.g., Schiappa et al., 2005). However, this study posited a different theoretical explanation for the effects of mediated contact, and provides preliminary evidence for that theory. Prior work focused on media's role in exposing people to an outgroup, facilitating learning about that outgroup and inculcating more positive attitudes. The social cognitive theory approach suggests that viewers observe and learn about engaging in intergroup contact. Thus, the shows provide models for positive intergroup interaction, models in which the behavior of the ingroup character involved in the interaction is at least as important as that of the outgroup member. Repeated exposure to positive intergroup interactions potentially provides a model (the relevant ingroup participant) who can be imitated when people find themselves in a similar intergroup setting. In support of a social cognitive theory explanation, identification with Grace was shown to be associated with more positive attitudes, suggestive of a modeling effect whereby viewers who are more immersed in Grace's character are also more inclined to adopt her orientation towards gay people. This effect is not predicted by the parasocial contact approach, although it is not inconsistent with it. If exposure to Will results in learning about the outgroup, it would not be a stretch to imagine that identification with Grace would enhance such learning. Again, the authors should acknowledge limits in the ability to infer causality here: Clearly Grace holds positive attitudes toward gays, and it is reasonable that viewers holding similar attitudes are likely to identify with Grace.

Some of these results were counter to expectations. The one significant moderator effect for character identification was not in the predicted direction: More exposure to Will & Grace was associated with lower anxiety and social distance toward gay people for those who identified less with Grace. This could be caused by a floor effect on the attitudes measures among high identifiers--high identifiers report very low mean levels of social distance and anxiety, which result in standard deviations that are lower for high identifiers than low identifiers. (4) This restricted variation would suppress the predicted strong correlation between viewing and attitudes among the high identifiers. Among low identifiers, the greater variation allows the viewing-attitudes association to show. Other explanations for this effect are certainly possible. For instance, some viewers might perceive a quasi romantic relationship between Will and Grace that is impeded by Will's sexuality. For these viewers, identification with Grace might manifest in mixed attitudes concerning gay sexuality, and hence an absence of the predicted relationship between exposure to the show and attitudes. More generally, somewhat restricted variance on the outcome measures probably impeded the ability to detect moderator effects across many of these analyses (McClelland & Judd, 1993). Future work should examine groups that demonstrate more variation on attitudinal measures (e.g., older adults, the mentally ill).

This study found fewer effects of Real World: Austin than Will & Grace. Real World features an ensemble cast that changes each season; hence, the per-episode exposure to the Black-White dyad on this show is relatively low, and the dyad only exists for a single season. The show also features less positive interaction than Will & Grace, a show in which the lead characters have a long-term intimate friendship. (5) Real World is also a reality show; it is possible that reality television does not incur identification with the characters to the same extent as fictional programs do, and hence that viewing intergroup relationships on reality shows will not conform to hypotheses premised on identification processes. Finally, portrayals of Blacks in the media are substantially more equivocal and common than portrayals of gay people. A single portrayal of an interracial relationship may be less influential than a portrayal of gay-straight relationship, simply due to the larger number of Blacks on television and the fact that a fairly wide array of clearly negative portrayals of interracial contact are apparent on TV (e.g., Mastro & Behm-Morawitz, 2005).

The authors acknowledge that the explanations above constitute somewhat piecemeal accounting for a set of results that is only partially supportive of the hypotheses. Together, however, they suggest a developing (if tentative) list of boundary conditions for the hypothesized effects. It is suspected that these hypotheses are most likely to be supported by media portrayals of deep, long-lasting intergroup relationships that receive maximal screen time. Television may, therefore, be more effective than one-time exposures from movies. The hypotheses may also be more relevant to intergroup contexts in which the outgroup is relatively infrequently portrayed in the media, and where attitudes are variable in the population (e.g., Asian Americans, fat people). Fictional portrayals may well have more impact than other portrayals. Of course, research is needed to substantiate the relevance of these boundary conditions.

Theoretical Considerations

As outlined in the introduction, previous work on contact, including mediated contact, has presented a variety of potential "mechanisms" for how these effects occur. The most common explanations focus on how contact provides information about diversity in the outgroup (Soliz & Harwood, 2003), increased knowledge concerning the outgroup (Schiappa et al., 2005), reduced anxiety and increased intimacy in dealing with outgroup members (Paolini et al., 2004), a "recategorization" of outgroup members into a new more inclusive ingroup (Gaertner & Dovidio, 2000), and exposure to counter-stereotypical (stereotype-disconfirming) outgroup members (Cook, 1978). The present approach, which is geared toward mediated contact, takes a somewhat different tack, focusing instead on ingroup identification and awareness, the modeling of appropriate intergroup behaviors, and the symbolic extrapolation from positive intergroup behavior to more general attitudes. Thus, from a social cognitive theory perspective, an association between exposure to positive intergroup interaction and intergroup attitudes has been shown. Similarly, preliminary indications of a link between identification with an ingroup member involved in intergroup contact and positive intergroup attitudes emerge from this research.

Of course, some of the results could be explained by other theories. Heider's (1958) balance theory, for instance, stipulates that individuals are motivated to maintain balance between their cognitions. If imbalance exists, a change will occur in one of the cognitions to achieve balance. Thus, if a viewer likes Grace, disliking Will creates imbalance because Grace likes Will. Balance is regained by developing a stronger affinity for Will. The extended contact effect (Wright et al., 1997) could also explain the results. According to the hypothesis, knowledge of an intergroup friendship involving an ingroup member is enough to lead to positive attitude change. Wright et al. demonstrate the effect, and note a variety of possible theoretical explanations (including mention of balance theory). They also suggest that an inclusion of other in the self effect may be occurring, which resembles closely the identification explanation. The extended contact effect is, therefore, consistent with these findings and the social cognitive theory explanation.

Notwithstanding the limitations of this study, the current work offers promise for reducing prejudice and enhancing tolerance. As noted earlier, intergroup contact is often anxiety laden, and in face-to-face contexts anxiety is exacerbated by high levels of group salience and perceived typicality--when outgroup members are seen as representative of their group, anxiety increases. However, typicality and salience are essential for generalization from specific contact situations to broader attitudes. This has presented an unpleasant conundrum: Encounters most likely to influence general attitudes are also those most likely to be imbued with anxiety. The media provide models for positive interaction that are viewed in an anxiety-free atmosphere. Unlike in face-to-face contexts, outgroup TV characters who are typical of their group need not be viewed with any more anxiety than other outgroup characters; similarly, in the scripted environment it is possible for encounters high in group salience to retain a positive tone. Will & Grace provides an example of an intimate, caring, and nuanced relationship between two characters that nonetheless retains Will's sexuality as a salient element. This is the perfect recipe for positive attitude change, and one that is challenging to achieve in face-to-face contact. Also, whereas other perspectives have focused on changing orientations to outgroup members, the present study emphasizes the ingroup member's role in the intergroup interaction. Intuition and theory suggest that it is easier to develop a positive orientation to an ingroup member than an outgroup member, and this work suggests that this may be productive (e.g., enhancing identification with the ingroup member has positive results for attitudes towards the outgroup). Thus, this approach offers some partial, yet novel routes to improving intergroup attitudes. These routes capitalize on the integration of observational learning and exposure to intergroup contact in the media context. More broadly, it is hoped that the current authors add to the momentum of work examining mass communication from an intergroup perspective (Harwood & Roy, 2005; Mastro, 2003): As central forces shaping culture, the media deserve attention for their role in influencing relations between social groups.

References

Aiken, L. R. (2002). Attitudes and related psychosocial constructs. Thousand Oaks, CA: Sage.

Aiken, L. S., & West, S. G. (1991). Multiple regression: Testing and interpreting interactions. Newbury Park, CA: Sage.

Allport, G. W. (1954). The nature of prejudice. New York: Addison-Wesley.

Bandura, A. (1977). Social learning theory. Englewood Cliffs, N J: Prentice Hall.

Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall.

Bandura, A. (1999). Social cognitive theory of personality. In D. Cervone & Y. Shoda (Eds.), The coherence of personality (pp. 185-241). New York: Guilford Press.

Bandura, A. (2002). Social cognitive theory of mass communication. In J. Bryant & D. Zillmann (Eds.), Media effects: Advances in theory and research (2nd ed., pp. 121-153). Hillsdale N J: Lawrence Erlbaum Associates, Inc.

Bogardus, E. S. (1925). Measuring social distances. Journal of Applied Sociology, 9, 299-308.

Canary, D. J., & Hause, K. S. (1993). Is there any reason to research sex differences in communication? Communication Quarterly, 4, 129-145.

Cohen, J. (2001). Defining identification: A theoretical look at the identification of audiences with media characters. Mass Communication & Society, 4, 245-264.

Cook, S. W. (1978). Interpersonal and attitudinal outcomes in cooperating interracial groups. Journal of Research in Developmental Education, 12, 97-113.

Esses, V. M., & Dovidio, J. F. (2002). The role of emotions in determining willingness to engage in intergroup contact. Personality and Social Psychology Bulletin, 28, 1202-1214.

Eyal, K., & Rubin, A. M. (2003). Viewer aggression and homophily, identification, and parasocial relationships with television characters. Journal of Broadcasting & Electronic Media, 47, 77-98.

Gaertner, S., & Dovidio, J. (2000). Reducing intergroup bias. Hove, UK: Psychology Press.

Greenland, K., & Brown, R. (1999). Categorization and intergroup anxiety in contact between British and Japanese nationals. European Journal of Social Psychology, 29, 503-521.

Harwood, J., Hewstone, M., Paolini, S., & Voci, A. (2005). Grandparent-grandchild contact and attitudes towards older adults: Moderator and mediator effects. Personality and Social Psychology Bulletin, 31, 393-406.

Harwood, J., & Roy, A. (2005). Social identity theory and mass communication research. In J. Harwood & H. Giles (Eds.), Intergroup Communication: Multiple perspectives (pp. 189-212). New York: Peter Lang.

Heider, F. (1958). The psychology of interpersonal relations. New York: Wiley.

Hewstone, M. (1996). Contact and categorization: Social psychological interventions to change intergroup relations. In N. Macrae, C. Stangor, & M. Hewstone (Eds.), Stereotypes and stereotyping (pp. 33-368). New York: Guilford Press.

Hewstone, M., & Brown, R. (1986). Contact is not enough: An intergroup perspective on the "Contact Hypothesis." In M. Hewstone & R. Brown (Eds.), Contact and conflict in intergroup encounters (pp. 1-44). Oxford, UK: Blackwell.

Hoffner, C. (1996). Children's wishful identification and parasocial interaction with favorite television characters. Journal of Broadcasting & Electronic Media, 40, 389-393.

Mastro, D. E. (2003). A social identity approach to understanding the impact of television messages. Communication Monographs, 70, 98-113.

Mastro, D. E., & Behm-Morawitz, E. (2005). Latino representation on primetime television. Journalism and Mass Communication Quarterly, 82, 110-130.

Mastro, D., Behm-Morawitz, E., & Ortiz, M. (2007). The cultivation of social perceptions of Latinos: A mental models approach. Media Psychology, 9, 347-365.

McClelland, G. H., & Judd, C. M. (1993). Statistical difficulties in detecting interactions and moderator effects. Psychological Bulletin, 114 (2), 376-390.

Paolini, S., Hewstone, M., Cairns, E., & Voci, A. (2004). Effects of direct and indirect cross-group friendships on judgments of Catholics and Protestants in Northern Ireland: The mediating role of an anxiety-reduction mechanism. Personality and Social Psychology Bulletin, 30, 770-786.

Pettigrew, T. F. (1998). Intergroup contact theory. Annual Review of Psychology, 49, 65-85.

Pettigrew, T. F., & Tropp, L. R. (2000). Does intergroup contact reduce prejudice? Recent meta-analytic findings. In S. Oskamp (Ed.), Reducing prejudice and discrimination (pp. 93-114). Hillsdale, N J: Lawrence Erlbaum Associates, Inc.

Schiappa, E., Gregg, P. B., & Hewes, D. E. (2005). The parasocial contact hypothesis. Communication Monographs, 72, 92-115.

Segrin, C., Taylor, M. E., & Altman, J. (2005). Social cognitive mediators and relational outcomes associated with parental divorce. Journal of Social and Personal Relationships, 22, 361-377.

Soliz, J., & Harwood, J. (2003). Perceptions of communication in a family relationship and the reduction of intergroup prejudice. Journal of Applied Communication Research, 31, 320-345.

Stephan, W. G., & Stephan, C. W. (1985). Intergroup anxiety. Journal of Social Issues, 41, 157-175.

Voci, A., & Hewstone, M. (2003). Intergroup contact and prejudice toward immigrants in Italy: The mediational role of anxiety and the moderational role of group salience. Group Processes and Intergroup Relations, 6, 37-54.

Wright, S. C., Aron, A., McLaughlin-Volpe, T., & Ropp, S. A. (1997). The extended contact effect: Knowledge of cross-group friendships and prejudice. Journal of Personality and Social Psychology, 73, 73-90.

Notes

(1) The remaining items are: "While viewing Will & Grace I can feel the emotions Grace portrays," "I think I have a good understanding of Grace," "I tend to understand the reasons why Grace does what she does," "When I watch Grace on Will & Grace, I feel I understand the way she feels," and "While viewing Will & Grace, I feel I can really get inside Grace's head."

(2) Only a small number of respondents could not think of any outgroup member with whom they have a relationship. These individuals were excluded from the analyses involving this measure (N = 7 for gay outgroup; N = 2 for Black outgroup).

(3) Gender was not included as a control because it did not flow from theory to include it as such. It is argued, in line with Canary and Hause (1993), that sex does not have any privileged status as a variable, and that examinations of sex differences should only proceed with sufficient theoretical rationale.

(4) High identifiers' anxiety SD = .66; social distance SD = .61. Low identifiers' anxiety SD = .78; social distance SD = 1.02; Levene's F = 3.75, p = .05, and F = 24.19, p < .001, respectively).

(5) Viewers of both shows rated the Will-Grace relationship (M = 4.39, SD = .56) as higher quality than Wes-Nehemiah's (M = 3.38, SD = .68; t(146) = 15.38, p < .001, (r.sup.2] = .62).

Michelle Ortiz (M.A., University of Arizona) is a doctoral student in the Department of Communication at the University of Arizona. Her research interests include intergroup processes and media effects.

Jake Harwood (Ph.D., University of California, Santa Barbara) is a Professor in the Department of Communication at the University of Arizona. His research interests include intergroup processes and aging. Table 1 Descriptive Statistics for All Variables

M SD Daily TV exposure (hours) 2.92 1.89 Exposure to specific

television show

Will & Grace 2.87 1.47

Real World: Austin 3.47 1.67 Identification

Grace 2.83 .99

Wes 2.62 .95 Perceived typicality

Will 4.23 1.19

Nehemiah 4.33 1.15 Intergroup anxiety

Gay 1.95 .77

Black 2.11 .71 Social distance

Gay 1.83 .97

Black 1.41 .48 Quality of intergroup

contact

Gay 3.70 1.01

Black 3.95 .81 Note: All variables are measured on a 1-5 scale except typicality (1-7) and exposure to specific shows (1-6). High scores indicate more of the construct (e.g., higher anxiety, greater typicality). Table 2 Hierarchical Regression Analyses Predicting Intergroup Anxiety and Social Distance from TV Exposure to Will & Grace and Character Identification with Grace (N = 163) or Real World. Austin and Character Identification with Wes (N = 147)

[r.sub.p.

[beta] t sup.2] Anxiety Step 1

Viewing--Will & Grace -.13 -1.56

Identification--Grace -.23 -2.76 ** .05 Step 2

Viewing 3 Identification .16 1.98 ([dagger]) .02 Social distance Step 1

Viewing--Will & Grace -.23 -2.87 ** .05

Identification--Grace -.25 -3.09 ** .05 Step 2

Viewing 3 Identification .20 2.73 ** .04 Anxiety Step 1

Viewing--Real World .13 1.53

Identification--Wes .00 -.01 Step 2

Viewing 3 Identification -.11 -1.16 Social distance Step 1

Viewing--Real World -.12 -1.43

Identification--Wes .18 2.11 * .03 Step 2

Viewing 3 Identification .00 .06

[R.sup.2] F Anxiety Step 1

Viewing--Will & Grace .08 8.13 ***

Identification--Grace Step 2

Viewing 3 Identification .10 6.82 *** Social distance Step 1

Viewing--Will & Grace .15 15.03 ***

Identification--Grace Step 2

Viewing 3 Identification .18 12.91 *** Anxiety Step 1

Viewing--Real World .00 1.21

Identification--Wes Step 2

Viewing 3 Identification .03 1.26 Social distance Step 1

Viewing--Real World .02 2.80

Identification--Wes Step 2

Viewing 3 Identification .02 1.85 ([dagger])p < .10, * p <.05, ** p <.01, *** P < .001 Note: In all cases, the viewing and identification measures were entered in the first step of the regression analysis, and the interaction (moderator) term was entered in the second step. Table 3 Hierarchical Regression Analyses Predicting Intergroup Anxiety and Social Distance from TV Exposure to Will & Grace and Perceived Typicality of Will (N= 164) as well as to Real World: Austin and Perceived Typicality of Nehemiah (N = 141)

[r.sub.p.

[beta] t sup.2] Anxiety Step 1

Viewing--Will & Grace -.19 -2.51 * .04

Perceived Typicality-Will -.14 -1.76 Step 2

Viewing 3 Perceived Typicality -.06 -.75 Social distance Step 1

Viewing--Will & Grace -.31 -4.23 *** .10

Perceived Typicality--Will -.15 -1.99 * .02 Step 2

Viewing 3 Perceived Typicality .05 .67 Anxiety Step 1

Viewing--Real World .14 1.62

Perceived Typicality--Nehemiah .08 .96 Step 2

Viewing 3 Perceived Typicality .14 1.51 Social distance Step 1

Viewing-Real World -.06 -.71

Perceived Typicality--Nehemiah -.06 -.68 Step 2

Viewing 3 Perceived Typicality .14 1.51

[R.sup.2] F Anxiety Step 1

Viewing--Will & Grace .05 5.10 **

Perceived Typicality-Will Step 2

Viewing 3 Perceived Typicality .05 3.58 * Social distance Step 1

Viewing--Will & Grace .12 11.70 ***

Perceived Typicality--Will Step 2

Viewing 3 Perceived Typicality .11 7.92 *** Anxiety Step 1

Viewing--Real World .01 2.02

Perceived Typicality--Nehemiah Step 2

Viewing 3 Perceived Typicality .04 2.11 Social distance Step 1

Viewing-Real World -.01 .56

Perceived Typicality--Nehemiah Step 2

Viewing 3 Perceived Typicality .00 1.13 * p < .05; ** p < .01; *** p < .001 Note: In all cases, the viewing and perceived typicality measures were entered in the first step of the regression analysis, and the interaction (moderator) term was entered in the second step.