Group-Derived Consumption: the Role of Similarity and Attractiveness in Identification With a Favorite Sports Team

Robert J. Fisher, University of Southern California
ABSTRACT - The present research evaluates the effects of two group characteristics (i.e., similarity and attractiveness) on group-derived identities and consumption choices. We study this topic within the context of sports fans’ identification with and support of their favorite sports team. Contrary to Kelman (1961), the results suggest that the individual’s perceived similarity with the group (rather than group attractiveness) is the most important factor leading to identification. The results have important theoretical and practical implications for group-related consumption behaviors.
[ to cite ]:
Robert J. Fisher (1998) ,"Group-Derived Consumption: the Role of Similarity and Attractiveness in Identification With a Favorite Sports Team", in NA - Advances in Consumer Research Volume 25, eds. Joseph W. Alba & J. Wesley Hutchinson, Provo, UT : Association for Consumer Research, Pages: 283-288.

Advances in Consumer Research Volume 25, 1998      Pages 283-288


Robert J. Fisher, University of Southern California


The present research evaluates the effects of two group characteristics (i.e., similarity and attractiveness) on group-derived identities and consumption choices. We study this topic within the context of sports fans’ identification with and support of their favorite sports team. Contrary to Kelman (1961), the results suggest that the individual’s perceived similarity with the group (rather than group attractiveness) is the most important factor leading to identification. The results have important theoretical and practical implications for group-related consumption behaviors.

Memberships in social groups are fundamental to both self-definition and self-esteem (Tajfel and Turner 1986; Turner 1985, 1982). Our feelings about ourselves depend on our affiliations with others and the emotional significance of these relationships. Inded, Tajfel (1982) argues that individuals are unable to form self-images in the absence of social identities derived from group affiliations. When people are asked to describe "who they are," the answer invariably reflects group associations related to family, work, religious, political, and social organizations.

Although we do not have much choice about our membership in groups defined by characteristics such as gender, age, family, and race, we do have options when it comes to many other types of groups. For example, we select our friends, and perhaps our neighbors and colleagues. We also choose to become members of political parties, religious organizations, and educational institutions. In turn, these group affiliations affect a variety of consumption choices ranging from the purchase of clothing which identifies us as a fan of the Dallas Cowboys, to the financial support we give our alma mater, and the neighborhood in which we live.

The impact of voluntary group associations on consumption makes it important to understand why individuals associate with specific groups. The present research investigates group characteristics that lead individuals to define themselves as group members and subsequently undertake group-derived consumption behaviors. We use research on identification to develop and test hypotheses about the effects of group similarity and attractiveness on the extent to which individuals are psychologically attached to (i.e., identify with) a social group. We examine this topic within the context of individuals’ identification with their favorite sports teams. The paper begins with a brief overview of the concept of identification, and then presents hypotheses related to three models of the effects of group similarity and attractiveness on self-definition. Finally, the implications of the study are discussed.


The degree to which a social group membership or affiliation affects self-definition can be defined as the strength of the individual’s identification with the group. Identification leads the individual to view the group as an extension of the self such that he or she experiences vicariously the successes and failures of the group (Kagan 1958). Group identification also motivates individuals to maintain their self-definition through group-derived behaviors (Dignan 1965; Freud 1949; Kelman 1961). Group-derived behaviors are mandated because they are part of a social role that the individual has accepted, and not necessarily because they are intrinsically satisfying.

Highly-identified sports fans provide an illustration of the psychological and behavioral aspects of identification. Sports fans who define themselves in terms of their relationship to their team feel ecstasy when the team wins and despair when they lose because the team is an extension of themselves. These fans are motivated to learn as much as possible about the team and its players. They are also likely to wear team-licensed products, watch and attend games, read the sports page frequently, and talk about the team whenever they get the chance. In the extreme they may even participate in acts of vandalism because they believe that it is an appropriate response to a playoff loss (or win). Identification leads these fans to both publicly and privately support their team in the ways that are mandated by group membership.


Kelman (1961) argues that the sole determinant of identification is the attractiveness of the referent. Attractive referents occupy a role desired by the individual, and have qualities or characteristics that the individual desires such as power, control, wealth, fame. In the Freudian view, children identify with their parents because of the influence parents exert over the child’s environment (Freud 1949; Kagan 1958). The parents’ ability to feed, restrict, reward, and punish leads the child to desire greater mastery over these outcomes. Identification also occurs when the individual defines him- or herself in relationship to another because of the satisfaction it provides (Kelman 1961). In this instance the individual is motivated to preserve his or her connection to an attractive target. For example, a patient may identify strongly with his or her doctor because of the physician’s skill, professionalism, and warm bedside manner. The patient doesn’t actually want to be the doctor, but rather sees the relationship as an important part of the self. Similar self-defining relationships may be found between fans and their favorite sports teams, students and their teachers, and clients and their consultants. Kelman argues that in each case it is the attractiveness of the target individual that leads to self-definition effects.

Nevertheless, some evidence suggests that attractiveness alone is insufficient to explain group identification. For example, professional sports teams vary dramatically in their performance (an important measure of attractiveness in this context) and yet the most successful teams do not necessarily generate the most fan support. Chicago Cubs fans are famous for their loyalty and perseverance after 50 years without a pennant: Wrigley Field routinely draws near-capacity crowds (even in 1994 when they were last in the division). The San Jose Sharks sold out all of their 64 luxury boxes at between $62,000 to $125,000 per year in their second season, despite establishing an NHL record for losses (Malone 1993). Conversely, some winning teams have major problems attracting fans. The New Jersey Devils, winners of the 1995 Stanley Cup routinely played to a half-filled home arena for much of the season. During the 1970’s the Oakland A’s won their division for five straight years and won the World Series three times, yet attendance was below league average for much of this period (Scully 1995).

One factor that may offer additional insights into the process of group identification is the perceived similarity of the group to the individual. The perspective is rooted in Freud’s original psychoanalytic treatment of identification and his distinction between primary and secondary forms (Freud 1949). Primary identification is defined in terms of an infant’s undifferentiated response to an external object such that it is viewed as part of the self. In this instance the self and other are indistinguishable (and therefore perfectly similar). Secondary identification occurs after the child becomes able to distinguish between self and other. Freud asserted that this state was designed to reduce anxieties over "anticipated aggression or rejection from the same-sex parent and obtain vicariously the affection of the opposite-sex parent" (Kagan 1958, p. 298). Identification with the same sex parent, that is, the most similar parent, leads to psychological and physical benefits.

Stotland, Zander, and Natsoulas (1961) demonstrate that similarity between individuals on one attribute leads to identification as measured by a generalized perception of similarity (see also Dignan 1965). The research found evidence that individualsidentified with another on the basis of both trivial attributes such as preferences for nonsense syllables (Stotland, Zander, and Natsoulas 1961) and factors of significant importance to the individual’s self-concept (Burnstein, Stotland, and Zander 1961). The greater the number of attributes that were similar between one individual and another, the stronger the identification. Thus, this research underlines the importance of similarity as a basis for identification.

Examples of the role of similarity in group identification abound in the sports realm. Perhaps the most impactful illustration is the importance of geographic location or community in fan support. Sports fans tend to cheer for (and by implication are more likely to identify with) the home team, i.e., the team that represents their country, state, community, university, or other group. The team is connected to the individual because they have a shared group affiliation. Moreover, as soon as an athlete puts on the home team uniform, he or she becomes "one of ours." Other dimensions of similarity might include the team’s character or "personality" (e.g., the blue-collar Cincinnati Reds of the mid-1990’s); ethnicity (e.g., a Croatian basketball player); or other factors (e.g., they wear my favorite colors).

Despite evidence which suggests that both attractiveness and similarity lead to group identification, the relationship between these variables has not been explored. Consequently, we propose three competing models in which there is the potential for a simultaneous effect of both variables on identification. Each model allows a significant association between the two group characteristics and identification, but proposes a different causal ordering of the variables.

Model 1: Similarity As Mediator. The first model positions similarity as mediator between attractiveness and identification. This model is consistent with the view that group attractiveness leads individuals to attribute greater similarity between themselves and the group (Marks and Miller 1982). For example, Marks, Miller, and Maruyama (1981) found that subjects assume a greater similarity between themselves and attractive others than between themselves and unattractive others. One explanation for this finding is that assuming similarity with attractive others is a way of enhancing self-esteem. In its purest form, this model submits that all of the effects of attractiveness on identification are mediated by similarity (i.e., there is no direct effect of attractiveness on identification when similarity is in the model).

Model 2: Attractiveness As Mediator. The second model posits that attractiveness mediates the effects of similarity on identification. This model is consistent with Byrne’s attraction paradigm which indicates that similarity enhances the attractiveness of a target individual (see Byrne 1971 for a review). Byrne’s "law of attraction" proposes that similarity leads to attraction because similarities (dissimilarities) are positive (negative) reinforcements. This perspective suggests that similarity is antecedent to both attractiveness and identification, and that there is no direct effect of similarity on identification when attractiveness is in the model.

Model 3: Similarity as Moderator. The final model suggests that similarity moderates the effects of attractiveness on identification. This model proposes that attractiveness is an important factor leading to identification, but only when a similarity "bond" exists with the group. In other words, the effect of attractiveness on identification is contingent on the perceived similarity between the individual and group. We examine all three models in the next sections.


We selected a convenience sample of 84 male and female undergraduate students from a required senior business class to test the hypotheses (60% of the sample was male). Respondents were contacted within a normal classroom situation. After being introduced to a test administrator, subjects were instructed to read a short cover sheet which outlined the objectives of the study and asked them to indicate the name of their favorite sports team. Respondents were then asked to answer the questions in the survey instrument in relation to that team.

We selected the sports context because of the pervasive nature of team-supportive consumption behaviors. On an annual basis in the 1990’s, professional football, baseball, basketball and hockey sports clubs take in approximately $6 billion in gate receipts, not including television rights in excess of $2 billion. Moreover, gross sales of licensed products related to these leagues are in excess of $12 billion, and even the retail value of sports trading cards surpasses $2 billion (see Survey of Current Business 1994; Ghorman and Calhoun 1994). Also, pretests indicated the students were highly involved in sports and that the vast majority had a favorite sports team.


All items other than those relating to team-supportive behaviors were measured on seven-point Likert scales anchored with "Strongly Disagree" and "Strongly Agree." A list of all items is found in the Appendix.

Group Identification. This variable was measured with eight items which reflect the extent to which the individual defined him- or herself as a fan of the team. Items included, "The team is part of my life," and "I think of myself as part of the team."

Group Attractiveness. Six items were used to measure the extent to which the respondent’s favorite sports team had qualities that others admired. Respondents were asked to indicate their level of agreement with a series of statements about the extent to which team members were popular, admired, and had desirable characteristics.

Group Similarity. Four items were used to measure the perceived similarity between respondents and their favorite sports team. Items included, "I have a lot in common with the members of the team," and "I have attitudes that are similar to those held by team members."

Team-Supportive Behaviors. Team-supportive behaviors were measured with the number of times in the past season respondents had attended, watched, or listened to a game played by their favorite sports team.


Scale Purification

A series of analyses were undertaken on the 18 multi-item Likert scale items (i.e., all measures other than those related to team-supportive behaviors). Item-to-total correlations were used to eliminate one measure of group identification that was below a cutoff point of .50, and a confirmatory factor model run on the remaining 17 items. Four additional items were eliminated on the basis of high normalized residuals. The resulting 13-item model fits the data adequately with a chi-square statistic of 75.26 (df=62; p>.10), a GFI of .89, and a Bentler and Bonett (1980) comparative fit index of .90. Nine of 91 standardized residuals exceed 2.0 (10%), with the largest of these residuals 2.7. The measures have convergent validity given that all lambdas are large and significant with all t-values in excess of 6.0 (p<.001). Discriminant validity is illustrated with all phi correlations significantly different from 1 (p<.001). Final scale characteristics are summarized in Table 1.



The Role of Similarity and Attractiveness

A series of OLS regression models were run to test the three models of the effects of similarity and attractiveness on group identification. [Before testing the hypotheses we performed a Box's M test for homogeneity of covariance across the sexes and found no significant difference (Box's M = 1.048, p>.30). We therefore pooled responses for males and females.] We used the procedures outlined by Baron and Kenny (1986) for tests of mediation. Evidence of similarity as a mediator exists if 1) variations in attractiveness account for a significant level of variation in similarity, 2) variations in similarity account for significant variation in identification, and 3) when the effects identified in points "1" and "2" are controlled, the previously significant relationship between attractiveness and identification is no longer significant. As summarized in Table 2, attractiveness was found to have a significant effect on similarity (b=.29, p<.05); similarity had a significant effect on identification (b=.62, p<.001); and when both similarity and attractiveness were used to predict identification, only the effect of similarity was significant (bsimilarity=.59, p<.001; battractiveness=.11, p>.20). This test is consistent with the hypothesis that similarity mediates between attractiveness and identification since the direct effect of attractiveness on identification is not significant when similarity is introduced as a mediator. The results indicate that all of the effects of attractiveness on identification are mediated by similarity.

Using an analogous approach to test the hypothesis that attractiveness is a mediator between similarity and identification revealed a significant effect of similarity on attractiveness (b=.29, p<.05) and a significant effect of attractiveness on identification (b=.28, p<.01). As reported earlier, when identification was regressed on similarity and attractiveness, only the similarity effect was significant (bsimilarity=.59, p<.001; battractiveness=.11, p>.20). This test does not support the hypothesis that attractiveness is a mediator given that the direct effect of similarity on identification remains when attractiveness is introduced. Moreover, the path between attractiveness and identification becomes nonsignificant when identification is regressed on both similarity and attractiveness.

A final model was estimated to test the potential for similarity to moderate the relationship between attractiveness and identification. The Ghiselli (1956) method was used to test this hypothesis because unlike subgroup analysis, this approach does not rely on the transformation of a continuous to dichotomous variable. Thus the Ghiselli method is a more powerful test of moderation. The method involves creating a difference score which reflects the absolute deviation between the standardized predictor and criterion variables. Then, the hypothesized moderator variable is used to predict the difference score. A positive relationship between the difference score and the moderator variable suggests that higher moderator scores result in larger differences between the predicted and actual values of the criterion. In the present context, similarity was not found to predict the difference score (b=-.12, p>.20), suggesting that similarity is not a moderator between attractiveness and identification.

Team-Supportive Behaviors

Strong support was found for the relationship between team identification and team-supportive behaviors. Team identification had a significant effect on the number of games attended (b=.20, p<.10), watched on television (b=.48, p<.001), and listened to on radio (b=.26, p<.05). The results suggest that the extent to which respondents identified with their favorite sports team affected their willingness to engage in consumption activities which supported the team. The final model is presented graphically in the Figure.




Consistent with Kelman’s (1961) view of identification, a significant association was found between group attractiveness and identification. Nevertheless, this relationship was significant only when similarity was not in the model. The results suggest that group attractiveness increases the willingness of individuals to define themselves as group members, but this effect only operates indirectly through similarity perceptions.

Even when similarity was not included in the model, the effect of attractiveness on identification was lower than would be expected given Kelman’s assertion that attractiveness is the sole determinant of identification. One explanation for the lack of a strong direct effect of attractiveness on identification is that favorite teams tend to be attractive, and hence a range restriction problem exists. However, Table 1 reveals a wide variation in attractiveness responses. The range restriction explanation appears, therefore, untenable.

In contrast, the direct effect of similarity on identification was significant and large with or without the inclusion of attractiveness in the model. A Fisher’s Z test reveals that similarity is significantly more strongly correlated with identification than player attractiveness (p<.05). On the basis of this evidence, it appears that similarity is a more important predictor of group-derived self-definition than attractiveness. Indeed, as noted earlier, wen both similarity and attractiveness were included in the model as predictors, the effect of attractiveness was not significant.

The practical and theoretical implications of the findings are important. From a practical perspective, the results suggest that sports marketers may benefit from emphasizing the similarities between the fans and their team rather than the attractiveness of the team’s players. Identification appears most likely when fans perceive that there is an essential similarity or link with the team. Although it is unclear what the critical dimensions of similarity might be, important similarity factors may be personality, value, or attitudinally based. For example, a team that is "crude, rude, and unattractive" may be appealing to fans who have the same qualities, but repulsive to fans who are more "civilized." Promotional campaigns that ignore the need for perceived similarity between fans and their team may be ineffective even if the team’s players have aspirational qualities.

From a theoretical perspective, the finding that the effect of attractiveness on identification is mediated by similarity is important. Prior studies which have found a significant attractiveness effect may have done so because similarity was not included in the model. The present research suggests that attractiveness influenced perceptions of similarity which subsequently had a direct effect on group identification. Thus, although attractiveness remains an important factor which contributes to group identification, it appears to operate differently than previously thought.

Finally, we found a significant link between group identification and three forms of team-supportive consumption behaviors. The degree to which individuals had a self-defining relationship with the team affected their willingness to attend games, watch them on television, and listen to them on the radio. The results highlight the importance of group identification as a variable explaining consumption behaviors that are derived from voluntary group associations.




The present research used global measures of similarity and attractiveness. It seems likely that both variables are driven by specific attributes of the team and its members. For example, perceived attractiveness might be a function of the extent to which team members were seen as famous, gifted athletically, and rich. However, given that the average fan is deficient on these attributes when compared to the average professional athlete, other factors may have greater inter-team variance and therefore be more important determinants of identification. In particular, future research is needed on the extent to which team-level "personality" factors such as sportsmanship, leadership, aggressiveness, and performance underlie the global measures of similarity and attractiveness we used.

As noted previously, the sports team context was selected because of the pervasiveness of team-related consumption and the high involvement of respondents in this area. Nevertheless, the favorite sports team context is only one of many possible voluntary group associations that affect consumption behaviors. Research is needed on identification related to alumni connections, product-related organizations (e.g., Harley owners’ groups), and musical groups.

Similarly, the area would benefit from further study of non-students. Although the present research context was meaningful to our respondents, students do not necessarily represent the ypical sports fan. Also, research in an experimental setting is warranted, although the strong relationship between the similarity and attractiveness makes orthogonal manipulations of these variables difficult (see Byrne 1971; Byrne and Nelson 1965).




Baron, Rueben M., and David A. Kenny (1986), "The Moderator-Mediator Variable Distinction in Social Psychological Research: Conceptual, Strategic, and Statistical Considerations," Journal of Personality and Social Psychology, 51, (6), 1173-1182.

Bentler, P.M., and Douglas G. Bonett (1980), "Significance Tests and Goodness of Fit in the Analysis of Covariance Structures," Psychological Bulletin, 88 (May), 588-606.

Brown, Steven P. (1989), "Moderators and Mediators: A Review of Concepts and Usage in Marketing Research," American Marketing Association Educators’ Proceedings, Paul Bloom et al. eds. 170-175.

Burnstein, Eugene, Ezra Stotland, and Alvin Zander (1961), "Similarity to a Model and Self-Evaluation," Journal of Abnormal and Social Psychology, 62 (March), 257-264.

Byrne, Donn E. (1971), The Attraction Paradigm, San Diego, CA: Academic Press.

Byrne, Donn E. and D. Nelson (1965), "Attraction as a Linear Function of Proportion of Positive Reinforcements," Journal of Personality and Social Psychology, 2, 884-889.

Dignan, Sister M. Howard (1965), "Ego Identity and Maternal Identification," Journal of Personality and Social Psychology, 1 (5), 476-483.

Freud, Sigmund (1949), Group Psychology and the Analysis of the Ego, London: Hogarth.

Ghiselli, Edwin E. (1963), "Moderating Effects and Differential Reliability and Validity," Journal of Applied Psychology, 47, 81-86.

Ghorman, Jerry and Kirk Calhoun (1994), The Name of the Game: The Business of Sports, New York: John Wiley.

Jellison, Jerald M., and Paul T. Zeisset (1969), "Attraction as a Function of the Commonality and Desirability of a Trait Shared with Another," Journal of Personality and Social Psychology, 11 (February), 115-120.

Kagan, Jerome (1958), "The Concept of Identification," Psychological Review, 65 (September), 296-305.

Kelman, Herbert C. (1961), "Process of Opinion Change," Public Opinion Quarterly, 25 (Spring), 57-78.

Malone, Michael S. (1993), "A Loser’s A Winner in Silicon Valley," New York Times, v 142, May 2, F25.

Marks, Gary, and Norman Miller (1982), "Target Attractiveness as a Mediator of Assumed Attitude Similarity," Personality and Social Psychology Bulletin, 8 (December), 728-735.

Marks, Gary, Norman Miller, and Geoffrey Maruyama (1981), "Effects of Targets’ Physical Attractiveness on Assumptions of Similarity," Journal of Personality and Social Psychology, 41 (December), 198-206.

Scully, Gerald W. (1995), The Market Structure of Sports. Chicago: University of Chicago Press.

Stotland, Ezra, Alvin Zander, and Thomas Natsoulas (1961), "Generalization of Interpersonal Similarity," Journal of Abnormal Psychology, 62 (March), 250-256.

Survey of Current Business (1994), U.S. Statistical Abstract.

Tajfel, H. (1982), "Introduction" in, Social Idetity and Intergroup Relations, H. Tajfel ed., New York: Academic Press, 1-11.

Tajfel, H. and John C. Turner (1986), "The Social Identity Theory of Intergroup Behavior," in Psychology of Intergroup Relations, S. Worchel and W. G. Austin, eds., Chicago: Nelson-Hall, 7-24.

Turner, John C. (1982), "Toward a Cognitive Redefinition of the Social Group," In J. Turner and H. Giles eds., Intergroup Behavior, Chicago: University of Chicago Press, 15-40.

Turner, John C. (1985), "Social Categorization and the Self-Concept: A Social Cognitive Theory of Group Behavior," In J.E. Lawler ed., Advances in Group Processes, Vol 2., Greenwich, CT: JAI Press, 77-122.