Physiology & Behavior, Vol. 65, No. 1, pp. 59 62, 1998 1998 Elsevier Science Inc.

c. All rights reserved. Printed in the U.S.A. 0031-9384/98 $19.00 .00

PII S0031-9384(98)00147 4

Testosterone Changes During Vicarious Experiences of Winning and Losing Among Fans at Sporting Events
PAUL C. BERNHARDT,* JAMES M. DABBS, JR., JULIE A. FIELDEN AND CANDICE D. LUTTER, *University of Utah, Department of Educational Psychology, MBH 327, Salt Lake City, UT 84112; and Georgia State University, Department of Psychology, Atlanta, GA 30303, USA Received 19 December 1997; Accepted 14 April 1998
BERNHARDT, P. C., J. M. DABBS, JR., J. A. FIELDEN, AND C. D. LUTTER. Testosterone changes during vicarious experiences of winning and losing among fans at sporting events. PHYSIOL BEHAV 65(1) 59 62, 1998.Basking in reected glory, in which individuals increase their self-esteem by identifying with successful others, is usually regarded as a cognitive process that can affect behavior. It may also involve physiological processes, including changes in the production of endocrine hormones. The present research involved two studies of changes in testosterone levels among fans watching their favorite sports teams win or lose. In the rst study, participants were eight male fans attending a basketball game between traditional college rivals. In the second study, participants were 21 male fans watching a televised World Cup soccer match between traditional international rivals. Participants provided saliva samples for testosterone assay before and after the contest. In both studies, mean testosterone level increased in the fans of winning teams and decreased in the fans of losing teams. These ndings suggest that watching ones heroes win or lose has physiological consequences that extend beyond changes in mood and self-esteem. 1998 Elsevier Science Inc. Testosterone Self-esteem Sports Spectators Fans BIRG

WHEN a sports team wins or loses an athletic contest, its fans respond much as the team athletes do: they bask in its glory and suffer in its defeat. If their team wins, they wear team colors the next day (4), talk about how we won (4,6,18), feel stronger (19) and more optimistic (17), and are prouder of themselves (10). If their team loses, the opposite happens, and fans feel defeated, depressed, and angry. These changes are cognitive and behavioral, they involving self-esteem, self-condence, and a readiness to face the future. They may also be associated with physiological changes that support tendencies to compete or retreat, to become strong or vulnerable. The present paper focuses upon changes in the endocrine system, specically in the hormone testosterone. Testosterone is central to sexual activity and male-male competition among animals. Among rhesus monkeys, testosterone both affects the outcome of a ght and is affected by the outcome. When two male monkeys ght, the one higher in testosterone is likely to win, and after the ght testosterone increases in the winner and decreases in the loser (1). Human males show similar patterns: testosterone levels in winners increase relative to losers in such physical contests as wrestling (9), martial arts (3), tennis (2), and in more cerebral contests, such as chess (14) and coin ips (12). Salvador, with others, found the winning record of ghters
1

correlated positively with changes in testosterone that was observed during the contests (16). Other researchers found increased testosterone in the hours before a contest (2,3) and found that initial levels predicted the outcome of the contest (3). Mazur (13) has emphasized the link between testosterone and dominance. He suggests that rising testosterone levels are tied to expressions of dominance, where dominance may be revealed in changes in posture, eye contact, and assertive speech. Mazur and Lamb (15) demonstrated in a number of situations (such as graduation from medical school) that increases in status were correlated with increases in testosterone. Mazur (13) proposes a biosocial theory of status, in which success increases testosterone levels, and higher testosterone levels increase the likelihood of being successful in the future. Failure lowers testosterone levels, which reduces the likelihood of success in future contests. Kemper (11) has hypothesized that fans at sporting events would have testosterone changes similar to those seen in the participants. Fans who identify with a team will feel its success or failures as their own, giving rise in them to vicarious experiences of success or failure. He hypothesized, therefore, that the teams success or failure should affect their testosterone levels, especially among fans who identify strongly with the team. Testosterone measurements are reliable, having a correlation of

To whom requests for reprints should be addressed. E-mail: Paul.Bernhardt@m.cc.utah.edu

59

60 about r 0.65 between single measurements taken on adjacent days (7). Measurements are also variable enough to change in response to external conditions, as indicated by the studies above. Testosterone levels can begin to increase in less than 20 min, making them suitable for following changes around sporting events. Testosterone levels decline across the day (7), typically dropping 30% 40% from early morning to late evening. This normal diurnal decline needs to be considered in designing research. Testosterone can be assayed reliably from saliva, and saliva samples are easy to collect in a variety of naturalistic settings (8). The present paper reports two studies of testosterone levels among sports fans watching their favorite teams in competition with traditional rivals. We measured testosterone levels before and after the competition, predicting that fans whose team won would increase in testosterone relative to fans whose team lost. Study 1: College Basketball The rst study, best considered a pilot study, examined testosterone changes among fans watching their favorite basketball team win or lose. Participants were fans at an annual contest between the University of Georgia and Georgia Tech. The schools are traditional rivals, and the year of this study, 1991, the game was played at a neutral site, ensuring that the crowd would not be dominated by fans from one school. The game turned out to be a close one, with the University of Georgia winning in the nal few seconds.
METHOD

BERNHARDT ET AL.

FIG. 1. Basketball fans testosterone levels before and after their team has won or lost.

During the hour before the game, the researcher circulated through the seating area in the arena and approached 15 males for participation in the study. Four refused to participate. Eleven initially participated but three failed to leave a post game sample. This left eight male participants. A questionnaire given to each fan showed that four supported each team, and they ranged in age from 20 42 years. Each participant chewed a stick of sugar-free gum to stimulate the ow of saliva and deposited 3 mL of saliva into a 20-mL polyethylene vial. The researcher took the sample and gave the participant a second vial and piece of gum to keep during the game. After the game the participant collected a second sample and either gave it to the researcher or left it at his seat for the researcher. All samples were collected within 15 min of the end of the game. Samples were stored frozen and assayed in duplicate in a single batch, using a radioimmunoassay with ether extraction, 125 ITestosterone tracer, and charcoal separation (7).
RESULTS

was held in the United States for the rst time. Matches were widely publicized, and fans displayed remarkable fervor. Many fans, especially the fans of foreign teams, felt that their national honor was at stake. We were unable to attend an actual game, so we studied fans watching the televised broadcast of the nal match of the tournament between Brazil and Italy. Both countries have a history of excellence in soccer. The outcome of the nal match remained uncertain until the last moment, when Brazil won on a penalty kick after the game had been played to a tie and an overtime had been played out with no one scoring. Method Participants were 26 male fans, ranging in age from 21 40 years, recruited at two Atlanta taverns. Twelve fans were Brazilian or of Brazilian descent, associated with the Brazilian American Society, and gathered to watch the match at a sports bar. Fourteen fans were Italian or of Italian descent, gathered at a pizzeria on a patio that was set aside for viewing the game. Two researchers arrived at each establishment about 1 h before the game and began to recruit participants. Fans were offered $10 each to participate (Brazilian fans donated their funds to the Brazilian American Society). The researchers collected saliva samples from each participant 30 10 min before the game and again 1525 min after the game. After the game they gave participants vials and gum to take home and asked them to collect saliva samples when they arose the next morning and mail them to us. As in the basketball study, fans chewed gum and deposited saliva into a plastic vial. Samples were stored frozen and assayed in duplicate in a single batch.
RESULTS

Mean testosterone scores were 6.3 ng/dL (SD 2.7) before the game and 6.5 ng/dL (SD 3.0) after the game, within the normal range (7). We analyzed the scores using repeated measures analysis of variance, with team (winners versus losers) a betweensubjects factor and time (pre- versus postgame) a within-subjects factor. There was a signicant team time interaction, F(1, 6) 9.94, p 0.05, reecting an increase in testosterone among winning fans and a decrease among losing fans. [Analysis of postgame difference via an ANCOVA, with pregame level as the covariate, yields similar results, F(1, 5) 8.86, p 0.05.] The partial 2 effect size for this interaction was 0.62. There was no signicant mean difference between the two teams, F(1, 6) 0.71, not signicant, nor between the two points in time, F(1, 6) 0.21, not signicant. The ndings are summarized in Fig. 1. Study 2: World Cup Soccer The second study extended the ndings to a larger sample and used fans in separate locations. In 1994, the World Cup of Soccer

Table 1 shows the number of fans who provided samples before the game, after the game, and the next day. Participation declined over the three points in time. The decline over time was greater among Italians than Brazilians, but the overall pattern was not statistically signicant, 2(2, n 59) 4.32, not signicant. However, the contrast between participation before the game and participation the next day, reecting a total accumulated loss of participants during the study, did show a signicant difference between the two groups, 2(1, n 37) 3.85, p 0.05. Italian fans dropped out of participation more than Brazilian fans did.

TESTOSTERONE CHANGES IN SPORTS FANS TABLE 1

SOCCER FANS AND STAGE OF DATA COLLECTION Collection Occasion Before Game After Game* Next Morning

61 Because the outcome of each game was not determined until the last few seconds, this effect was likely sudden, rather than building up gradually during the game. These contests were particularly meaningful to fans and might reasonably be expected to produce strong effects. The University of Georgia and Georgia Tech are traditional rivals, and rivalry among countries competing in World Cup soccer is intense. In particular, Italy and Brazil have long histories of excellence in World Cup soccer. Fans viewed the soccer match on television with an excited and festive manner. After the game, researchers observed a striking difference in mood between the two groups: Italian fans were despondent; Brazilian fans were exuberant. The greater failure of Italian fans to provide samples the next day suggests that mood changes after a contest is much more than just a eeting change. Cialdini et al. (5) argues that fans feel involved in a contest and bask in the reected glory of a win because the win boosts their self-esteem. We have no evidence that our fans were low in self-esteem. However, they were very much involved in the contests, and they seemed to gain positive feelings from seeing their teams win. A feeling of involvement is probably essential for basking in reected glory to take place, but it is not clear how much involvement is needed for the observed testosterone changes to take place, suggesting an avenue for further research. Our ndings bring together processes of social cognition and physiology. Basking in reected glory is part of an over-arching process in which we maintain our self-esteem through comparing ourselves to others. Self-esteem maintenance theory predicts we will like or dislike others, and be proud or envious of their success, depending upon their relevance to our self-images (20). The theory emphasizes: 1) psychological closeness; 2) relevance to ones self-denition; and 3) performance of the other persons. The results of these studies suggest these factors may also moderate hormonal responses among fans. The link between self-esteem and the effects of winning and losing on testosterone levels deserves further exploration. Kemper offers that periodic testosterone surges in individuals are important for maintaining stable social order, sports spectatorship being one possible way to attain those surges (11). The results support a portion of his proposition, that testosterone surges do occur in winning fans. Testing the remaining portion of his proposition, that these surges result in contentment with the social order, is certainly possible via the techniques used for these studies. Our ndings also bear upon the idea that changes in testosterone level in competitive encounters are adaptive. The biosocial theory articulated by Mazur et al. (13,14) treats testosterone as a physiological aspect of the interaction between competition and status. Success in competition leads to an increase in status and an increase in testosterone. Higher testosterone levels increase the likelihood of approaching and winning the next encounter, thereby perpetuating a status difference produced by the rst encounter. Similarly, failure leads to a drop in status and a drop in testosterone, which decreases the likelihood of approaching and winning the next encounter. The process is adaptive to both parties because it helps resolve ongoing conict quickly, with a single encounter, avoiding the repeated confrontations that could debilitate both winner and loser without producing a meaningful change in their relationships. The ndings are consistent with a biosocial model, but they are also incomplete. They tell us little about the effects of changes in testosterone. The dejection among Italian fans and enthusiasm among Brazilian fans might possibly be due to testosterone changes, but may not be the extent of testosterone effects. It might affect posture, demeanor, or behavior in ways that set one up to

Brazilian Italian

12 14

12 10

9 2

* After game sample includes one participant whose sample was not usable. Samples collected the next morning were not included in the analysis of testosterone scores.

Because there were so few saliva samples from the second day, we limited our analysis of testosterone scores to samples collected on the day of the game. One postgame Italian sample was contaminated with blood and could not be used, leaving us with 12 Brazilian and 9 Italian fans before and after the game. Mean testosterone scores were 7.7 ng/dL (SD 3.5) before the game and 7.3 ng/dL (SD 2.8) after the game, within the normal range. As in the basketball study, we analyzed the scores using repeated measures ANOVA, with team (winners versus losers) a between-subjects factor, and time (pre- versus postgame) a withinsubjects factor. There was a signicant team time interaction, F(1, 19) 13.01, p 0.01, reecting an increase in testosterone among winners and a decrease among losers. [Analysis of postgame difference via an ANCOVA, with pregame level as the covariate, yields similar results, F(1, 18) 15.05, p 0.01.] The partial 2 effect size for this interaction was .42. There was no signicant mean difference between the two teams, F (1,19) 0.25, ns, nor between the two points in time, F(1, 19) 1.38, not signicant. The ndings are summarized in Fig. 2.
DISCUSSION

In both studies, testosterone levels increased among fans of winning teams and decreased among fans of losing teams. The effect among winning fans was strong enough to reverse the normal pattern of decline in testosterone levels across the day.

FIG. 2. Soccer fans testosterone levels before and after their team has won or lost.

62 win or lose future contests. Detailed studies of fans whose testosterone levels have been affected could address a number of questions. Do changes in testosterone among fans play a role in domestic violence after a contest (21)? Do changes in testosterone promote changes in sexual activity? Are subsequent everyday manifestations of dominance, such as taking charge in a conversation or meeting, affected by changes in testosterone levels? Do the changes in testosterone and feelings of self-worth after victory or defeat affect ones health, including the health of ones immune system? Would female fans show the same effects as male fans, even though pathways of testosterone production are different in the two sexes?

BERNHARDT ET AL. Finally, could small changes in testosterone in individuals lead to large cognitive and behavioral changes in a group? The mood of a community changes when its team wins or loses a game, and the mood of a nation changes when its army wins or loses a battle. Physiological factors in individuals, including changes in hormone levels, may affect these group phenomena.

ACKNOWLEDGMENTS

This research was supported by National Science Foundation Grant SBR-9511600.

REFERENCES
1. Bernstein, I. S., Rose, R. M.; Gordon, T. P. Behavioral and environmental events inuencing primate testosterone levels. J. Hum. Evol. 3:517525; 1974. 2. Booth, A., Shelley, G., Mazur, A., Tharp, G.; Kittok, R. Testosterone, and winning and losing in human competition. Horm. Behav. 23:556 571; 1989. 3. Campbell, B., ORourke, M.; Rabow, M. Pulsatile response of salivary testosterone and cortisol to aggressive competition in young males. Paper presented at the annual meeting of the American Association of Physical Anthropologists, Kansas City, Kansas, 1988. 4. Cialdini, R. B., Borden, R. J., Thorne, A., Walker, M. R., Freeman, S.; Sloan L. R. Basking in reected glory: Three (football) eld studies. J. Pers. Soc. Psychol. 34:366 375; 1976. 5. Cialdini, R. B., Finch, J. F., De Nicholas, M. E. egic self-presentation: The indirect route. Cody, M. J.; McLaughlin, M. L., eds. The psychology of tactical communication. London: Multilingual Matters; 1990: 194 206. 6. Cialdini, R. B.; Richardson, K. D. Two indirect tactics of image management: Basking and blasting. J. Pers. Soc. Psychol. 39:406 415; 1980. 7. Dabbs, J. M., Jr. Salivary testosterone measurements: Reliability across hours, days, and weeks. Physiol. Behav. 48:83 86; 1990. 8. Dabbs, J. M., Jr., Campbell, B. C., Gladue, B. A., Midgley, A. R., Navarro, M. A. Read, G. F., Susman, E. J., Swinkels, L. M. J. W.; Worthman, C. Reliability of salivary testosterone measurements: A multicenter evaluation. Clin. Chem. 41:15811584; 1995. 9. Elias, M. Serum cortisol, testosterone, and testosterone-binding globulin responses to competitive ghting in human males. Aggressive Behav. 7:215224; 1981. 10. Hirt, E. R., Zillmann, D., Erickson, G. A.; Kennedy, C. Costs and benets of allegiance: Changes in fans self-ascribed competencies after team victory versus defeat. J. Pers. Soc. Psychol. 63:724 738; 1992. Kemper, T. Social structure and testosterone. New Brunswick, NJ: Rutgers University Press; 1990. McCaul, K. D., Gladue, B. A.; and Joppa, M. Winning, losing, mood, and testosterone. Horm. Behav. 26:486 504; 1992. Mazur, A. A biosocial model of status in face-to-face primate groups. Social Forces 64:377 402; 1985. Mazur, A., Booth, A.; Dabbs, J. M., Jr. Testosterone and chess competition. Soc. Psychol. Q. 55:70 77; 1992. Mazur, A.; Lamb, T. A. Testosterone, status, and mood in human males. Horm. Behav. 14:236 246; 1980. Salvador, A., Simon, V., Suay, F.; Llorens, L. Testosterone and cortisol responses to competitive ghting in human males: A pilot study. Aggressive Behav. 13:9 13; 1987. Schweitzer, K., Zillmann, D., Weaver, J. B.; Luttrell, E. S. Perception of threatening events in the emotional aftermath of a televised college football game. J. Broadc. El. 36:75 82; 1992. Snyder, C. R., Lassegard, M.; Ford, C. E. Distancing after group success and failure: Basking in reected glory and cutting off reected failure. J. Pers. Soc. Psychol. 51:382388; 1986. Tesler, B. S.; Alker, H. A. Football games: Victory, defeat, and spectators power preferences. J. Res. Pers. 17:72 80; 1983. Tesser, A. Toward a self-evaluation maintenance model of social behavior. In: Berkowitz, L. ed. Adv. Exp. Soc. Psychol. New York: Academic Press. 21:181227; 1988. White, G. F., Katz, J.; Scarborough, K. E. The impact of professional football games upon violent assaults on women. Violence Vict. 7:157 171; 1992.

11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.