EXPLANATORY STYLE AS a MECHANISM of DISAPPOINTING ATHLETIC PERFORMANCE Martin E.P

PSYCHOLOGICAL SCIENCE

Research Report

EXPLANATORY STYLE AS A MECHANISM OF DISAPPOINTING ATHLETIC PERFORMANCE Martin E.P. Seligman, 1 Susan Nolen-Hoeksema,2 Nort Thornton,3 and Karen Moe Thornton3 1University ofPennsylvania, 2Stanford University, and 3 University ofCalifornia at Berkeley

Abstract-Two university varsity swim- study to date has tested the mechanism reviews see Peterson & Seligman, 1984; ming teams took the Attributional Style by which pessimistic style impairs Sweeney, Anderson, & Bailey, 1986.) Questionnaire (ASQ) at the start of the achievement. season. Swimmers with a pessimistic ex- We asked if pessimistic explanatory Coaches' judgments. The coaches planatory style wellt on to show more un- style predicts poorer than expected ath- rated each of their swimmers in the be- expected poor performances during letic performance and if it works by the ginning of the season on a I to 7 scalb, competition titan optimistic swimmers. mechanism of lowered response initia- judging how the swimmers would per- We then testedthe purported mechanism tion specifically following defeat. form after a defeat, where I was much ofthis effect by experimentally simulat- worse than average and 7 was much bet- ing defeat, giving each swimmer falsely ter than average. There was one coach negative' times. Performance deterio- STUDY 1 per swimmer: Nort Thornton for the rated for tllOse swimmers with a pessi- men, and Karen Thornton for the mistic explanatory style for bad events .Method women. Throughout the studies the on their next swim, whereas perfor- coaches remained blind to explanatory mance for those swimmers with an opti- Subjects style scores. mistic style did not. The subjects were members of the varsity men's and women's swim teams at the University of California at Berke- Swim season ratings. Following Some people habitually explain bad ley in 1987-1988. Both ofthese teams are each competitive swim during the sea- events by causes that are stable in time, nationally ranked and several swimmers son, the coach rated each swimmer's global in effect, and internal, but explain on the teams hold national or world performance on a I to 7 scale, where 1is good events by causes that are unstable, records. Twenty-one men and twenty- much worse than expected, 4 is as ex- specific, and external. The reformulation six women participated in Study I. pected, and 7 is much better than ex- of the learned helplessness model of de- pected. Every swim on which a swimmer received a rating of I, 2, or 3 was pression (Abramson, Seligman, & Teas- Instruments dale, 1978) predicts that individuals who counted as a poor swim. Each swimmer have this pessimistic style of explaining also rated pis or her performance using To measure ex- events sh0!1ld do worse than expected in Explanatory style. the same scale. Since the correlation be- planatory style, we administered the At- achievement situations, whereas people tween coach and swimmer approached tributional Style Questionnaire (ASQ; 'with a more optimistic style should do 1.00, we discontinued the swimmers' rat- Peterson, Semmel, von Baeyer, Abram- better. A pessimistic explanatory style is ings after a few meets. son, Metalsky, & Seligman, 1982; Selig- claimed to lead to poor performance be- man, Abramson, Semmel, & von cause it contributes to the expectation Baeyer, 1979). This self-report instru- Results that bad events will recur in several do- ment yields scores for explanatory style mains. In tum, this expectation leads to for bad and good events along three cau- ASQ Results lowered voluntary response initiation sal dimensions: stable versus unstable, following failure (Seligman, 1975). The global versus specific, internal versus prediction that people with a pessimistic Regression analyses (prediction external. A composite score for explana- We performed a se- explanatory style"will show poorer ofperformance). tions of bad events (CN) is obtained by ries of regression analyses to test our achievement than"people with an opti- summing the subject's score on the three predictions about the main independent mistic style has been supported in field dimensions for the bad events. A com- variables predicting the number of poor studies of academiC performance posite score for good events (CP) is ob- swims. In the main analysis CPCN, sex, (Dweck & Licht; i980; Nolen-Hoekse- tained by summing the subject's scores and coaches' judgment were simulta- ma, Girgus, & Seligman, 1986; Peterson on the three dimensions for good events. neously regressed against percentage of & Barrett, 1987) and the workplace (Se- A full scale score (CPCN) is obtained by poor swims. The squared multiple R was ligman & Schulman, 1987). But no field subtracting the composite score for bad .53 and each of the predictor variables events from the composite score for contributed significantly: CPCN (t = Send correspondence and reprint requests good events. Cronbach's alphas (a mea- 2.62, p < .012), sex (t = 2.51, p < .016), to: Martin E.P. Seligman, Department ofPsy- sure of inter-item consistency) in this and coaches' judgment (t = 3.28, p < chology, University of Pennsylvania, 3815 sample were .71, .76, and .79 for CPCN, .(02). Additional regression analyses us- Walnut Street, Philadelphia, PA 19104. CN, and CP, respectively. (For validity ing CN and CP showed that each con-

., ¥ Swimming and Explanatory Style

tributed (1 = 1.89, p < .066; 1 = 2.51, p better following defeat than persons with victories and defeats. The analyses for < .012 respectively), but the strongest a pessimistic explanatory style. Study 2 focus on composite explanatory effect was with CPCN. Thus we imposed defeat on all of the style scores for negative events (CN) swimmers and tested their performance only, because here we were trying to following this defeat. Specifically, each predict reactions to a defeat only. Correlations of ASQ and coach- of the swimmers swam a time trial of his In general, the swimmers with an op- The coaches' judg- es' judgments. or her best event, and were given a timistic explanatory style for negative ments were not significantly correlated falsely slow score. After a rest, the events did at least as well after defeat as with any of the three ASQ composite swimmers swam the event again. they did in their first swim, but the pes- scores. This suggests that the coaches' simists' performances deteriorated. We predictions of swimmers' performances divided the swimmers into two groups are based on something other than an Method based on a median split of CN scores. awareness ofthe swimmers' explanatory The mean ratio ofthe time after the false style (,[CPCN] = .14, ,[CN] = .;... .22, Subjects feedback to the actual first time was .995 ,[CP] = .14). The subjects were 33 swimmers from the original 46. These tended to be the for the swimmers with optimistic explan- best swimmers on the teams who had re- atory style scores, whereas the mean for Sex differences. On the whole, the mained at school after the end ofthe sea· the swimmers with pessimistic scores men- had greater optimistic explanatory son to prepare for the 1988 Olympic tri- was 1.016 (1 = 1.96, p < .059). These are style than the women. The total scale als. Nineteen were women and 14 were not trivial changes as swimming times score,CPCN, was significantly more op- men. go. The absolute improvement and dete- timistic for the men (men X = 6.44, SD rioration in many of the cases would be = 3.15; womenX = 3.15, = 1.96; 1 the difference between winning and los- SD Procedure = 3.46, < .001). The men's composite ing an event. On coaches' ratings of the p Each swimmer was asked by the score for negative events (X = 11.60, quality ofthe second swim, the same dif- SD coach to swim in his or her best event, = 3.12) (CN) was marginally more opti- ference held: Changes in the coaches' and at the end of the swim was given a mistic than the women's (X = 12.81, judgments of the quality of the swims SD slower time. For loo-meter events, the = 1.72; 1 = 1.69, p < .097) and their from time I to time 2 showed a small time given to the swimmers was 1.5 sec- composite score for postive events (CP) deterioration (- .094) among the swim- onds slower than they had actually was much more optimistic than the wom- mers with an optimistic explanatory swum, for 200 meter events it was 2 sec- en's (men X = 18.03, = 1.89; style, but a more substantial deteriora- SD onds slower, for 400 meter events it was women X = 15.96, = 1.41; 1 = 4.3, tion of - .833 among the pessimists (t = SD 4 seconds slower, and for 500 meter < .00(1). These sex differences are no- 2.07, p < .047). It is also relevant that p events it was 5 seconds slower. These table. On average, the men's absolute coaches' judgments of the first swim times were chosen to accomplish two scores are as optimistic as any group we ~ were not different for the optimists and goals: (a) to produce serious disappoint- have tested except for life insurance pessimists, meaning that the optimists ment, and (b) be small enough to be un- agents, whereas the women's scores and the pessimists did as well as each detectable. Both goals seem to have look like typical college student female other initially (means = 3.53 versus been accomplished, although no formal means. Remember that many of the 3.50). The effect of explanatory style ap- manipulation check was made. Swim- women (as well as the men) are world- peared only after defeat. The effect was mers looked disheartened and in some class athletes and campus heroines, specific to explanatory style for negative cases, dejected, after being given the however. This suggests that women tend events; CP (and CPCN) did not show sig- slow times, and no swimmer voiced sus- to be more pessimistic than men even nificant effects on time 2 performance picion either during the study or during when they have high status and high relative to time 1. debriefing. After an average of 30 min- achievement (Nolen-Hoeksema, 1987). Regression analyses, using CN utes of rest, chosen to ensure fresh per- scores, sex, and coaches' judgments of formance with no fatigue, each swimmer ability to rebound after defeat to predict '7' swam the event again. The coaches rated "STUDY2 second swim times confirmed the find- each performance on the 1-7 scale from ings above. When we used the ratio of much worse than expected to much bet- Study Lshowed that explanatory the times before and after feedback -as style and coaches' judgments of swim- ter than expected. We also computed the - the dependent variable, the model ratio of the swimmer's time on the sec- mers' resilience after defeat predicted yielded a squared multiple correlation of how many unexpectedly poor swims the ond swim to his or her time of the first .250; CN scores accounted for signifi- swim. team members would go on to show over cant variance (t = 3.06, p < .005), but the season. Although these results are in- neither sex nor coaches' judgments ac- triguing, because the study is naturalistic Results - counted for any significant portion ofthe we could not directly test the mechanism variance. When we used the difference by which explanatory style is alleged to In Study I, the most relevant explan- in coaches' judgments of the two times affect performance: Persons with an op- atory style variable was CPCN, since as the dependent variable, the model timistic explanatory style will recover swims over the whole season follow both yielded a squared multiple, of .237; CN

144 VOL. 1, NO.2, MARCH 1990 Downloaded from pss.sagepub.com at UNIV OF PENNSYLVANIA on July 13, 2015 PSYCHOLOGICAL SCIENCE

M.E.P. Seligman et at. scores accounted for variance approach- variance in poor swims over and above terns of subjective beliefs about the ing significance (t = 1.96, p < .06), but coaches' judgments in the regression causes of events. What should one do neither sex nor coaches' judgments of analyses. This suggests that knowing with a very talented person who has pes- ability to comeback accounted for signif- swimmers' explanatory styles may be simistic explanatory style? It may be im- icant variance, practically useful for coaches in making portant that cognitive therapy reliably decisions about who should swim, par- and stably changes pessimism into opti- ticularly when the swimmers have suf- mism among depressed patients (Selig- DISCUSSION fered recent defeats or the event is very man, Castellon, Cacciola, Schulman, important. Those swimmers who have Luborsky, Ollove, & Downing, 1988; Three major findings emerged from optimistic explanatory styles should on DeRubeis, Evans, Hollon, Garvey, these two studies: (I) Swimmers with a average do better under pressure than Grove, & Tuason, 1988). It is plausible pessimistic explanatory style were more those with pessimistic styles. that similar techniques could be used likely to go on to perform below expec- In the theory oflearned helplessness, with normal, pessimistic adults to help tations during the season than swimmers the expectation of future failure works them perform at capacity. with an optimistic explanatory style. (2) , by undermining the incentive to try, After a simulated defeat, swimmers with thereby lowering the probability of vol- Acknowledgments-This work was sup- a pessimistic explanatory style showed untary response initiation (Seligman, ported in part by NIMH grant 1%04, NIA deteriorated performance, whereas 1975). The performance of swimmers af- grant AG05590, and a grant from the Mac- swimmers with an optimistic style did ter experimentally manipulated defeat in Arthur Foundation Research Network on Determinants and Consequences of not. (3) Explanatory style predicted per- Study 2 suggests that lowered response Health-Promoting and Health-Damaging formance by the swimmer even after initiation following defeat may be the Behavior to M. Seligman. We thank the' coaches' judgments of ability to come mechanism by which pessimistic explan- members of the varsity swimming teams back was taken into account. "ato'iy style hurts achievement in natural at the University ofCalifornia at Berkeley Is it possible that swimmers with an settings. There are other possibilities as whose cooperation made this study possi· ble. optimistic explanatory style are simply well, however. Explanatory style may it- more talented swimmers than swimmers self merely be a correlate ofa more basic with a pessimistic style, and this is why dimension. A number of cognitive styles they showed fewer poor swims in Study , such as optimism (Scheier & Carver, REFERENCES 1 and better comeback times in Study 2? 1987), hardiness (Hull, Van Treuen, & There are four lines ofevidence that sug- Virnelli, 1987), self-handicapping Abramson, L.Y, Seligman, M.E.P., & Teasdale, gest that explanatory style had predic- (Rhodewalt, Saltzman, & Wittmer, 1984) J.D. (1978). Learned helplessness in humans: tive power over and above swimming and negative affectivity (Watson & Critique and reformulation. Journal ofAbnor- mal Psychology, 87, 49-74. ability. First, explanatory style did not Clark, 1984) have been related to Dweck, C.S., & Licht, B. (1980). Learned helpless- correlate with the coaches' judgments of achievement. Could explanatory style be ness and intellectual achievement. In J. Garber how well a swimmer would rebound subsumed under these more basic & M.E.P. Seligman (Eds.), Human helplessness (pp. 197-222). New York: Academic from defeat---one measure of a swim- styles? Perhaps, but unlike these vari- Press. mer's talent. Second, explanatory style ables, explanatory style in reformulated DeRubeis, R.I., Evans, M.D., Hollon, S.D., Garvey, M.J., Grove, W.M., & Tuason, V.B. did, not correlate with the number of helplessness theory specifies the mecha- (1988). Active components and mediating times the coach allowed the swimmer to nism by which achievement is impaired: mechanisms in cognitive therapy, pharmaco- compete during the season, a more ob- lowered response intiation after defeat. therapy, and combined cognitive-pharmaco- therapy for depression: III. Processes of jective index of how good the coach Thus it provides a fairly precise and test- change in the CPT project. Manuscript submit- thought the swimmer was. Third, the able account ofhow a cognitive style im- ted for publication. main dependent variable in Study 1 was pairs performance. Indeed Study 2 Hull, J., Van Treuen, R., & VirneIli, S. (1987). Har- diness and health: A critique and alternative not absolute time, but doing better or tested and confirmed the role ofthis pur- approach. Journal of Personality and Social worse than expected, a judgment which ported mechanism. We suggest that even Psychology. 53. 518-530. takes into account high expectations for if explanatory style is correlated with Nolen-Hoeksema, S. (1987). Sex differences in depression: Theory and evidence. Psychological highly talented swimmers. Finally, in these other styles, there is conceptual Bulletin, 101. 259-282. Study 2 improvement or deterioration in and empirical justification for focusing Nolen-Hoeksema, S., Girgus. J., & Seligman, on explanatory style as a predictor of M.E.P. (l986). Learned helplessness in chil- time relative to a' swimmer's prior time , dren: A longitudinal study of depression, was the target; so, swimming was judged achievement. achievement, and explanatory style. Journal relative to the swimmer's own time, not In conclusion, we found that explan- ofPersonality and Social Psychology. 51. 435- 442. by an absolute standard ofquality. Thus, atory style predicted swimming perfor- Peterson, C., & Barrett, L.C. (1987). Explanatory explanatory style does not appear to be a mance. Optimists performed better than style and academic performance among uni- mere reflection of athletic talent. expected and pessimists worse than ex- versity freshmen. Joumal of Personality and Social Psychology. 53, 60~7. Nor is it a dimension that the coach pected, particularly after defeat. Explan- Peterson, C., & Seligman, M.E.P. (1984). Causal ex- already knows about and takes into ac- atory style predicted swimming perfor- planations as a risk factor for depression: The- count. Explanatory style did not corre- mance beyond measures of talent, sug- ory and evidence. Psychological Re,·iew. 91. 347-374. late with the coaches' judgments. But it gesting that actual performance is jointly Peterson, C., Semmel, A., von Baeyer, c., Abram- accounted for a significant amount ofthe determined by talent and habitual pat- son, L.Y., Metalsky, G.!., & Seligman,

VOL. I, NO.2, MARCH 1990 145 Downloaded from pss.sagepub.com at UNIV OF PENNSYLVANIA on July 13, 2015 PSYCHOLOGICAL SCIENCE

Swimming and Explanatory Style

M.E.P. (1982). The Attributional Style Ques- sion, dewlopment, and death. San Francisco: atory style as a predictor of performance as a tionnaire. Cognitive Therapy and Research, 6, W.H. Freeman. life insurance sales agent. Journal ofPerson- 287-299. Seligman, M.E.P., Abramson, L., Semmel, A., & ality and Social Psychology, 50, 832-838. Rhodewalt, F., Saltzman, A., & Wittmer, J. (1984). von Baeyer, C. (1979). Depressive attribu· Sweeney, P.D., Anderson, K., & Bailey, S. (1986). Self-handicapping among competitive athletes: tional style. Journal ojAbnormal Psychology, Attributional style in depression: A meta- The role of practice and self-esteem protec- 88, 242-247. analytic review. Journal oj Personality and tion. Basic and Applied Social Psychology, 5, Seligman, M.E.P., Castellon, C., Cacciola, J., Social Psychology, 50, 974-998. 197-209. Schulman, P., Luborsky, L., Ollove, M., & Watson, D., & Clark, L. (1984). Negative affectiv- Scheier, M.F., & Carver, C.S. (1987). Dispositional Downing, R. (1988). Explanatory style change ity: The disposition to experience aversive optimism and physical well-being: The influ- during cognitive therapy for unipolar depres- emotional states. Psychological Bulletin, 96, ence of generalized outcome expectancies on sion. Journal of Abnormal Psychology, 97, 465-490. health. Journal ojPersonality, 55, 169-210. 1-6. Seligman, M.E.P. (1975). Helplessness: On depres- Seligman, M.E.P., & Schulman, P. (1987). Explan- (RECEIVED 8/3/89; ACCEPTED 10113/89)

146 VOL. I, NO.2, MARCH 1990 Downloaded from pss.sagepub.com at UNIV OF PENNSYLVANIA on July 13, 2015