03 Pates - Semantic Scholar

The Sport Psychologist, 2003, 17, 55-76 © 2003 Human Kinetics Publishers, Inc.

Self-Efficacy, Causal Attribution, and Track Athletic Performance Following Unexpected Success or Failure Among Elite Sprinters Christophe Gernigon National Institute of Sport and Physical Education, Paris Jean-Baptiste Delloye University of Paris X The influence of an unexpected outcome in a first sprint trial on athletes’ selfefficacy and performance, and the relationships between outcome, causal attribution, self-efficacy, and performance were examined. Sixty-two national level competition sprinters assessed self-efficacy, ran a first 60 m trial with manipulated time feedback (success vs. failure), expressed causal attributions, assessed self-efficacy again, and ran a second 60 m trial. Success and failure, respectively, increased and decreased self-efficacy. Stability of causes mediated the feedback, self-efficacy relation for males. Personal control predicted self-efficacy for females. Performance was not influenced by feedback but was weakly predicted by self-efficacy. This study sheds light on some of the cognitive and motivational processes that are involved in serial sports events.

One of the most investigated types of thought that strongly predicts achievement behaviors, such as effort expenditure, persistence, and performance, is selfefficacy expectation (Bandura, 1977, 1986, 1997). Self-efficacy refers to “beliefs in one’s capabilities to organize and execute the courses of action required to produce given attainments” (Bandura, 1997, p. 3). Recent narrative (Feltz & Lirgg, 2001) and meta-analytic (Moritz, Feltz, Fahrbach, & Mack, 2000) reviews conducted in the domain of sport have shown clear evidence for a significant relationship between self-efficacy beliefs and performance. Because of the importance of this relationship, there is an obvious need to understand the mental processes that

Christophe Gernigon is now with the Faculty of Sport Sciences and Physical Education at the University of Montpellier 1, Montpellier, France. E-mail: [email protected]. Jean-Baptiste Delloye is with the Department of Social Psychology of the University of Paris X - Nanterre, France. 55

56

• Gernigon and Delloye

take place from the perception of a specific sport outcome unto athletes’ sense of future efficacy. Bandura’s (1986, 1997) self-efficacy theory lies in the broader framework of social-cognitive theory, according to which human agency operates within an interdependent causal structure involving triadic reciprocal causation between personal factors, external environment, and behavior. For Bandura, self-efficacy beliefs are constructed from four major sources of information: past enactive mastery experiences, vicarious experiences, verbal persuasion, and physiological and affective states. Past inactive mastery experiences, the most influential sources of self-efficacy, refer to the personal history of previous successes and failures encountered in a given situation. Prior successes build robust beliefs in one’s personal efficacy and favor future achievement, whereas failures entail the opposite effects. Vicarious experiences correspond to the appraisal of one’s capabilities to achieve, based on the observation of the performances of others. Verbal persuasion that can be conveyed through evaluative feedback provided by knowledgeable persons also influences self-efficacy beliefs, but to a lesser extent than inactive and vicarious experiences. Finally, the least operating sources of self-efficacy are physiological (e.g., activation, fitness, fatigue, pain) and affective states (e.g., mood). Additionally, Maddux (1995) proposed imaginal experiences as a separate source of self-efficacy. Imaginal experiences refer to imagining oneself or others succeeding or failing in the future. In order to test the influence of mastery experiences on self-efficacy, sport psychology researchers have sometimes experimentally induced failure or success by manipulating feedback. In a study conducted by Fitzsimmons, Landers, Thomas, and van der Mars (1991), experienced weightlifters were told that they had lifted more or less than they actually had. Weightlifters’ self-efficacy beliefs and performances then appeared to be increased by the false positive feedback. Recently, Escartí and Guzmán (1999) used a similar approach to manipulate college students’ 70 m hurdles performance. The participants who received positive feedback showed higher self-efficacy and performance than those who had received negative feedback. These findings support, in the domain of sport, Bandura’s (1986, 1997) contention that past experiences can effectively influence self-efficacy. The causes that are used by individuals to explain success or failure in achievement contexts are also considered to be effective determinants of their expectations and future behavior. According to Weiner’s (1985a, 1986, 1992) causal attribution theory, the influence of causal attribution originates less in the causes themselves than in the dimensions that underlie them: locus of causality, stability, and controllability. Locus of causality involves causes that are internal or external to the person (i.e., attribution to personal or situational factors, respectively). Stability refers to the extent that a cause is modifiable (i.e., stable or unstable). Controllability indicates whether or not a cause can be modified by the person (i.e., controllable or uncontrollable). Central to Weiner’s theoretical framework is that causal dimensions that characterize the causes result from individuals’ own attributional thinking. For example, ability is a cause that can be either stable if perceived as a gift or unstable if seen as improvable by learning. How causes are attributed to a specific event depends on both personal (e.g., Peterson & Seligman, 1984) and situational (e.g., Weiner, 1985b) factors. One typical situational characteristic that influences behavior is the outcome of a task, according to which individuals tend to adopt different attributional strategies. These

Self-Efficacy, Causal Attribution, Athletic Performance

•

57

strategies called “self-serving bias” (Miller & Ross, 1975) are used to protect or enhance self-esteem. Attributing a failure to external causes such as the difficulty of the task allows individuals to avoid depreciating their self-worth. On the other hand, there is a preference to attribute success to internal and stable causes such as fixed ability because these causes are more gratifying. Empirical support of the self-serving bias has been provided in sport settings. For example, McAuley and Gross (1983) found that winners of table tennis matches explained their outcomes with more internal, stable, and controllable attributions than losers did, whereas Grove, Hanrahan, and McInman (1991) observed that basketball winning situations entailed more stable and controllable attributions than losing situations did. In the area of competitive sport, McAuley (1985) found that female intercollegiate gymnasts who performed well and perceived their performance in competition as highly successful made more internal, stable, and controllable attributions than those who scored lower and perceived their performance as less successful. According to Weiner (1985a, 1986, 1992), the action of causal attribution is motivational, since causal dimensions influence choice, intensity, and persistence of behavior. Motivational effects of causal attributions on motor or sport performance have been observed. Results of laboratory studies carried out in the area of motor tasks showed that performance is enhanced when outcomes are attributed to unstable cause such as effort (Grove & Pargman, 1986) or when perceived failures are attributed to internal, unstable, and controllable causes (Rudisill, 1989; Rudisill & Singer, 1988). In sport, Orbach, Singer, and Murphey (1997) found that adults who were trained to attribute their failures to unstable and controllable causes actually made unstable and controllable ascriptions and improved their performance in a task of basketball dribbles, whereas those who were trained to stable and uncontrollable causes did not. Links can be established between causal attribution theory and self-efficacy theory, since they are both founded on self-referent thoughts involving personal beliefs about one’s control over the environment (McAuley, 1992). These two conceptual frameworks of human behavior both emphasize the role played by different types of perception of control as a key determinant of motivation and action (Biddle, 1999; Skinner, 1996). According to Skinner’s (1996) classification of the different theoretical constructs of perceived control, causal attributions and their dimensions are involved in the “means-ends” relation of the control process. Indeed, causal attributions are one of the means that can be seen by individuals as potentially leading to particular outcomes. On the other hand, self-efficacy expectations are perceptions of control that are involved in the “agent-means” relation of the process of control, because they refer to the personal belief that one can act or not on the means that is assumed to lead to the desired outcome. Thus, causal attributions and self-efficacy are linked, since the causes perceived as explaining the occurrence of an event (i.e., means-ends relation) can then be deemed as more or less manageable by the person him or herself (i.e., agent-means relation) in the future. For Weiner (1985a, 1986), the link between causal attributions and expectations is ensured by the stability dimension of causes. If the causes of a given outcome are perceived as stable, then the person will think that there are few reasons that they will change. Consequently, future outcomes will be expected to be similar to the previous one. In these circumstances, a success would increase the anticipation of future success, and a failure would increase the perceived likelihood

58


of failing again. This explains why some training based on success experiences provided with internal and stable attributional feedback such as ability have been found to develop a sense of efficacy in academics (Schunk, 1983, 1984). If the causes are seen as unstable, then the person cannot make any prediction, and future outcomes can either be expected to be similar to or to differ from the earlier one. However, perceiving unstable causes as also internal and under his or her own control (e.g., effort) is particularly useful after a failure, since the undesired outcome can be expected to change in the future. For example, a weak achievement attributed to an unstable and controllable factor such as low effort could be seen as improvable by increasing effort. Thus, internal, unstable, and controllable causes such as effort are also worthy of being trained (Försterling, 1985). The existence of relationships between causal attribution, self-efficacy, and performance is well documented by research in academics. For instance, didactic training with effort and/or ability attributional feedback proved to have positive effects on children’s self-efficacy and arithmetic skill development (Schunk, 1982, 1983, 1984; Schunk & Cox, 1986). However, as pointed out by McAuley (1992), few investigations dealing with these relationships have been conducted in the area of motor tasks, and even fewer in the context of competitive sport. In the domain of motor tasks, Rudisill (1989) gave various attributional instructions to junior high school students who then experienced failure in a stabilometer balancing task. Students who were oriented to perceive their performance as due to internal, unstable, and controllable causes revealed higher self-efficacy and performed better than both those who were oriented to attribute their performance to internal, stable, and uncontrollable causes and those who were provided with no instructions. In sport, collegiate beginning tennis players who were trained to attribute their failures in tennis-ball returns to unstable and controllable causes developed higher self-efficacy than those who were trained to stable and uncontrollable causes (Orbach, Singer, & Price, 1999). Bond, Biddle, and Ntoumanis (2001) examined the relationships between causal attribution and pre and postcompetition self-efficacy among experienced female golfers. Under conditions of perceived success, stability was found to predict postcompetition self-efficacy. Furthermore, golfers whose efficacy increased made more internal and stable attributions than those whose self-efficacy decreased. Of particular interest would be the role of causal attributions in sports that include several events at close intervals (e.g., semi-final and final events). According to Bandura’s (1977, 1986, 1997) self-efficacy theory, it is likely that the outcome in an earlier event may have some cognitive and behavioral consequences on the next one. However, these consequences are assumed to be mediated by attributional activity (Weiner, 1985a, 1986). Furthermore, because a causal search is particularly elicited from unexpected events (Weiner, 1985b), the mediating role of causal ascriptions is more likely to be salient when an athlete’s performance gives unexpected results. The primary purpose of this study was to examine the influence of an unexpected success or failure on a first sprint trial on elite sprinters’ self-efficacy and performance on a second trial immediately following. The relationship between unexpected outcome, causal attributions, self-efficacy, and performance were also explored and more specifically, whether causal dimensions mediated the relation between outcome and self-efficacy. It was hypothesized that the announcement of an unexpected success on the first trial would increase self-efficacy and performance


•

59

on the following trial, whereas the announcement of an unexpected failure would entail the opposite effects (e.g., Bandura, 1997). If there is a mediating role of causal dimensions (e.g., Weiner, 1985a), success should be related to internal and stable attributions (Miller & Ross, 1975), which in turn should be related to high subsequent self-efficacy expectations (e.g., Weiner, 1985a), while failure should be related to external attributions (Miller & Ross, 1975), which in turn should be related to low subsequent self-efficacy expectations (e.g., Weiner, 1985a). Moreover, self-efficacy for the second trial should positively predict the second performance (e.g., Bandura, 1997).

Method Participants Sixty-two national level French sprinters (42 males, 20 females) whose average age was 19.9 years (SD = 3.1) voluntarily participated in the study. They were all specialists of short distance track events (inferior to 400 m) and were enlisted in the national indoor championship. They were recruited from a score of clubs and training centers whose presidents and coaches gave their consent.

Task and Apparatus The task consisted in running 60 m alone leaving from a starting block. The starting orders followed the rules of the International Athletic Amateur Federation (IAAF). The distance of 60 m was chosen on the basis of expert coaches’ recommendations. Two major reasons warranted this choice. First, for the athletes’ sake, a task that was as minimally physiologically disruptive as possible had to be chosen. Short sprint running uses more quickly recoverable physiological substrata and produces lower levels of blood lactate concentrations than long sprint running does. Second, it was important to enroll the greatest possible number of participants who were familiar with the task. Both short and long sprint runners are used to training intensively on short distances and know their best time on 60 m, whereas only long sprint runners are used to training on longer distances, such as 400 m. The apparatus included a computer connected to a starting gun and to a photoelectric cell. This computer was installed at the track side, near the finish line. The starting gun was used near the start line. The photoelectric cell and its reflector were placed on top of telescopic tripods, 1.30 m above the finish line. The computer ran a software that timed the participant from the starting signal until he or she passed the photoelectric cell. Another program manipulated the feedback provided at the end of the first trial, according to the experimental conditions of success or failure. The manipulation of the time clocked on the first trial needed to be equivalent for all the athletes, whatever their performance levels. Indeed, improving one’s time by 0.10 s does not represent the same performance variation for an athlete who runs 60 m in 6.50 s and for an athlete who runs this distance in 7.45 s. Consequently, the time manipulation was based on the IAAF Scoring Table for Indoor Athletics (Spiriev, 1998). This table was constructed from international athletic performance statistics and is set up in such a way that time scores are translated to a corresponding number of points. By translating time into points on the table, the manipulation of the scores in numbers of points could be held constant, whatever the level of the athletes in the different feedback conditions.

60


Thus, a given correction in terms of points corresponds to a smaller time correction for the fastest participants and to a larger correction for the slowest ones. In the success condition, the number of points corresponding to the time that was announced to the participant as feedback was the number of points corresponding to the time he or she expected to carry out, plus 60. In the failure condition, the number of points was reduced by 60 points. Finally, the time gap between the expected performance and the announced performance was contained between + (failure) or - (success) 0.13 s for the fastest participants, and + or - 0.21 s for the slowest participants.

Measures Self-Efficacy. Self-efficacy expectations were assessed using a procedure adapted from Bandura and Adams (1977), which distinguishes level and strength of expectations. The level of self-efficacy was measured in terms of points on the IAAF scoring table. These points corresponded to the time that the participant expected to make for the next 60 m trial, to the nearest of 0.01 s. To indicate this time, he or she answered the following item: “For this first (or second) 60 m test, please, indicate what time performance (to the nearest of 0.01 s) you expect to achieve.” The strength of self-efficacy was the degree of confidence of making the expected time, which each participant indicated by answering the following question: “What is your degree of certainty to achieve the performance that you have just stated?” The score of strength of self-efficacy was then the percentage the participant had to circle on a scale ranging from 10% (not sure) to 100% (totally sure). Causal Attributions. Causal attributions and dimensions were assessed using a French version of the Revised Causal Dimension Scale (CDS II; McAuley, Duncan, & Russell, 1992). Participants first wrote an answer to an open-ended question regarding what they perceived was the main cause for their performance. They then assessed the causal dimensions. Three bipolar items were used to measure each dimension of locus of causality (internal vs. external), stability (stable vs. unstable), personal control (personally controllable vs. not personally controllable), and external control (controllable by others vs. not controllable by others) of this cause along 9-point Likert-type scales. The method of back-translation (Brislin, Lonner, & Thorndike, 1973) was used. It consisted in first translating the items from English to French by a bilingual researcher. This was followed by a second translation from French to English by an independent bilingual translator. The final English version thus obtained was then submitted to the first author of the CDS II who acknowledged its conformity to the original. For each causal dimension scale of the French questionnaire, the coefficients alpha (Cronbach, 1951) were respectively .36 (locus), .63 (stability), .62 (personal control), and .75 (external control). The locus scale displayed internal consistency that was too low and was then excluded from the analyses. The stability and the personal control scales had somewhat low internal consistencies. However, according to Nunnally and Bernstein (1994), coefficients alpha can be affected by underestimates of scale item intercorrelations due to a small number of items per scale. So, marginally acceptable internal consistencies with coefficients alpha exceeding .60 can be tolerated for questionnaires with a limited number of items per scale. This is common for the CDS II (see Biddle & Hanrahan, 1998 for a review), which has three items per scale. Consequently, and also because of the novelty of the French version of the CDS II, stability and personal control scales were retained for the present investigation.


•

61

Performance. Performance scores were the number of points of the IAAF scoring table that corresponded to the actual time to run each 60 m trial. For example, scores of 600, 700, and 800 points corresponded to times of 7.88 s, 7.62 s, and 7.38 s, respectively.

Procedure The study was conducted by two male experimenters. It unfolded during the two months preceding the indoor national championship at a time when the participants were highly motivated to perform well. The participants were enrolled in the study on the spot during their training sessions. Males and females were randomly assigned to one of the two conditions of feedback (success vs. failure). The procedure consisted of the following phases: presentation of the study, warm-up, first trial, recovery, second trial, first debriefing, second debriefing. Presentation of the Study. To create a challenge, the study was presented to the athletes as a test ordered by the Track and Field Federation to assess the level of their form before the championship. Athletes were informed that they were free to participate or not and that they could quit at any moment of the test, if they wished. It was emphasized that those who would agree to do the test were expected to run with the same determination as in competition. It was also explained that the test consisted in running 60 m twice with a 10-min recovery period between the two trials and that the starting conditions would respect the usual rules of track running. It was announced that this test would also lend support to a study about how athletes explain their performance and that participants would have to fill in questionnaires. The participants were assured of the confidentiality of their responses. Warm-Up. After signing an informed consent form, each participant had to warm up as for an actual competition, for about 45 min. At the end of the warmup, he or she was invited to fill in the self-efficacy questionnaire concerning the first 60 m trial. The level of self-efficacy expectations (i.e., the participant’s expected time) was then keyed in to the computer so to convert the time as described above. First Trial. Each participant ran a first 60 m and went to read his or her time on the computer screen right after crossing the finish line. The performance that was displayed was the falsified time according to the conditions of unexpected success or failure. Recovery. Each participant was then invited to get dressed to stay warm and to rest for 10 min. During recovery, the participant filled in the causal attribution questionnaire for the first trial and the self-efficacy questionnaire for the second trial. Second Trial. The second 60 m trial was then run in the same conditions as the first, except that the feedback that was provided by the computer was not manipulated. First Debriefing. Because the sessions of the study (i.e., one session per club or training center) could not take place at the same time, the participants could not be completely debriefed until all the sessions were finished. Had they been completely debriefed immediately, the participants of one session could have informed other athletes from other sites about the manipulations of the study during regional competitions. Therefore, when all the participants of one session had been tested, a first (partial) debriefing took place. During this immediate debriefing,

62


the participants were informed that the timing device was very accurate but had not been standardized according to an official time clock so that they ought not to lend too much importance to their times and not modify their preparation for the championship because of them. It was added that their actual outcomes would be mailed to them as soon as the device was checked. Second Debriefing. When all the sessions of the study were finished, a second debriefing was carried out by mail. Each participant was informed of the actual outcomes he or she obtained during the study. The letter also emphasized how much the factors being studied could have an impact on athletes’ performance. It explained in detail how the actual times were manipulated and gave the main results of the study accompanied by its implications when applied to preparing for competitions.

Data Analyses Because the IAAF table is common to males and females, gender comparisons pertaining to the scores of level of self-efficacy and performance would have been biased by the usual gender differences observed in raw athletic performances. Consequently, the analyses were conducted separately for males and females. To examine the influence of feedback on subsequent self-efficacy and performance, multivariate and univariate feedback (success, failure) by trial analyses of variance, with repeated measures on the last factor, were processed for selfefficacy variables and performance. Scheffé tests were used as post hoc comparisons. Furthermore, to assess the meaningfulness of the observed differences, effect sizes were calculated using pooled standard deviations (Hedges & Olkin, 1985). Finally, the relations between feedback, causal dimensions, self-efficacy variables, and performance were examined using separate simple and stepwise multiple regression analyses. According to Baron and Kenny’s (1986) recommendations, the test of the assumed mediating role of causal dimensions in the relation between feedback and self-efficacy followed a four-step procedure: (a) regression of each causal dimension on feedback, (b) regression of each self-efficacy variable on causal dimensions, (c) regression of each self-efficacy variable on feedback, and (d) regression of each self-efficacy variable on feedback when the mediator (i.e., a causal dimension) was controlled. There is a mediating effect of a particular causal dimension if significant links are found at the first three steps and if the relation between feedback and self-efficacy is significantly reduced when the mediator is introduced into the analysis (step d). Additional regression analyses were processed to examine the relations between self-efficacy and performance.

Results Factorial Analyses The means and standard deviations of the scores obtained for each category of participants for self-efficacy variables and performance are presented in Table 1. The MANOVA for males yielded a significant main effect for trial, Wilks’s Lambda = .62, F(3, 38) = 7.66, p < .001, and a significant feedback by trial interaction, Wilks’s Lambda = .37, F(3, 38) = 21.90, p < .001, but no main effect for feedback. The MANOVA for females only revealed a significant feedback by trial interaction, Wilks’s Lambda = .39, F(3, 16) = 8.37, p < .01. Follow-up ANOVAs

557.5 582.3 532.7

Females (n = 20) Success (n = 10) Failure (n = 10)

SD

112.4 89.5 131.5

100.2 114.3 89.1

Note. SEE = self-efficacy expectations.

868.4 876.1 862.0

M

Males (n = 42) Success (n = 19) Failure (n = 23)

Groups

SEE (level)

62.00 63.00 61.00

71.19 65.26 76.09

M

14.73 13.38 16.63

14.56 14.76 12.70

SD

SEE (strength)

Trial 1

483.8 499.2 468.3

780.8 797.1 767.3

M

115.9 121.3 114.6

91.7 104.7 79.2

SD

Performance

549.1 611.6 486.5

886.7 932.1 849.3

M

122.0 85.5 124.2

114.3 113.9 102.4

SD

SEE (level)

65.55 71.10 60.00

63.93 68.16 60.44

M

16.71 12.86 18.86

17.72 18.50 16.65

SD

SEE (strength)

Trial 2

Table 1 Means and Standard Deviations of Self-Efficacy Expectations and Performance According to Feedback for Males and Females

473.2 492.9 453.4

783.9 802.7 768.3

M

118.7 123.6 116.5

91.8 99.7 83.8

SD

Performance

Self-Efficacy, Causal Attribution, Athletic Performance • 63

64


were then conducted separately for level and strength of self-efficacy and performance. Level of Self-Efficacy. A significant main effect for trial was found for males, F(1, 40) = 19.18, p < .001. Males’ level of self-efficacy slightly increased from the first to the second trial (ES = .17). However, this main effect was superseded by a significant feedback by trial interaction effect, F(1, 40) = 48.27, p < .001. Scheffé tests indicated that in the success condition males’ level of selfefficacy moderately increased from the first to the second trial (p < .001, ES = .49). Moreover, males’ level of self-efficacy for the second trial was largely higher for participants in the success condition than for those in the failure condition (p < .001, ES = .77). For females, the feedback by trial interaction was also significant at the univariate level, F(1, 18) = 27.83, p < .001. Scheffé tests revealed that in the failure condition, females’ level of self-efficacy slightly decreased from the first to the second trial (p < .01, ES = .36). In the success condition, females’ level of selfefficacy tended to increase slightly from the first to the second trial, but this trend did not reach significance (p < .069, ES = .34). As a result, females’ level of selfefficacy for the second trial was largely higher for participants in the success condition than for those in the failure condition (p < .001, ES = 1.17). Strength of Self-Efficacy. A significant main effect for trial was found for males, F(1, 40) = 5.36, p < .05. Males’ strength of self-efficacy moderately decreased from the first to the second trial (ES = .45). However, this main effect was superseded by a significant feedback by trial interaction effect, F(1, 40) = 11.33, p < .01. Scheffé tests indicated that this decrease occurred only in the failure condition and was large (p < .01, ES = 1.06). No significant effect was found for strength of self-efficacy at the univariate level for females. Performance. The ANOVAs processed for performance did not reveal any significant effects for males or for females.

Regression Analyses For the regression analyses, feedback was coded 1 (success) versus -1 (failure) and all the variables were standardized to reduce multicollinearity (Jaccard, Turrisi, & Wan, 1990). Each time a dependent variable pertained to the second trial (i.e., level and strength of self-efficacy and performance for the second trial), the corresponding variable for the first trial was controlled by first entering it into the analysis (i.e., level and strength of self-efficacy and performance for the first trial, respectively). It should be noted that the number of female participants was marginally sufficient in regard to the number of variables of some regression analyses which were processed. Consequently, caution should be exercised when interpreting the findings for females yielded by these analyses. The results of the regression analyses are summarized in Table 2 (males) and Table 3 (females). Relations Between Feedback and Causal Dimensions. Among the different causal dimensions that were retained, only stability was significantly predicted by feedback for both males (␤ = .45, p < .01) and females (␤ = .55, p < .01), with 21% and 30% of the variance explained, respectively. Consequently, none of the other causal dimensions could have played a mediating role in the relation between feedback and self-efficacy. Relations Between Causal Dimensions and Self-Efficacy. A first set of analyses was conducted using the level of self-efficacy for the second trial as the predicted variable. For males, the level of self-efficacy for the first trial was entered

Causal dimensions Æ Strength of SEE 2

FB Æ Level of SEE 2 with potential mediator controlled

FB Æ Level of SEE 2

Causal dimensions Æ Level of SEE 2

FB Æ Causal dimensions

Strength of SEE 2

Level of SEE 2

Stability Personal control External control

Predicted variable

Tested relation

1 2 3 4

1 2 3

1 2

1 2 3 4

1 1 1

Step

Summary of Regression Analyses for Males

Table 2

Strength of SEE 1 Stability External control Personal control

Level of SEE 1 Stability FB

Level of SEE 1 FB

Level of SEE 1 Stability External control Personal control

FB FB FB

Predictor variable

.29 .38 .11 .01

.89 –.01 .30

.89 .30

.91 .14 .06 .02

.45 .04 –.01

␤

.06 .19 .20 .20

.83 .85 .92

.83 .92

.83 .85 .85 .85

.21 .00 .00

R (cum.)

2

.13 .01 .00

.02 .07

.09

.02 .00 .00

R2 change

Females

•

(continued)

1.90 2.52* .70 .04

20.02*** –.12 6.09***

20.33*** 6.85***

13.65*** 2.19* .90 .34

3.26** .25 –.07

t–value

Self-Efficacy, Causal Attribution, Athletic Performance 65

Strength of SEE 2

FB Æ Strength of SEE 2

Performance 2

Strength of SEE 2 Æ Performance 2 1 2 3

1 2 3

1 2 3

1 2

Step

Performance 1 Strength of SEE 1 Strength of SEE 2

Performance 1 Level of SEE 1 Level of SEE 2

Strength of SEE 1 Stability FB

Strength of SEE 1 FB

Predictor variable

.92 .08 .04

.73 .12 .17

.37 .26 .24

.39 .36

␤

.87 .88 .88

.87 .91 .91

.06 .19 .23

.06 .18

R (cum.)

2

Males

.01 .00

.04 .00

.13 .04

.12

R2 change

16.07*** 1.38 .65

10.01*** .87 1.44

2.42* 1.63 1.41

2.52* 2.35*

t–value

Note. For the regression analyses including several steps, R2–values are cumulative, with each incremental step adding to the variance explained. FB = feedback; SEE 1 = self–efficacy expectations for the first trial; SEE 2 = self–efficacy expectations for the second trial; Performance 1 = first performance; Performance 2 = second performance. * p < .05; ** p < .01; *** p < .001.

Performance 2

Level of SEE 2 Æ Performance 2

FB Æ Strength of SEE 2 with potential mediator controlled

Predicted variable

(continued)

Tested relation

Table 2

66 • Gernigon and Delloye

Causal dimensions Æ Strength of SEE 2

FB Æ Level of SEE 2

Causal dimensions Æ Level of SEE 2

Strength of SEE 2

Level of SEE 2

Stability Personal control External control

Predicted variable

1 2 3 4 Strength of SEE 1 Stability Personal control External control

Level of SEE 1 FB

Stability External control

3 4 1 2

Level of SEE 1 Personal control

FB FB FB

Predictor variable

1 2

1 1 1

Step

Summary of Regression Analyses for Females

FB Æ Causal dimensions

Tested relation

Table 3

.49 .22 –.13 –.10

.84 .34

.05 .01

.89 .21

.55 .36 .21

␤

.33 .35 .36 .37

.83 .94

.88 .88

.83 .88

.30 .13 .04

R (cum.)

2

.02 .01 .01

.11

.00 .00

.05

R2 change

Females

(continued)

2.21* .99 –.59 –.47

14.31*** 5.75***

.57 .10

9.12*** 2.37*

2.84** 1.70 .93

t–value Self-Efficacy, Causal Attribution, Athletic Performance • 67

Performance 2

Strength of SEE 2 Æ Performance 2 1 2 3

1 2 3

1 2

Step

Performance 1 Strength of SEE 1 Strength of SEE 2

Performance 1 Level of SEE 1 Level of SEE 2

Strength of SEE 1 FB

Predictor variable

1.00 –.02 .03

.99 –.13 .14

.55 .30

␤

.98 .98 .98

.98 .98 .98

.33 .42

R2 (cum.)

.00 .00

.00 .00

.09

R2 change

Females

25.27*** –.41 .70

19.34*** –1.52 1.84

3.05** 1.68

t–value

Note. For the regression analyses including several steps, R2–values are cumulative, with each incremental step adding to the variance explained. FB = feedback; SEE 1 = self–efficacy expectations for the first trial; SEE 2 = self–efficacy expectations for the second trial; Performance 1 = first performance; Performance 2 = second performance. * p < .05; ** p < .01; *** p < .001.

Performance 2

Strength of SEE 2

FB Æ Strength of SEE 2

Level of SEE 2 Æ Performance 2

Predicted variable

(continued)

Tested relation

Table 3

68 • Gernigon and Delloye


•

69

first into the analysis and significantly predicted the level of self-efficacy for the second trial (␤ = .91, p < .001), accounting for 83% of the variance. The following steps revealed that stability was the only variable that significantly predicted males’ level of self-efficacy for the second trial (␤ = .14, p < .05), adding 2% to the variance. For females, when entered first, the level of self-efficacy for the first trial significantly predicted the level of self-efficacy for the second trial (␤ = .89, p < .001), accounting for 83% of the variance. Then, only personal control emerged as a significant predictor of females’ level of self-efficacy for the second trial (␤ = .21, p < .05), adding 5% to the variance. A second set of analyses was conducted using the strength of self-efficacy for the second trial as the predicted variable. For males, the strength of self-efficacy for the first trial was entered first and tended to predict the strength of selfefficacy for the second trial (␤ = .29, p < .065), explaining 6% of the variance. The following steps revealed that stability was the only variable that predicted males’ strength of self-efficacy for the second trial (␤ = .38, p < .05), accounting for 13% of the variance. For females, when entered first, the strength of self-efficacy for the first trial significantly predicted the strength of self-efficacy for the second trial (␤ = .49, p < .05), accounting for 33% of the variance. None of the causal dimensions that were entered at the subsequent steps emerged as significant predictors of females’ strength of self-efficacy for the second trial. As a result, no attributional mediator of the relation between feedback and subsequent level or strength of self-efficacy was identified for females. Relations Between Feedback and Self-Efficacy. Regarding the level of self-efficacy for males, when entered first, the level of self-efficacy for the first trial significantly predicted the level of self-efficacy for the second trial (␤ = .89, p < .001), accounting for 83% of the variance. Entered next, feedback appeared to significantly predict the level of self-efficacy for the second trial (␤ = .30, p < .001), adding 9% to the variance. For females, entered first, the level of self-efficacy for the first trial significantly predicted the level of self-efficacy for the second trial (␤= .84, p < .001), explaining 83% of the variance. Then, feedback significantly predicted the level of self-efficacy for the second trial (␤ = .34, p < .001), adding 11% to the variance. Regarding the strength of self-efficacy for males, when entered first, the strength of self-efficacy for the first trial significantly predicted the strength of self-efficacy for the second trial (␤ = .39, p < .05), explaining 6% of the variance. Entered next, feedback also significantly predicted the strength of self-efficacy for the second trial (␤ = .36, p < .05), adding 12% to the variance. For females, when entered first, the strength of self-efficacy for the first trial significantly predicted the strength of self-efficacy for the second trial (␤ = .55, p < .01), explaining 33% of the variance. However, the next step revealed that feedback did not significantly predict the strength of self-efficacy for the second trial (p > .10). Relations Between Feedback and Self-Efficacy, With Causal Dimensions Controlled. The only causal dimension that previously met the first conditions of the potential status of a mediating variable in the relation between feedback and self-efficacy was stability. This concerned males only. Introducing stability into the regression analysis prior to feedback did not alter the relation previously found between feedback and males’ level of self-efficacy for the second trial, when the level of self-efficacy for the first trial was controlled (␤ = .30, p < .001). Thus, stability did not mediate the relation between feedback and subsequent males’ level

70


of self-efficacy. The same procedure revealed that the link previously found between feedback and males’ strength of self-efficacy became nonsignificant when stability was entered prior to feedback into the analysis (␤ = .24, p > .17). Therefore, stability was a mediator of the relation between feedback and subsequent males’ strength of self-efficacy. Relations Between Self-Efficacy and Performance. For both males and females, no significant links were found between the level or the strength of selfefficacy for the second trial and the second performance, when the first performance and the level or the strength of self-efficacy for the first trial were controlled. Additional analyses were then processed with males’ and females’ standardized data pooled together to increase the number of observations. Considering the whole population of participants, when entered first into the analysis testing the relation between the level of self-efficacy and performance, the first performance significantly predicted the second performance (␤ = .81, p < .001), accounting for 90% of the variance. Entered second, the level of self-efficacy for the first trial did not predict the second performance (p > .72). Entered at the third and final step, the level of self-efficacy for the second trial slightly predicted the second performance (␤ = .18, p > .05), adding 3% to the variance. When testing the relation between the strength of self-efficacy and performance, the first performance significantly predicted the second performance (␤ = .92, p < .001), explaining 87% of the variance. Entered then, neither the strength of self-efficacy for the first trial predicted the second performance (p > .17), nor did the strength of selfefficacy for the second trial (p > .52).

Discussion The primary purpose of this study was to examine the influence of an unexpected first outcome on self-efficacy and performance on the next trial. As expected, the regression analyses revealed that self-efficacy was related to previous feedback. More specifically, the analyses of variance showed that success feedback entailed an increase in the level of self-efficacy for the next trial, for both males and females, although this increase was only a trend for females. The changes of level of self-efficacy between the first and second trials resulted in a large difference on the second trial between the two feedback conditions: Participants in the success group expressed higher level of self-efficacy for the second trial than those in the failure group did. While success feedback did influence the level of self-efficacy, it did not produced any changes in the strength of self-efficacy. Failure feedback entailed a decrease in the level of self-efficacy only for females and a decrease in the strength of self-efficacy only for males. Taken as a whole, these findings support Bandura’s (1977, 1986, 1997) contention that past experiences are an important source of information used by individuals to construct their self-efficacy beliefs. In the present study, as well as in Fitzsimmons et al.’s (1991) and Escartí and Guzmán’s (1999), a trial with falsified feedback constituted such an experience. The difference that was observed between the patterns of changes of level and strength of self-efficacy requires further explanation concerning the natures of these two dimensions. According to Bandura (1997), the level of self-efficacy refers to the range of perceived capability to answer certain levels of task demands. The strength of self-efficacy addresses confidence only (i.e., the degree of certainty),


•

71

but not the object (i.e., the level of performance in answering task demands) the confidence is about: It should be noted that the construct of self-efficacy differs from the colloquial term confidence, which is widely used in sports psychology. Confidence is a nondescript term that refers to strength of belief but does not necessarily specify what the certainty is about. (p. 382) Consequently, expecting to perform well with more or less certainty is not the same as being highly certain to perform well or to perform badly. According to the present findings, after a success, athletes expected to improve their performance while not altering their confidence. After a failure, males’ confidence to run as fast as the first time decreased, whereas females expected to perform less well than for the first trial while maintaining their confidence. In sum, except for males in the failure condition, the level of self-efficacy was influenced by the prior performance as hypothesized, whereas confidence remained stable whatever the new level of performance expected by the participants. These findings lend support to the assumption that the feedback manipulations were credible for the participants, high level athletes that they were. Males not changing their level of expectations but lowering their confidence after failure could be interpreted as a gender specific resilient pattern of self-efficacy against failure. However, because the present study was not specifically designed to compare males’ and females’ cognitive reactions to success and failure, such an interpretation remains speculative. Contrary to our hypotheses, the effect of feedback did not extend to the second performance. In their study, Escartí and Guzmán (1999) also used a sprint running task and succeeded to demonstrate direct effects of feedback on performance. Two differences between Escartí and Guzmán’s study and the present one could explain this inconsistency. First, the participants enrolled by Escartí and Guzmán were college students, whereas our research bore on elite sprint athletes. Second, the task used by Escartí and Guzmán included technical skills to master in the form of hurdles to be gotten over. Because of this difficulty, the task may be considered, for nonspecialists of hurdle racing, as similar to a learning task. In such a condition, individuals’ lack of experience makes their self-efficacy beliefs and subsequent achievements more easily influenced by a single experience (Bandura, 1997). Conversely, in the present study, elite athletes had to perform in conditions that reproduced some of the characteristics of a natural sport environment. Consequently, it was necessary to use a task that stemmed from the athletes’ regular sport activity. Elite athletes obviously have strong experience in such a task and have probably developed a high and little malleable sense of efficacy in it. Although changes of self-efficacy due to feedback conditions were observed, it can be assumed that these changes occurred within a high level of self-efficacy. Reported self-efficacy differences may really be reflecting just degrees of high self-efficacy, so that it could not be easily followed by significant performance differences. This assumption is consistent with Bandura’s (1986, 1997) contention that resilient self-efficacy acquired by repeated experiences buffers individuals from the effects of occasional outcomes. The lack of effect of feedback on performance could also be due to the short duration of 60 m running. Performance has been shown to be sensitive to selfefficacy manipulations in tasks such as muscular endurance task (e.g., George, Feltz, & Chase, 1992; Gould & Weiss, 1981) that require persistence or even

72


resistance to pain or exhaustion. It can be assumed that the shortness of the 60 m running task that was used for the present study might offer a small margin of action for the motivational effects of self-efficacy. This interpretation is supported by the fact that the level of self-efficacy for the second trial appeared to weakly predict the second performance. To solicit the motivational power of self-efficacy more, long exposures to painful situations, such as running events where performance is mainly determined by the athletes’ capacity to withstand exhaustion, would have been more relevant (Tenenbaum et al., 2001). To create such conditions in track running, it would have been necessary to have the participants run for distances of at least 400 m. This option was not chosen because of the muscle lactate accumulation that this type of exercise would have induced. High levels of blood lactate concentrations could have disturbed the training sessions just before the national championship. One of the solutions to this issue would have been to collect data right after actual successes and failures in serial competitive events such as 400 m. However, it would have been unlikely to get athletes’ and coaches’ consents for such an invasive investigation. This study also aimed at examining the relationships between unexpected success or failure, causal attributions, and self-efficacy. Regarding the relations between feedback and causal attribution, the self-serving bias hypothesis (Miller & Ross, 1975) could not have been verified as to the locus of causality because of the lack of reliability of the scale measuring this causal dimension. However, a positive relation was found between feedback and stability for both males and females, feedback accounting for a large part of the variance of stability. This finding is consistent with previous observations that winning situations lead to more stable attributions than losing ones do (Grove et al., 1991; McAuley & Gross, 1983). Success being more likely to be attributed to stable causes supports the selfserving hypothesis, because stability often refers to gratifying causes of success, such as ability. With regard to the relations between causal attributions and self-efficacy, the results of the present study show that self-efficacy for the second trial was positively predicted by stability for males. While stability contributed to only a small part of the variance of the level of self-efficacy for the second trial, this causal dimension appeared to be the strongest predictor of males’ strength of selfefficacy for the second trial. Furthermore, stability significantly mediated the relation that was found between feedback and males’ strength of self-efficacy for the second trial. These findings support Weiner’s (1985a, 1986) assumption that causal attribution is a key process through which past outcomes influence future expectations. They also directly confirm that stability does predict these expectations. However, the fact that only the strength of self-efficacy was concerned here gives additional precision to Weiner’s model when applied to sport settings. Stability seems to preferentially influence athletes’ confidence toward their expected goals. Contrary to the results of Bond et al.’s (2001) study, the only causal dimension that predicted females’ self-efficacy for the second trial was personal control. This relation only concerned the level of the expectations. This finding parallels the results of previous studies that have demonstrated the adaptive value of controllable attributions toward self-efficacy in motor task (e.g., Rudisill, 1989). However, no mediating role of any causal dimension was found for females. As previously mentioned, the present study does not permit any reliable interpretations in terms of gender specificities. Further investigations are needed to examine such potential


•

73

specificities concerning the links between causal attribution and self-efficacy. Moreover, because only a small number of females took part in the present study, findings derived from the regression analyses for females should be interpreted cautiously. This study offers particular insights about cognitive and motivational processes that are involved in serial sports events, especially when series follow each other closely. The results provide information about the impact of an unexpected outcome on self-efficacy. They shed light on the relations between causal attribution and self-efficacy, showing that self-efficacy can be predicted by attributions to stable or personally controllable causes. Furthermore, the present findings confirm that the stability of the causes may mediate the influence of outcome on subsequent self-efficacy. Finally, even if they were thin, some links between selfefficacy and high level athletic performance have been found. The practical implications of these findings mainly concern the nature of the causes that should be privileged as relevant coaching strategies to apply during recovery times. For example, attributing success to stable causes, such as ability, can be gratifying and favor self-efficacy. However, it should be considered that this positive influence can be a double-edged sword if it leads athletes to believe that because of the stability of such causes, future outcomes will probably be similar to the earlier one and effort is not necessary. Rather, attributing outcomes to personally controllable causes seems to be both adaptive toward self-efficacy and less risky. Because personal control depends on intention and effort, these latter factors are more likely to be considered as useful by the athlete, when attribution to personally controllable causes is privileged. Therefore, it can be assumed the athlete will be more likely to strongly exert effort in this case. This interest of focusing on the controllable causes has already been demonstrated in sport settings (Orbach et al., 1997; Orbach et al., 1999). Because the present study has some limitations, the previous recommendations are to be considered cautiously. First, the questionnaire that was used to assess the causal dimensions (i.e., CDS II) presents a marginally acceptable reliability. Particularly, the locus scale was not valid. Future research focusing on athletes’ reactions to an unexpected outcome should overcome this problem by improving the psychometric value of the CDS II (particularly its French version) with a greater number of participants. The locus of causality deserves to be considered in future studies that might deal with athletes’ psychological reactions as extended to emotions. Emotions certainly play a major role in these reactions. According to Weiner (1985a, 1986), the relation between outcome and emotions is precisely assumed to be mainly mediated by the locus of causality. Second, even though the patterns of relations that were found between causal attributions and self-efficacy were different for males and for females, it was not possible to ascertain significant gender differences in this matter. Further investigations need to be especially designed to compare males’ and females’ reactions to unexpected outcomes, using equivalent measures of performance and self-efficacy. Moreover, a greater number of female participants is necessary if the interpretation of gender specificities is to be considered reliable. Third, because of its short duration, the task that was used for the present research hardly let the motivational factors of performance, such as striving, act effectively. However, experimental manipulations in longer effort tasks are too invasive to be undertaken shortly before important competitions. One solution for further research on long effort tasks might be to give up the experimental

74


paradigm and to track cognitions and emotions that flow between serial events in a more natural competitive context.

References Bandura, A. (1977). Self efficacy: Toward a unifying theory of behavioral change. Psychological Review, 84, 191-215. Bandura, A. (1986). Social foundations of thought and action: A social cognitive theory. Englewood Cliffs, NJ: Prentice-Hall. Bandura, A. (1997). Self-efficacy: The exercise of control. New York: Freeman. Bandura, A., & Adams, N.E. (1977). Analysis of self-efficacy theory of behavioral change. Cognitive Therapy and Research, 1, 287-310. Baron, R.M., & Kenny, D.A. (1986). The moderator-mediator variable distinction in social psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality and Social Psychology, 51, 1173-1182. Biddle, S.J.H. (1999). Motivation and perceptions of control: Tracing its development and plotting its future in exercise and sport psychology. Journal of Sport & Exercise Psychology, 21, 1-23. Biddle, S.J.H., & Hanrahan, S.J. (1998). Attributions and attributional style. In J.L. Duda (Ed.), Advances in sport and exercise psychology measurement (pp. 3-19). Morgantown, WV: Fitness Information Technology. Bond, K.A., Biddle, S.J.H., & Ntoumanis, N. (2001). Self-efficacy and causal attribution in female golfers. International Journal of Sport Psychology, 31, 243-256. Brislin, R.W., Lonner, W.J. and Thorndike, R.M. (1973). Cross-cultural research methods. New York: Wiley. Cronbach, L.J. (1951). Coefficient alpha and the internal structure of tests. Psychometrika, 16, 297-334. Escartí, A., & Guzmán, J.F. (1999). Effects of feedback on self-efficacy, performance, and choice in an athletic task. Journal of Applied Sport Psychology, 11, 83-96. Feltz, D.L., & Lirgg, C.D. (2001). Self-efficacy beliefs of athletes, teams, and coaches. In R.N. Singer, H.A. Hausenblas, & C.M. Janelle (Eds.), Handbook of sport psychology (2nd ed., pp. 340-361). New York: Wiley. Fitzsimmons, P.A., Landers, D.M., Thomas, J.R., & van der Mars, H. (1991). Does selfefficacy predict performance in experienced weighlifters? Research Quarterly for Exercise and Sport, 62, 424-431. Försterling, F. (1985). Attributional retraining: A review. Psychological Bulletin, 98, 495512. George, T.R., Feltz, D.L., & Chase, M.A. (1992). Effects of model similarity on self-efficacy and muscular endurance: A second look. Journal of Sport & Exercise Psychology, 14, 237-248. Gould, D., & Weiss, M.R. (1981). The effects of model similarity and model talk on selfefficacy and muscular endurance. Journal of Sport Psychology, 3, 17-29. Grove, J.R., Hanrahan, S.J., & McInman A. (1991). Success/failure bias in attributions across involvement categories in sport. Personality and Social Psychology Bulletin, 17, 93-97. Grove, J.R., & Pargman, D. (1986). Attributions and performances during competition. Journal of Sport Psychology, 8, 129-134. Hedges, L.V., & Olkin, I. (1985). Statistical methods for meta-analysis. Orlando, FL: Academic Press.


•

75

Jaccard, J., Turrisi, R., & Wan, C.K. (1990). Interaction effects in multiple regression. Newberry Park, CA: Sage. Maddux, J.E. (1995). Self-efficacy theory: An introduction. In J.E. Maddux (Ed.), Selfefficacy, adaptation, and adjustment: Theory, research, and application (pp. 3-33). New York: Plenum Press. McAuley, E. (1985). Success and causality in sport: The influence of perception. Journal of Sport Psychology, 7, 13-22. McAuley, E. (1992). Self-referent thought in sport and physical activity. In T.S. Horn (Ed.), Advances in sport psychology (pp. 101-118). Champaign, IL: Human Kinetics. McAuley, E., Duncan, T.E., & Russell, D.W. (1992). Measuring causal attributions: The revised Causal Dimension Scale (CDS II). Personality and Social Psychology Bulletin, 18, 566-573. McAuley, E., & Gross, J.B. (1983). Perceptions of causality in sport: An application of the Causal Dimension Scale. Journal of Sport Psychology, 5, 72-76. Miller, D.T., & Ross, M. (1975). Self-serving biases in the attribution of causality: Fact or fiction? Psychological Bulletin, 82, 213-225. Moritz, S.E., Feltz, D.L., Fahrbach, K.R., & Mack, D.E. (2000). The relation of self-efficacy measures to sport performance: A meta-analytic review. Research Quarterly for Exercise and Sport, 71, 280-294. Nunnally, J.C., & Bernstein, I.H. (1994). Psychometric theory (3rd ed.). New York: McGraw Hill. Orbach, I., Singer, R.N., & Murphey M. (1997). Changing attributions with an attribution training technique related to basketball dribbling. The Sport Psychologist, 11, 294304. Orbach, I., Singer, R.N., & Price, S. (1999). An attribution training program and achievement in sport. The Sport Psychologist, 13, 69-82. Peterson, C., & Seligman, M.E.P. (1984). Causal explanations as a risk factor for depression: Theory and evidence. Psychological Review, 91, 347-374. Rudisill, M.E. (1989). Influence of perceived competence and causal dimension orientation on expectation, persistence, and performance during perceived failure. Research Quarterly for Exercise and Sport, 60, 166-175. Rudisill, M.E., & Singer, R.N. (1988). Influence of causal dimension orientation on persistence, performance and expectations of performance during perceived failure. Journal of Human Movement Studies, 15, 215-228. Schunk, D.H. (1982). Effects of effort attributional feedback on children’s perceived selfefficacy and achievement. Journal of Educational Psychology, 74, 548-556. Schunk, D.H. (1983). Ability versus effort attributional feedback: Differencial effects on self-efficacy and achievement. Journal of Educational Psychology, 75, 848-856. Schunk, D.H. (1984). Sequential attributional feedback and children’s achievement behaviors. Journal of Educational Psychology, 76, 1159-1169. Schunk, D.H., & Cox, P.D. (1986). Strategy training and attributional feedback with learning disabled students. Journal of Educational Psychology, 78, 201-209. Skinner, E.A. (1996). A guide to constructs of control. Journal of Personality and Social Psychology, 71, 549-570. Spiriev, B. (1998). Scoring tables of indoor athletics. Budapest: International Athletic Amateur Federation. Tenenbaum, G., Hall, H.K., Calcagnini, N., Lange, R., Freeman, G., & Lloyd, M. (2001). Coping with physical exertion and negative feedback under competitive and selfstandard conditions. Journal of Applied Social Psychology, 31, 1582-1626.

76


Weiner, B. (1985a). An attributional theory of achievement motivation and emotion. Psychological Review, 92, 548-573. Weiner, B. (1985b). “Spontaneous” causal thinking. Psychological Bulletin, 97, 74-84. Weiner, B. (1986). An attributional theory of motivation and emotion. New York: SpringerVerlag. Weiner, B. (1992). Human motivation: Metaphors, theories, and research (2nd ed.). Thousand Oaks, CA: Sage.

Authors’ Notes The present research was carried out while the first author was at the National Institute of Sport and Physical Education, Paris. A part of the present paper was presented at the Annual Conference of the North American Society for the Psychology of Sport and Physical Activity, San Diego, CA, June 2000. This study was ordered by the Track and Field French Federation and supported by a grant from the French Ministry of Youth and Sports. We are grateful to Bruno Reine for constructing and programming the electronic apparatus. Manuscript submitted: March 12, 2001 Revision received: October 17, 2001