Influences of perfectionism on motor performance, affect, and causal attributions in response to critical information feedback.

Anshel, Mark H. ; Mansouri, Hossein

Perfectionism is formally defined as "the setting of excessively high standards of performance in conjunction with a tendency to make overly critical self-evaluations" (Frost, Marten, Lahart, & Rosenblate, 1990, p. 450). It is a trait (dispositional, stable), not state measure. To Antony and Swinson (1998), stability implies that characteristics, such as perfectionism, affect behavior across situations and over time. In extreme cases, perfectionists evaluate their experiences dichotomously, what Bums (1980) calls "all-or-none" thinking, usually labeled good or bad. They are also characterized as overemphasizing precision, neatness, order, and organization. In a somewhat distorted view of others, perfectionists contend that successful individuals achieve their goals with minimal effort, few (if any) errors, high self-confidence, and little or no emotional distress (Burns, 1980). All of these characteristics are relevant for competitive athletes.

In the general psychology literature, perfectionism has been studied with respect to both constructive, also called normal, healthy, and positive, and destructive manifestations, also referred to as neurotic, unhealthy, self-destructive, and negative (Hamachek, 1978; Hewitt and Flett (1990, 1991). Healthy aspects of this disposition include positive achievement striving, self-actualization, personal adjustment, high goal orientation, conscientiousness, self-confidence, and success (Flett, Hewitt, Blankstein, & Mosher, 1991). Negative perfectionists, on the other hand, are overly self-critical, rarely feel competent in carrying out their responsibilities and duties, and consistently doubt the quality of their performance (Hewitt & Flett, 1990, 1991). High perfectionists coped more poorly with stress and suffered greater depression than their less perfectionist peers. In their review of related literature, Hewitt and Flett (1996) concluded that perfectionism was significantly associated with maladaptive coping and heightened negative affect in educational and clinical settings.

In their comparison of normal and neurotic perfectionism, Frost et al. (1990) contend that "Normal perfectionists are those who set high standards for themselves yet feel free to be less precise as the situation permits. Neurotic perfectionists, on the other hand, set high standards but allow little latitude for making mistakes; thus they never feel that anything is done completely enough or well enough" (p. 450). The authors conclude that "the psychological problems associated with perfectionism are probably more closely associated with these critical evaluation tendencies than with the setting of excessively high standards" (p. 450). Whether these characteristics are associated with competitive athletes has yet to be determined.

While perfectionism has been studied extensively in the general psychology literature, to date, it has received surprisingly little attention by theorists and researchers in sport psychology. In the context of competitive sport, it is feasible to surmise that extreme perfectionism can undermine athletes' efforts to maintain their participation in sport, because it is a condition that consistently reinforces failure and perceived low competence. In general, manifestations of neurotic perfectionism might have deleterious effects on the athlete's enjoyment, emotions, and performance (Frost & Henderson, 1991; Zinsser Bunker, & Williams, 1998).

Another plausible hypothesis about the deleterious effects of perfectionism in sport that perfectionistic athletes may be more likely than normal perfectionists or nonperfectionists to feel less gratification from sport and, therefore, are more likely to experience chronic stress and withdraw from sport participation. This is because perfectionistic behavior increases "both the frequency of stressful events or failures and the negative psychological impact of stressful events" (Hewitt & Flett, 1996, p. 419), and is highly associated with trait and state anxiety (Flett et al., 1994). These characteristics have been identified as predictors of low enjoyment in and eventual withdrawal from competitive sport (Anshel, 2003; Zinsser, et al., 1998). It is important to identify athletes who are at risk for engaging in negative self-evaluation in sport settings as has been addressed in the general psychology literature (e.g., Alden, Bieling, & Wallace, 1994; Flett, Hewitt, Endler, & Tassone, 1994). Negative self-appraisal has been linked to increased intensity and greater frequency of perceived stress, particularly in response to making mistakes (Frost & Henderson, 1991; Frost & Marten, 1990). While the current study has not focused on identifying neurotic, as opposed to normal perfectionism, it is important to describe the potential harm of this disposition in sport environments.

The most common inventory used to measure perfectionism is the Multiple Perfectionism Scale (MPS). The MPS is a 35-item inventory that consists of six subscales, concern over mistakes (CM), personal standards (PS), parental expectations (PE), parental criticism (PC), doubts about actions (D), and organization/order (O). Frost et al. (1990) define the multiple dimensions of perfectionism in the following manner. The CM dimensions refers to "negative reactions to mistakes, a tendency to interpret mistakes as equivalent to failure, and a tendency to believe that one will lose the respect of others following failure" (p. 453). PS reflects the person's tendency to set "very high standards and the excessive importance placed on these high standards for self-evaluation" (p. 452). PE is defined as "the tendency to believe that one's parents set very high goals and are overly critical" (p. 453). PC consists of the tendency of parents to engage in consistent, often critical, evaluation of their child, coupled with the child's tendency to place considerable value on these evaluations. D is "the tendency to feel that projects are not completed to satisfaction" (p. 453). Finally, O refers to "the importance of and preference for order" (p. 453). While each sub-component of perfectionism may be independently determined, the sum of the latter five sub-scales reflects a general perfectionism score.

To date, there has been a virtual absence of research on perfectionism among competitive athletes. In one rare study in this area, Frost and Henderson (1991) examined the relationship between perfectionism and the athletes' negative reactions to mistakes experienced during athletic competition using the MPS. A secondary focus of their study was to determine links between perfectionism and the athlete's thoughts 24 hours before a "major" competition. The athletes' thoughts following errors were identified and then related to their scores on the dimensions CM and DA scales on the MPS. The results indicated high associations between athletes who rated high on CM and competitive anxiety (r = .47), between CM and failure orientation (r = .70), while a negative relationship was found between CM and trait sport confidence (r = -.61). Further, high scorers on the CM dimension were more likely to react negatively to mistakes during competition and to experience more negative thoughts prior to competition than low CM scorers. The researchers concluded that perfectionistic athletes feel threatened, resulting in increased state anxiety, because they perceive evaluative situations as opportunities for failure. In two other sports studies using the MPS to measure perfectionism, Gould, Udry, Tuffey, and Loehr (1996) found that selected dimensions of perfectionism were associated with burnout among young (junior) competitive tennis players. Hall, Kerr and Matthews (1998) found that perfectionism was a significant predictor of state cognitive anxiety among competitive runners.

The study of perfectionism, in both positive and negative forms, has strong implications for sports competitors. In a favorable direction, perfectionist athletes (and coaches, for that matter) tend to have very high, yet realistic, self-expectations, set challenging, yet attainable, goals, and make internal causal attributions (Anshel, 2003). The tendency to set high standards infers a strong association between high quality performance and making causal attributions of high effort. Failure outcomes are interpreted as a need to improve future performance rather than as a manifestation of low ability.

On the negative side, however, unhealthy perfectionists will likely experience more frequent and highly intense stress due to harsh self-criticism and difficulty in feeling gratification from performance improvement and meeting realistic expectations (Hall et al., 1998). As Hall et al. have concluded, since perfectionists set excessively high standards, "they do not allow themselves the flexibility to make mistakes while performing. The adoption of an overly self-critical perspective on performance means that they are rarely satisfied with their level of achievement" (pp. 196-197). The implication in sport is that highly perfectionistic competitive athletes are very sensitive to feedback from, and need of the approval of, significant others (e.g., coaches, parents), and are concerned about their coach's or parent's responses to their performance (Gould et al., 1996). The result is heightened stress and state anxiety, and greater likelihood of distraction from the task at hand.

From a clinical perspective, Zinsser, et al. (1998) contend that perfectionistic athletes are so afraid of failure and of making mistakes that their enjoyment of sport is greatly reduced. According to the authors, "athletes who believe they should be perfect will blame themselves for every defeat, every setback. Their self-concept will likely suffer and they may start a fear-of-failure syndrome" (p. 286). In addition, Frost and Henderson (1991) assert that perfectionistic athletes focus their attention on mistakes that they commit during competition, and have difficulty forgetting about them. Their distraction from task-relevant thoughts often results in poor performance.

An additional conceptual framework that captures a psychological characteristic of perfectionism and is entrenched in the sport psychology area is attribution theory. The research question, with respect to the present study, is the extent to which perfectionism influences a performer's causal attributions to perceived failure or success, and the effects of these attributions on subsequent motor performance. According to Thompson, Davis, and Davidson (1998), "past failures rather than past successes form the basis for predicting future achievement outcomes" (p. 383), in which failure is operationally defined as "doing worse than expected" (p. 383). Self-blame following perceived failure, then, is a common attribute of perfectionists. Perfectionists externalize their success, while perceived failure is internalized. Extreme perfectionists who, in fact, are high achieving individuals, suffer from a condition called "the imposter phenomenon," in which the individual harbors feelings of resilient doubts of their own abilities, which they believe are overestimated by others. These individuals fear that others will "discover" that they are not truly intelligent or talented, but are in fact, "imposters" (Cozzarelli & Major, 1990; Ferguson & Rodway, 1994).

The likely causes of this bias toward self-blame following perceived failure, coupled with the difficulty of self-congratulation following success, include the combination of perfectionists to set unrealistically high standards and expectations, and their intolerance to meet these standards. The results are reduced state self-esteem, low expectations of success toward future performance, diminished self-estimates of ability (i.e., talent) or skill, and negative affect (Antony & Swinson, 1998; Thompson et al., 1998). These outcomes have significant implications in competitive sport settings for motivation, confidence, concentration, and persistence on task, particularly after performance errors. The extent to which perfectionism influences causal attributions following perceived failure after performing a motor task has apparently heretofore been ignored in the extant sport psychology research literature.

Competitive sport situations are replete with inevitable opportunities to experience negative feedback and perceived performance failure. Of primary concern in this study is the extent to which heightened perfectionism has a deleterious effect on the athlete's emotions (affect) and subsequent performance after experiencing critical verbal feedback on performance outcome. The purpose of this study, then, was to examine the extent to which perfectionism predicted or was associated with motor performance, explaining the causes of performance outcome and positive and negative affect. Given the apparent predisposition of perfectionists to set excessively high standards, be anxious in achievement situations, and to be overly self-critical, it would seem plausible to hypothesize that high perfectionists would suffer from greater negative affect, engage in more self-blame (i.e., making internal causal attributions after perceived failure), and experience a deterioration in motor performance after receiving critical verbal feedback on performance outcome as opposed to their reactions to no feedback.

Method

Participants

The study consisted of 30 male undergraduate university students who attended a university in the southwest U.S., ranging in age from 19.6 to 22.8 yrs. Participants were former competitive athletes for their high school team, and engaged in the study for course credit. The university's Institutional Review Board approved the study, and participants, each of whom signed a consent form, were informed that they could withdraw from the study at any time without negative ramifications.

Materials

The Multiple Perfectionism Scale (MPS; Frost et al., 1990) is a 35-item inventory that was used in this study to ascertain the athletes' perfectionistic thinking for each of six dimensions (described earlier). Internal consistencies (alphas) for the subscales ranged from .77 to .93. In their review of previous studies using the MPS, Enns and Cox (2003) concluded, "Collectively the studies reviewed ... provide compelling evidence of the construct, concurrent, and discriminant validity of the Frost MPS" (p. 42). Adapted versions of this inventory, adapted for sport, have been used to examine relationships between perfectionism and goal orientation (Dunn, Dunn, & Syrotuik, 2002) and in an attempt to identify components of perfectionism among competitive athletes (Anshel & Eom, 2003). In both studies, high reliability and construct and predictive validity were established.

In summary, the MPS was administered to the participants to assess the extent to which they engaged in perfectionistic thinking, with each item on the Likert-type scale ranging from 1 (strongly disagree) to 5 (strongly agree). The MPS has received extensive psychometric scrutiny, with reliability and validity of the instrument established in several studies (e.g., Frost et al., 1990, 1993; Frost & Marten, 1990), and used in at least two published studies in the sport psychology research literature (Frost & Henderson, 1991; Hall et al., 1998). A stabilometer (Lafayette Company, Model no. 16020) was used to measure total body time on balance (msec). The athletes' positive and negative affect was obtained with an adult version of the Children's Arousal Scale (CAS-A), an inventory that measures positive (4 items, such as "enthusiastic," "happy," "relaxed") and negative (7 items, such as "worried," "nervous," "sad," and "upset") affect, based on a 7-point semantic differential scale ranging from 1 (very much) to 7 (not at all).

The CAS-A was used in this study as opposed to other measures of affect because it can be completed quickly, within 15 sec., thus, capturing the individual's emotional status more accurately than other inventories that consume far more time. This was an important consideration when identifying a person's negative affect following a stressful episode. In addition, the CAS-A reflects changes in positive affect, often missing in sport-related stress research, in addition to measuring negative affect.

The internal consistency of the total CAS scale in Anshel's (1985) study was alpha = 0.79. Alphas for the positive and negative dimensions of the scale were .81 and .77, respectively. In a more recent study, Anshel and Martin (1996) used the CAS-A to determine negative affect as a function of negative self-monitoring strategies while performing a motor task (alphas = .77, .75 and .79 for total, positive, and negative dimensions, respectively). The scale's construct validity was supported in both of these studies based on significant changes in negative affect as a function of the respective treatments. In the present study, pre-treatment alphas were .69 for positive affect and .89 for negative affect. Post-treatment alphas were .29 for positive affect and .82 for negative affect. Thus, the measure of affect was reliable with the exception of post-treatment positive affect (discussed later).

Procedure

Participants were administered the MPS one week prior to the study. Motor performance testing was conducted individually. Participants engaged in two sessions, each consisting of performing 20 trials, with each trial including massed practice intervals of 20 sec. work and 5 sec. rest. As indicated earlier, the criterion task consisted of a total-body-balance task on the stabilometer, with performance measured as time (.01 sec) on balance.

The experiment consisted of a within-subjects randomized design in which participants were assigned to each of two conditions, receiving negative (bogus) verbal feedback (experimental) or no verbal feedback (control), experienced in counterbalanced order one week apart. Thus, for example, participants who received negative feedback on the first experimental session received no feedback in the second session. One week was viewed as necessary to prevent a learning effect on task performance for 20 trials.

Verbal information feedback was standardized and specific. In the negative feedback condition, after establishing eye contact, the experimenter informed the participant, "you are failing to reach your previous best," while in the control condition, performance was conducted in the absence of feedback. Under both conditions, actual time-on-balance scores were not visually observable to the performers. This procedure, consistent with previous studies related to feedback effects on motor performance (e.g., Anshel, 1979, 1988; Anshel & Hoosima, 1989), ensured that each treatment was carried out consistently. One weakness of the Anshel and Hoosima study was absence of a no feedback (control) condition, a flaw that was overcome in the present research design.

There were several reasons this study did included a negative, but not a positive, feedback condition. First, the research question concerned the extent to which perfectionism influences motor performance, among other variables, after receiving unpleasant, critical information feedback. High perfectionists feel threatened, resulting in increased state anxiety, because they perceive evaluative situations as opportunities for failure (Hall, Kerr, & Matthews, 1998). According to Hall et al., since perfectionists set excessively high standards, "they do not allow themselves the flexibility to make mistakes while performing. The adoption of an overly self-critical perspective on performance means that they are rarely satisfied with their level of achievement" (pp. 196-197). The research problem addressed in this study was whether negative information feedback, in contrast to no feedback, would exacerbate this effect as a function of perfectionistic thinking.

The second factor that lent credence to the use of only negative and control conditions was the mediating influence of experiencing a third set of 20 trials (i.e., a positive feedback condition) on task mastery, even with one-week intervals. Practice enhances organization of movement components, thereby increasing automatization of motor commands (Wrisberg, 1993, p. 68). Novices, however, are more susceptible to environmental factors (e.g., presence of observers, critical feedback). As Wrisberg concluded from his extensive review of this literature, learning a motor skill is accompanied by "speeding up" certain aspects/stages of information processing, and the ability to selectively interpret or filter out relevant or irrelevant information, respectively. Advanced performers selectively attend to relevant information, while remaining impervious to input perceived as intrusive or irrelevant. Thus, it is plausible to surmise that experiencing task mastery by experiencing all three conditions--a total of 60 trials--would likely have masked the effects of critical feedback on motor performance as a function of perfectionistic thinking.

Finally, while error information is more effective for encouraging skill improvement, unknown is the extent to which selected dispositions, in this study, perfectionism, enhance or impede this process. Positive feedback, on the other hand, has a motivational role early in the skill acquisition process (Magill, 2001).

Participants performed a series of 20-sec. trials in attempting to maintain their balance on the stabilometer over 20 trials. They were aware that a clock measured their time-on-balance. To establish a performance baseline, against which to compare and evaluate subsequent performance, two "practice" trials were completed. Then, only the performers in the experimental condition were informed that their scores on the following (first) set of five trials would be compared to their best score from the practice trials. This formed the basis on which failure feedback could be imposed in the experimental condition. After completing the first set of five trials, participants received "failure" information feedback (experimental) or no feedback (control) regarding their average time on balance. After performing the second set of trials, participants were again given the same form of feedback (negative or none) on their performance. To enhance the credibility of critical feedback, however, the criterion for meeting their previous best performance was the previous set of five trials throughout the study. For example, rather than compare performance against the initial practice trials and to recognize naturally improved performance due to a practice effect, participants were told that performance for the second set of five trials (Ts 6-10) would be compared to the previous best performance from the first set of five trials (Ts 1-5). Similarly, performers were informed that their performance on the third set of trails (Ts 11-15) would be compared to their best score on the second set of trials (Ts 6-10), and so on.

After the each set of five trials, participants were again given negative feedback or no feedback. After the third set of trials (Ts 11-15), the CAS-A was completed as an immediate measure of affect, which served as a manipulation check of the experimental (negative feedback) condition. Either no feedback or negative feedback was given again after the fourth and fifth sets of trials. Participants were given their last performance feedback and completed the attribution scale to ascertain their perceived cause of performance outcome immediately after their last performance trial. At this time, they were debriefed about the purpose of the experiment, informed that the feedback they were given was false, and that they had, in fact, performed quite well. They were also reminded of the importance of the study, and they agreed not to disclose any information about the experiment to others. They were reminded that they would receive one credit point toward their final grade. Finally, after being thanked for their participation, the experimenter verbally ascertained their feelings about the treatment they had received and assured them that he (the experimenter) was pleased with their performance. This was to ensure that they were in a positive state of mind and harbored no ill feelings about the treatment they had received. No individual indicated ill feelings about the study.

Results

The perfectionism data consist of measurements on the same participant performing under each of two conditions, no feedback (control) and critical feedback (experimental). Thus, the respective scores are correlated. The analysis of data involving repeated measures depends on the structure of the correlation matrix. In practice, several plausible correlation structures are considered. One or more information criteria are used to determine which model best fits the data. Three of the most widely used criteria are the Akaike Information Criterion (AIC), Schwarz Bayesian Criterion (SBC), and the Loglikelihood Criterion (see Verbeke & Molenberghe, 2000). In general, a better model has a higher value of the information criterion. AIC = (maximized log likelihood)-(effective number of covariance parameters), is the most frequently used criterion. It is used for comparing models with the same fixed effects and different covariance structures. A better model has a higher value of this information criterion. In analyzing the present data, we fit linear mixed models to the data using PROC MIXED of the SAS System (version 8).

Motor Performance

For each participant, the performance scores consisted of the mean of four trial blocks for two conditions, receiving critical feedback (experimental) and no feedback (control). A first-order autoregressive (AR1) model, in which it is assumed that the correlation structure of the performance scores under the two conditions may differ, was superior to a compound symmetric model that assumes that all trials are equally correlated. The correlation matrix for each condition is presented in Table 1.

Under the negative feedback condition the correlation between trials was much weaker than that under the no feedback condition. In fact, the correlation between trials more than one lag apart was negligible for performance with negative feedback. For instance, while the correlation under adjacent trials under the control condition was r = 0.79, it was r = 0.42 under the negative feedback condition. The correlation between trials two lags apart was rs = 0.62 and 0.18 under the control and negative feedback conditions, respectively. Finally, the correlation between the first and fourth trials was rs = 0.49 and 0.07 under the control and negative feedback conditions, respectively. The pattern of these varying autocorrelations indicated differences in motor skill performance under the negative feedback and control conditions.

Using the AR (1) correlation structure, a linear model of performance on perfectionism scores adjusted for conditions, trial blocks, and practice performance was fitted to determine the relationship between performance and perfectionism scores. The best model was with non-parallel equations across conditions, as indicated by the model having statistically significant interactions between the corresponding perfectionism score and condition. Table 2 contains the partial F-tests for the model. A factor was significant if p value was < = 0.05, marginally significant for 0.05 < p < = 0.10, and not significant if p > 0.10.

To analyze the information contained in Table 2, adjusted for other predictors, performance did not differ significantly under the two conditions. However, because of significant interactions between conditions and trials, as well as between conditions and perfectionism scores, it was concluded that performance in different trials was affected by the experimental condition of receiving critical information feedback. This is evident from the default SAS PROC MIXED output located in Table 3. Under the control condition (condition 1), performance in trial blocks 1, 2, and 3 were significantly different from performance in trial block 4 (ps = 0.01, 0.01, 0.004, respectively), while under the experimental condition, the performance in trial block 2 was significantly different from the performance in trial block 4 (p < 0.04). As is evident from Table 3, these estimates are negative, indicating an increase in performance scores (see entries corresponding to Trial Block[cond]). As seen in Table 2, perfectionism scores were important predictors of performance under the two conditions, while the dimension, concerns over mistakes, was marginally significant (p = .059), and organization was not significant (p >. 10).

Since all of perfectionism scores in the model are in the form of interactions, it was important to determine how each score influenced performance under each of the two conditions. These results are summarized in Table 3. As shown in Table 3, partial t-tests clearly indicate that perfectionism scores were significant under the experimental, but not the control condition for each of the six measures of perfectionism (all ps < .01) except organization (p > .05). In addition, under the experimental condition, coefficients of the perfectionism scores under the experimental condition were negative. This indicated a deterioration of performance under the experimental condition (critical feedback) for those participants with higher perfectionism scores. Table 4 contains descriptive statistics of performance scores.

Affect

The positive and negative affect scores for each performer were obtained under the control and experimental conditions. Pre-treatment scores served as a covariate, while post-treatment scores formed the participants' responses. The full model containing the perfectionism scores, adjusted for condition and affect types showed no significant relationship with perfectionism. The most effective model included condition, affect type, and a condition x affect type interaction. The results for this model (model 1) are summarized in Table 5.

Organization is the most important perfectionism characteristic that may influence affect. However, inclusion of this factor in the model in addition to those explanatory variables already present in model 1 was not statistically significant. For models 1 and 2, AICs were -157.8 and -157.9, respectively. This indicated a lack of statistical significance in adding Organization to the model. The partial F-tests (model 2) are summarized in Table 6. Affect, then, did not appear to be an important component in explaining links between perfectionism, conditions, and performance quality. Table 7 includes the descriptive statistics for post-treatment affect.

The statistical data was supported by a manipulation check, conducted privately and individually in a personal interview immediately after the participants' involvement in the study. The purpose of the interview was to determine both the credibility and the source of the negative feedback they received in producing acute stress. Two questions, each requiring a response on a scale from 1 (not at all) to 10 (always), were: "To what extent did you believe the content of the feedback you received?," and "To what extent did the person who provided the feedback seem believable?" The participants' responses indicated that the feedback content was highly credible (M = 8.89, SD = 0.46), and that the source of feedback was believable (M = 8.72, SD = 0.61). However, it should be noted that alphas (rs) for positive affect following the treatment was low for both groups. This suggested that individual differences exist in the manner in which positive affect is experienced and may partially explain the lack of congruity of self-reporting positive affect, at least in the laboratory experimental setting of this study.

Tenenbaum, Lloyd, Pretty and Hanin (2002) contend that negative affect may be reported more consistently than positive affect because the former more accurately reflects a "general schema," or overlearned set of emotions associated with performance under stressful conditions. In the present study, participants received either negative (critical) or no information during performance trails; no positive feedback was offered. Under these conditions, general schema development may have been exacerbated when performing a relatively complex task (total body balancing) in the presence of an evaluative audience. The general schema for negative affect may have been more substantial than for positive affect, resulting in more reliable self-report for negative than positive affect. The result is that reporting certain emotions may be stronger, more accurate, and more consistent for negative than positive emotions, or for some positive emotions (e.g., happy, excited, eager) than others (e.g., relaxation) in response to the same event.

Causal Attribution

Four types of attribution classification were considered under both conditions, ability, difficulty, effort, and luck, in terms of the ways in which participants explained their performance. While performers received no feedback in the control condition, their causal attributions could be compared to the experimental (negative feedback) condition. The extent to which participants in the control and experimental groups differed on their causal attributions was determined by examining interactions among the variables, perfectionism scores (i.e., personal standard, parental expectation, and criticism), condition, and attribution classification (ability, effort, task difficulty, and luck). The results are summarized in Table 8.

Differences between attribution scores in the control and experimental conditions (p = 0.006), and the condition x attribution classification interaction (p = 0.03), were highly significant. This indicates that causal attributions differed markedly between conditions. Perfectionism scores that weakly contributed to causal attributions were personal standards and parental expectation, ps = 0.05 and 0.06, respectively. However, attribution was significantly influenced by criticism, and this influence differed under the two conditions based on the highly significant condition by criticism interaction (p = 0.007).

Further breakdown of this particular relationship revealed that the participants' attributions were not influenced under the control condition (p = 0.72), while attributions were significantly influenced by the experimental condition in which they received critical feedback (p = 0.01). Further analyses indicate that differences between conditions in causal attributions were statistically significant for the internal causal attributions of ability and effort under the no-feedback condition. No significant differences between conditions existed, however, for the external attributions of task difficulty and luck (see Table 9). A comparison of means indicated that the no-feedback condition elicited stronger internal causal explanations (i.e., ability and effort), as compared to the negative feedback condition (see Table 10 for descriptive statistics).

Discussion

The primary focus of the present study was to determine the extent to which the dimensions of perfectionism contributed to the motor performance, affect, and causal attributions in response to critical feedback. It was predicted that individuals who scored high on any of the perfectionism dimensions would perform poorer on a total body balancing task, exhibit more negative affect, and be more internal in their causal explanations of perceived failure than their less perfectionistic peers. The results of this experiment partially supported these hypotheses. The results on the three measures of performance, affect, and attributions will be reported jointly in order to provide a more coherent explanation for the outcomes.

The findings indicated that the balancing performance of individuals with higher measures of personal standards, concerns over mistakes, doubts about actions, parental expectations, total perfectionism, and to a lesser degree, criticism, significantly deteriorated under negative feedback, the experimental condition. The significant perfectionism-by-conditions interactions clearly indicated that performance differed significantly under the experimental, but not under the control, condition for each of the six measures of perfectionism except organization. In addition, under the experimental condition, the negative coefficients of perfectionism scores indicated poorer performance by high perfectionists. When performance quality was assessed according to trial blocks, it was found that critical feedback elicited marked performance differences between trial blocks two and four, whereas in the control (no feedback) condition, differences existed between each of the first three trial blocks and trial block four. An examination of means for each trial block indicated that significant performance increments occurred sooner under the control (no feedback) condition as compared to the experimental (negative feedback) condition.

With respect to affect, a meaningful statistical relationship between affect and perfectionism was not established. Participants did not appear to experience any marked changes in negative affect as a function of their perfectionism scores nor in response to the experimental and control conditions. The manipulation check, however, indicated that the feedback content and its source were credible. Still, the extent to which perceptions of credibility actually influenced affect is less certain. Perhaps the contrived conditions under which feedback was administered, the fact that the source of feedback was known to the performers, and the meaningfulness of the criterion task--whole body balancing--may have jointly influenced the performer's affect.

In two previous studies employing contrived critical negative feedback, Anshel (1979, 1988) found that task meaningfulness and the perceived authenticity (credibility) of the feedback strongly influenced the extent to which feedback influenced motor performance. While the experimental treatment did not appear to heighten cognitive anxiety in the current study, it is plausible to surmise that performers experienced heightened somatic anxiety during task performance. Given the nature of a total body-balancing task, somatic anxiety, not measured in this study, may have markedly affected performance quality. This explanation is supported by the results of at least three previous perfectionism studies (Flett et al., 1994; Hall et al., 1998; Magnusson, Nias, & White, 1996).

In a Canadian non-sport study, Flett et al. showed that heightened perfectionism was associated with self-reported increased levels of autonomic-emotional components of state anxiety for both males and females. In the context of competitive sport, Hall et al. (1998) found that concern over mistakes predicted cognitive anxiety, doubts about action predicted somatic anxiety, and personal standards predicted confidence among high school athletes in the UK. The researchers surmised that "because perfectionists tolerate little discrepancy between their achievement standards and performance, they are considered to experience an almost continuous state of self-focused attention until they perceive that acceptable standards of performance have been achieved" (p. 212). In addition, according to Frost and Henderson (1991), perfectionists often experience state anxiety and emotional exhaustion prior to performing in achievement settings. The researchers contend that the poor mental state of perfectionists is further exacerbated by their tendency to worry in the presence of an audience, including an experimenter who provided critical feedback in the present study.

Not surprisingly, then, perfectionists, generally react poorly to mistakes. In a study of female nurses, Magnusson et al. found that heightened perfectionism, particularly the dimension doubts about actions, was significantly associated with mental and physical (trait and state) fatigue. Thus, perhaps any deleterious effect of the critical feedback condition on performance was partly due to either cognitive (e.g., distraction, heightened self-consciousness, reduced concentration) or somatic (e.g., increased muscle tension) forms of anxiety.

Another likely explanation for the influence of critical feedback on performance among high perfectionists is "the sense of doubt perfectionistic people have about the quality of their performance. Perfectionists believe their efforts are never quite good enough to warrant feelings of satisfaction" (Frost & Marten, 1990, p. 562). To these researchers, perfectionists are motivated by a fear of failure, and "new tasks are viewed as opportunities for failure rather than accomplishment" (p. 561). Perhaps the self-critical nature of high perfectionists in the present study created heightened self-evaluation and tension, thereby inhibiting performance quality.

Finally, the performers' causal attributions was reflected by their perfectionistic tendencies and the experimental condition in which negative feedback was administered. The relationship between causal attribution and perfectionism scores was meaningful, in which the performers' attributions were influenced as a function of condition. In particular, following negative feedback, performers were markedly less likely to take responsibility for their performance by making external attributions. It would appear that perfectionists are less willing to take personal responsibility for their performance following failure feedback as opposed to no feedback in the control condition. A perusal of the perfectionism data suggests that the perfectionism dimension of "concern about criticism" may be the most important contributor to a performer's causal attribution as opposed to the other perfectionism dimensions. This is understandable given the propensity of high perfectionists to possess high goal orientation (Dunn et al., 2002).

One possible explanation for this finding is the tendency of perfectionists, particularly if they perceive the criterion task as difficult, to engage in the process of self-handicapping (SH). SH is the process by which a person "may purposefully procure impediments to performance if they feel uncertain of their ability to succeed" (Hobden & Pliner, 1995). In the present experiment, participants were engaging in a task that would likely be perceived as complex, or difficult. This perception would have been magnified under the experimental condition, in which critical feedback was offered regularly. High perfectionists, then, might have taken on greater responsibility to manage impressions of the experimenter, who was known to the participants.

Hobden and Pliner found that high perfectionists are more likely to have heightened fear of failure and, in accordance with their prediction, were more likely to self-handicap. According to the results of a study by Saboonchi and Lundh (1997), this may be due to a greater tendency of perfectionists to be more self-conscious and anxious, especially under duress. Concern over mistakes was the dimension of perfectionism that most reliably predicted self-consciousness. Ferrari (1992) found that perfectionism was significantly related to protective self-presentations and SH. He surmised that "in self-handicapping, a person protects self-esteem by distancing performance from association with failure, particularly if the performance would be known to others" (p. 77).

Further research is clearly needed to identify high perfectionists among competitive athletes to determine the manner in which the undesirable qualities of perfectionism interfere with cognitive and somatic processes that occur prior to and during sport performance, to compare the effects of negative and positive forms of information feedback on motor performance as a function of perfectionistic thinking, and to examine the effectiveness of interventions that might attenuate the deleterious effects of "negative" (unhealthy) forms of perfectionism in competitive sport.

Table 1.
Correlations between selected stress trials
for each condition on motor performance scores.

 No Feedback Negative Feedback
Trial (Control) (Experimental)
Blocks
 2 3 4 2 3 4

1 0.79 0.62 0.49 0.42 0.18 0.07
2 0.79 0.62 0.42 0.17
3 0.79 0.42

Table 2.
Partial F-tests for performance scores

Effect NDF DDF F p

Condition 1 29 0.12 0.73
Trial(condition) 6 174 2.64 0.02
Practice x Condition 2 187 81.77 <.0001
PS x Condition 2 187 3.91 0.02
CM x Condition 2 187 3.38 0.04
D x Condition 2 187 4.29 0.02
PE x Condition 2 187 4.63 0.01
PC x Condition 2 187 2.86 0.06
O x Condition 2 187 0.93 0.39
TP x Condition 2 187 4.03 0.02

NDF = Numerator Degrees of Freedom
DDF = Denominator Degrees of Freedom
PS = Personal standards; CM = Concern about mistakes;
D = Doubts about actions; PE = Parental expectations;
PC = Criticism; 0 = Organization; TP =Total perfectionism.

Table 3.
Partial t-tests for model parameters for motor performance

 Standard
Effect condition trial Estimate Error

Intercept 3.11 1.17
Control 1 -0.49 2.35
Experimental 2 0.00 .
Trial Block(cond) 1 1 -0.97 0.35
Trial Block(cond) 1 2 -0.79 0.30
Trial Block(cond) 1 3 -0.64 0.23
Trial Block(cond) 1 4 0.00 .
Trial Block(cond) 2 1 -0.36 0.34
Trial Block(cond) 2 2 -0.66 0.32
Trial Block(cond) 2 3 -0.14 0.27
Trial Block(cond) 2 4 .000 .
Practice x Cond 1 1.13 0.15
Practice x Cond 2 0.85 0.08
PS x Cond 1 0.38 3.47
PS x Cond 2 -5.34 1.91
CM x Cond 1 0.35 4.79
CM x Cond 2 -6.87 2.65
DA x Cond 1 0.95 1.93
D x Cond 2 -3.09 1.07
PE x Cond 1 0.88 2.64
PE x Cond 2 4.36 1.44
Crit x Cond 1 -0.05 2.02
Crit x Cond 2 -2.64 1.10
Org x Cond 1 -0.21 0.48
Org x Cond 2 -0.35 0.27
TP x Cond 1 -2.81 14.63
TP x Cond 2 22.72 8.02

Effect DF t p

Intercept 29 2.65 0.01
Control 29 -0.21 0.86
Experimental . . .
Trial Block(cond) 174 -2.79 0.01
Trial Block(cond) 174 -2.64 0.01
Trial Block(cond) 174 -2.88 0.004
Trial Block(cond) . . .
Trial Block(cond) 174 -1.08 0.28
Trial Block(cond) 174 -2.09 0.04
Trial Block(cond) 174 -0.52 0.61
Trial Block(cond) . . .
Practice x Cond 187 7.62 <0.0001
Practice x Cond 187 10.27 <0.0001
PS x Cond 187 0.11 0.91
PS x Cond 187 -2.79 0.006
CM x Cond 187 0.07 0.94
CM x Cond 187 -2.60 0.01
DA x Cond 187 0.49 0.62
D x Cond 187 -2.89 0.004
PE x Cond 187 0.33 0.74
PE x Cond 187 -3.02 0.003
Crit x Cond 187 -0.03 0.97
Crit x Cond 187 -2.39 0.02
Org x Cond 187 -0.44 0.66
Org x Cond 187 -1.29 0.19
TP x Cond 187 -0.19 0.85
TP x Cond 187 2.83 0.005

Conditions: 1 =Control; 2=Experimental;
PS = Personal Standards; CM = Concern About Mistakes;
D = Doubts About Actions; PE = Parental Expectations;
Crit = Criticism; Org = Organization; TP = Total
Perfectionism.

Table 4.
Descriptive statistics on motor performance

 Condition
Trial
Block Control Experimental Combined

1 Mean 9.07 9.18 9.13
 Std.Dev. 2.95 2.34 2.64
2 Mean 9.25 8.89 9.07
 Std.Dev. 2.81 2.26 2.53
3 Mean 9.40 9.41 9.40
 Std.Dev. 2.97 2.26 2.62
4 Mean 10.04 9.55 9.79
 Std.Dev. 3.29 2.26 2.81
Combined Mean 9.44 9.26 9.35
 Std.Dev. 2.99 2.27 2.65

Table 5.
Partial F-Test for Affect Scores

Model 1

Effect NDF DDF F p

Condition 1 29 9.82 0.004
Pre-affect 1 86 65.84 <.0001
Affect-type 1 29 4.08 0.05
Condition x Affect 1 29 4.60 0.04

Table 6.
Partial F-test for affect scores

Model 2

Effect NDF DDF F p

Condition 1 29 3.51 0.07
Pre-affect 1 84 67.14 <0.0001
Affect 1 29 4.09 0.05
Condition x Affect 1 29 4.61 0.04
Organization x Condition 2 84 2.19 0.11

Table 7.
Descriptive statistics of post-treatment affect

 Conditions

Affect Control Experimental Combined

Negative Mean 5.94 3.68 4.81
 Std.Dev. 1.07 0.98 1.53

Positive Mean 5.18 3.80 4.49
 Std.Dev. 1.31 0.91 1.32

Combined Mean 5.56 3.74 4.65
 Std.Dev. 1.25 0.94 1.43

Table 8.
Partial F-tests for attribution scores

Effect NDF DDF F P

Condition 1 29 14.66 0.0006
Attribution-class 3 87 60.07 <.0001
Condition x Attribution 3 87 2.91 0.040
Personal Standards 1 26 4.18 0.050
Parental Expectations 1 26 3.64 0.070
Condition x Criticism 2 201 2.35 0.007

Table 9.
Comparing Attribution Scores Between Groups

 Std.
Attribution Estimate Error df t p

Ability 1.1373 0.4027 87 3.82 0.006
Effort 1.8373 0.4027 87 4.56 <.001
Difficulty 0.9707 0.4027 87 2.41 0.12
Luck 0.9040 0.4027 87 2.24 0.33

Table 10.
Descriptive statistics of causal attributions

 Conditions

 Control Experimental
Attribution

Internal
 Ability Mean 3.97 3.07
 Std.Dev. 1.10 1.01
 Effort Mean 4.83 3.03
 Std.Dev. 1.07 1.40

External
 Task
 Difficulty Mean 3.73 3.60
 Std.Dev. 1.05 1.22
 Luck Mean 1.67 1.60
 Std.Dev. 0.76 0.89

References

Alden, L.E., Bieling, P.J., & Wallace, S.T. (1994). Perfectionism in an interpersonal context: A self-regulation analysis of dysphoria and social anxiety. Cognitive Therapy and Research, 18, 297-316.

Anshel, M.H. (2003). Sport psychology: From theory to practice (4th ed.). San Francisco, CA: Benjamin-Cummings.

Anshel, M.H. (1988). Effect of mood and pleasant versus unpleasant information feedback on performing a motor skill. Journal of General Psychology, 115, 117-130.

Anshel, M.H. (1985). Effect of arousal on warm-up decrement. Research Quarterly for Exercise and Sport, 56, 1-9.

Anshel, M.H. (1979). Effect of age, sex, and type of feedback on motor performance and locus of control. Research Quarterly, 50, 305-317.

Anshel, M.H., & Eom, H.J. (2003). Identifying dimensions of perfectionism in sport. International Journal of Sport Psychology, 34, 255-266.

Anshel, M.H., & Hoosima, D.E. (1989). The effect of positive and negative feedback on causal attributions and motor performance as a function of gender and athletic participation. Journal of Sport Behavior, 12, 119-130.

Anshel, M.H., & Martin, M.B. (1996). Effect of divergent self-monitoring strategies on motor performance and affect as a function of task complexity. Journal of Sports Sciences, 14, 233-242.

Antony, M.M., & Swinson, R.P. (1998). When perfect isn't good enough: Strategies for coping with perfectionism. Oakland, CA: New Harbinger Publications.

Arbuckle, J. L. (1997). Amos users 'guide, Version 3.6. Chicago, Il: SPSS Inc.

Burns, D. (1980). The perfectionist's script for self-defeat. Psychology Today, pp. 34-51.

Cozzarelli, C., & Major, B. (1990). Exploring the validity of the impostor phenomenon. Journal of Social and Clinical Psychology, 9, 401-417.

Dunn, J.G.H., Dunn, J.C., & Syrotuik, D.G. (2002). Relationship between multidimensional perfectionism and goal orientations in sport. Journal of Sport & Exercise Psychology, 24, 376-395.

Enns, M.W., Cox, B.J. (2003). The nature and assessment of perfectionism: A critical analysis. In G.L. Flett & P.L. Hewitt (Eds.), Perfectionism: Theory, research, and treatment (pp. 33-62). Washington, DC: American Psychological Association.

Ferguson, K., & Rodway, M.R. (1994). Cognitive behavioural treatment of perfectionism: Initial evaluation studies. Research on Social Work Practice, 4, 283-308.

Ferrari, J.R. (1992). Procrastinators and perfect behavior: An exploratory factor analysis of self-presentation, self-awareness, and self-handicapping components. Journal of Research in Personality, 26, 75-84.

Flett, G.L., Hewitt, P.L., Blankstein, K.R., & Koledin, S. (1991). Dimensions of perfectionism and irrational thinking. Journal of Rational-Emotive & Cognitive-Behavior Therapy, 9, 185-201.

Flett, G.L., Hewitt, P.L., Blankstein, K.R., & Mosher, S.W. (1991). Perfectionism, self-actualiza tion, and personal adjustment. Journal of Social Behavior and Personality, 6, 147-160.

Flett, G.L., Hewitt, P.L., & De Rosa, T. (1996). Dimensions of perfectionism, psychosocial adjustment, and social skills. Personality and Individual Differences, 20, 143-150.

Flea, G.L., Hewitt, P.L., Endler, N.S., & Tassone, C. (1994). Perfectionism and components of state and trait anxiety. Current Psychological Research and Reviews, 13, 326-350.

Flett, G.L., Hewitt, P.L., & Singer, A. (1995). Perfectionism and parental authority styles. Individual Psychology, 51, 50-60.

Frost, R.O., Heimberg, R.G., Holt, C.S., Mattia, J.I., & Neubauer, A.L. (1993). Acomparison of two measures of perfectionism. Personality and Individual Differences, 14, 119-126.

Frost, R.O., & Henderson, K.J. (1991). Perfectionism and reactions to athletic competition. Journal of Sport and Exercise Psychology, 13, 323-335.

Frost, R.O., Lahart, C., & Rosenblate, R. (1991). The development of perfectionism: A study of daughters and their parents. Cognitive Therapy and Research, 15, 469-489.

Frost, R.O., & Marten, P.A. (1990). Perfectionism and evaluative threat. Cognitive Therapy and Research, 14, 559-572.

Frost, R.O., Marten, P.A., Lahart, C., & Rosenblate, R. (1990). The dimensions of perfectionism. Cognitive Therapy and Research, 14, 449-468.

Gould, D., Udry, E., Tuffey, S., & Loehr, J. (1996). Burnout in competitive junior tennis players: I. A quantitative psychological assessment. The Sport Psychologist, 10, 322-340.

Hall, H.K., Kerr, A.W., & Matthews, J. (1998). Precompetitive anxiety in sport: The contribution of achievement goals and perfectionism. Journal of Sport and Exercise Psychology, 20, 194-217.

Hamachek, D.E. (1978). Psychodynamics of normal and neurotic perfectionism. Psychology: A Journal of Human Behavior, 15, 27-33.

Hardy, L., Jones, G., & Gould, D. (1996). Understanding psychological preparation for sport: Theory and practice of elite performers. New York: John Wiley.

Hewitt, P.L., & Flea, G.L. (1996). Personality traits and the coping process. In M. Zeidner & N.S. Endler (Eds.), Handbook of coping: Theory, research, applications (pp. 410-433). New York: John Wiley & Sons.

Hewitt, P.L., & Flett, G.L. (1991). Perfectionism in the self and social contexts: Conceptualization, assessment, and association with psychopathology. Journal of Personality and Social Psychology, 60, 456-470.

Hewitt, P.L., & Flett, G.L. (1990). Perfectionism and depression: A multidimensional analysis. Journal of Social Behavior and Personality, 5, 423-438.

Hewitt, P.L., Flett, G.L., & Endler, N.S. (1995). Perfectionism, coping, and depression symptomatology in a clinical sample. Clinical Psychology and Psychotherapy, 2, 47-58.

Hobden, K., & Pliner, P. (1995). Self-handicapping and dimensions of perfectionism: Self-presentation versus self-protection. Journal of Research in Personality, 29, 461-474.

Jackson, D.N. (1984). Personality research form manual (3rd Ed.). Port Huron, MI: Research Psychologist Press.

Juster, H.R., Heimberg, R.G., Frost, R.O., Holt, C.S., Mattia, J.I., & Faccenda, K. (1996). Social phobia and perfectionism. Personality and Individual Differences, 21, 403-410.

Magill, R.A. (2001). Motor learning: Concepts and applications. New York: McGraw-Hill.

Magnusson, A.E., Nias, D.K.B., & White, P.D. (1996). Is perfectionism associated with fatigue? Journal of Psychosomatic Research, 41, 377-383.

Saboonchi, F., & Lundh, L.G. (1997). Perfectionism, self-consciousness, and anxiety. Personality and Individual Differences, 22, 921-928.

Tabachnick, B.G., & Fiddell, L.S. (1989). Using multivariate statistics (2nd ed.). New York: Harper & Row.

Tenenbaum, G., Lloyd, M., Pretty, G., & Hanin, Y.L. (2002). Congruence of actual and retrospec tive reports of precompetition emotions in equestrians. Journal of Sport & Exercise Psychology, 24, 271-288.

Terry-Short, L.A., Owens, R.G., Slade, P.D., & Dewey, M.E. (1995). Positive and negative per fectionism. Personality and Individual Differences, 18, 663-668.

Thompson, T., Davis, H., & Davidson, J. (1998). Attributional and affective responses of imposters to academic success and failure outcomes. Personality and Individual Differences, 25, 381-396.

Verbeke, G., & Molenberghe, G. (2000). Linear models for longitudinal data. New York: John Wiley & Sons.

Wrisberg, C.A. (1993). Levels of performance skill. In R.N. Singer, M. Murphey, & L. K. Tennant (Eds.), Handbook of research on sport psychology (pp. 61-72). New York: Macmillan.

Zinsser, N., Bunker, L., & Williams, J.M. (1998). Cognitive techniques for building confidence and enhancing performance. Applied sport psychology: Personal growth to peak performance (pp. 270-295). Mountain View, CA: Mayfield.

Authors' Notes

The researchers extend their gratitude to Rogelio Puente for his assistance in collecting data for this study.

Mark H. Anshel

Middle Tennessee State University

and

Hossein Mansouri

Texas Tech University

Address Correspondence To: Dr. Mark Anshel, Department of HPERS, Box 96, Middle Tennessee State University, Murfreesboro, TN 37132 or [email protected]