آیا مقیاس خودناتوان سازی در حوزه دستاورد غیر علمی قابل اعتماد است؟

Abstract The present study investigated whether the short version of the Self-Handicapping Scale (SHS; Rhodewalt, 1990) is a reliable measure of self-handicapping outside of academic settings. Across 3 samples of athletes, analyses indicated that SHS items were not very meaningful to respondents and the subscales lacked internal consistency, particularly the effort subscale (α and composite reliability estimates ranged from 0.36 to 0.49). Confirmatory factor analyses revealed that the SHS had an unstable factor structure as the majority of fit indices were well below the 0.90 level of acceptability. These findings suggest that the SHS is not a reliable instrument for identifying trait self-handicappers in sport achievement settings. Rather, the SHS may be a domain-specific measure of academic self-handicapping.

Introduction Self-handicapping involves adopting or claiming possession of impediments that reduce the probability of success but provide a plausible excuse for failure (Leary, 1995). Individuals vary in their tendencies toward the use of self-handicapping strategies. Although some people may never self-handicap, others may self-handicap only occasionally and others may be chronic self-handicappers (e.g. Berglas, & Jones, 1978, Snyder, & Smith, 1982, Higgins, & Berglas, 1990 and Tice, 1991). An interest in identifying people who are prone to self-handicapping led Jones, and Rhodewalt (1982) to develop the Self-Handicapping Scale (SHS). In its original published form, the SHS consisted of 25 items that probed respondents' tendencies to use such self-handicaps as lack of effort, illness, procrastination or emotional upset in conjunction with evaluative performances. The scale also included items designed to assess concerns about achievement. In 1990, Rhodewalt re-analyzed the scale's factor structure and found that only 14 items loaded significantly (greater than 0.40) on one of two scale factors. He labelled these factors `excuse making' and `effort'. The excuse making subscale assesses the tendency to make excuses prior to evaluative performances (i.e. a claimed self-handicapping strategy) and the effort subscale taps an individual's willingness to withhold effort in achievement situations (i.e. a behavioral self-handicapping strategy). As a result of his findings, Rhodewalt (1990) recommended that researchers employ the revised 14-item SHS in lieu of the original 25-item instrument. Social psychologists have frequently used both the original and the revised SHS in studies of self-handicapping. These studies have demonstrated the utility of the SHS in predicting the use of various self-handicapping strategies such as claiming symptoms (Baumeister, Kahn, & Tice, 1990) and extenuating circumstances as self-handicaps (Strube, 1986, study 2), procrastinating (Ferrari, 1992) and failing to practice (Deppe, & Harackiewicz, 1996) or withholding practice effort on an evaluative task (Rhodewalt, & Fairfield, 1991). Researchers have also shown that the SHS is significantly correlated with constructs that are theoretically linked with self-handicapping such as public self-consciousness, social anxiety and self-esteem Strube, 1986 and Rhodewalt, 1990. Although it has been suggested that the SHS assesses trait self-handicapping across a variety of domains such as sport competition and interpersonal relationships (Rhodewalt, Saltzman, & Wittmer, 1984; Rhodewalt, 1990), the scale's validity and reliability have not been examined in these contexts. Rather, reports of the SHS's psychometric properties have been limited to the domain in which the SHS was originally developed-intellectual achievement among college undergraduates Strube, 1986 and Rhodewalt, 1990. To assess the reliability and factor structure of the SHS in non-academic achievement domains, the present study examined these properties of the 14-item SHS when administered to competitive athletes. Athletes were considered an ideal population in which to examine the psychometrics of the SHS for two reasons. First, self-handicapping is most likely to occur in publicly evaluated situations that are important to the individual's self-concept and where the individual is evaluated against very high standards or relative to another person's performance Berglas, & Jones, 1978 and Self, 1990. These conditions are inherent to sport competition where an athlete's self-concept can be strongly influenced by self- and public-evaluations of his or her physical abilities (cf. Vallerand, Pelletier, & Gagné, 1991). Hence, self-handicapping is predicted to occur in the sport domain (Rhodewalt et al., 1984). In their seminal paper, Jones, and Berglas (1978, p. 201) noted the strong potential for self-handicapping among athletes when they stated that “self-handicappers are legion in the sports world from the tennis player who externalizes a bad shot by adjusting his racket strings, to the avid golfer who systematically avoids taking lessons or even practicing on the driving range”. If self-handicapping is as common among athletes as Jones, and Berglas (1978) believed it to be, then the SHS should identify athletes with a tendency to use self-handicaps. Second, preliminary studies of self-handicapping in sport have identified two types of self-handicapping strategies used by athletes. Some athletes self-handicap by failing to increase their practice effort prior to an important competition (Rhodewalt et al., 1984) or by providing excuses for potential failure prior to an upcoming event (Carron, Prapavessis, & Grove, 1994; Hausenblas, & Carron, 1996). These two self-handicapping strategies-withholding effort and excuse making-are assessed by the two subscales of the 14-item SHS (Rhodewalt, 1990). Hence, if it is valid in sport achievement settings, the SHS should detect and measure athletes' proclivity to use these two types of self-handicaps. In summary, because athletes are believed to frequently use self-handicaps (Berglas, & Jones, 1978) by withholding practice effort (Rhodewalt et al., 1984) and making excuses Carron et al., 1994 and Hausenblas, & Carron, 1996, they were considered an ideal population in which to assess the reliability and factor validity of the 14-item SHS. If the SHS is reliable and valid in achievement domains other than those related to academics, then it should be psychometrically sound when administered to samples of competitive athletes. To address this issue, the psychometric properties of the SHS were assessed across three samples of athletes. In addition, confirmatory factor analyses were performed to examine the validity of the two-factor model proposed by Rhodewalt (1990).

. Results 3.1. Descriptive statistics The descriptive statistics for the effort and excuse making subscales of the SHS are presented for each data set in Table 1. The mean, median and mode for the subscales were very similar for each of the three samples suggesting that the distribution of scores was nearly symmetric. Table 1. Descriptive statistics for the excuse making and effort subscales of the SHS. There are nine items on the excuse making subscale and five items on the effort subscale. All items are scored on a six-point scale: 1=disagree very much, 6=agree very much. CRCMM=composite reliability estimates Sample Canadian athletes (n=245) Australian athletes (n=221) Sport-relevant instructions (n=136) excuse effort excuse effort excuse effort M for scale 26.75 15.73 24.19 17.27 25.56 15.07 S.D. 6.92 3.72 7.2 4.29 6.84 3.80 Median 26 15 25 17 25 15 Mode 25 15 26 18 21,26 13 Skew 0.29 −0.06 0.38 −0.26 0.18 −0.29 Kurtosis −0.16 −0.39 −0.40 −0.18 0.02 −0.29 Range 11–47 6–27 12–46 5–26 10–47 5–23 α 0.70 0.42 0.70 0.49 0.68 0.38 CRCMM 0.69 0.36 0.73 0.48 0.68 0.42 M per item 2.97 3.15 2.69 3.45 2.84 3.01 Table options The normality of the scoring distributions was assessed by conducting significance tests on the obtained values of skewness and kurtosis against the null hypotheses of zero (Tabachnick, & Fidell, 1989). Consistent with Tabachnick and Fidell's recommendations for evaluating the significance of skewness and kurtosis with small to moderate samples, a conservative alpha level of 0.001 was used for all significance tests. For the skewness values, z-scores ranged from −1.58 to 2.30 and were non-significant (all p's>0.01). For the kurtosis values, z-scores ranged from −1.26 to 0.05 and were nonsignificant (all p's>0.10). Given the absence of significant skewness or kurtosis, it appears that across the three samples, scores on the effort and excuse making subscales were normally distributed. Although the scoring distributions were normal, across all three samples the distribution of scores fell into a restricted range at the low end of the scale (i.e. scores that reflected a low level of personally reported self-handicapping tendencies among athletes). Closer examination revealed that on average, only 6.5% of participants had scores that fell into the top third of the range of possible excuse making subscale scores. Similarly, an average of just 7% of participants scored within the top third of the range of possible effort subscale scores. Thus, the vast majority of respondents scored within a restricted range that did not reflect the full range of possible SHS scores. If the SHS detects only a small proportion of individuals in that extreme, high self-handicapping category, then either there are few self-handicappers in sport, or the measure is not sufficiently sensitive or relevant to elicit responses that identify dispositional self-handicappers among athletes. Inspection of the mean responses per item provided further evidence that athletes tended to give responses to SHS items that fell within a truncated range. Across the three samples, the average item ratings for the excuse making and effort subscales were below the scale midpoint of 3.5 (i.e. between 2.69 and 3.45). In addition to demonstrating the restricted range of SHS responses, this observation also suggests that participants did not consider the content of the SHS items to reflect characteristics of them. 3.2. Reliability Cronbach's alpha coefficient was used to estimate the internal consistency of the excuse making and effort subscales (see Table 1). Across the three samples, alpha coefficients for the excuse making subscale ranged from 0.68 to 0.70, indicating an adequate level of internal consistency (Nunnally, 1978) but one that could be improved. The internal consistency of the effort subscale was poor, ranging from 0.23 to 0.49. In addition, the composite reliability for congeneric measures model (CRCMM; Raykov, 1997a) was used to estimate composite reliability for the effort and excuse making subscales (see Table 1). This approach was taken because recent research has demonstrated that Cronbach's alpha coefficient tends to underestimate test reliability except when estimating reliability for scales with items that load highly on a single underlying factor and that have uncorrelated errors Miller, 1995, Komaroff, 1997 and Raykov, 1997b. For more general, congeneric tests-tests such as the SHS-with items that measure the same latent dimension in possibly different units of measurement and with possibly different precision, the CRCMM is a more appropriate method for estimating composite reliability. Composite reliability estimates for the excuse making subscale were low but considered acceptable, and ranged from 0.68 to 0.73. Estimates for the effort scale were unacceptable and ranged from 0.36 to 0.48. Corrected item-total correlations were also calculated for each subscale (see Table 2). For the excuse making subscale, corrected item-total correlations were low to moderate, ranging from 0.14 to 0.53. For the effort subscale, item-total correlations were between 0.05 and 0.37 indicating that effort subscale items were poorly related to the subscale total. Table 2. Corrected item-total correlations for the excuse making and effort subscales of the SHS Item Sample Canadian athletes (n=245) Australian athletes (n=221) Sport-relevant instructions (n=136) excuse effort excuse effort excuse effort 1 0.26 0.34 0.46 2 0.16 0.10 0.13 3 0.41 0.44 0.35 4 0.22 0.14 0.23 5 0.28 0.33 0.40 6 0.17 0.15 0.05 7 0.29 0.37 0.29 8 0.22 0.23 0.30 9 0.35 0.40 0.33 10 0.40 0.39 0.42 11 0.48 0.40 0.50 12 0.36 0.39 0.35 13 0.27 0.37 0.14 14 0.45 0.53 0.29 Table options Given the low alpha coefficients, low reliability estimates and low item-total correlations for the subscales, the SHS, particularly the effort subscale, does not appear to be sufficiently reliable when administered to athletes. This weakness may be due, in part, to heterogeneous item content. Because factor analysis provides another means of assessing the homogeneity of scale items (Schutz, & Gessaroli, 1993), the scale's item structure was examined via factor analytic procedures. 3.3. Factor structure A series of confirmatory factor analyses were conducted to test the fit of the Rhodewalt (1990) two-factor model to the three data sets. The two factors (effort and excuse making) were allowed to correlate with one another (i.e. oblique factor structure). Using the computer program AMOS (Arbuckle, 1994), maximum likelihood estimation methods were employed. These confirmatory factor analyses were based on the variances and covariances of the items on each subscale. To evaluate the fit of the two-factor structure to the data, the AMOS program provided a chi-square test. Yet the chi-square value may be an inappropriate test of model fit in applications of confirmatory factor analysis that do not employ very large samples (Raykov, 1998). Consequently, the following goodnesss-of-fit values were also examined: the root-mean-square error of approximation (RMSEA), the adjusted goodness of fit index (AGFI) and the comparative index of fit (CFI). Results of the confirmatory factor analyses are presented in Table 3. The chi-square statistics were all significant (p<0.001). Although there are no tests of significance for the other goodness-of-fit criteria, researchers have suggested that an RMSEA≤0.05 indicates a close fit, ≤0.08 a fair fit, ≤0.10 a mediocre fit and >0.10 indicates a poor fit (Browne, & Cudeck, 1993). Inspection of the RMSEA values and their confidence intervals indicated that the goodness of fit varied across data sets and, for the most part, was in the mediocre range. The best fit was for the sport-relevant data set (RMSEA=0.07, CI=0.05; 0.09). Interestingly, this is the data set that produced the lowest reliability estimates. Turning to the other fit indices, based on the convention that AGFI and CFI values of 0.90 or higher indicate a good model fit (Schumacker, & Lomax, 1996), the observed AGFI and CFI values suggested that the model was a poor fit to the data. Taken together, these results indicate that the two-factor model of the SHS does not reliably produce a good fit when the scale is administered to athlete samples. Table 3. Results of the confirmatory factor analyses for the 2-factor model of the SHS. *p<0.0001 Index Data Set Canadian athletes Australian athletes Sport-relevant χ2 214.14* 183.20* 123.40* RMSEA (90% confidence interval) 0.09 (0.07; 0.13) 0.08 (0.07; 0.10) 0.07 (0.05; 0.09) AGFI 0.84 0.86 0.84 CFI 0.68 0.74 0.79 Table options Table 4 presents the factor loadings for each sample. For each sample, several items loaded below 0.40 (this was the loading criterion used by Rhodewalt, 1990), particularly for items on the effort subscale (factor I). Factor loadings for the excuse making subscale (factor II) were higher, with a majority of these scale items loading ≥0.40 across the samples. Nonetheless, the patterns of association between each item and the overall factor structure varied and items showed differential sensitivities across samples. Table 4. Factor loadings for the three samples. Factor I=effort; factor II=excuse making. Factor loadings <0.40 were evidenced mainly for the effort subscale Item Canadian sample (n=245) Australian sample (n=221) Sport-relevant sample (n=136) I II I II I II (1) Excuse 0.31 0.39 0.61 (2) Effort 0.27 0.43 0.32 (3) Excuse 0.44 0.48 0.43 (4) Effort 0.36 0.15 0.18 (5) Excuse 0.32 0.43 0.51 (6) Effort 0.38 0.24 0.10 (7) Effort 0.33 0.36 0.54 (8) Effort 0.41 0.40 0.61 (9) Excuse 0.48 0.48 0.42 (10) Excuse 0.50 0.47 0.47 (11) Excuse 0.60 0.50 0.65 (12) Excuse 0.44 0.45 0.43 (13) Excuse 0.32 0.44 0.22 (14) Excuse 0.54 0.66 0.32 Correlation between factors 0.95 0.87 0.70