توانایی پیش بینی خود ناتوان سازی و عزت نفس در زمینه دستیابی به فعالیت های فیزیکی

Abstract Self-handicapping and self-esteem were assessed as predictors of completion of or withdrawal from military recruit training. Three hundred and fifty three male recruits completed trait self-handicapping and self-esteem measures. Discriminant function analysis indicated that successful completion of training could be correctly predicted from these two constructs, with 92.6% accuracy. Moreover, recruits voluntarily withdrawing from training were correctly predicted with 58.8% accuracy. These prediction rates were significantly better than chance, based on known pass rates, and represent an improvement over current prediction based solely on physical fitness testing. Results show that psychological measures can provide a significant source of information from which to predict performance in real, multifaceted achievement domains. The relevance of the results to theoretical and applied perspectives is discussed.

Introduction In demanding situations, where performance occurs under conditions of threat or challenge and where the outcome has a high degree of personal salience, individuals are likely to employ behavioural and cognitive strategies in order to maximise potential benefits and minimise potential costs. Kelley (1972) explained how specific attributions could be made to either augment or discount the effect of an outcome. To achieve maximum benefits after a successful outcome, attributions would be made to internal and stable factors (e.g. skill). Conversely, to minimise the cost of an unsuccessful outcome, attributions would be made to external and unstable factors (e.g. bad luck or task difficulty). However, instead of awaiting the outcome before making self-serving attributions, individuals can also employ strategies before the outcome is known, or self-handicap, in order to maximise benefits or minimise costs. In his review, Rhodewalt (1990) outlined research that supported the existence of individual differences in general disposition to use self-handicaps, across a range of situations. The construct of self-handicapping is closely associated with self-esteem and different levels of esteem have been shown to determine whether the motive to self-handicap was to augment or discount an outcome (Tice, 1991). 1.1. Self-handicapping and self-esteem Self-handicapping has been defined as “any action or choice of performance setting that enhances the opportunity to externalise (or excuse) failure and to internalise (or reasonably accept credit for) success” (Berglas & Jones, 1978, p. 406). Actions that constitute self-handicapping can be construed on a scale with gradually increasing protection often associated with commensurate loss of opportunity for enhancement. Thus, an individual who falsely complains of flu before giving an important presentation has a ready excuse for poor performance. However, if the presentation is good then the individual accrues even more credit for performing well, against the odds. It is important to note that self-handicapping by symptom report can be achieved without any actual detrimental effect upon performance. By contrast other strategies such as alcohol consumption (Kolditz & Arkin, 1982), lack of practice or reduced effort (Rhodewalt, Saltzman, & Wittmer, 1984) and physical over-training all offer similar protection from failure but at the cost of reduced chance of success, since performance will be negatively affected by such behaviours. It is suggested that this repertoire of self-handicapping behaviours could be extended and that ultimately, individuals seeking maximum protection from failure could withdraw before the outcome is decided (Zuckerman, Kieffer, & Knee, 1998) or engineer situations that enable them to avoid the evaluative performance demands altogether (Snyder & Smith, 1982). Under these circumstances the high level of protection from potential costs is achieved by relinquishing any possible chance of success. Jones and Rhodewalt (1982) developed the Self-Handicapping Scale (SHS) to assess the general disposition to use self-handicapping strategies and to distinguish between individuals with different propensities to use behavioural and self-reported handicaps (see Rhodewalt, 1990, for review). The original SHS showed negative correlations (r's 0.30 to 0.50) with measures of self-esteem that suggested a straightforward inverse relationship between the two. However, in a review of the literature Baumeister, Tice, and Hutton (1989) revealed that the relationship between self-handicapping and self-esteem was more complicated and could be explained in terms of the different motives to engage in self-handicapping. They suggested that high esteem individuals were driven to enhance their public image whilst low esteem individuals were more motivated to protect their self-image. Tice (1991) confirmed this, demonstrating that high-self esteem individuals used self-handicapping strategies to enhance their public image whereas individuals with low self-esteem used the same strategies to protect their public image. The significant role of self-esteem in determining motivation for self-handicapping behaviour is consistent with its central role in other established theories in psychology (e.g. Heider, 1958 and Tajfel and Turner, 1979). Global self-esteem, the evaluative component of self-concept, has also been associated with performance in academic, social, and physical settings (see Sonstroem, 1984, for review). Empirically supported models of self-concept tend to agree that it is a multifaceted psychological construct (Fox, 1997 and Shavelson et al., 1976) and consequently, it is recommended that measures of self-esteem be selected to be appropriate to the domain of interest. However, outside of controlled, experimental designs, naturalistic research frequently encounters situations in which performance is associated with many components of self-esteem together. In these situations a global measure of self-esteem such as Rosenberg's (1965) Self-Esteem Scale may be more appropriate than a specific domain measure. 1.2. Contextual factors influencing the use of self-handicapping A number of contextual factors, which have a mediating influence on the deployment of self-handicapping strategies, have emerged from the literature. Firstly, the extent to which performance is considered public greatly increases the self-presentational motivations (Tedeschi, 1981) and consequently the desire to protect or enhance self-esteem. The influence of others present can be considered to operate on self-esteem through several processes including raised self-awareness as one attempts to determine how one is perceived by others in order to create a socially desirable public image (Baumeister, 1982) and deriving information about oneself from information from others (Self-Perception Theory, Bem, 1972). This influence will be further increased according to the principles of strength, immediacy and number outlined by Social Impact Theory (Latané & Nida, 1980), and by the formal capability of others to evaluate performers on salient dimensions (Evaluation Apprehension, Cottrell, 1972). A second contextual factor is the perceived salience of the performance to the individual, which is critical in determining whether self-handicapping strategies might be used to protect or enhance self-esteem (Hirt, Deppe, & Gordon, 1991). Performance situations that are not relevant to self-concept will be much less likely to stimulate self-handicapping behaviour since success or failure at a low salience task will have little or no impact on self-esteem (Tice, 1991). Individuals will assess situations through a process of appraisal (Lazarus & Folkman, 1984) in which they determine whether there is a threat or challenge present and whether it is relevant to them or not. It is important to recognise that through this process of appraisal, whilst a situation might appear to present a threat or challenge it may actually have no relevance to an individual. This point emphasises the limitation of experimental work, utilising laboratory based tasks which rarely generate the genuine personal salience found in real world tasks that individuals self-select and to which they commit considerable resources. In an attempt to reduce this limitation research on self-handicapping has used tests of intelligence with student populations in order to ensure task salience (Harris and Snyder, 1986, Tice, 1991 and Tice and Baumeister, 1990). However, even where the task may appear relevant, participants are usually quite cognisant that their performance in an artificial experiment has no ramifications to them outside of the situation. This is particularly true if they have recently completed informed consent documents assuring them that any information pertaining to them will not be revealed. Naturalistic research, where personal salience is high, can offer important contributions to the testing and development of theoretical constructs such as self-handicapping. The nature of the task and consequent feedback is a further contextual factor that has a potentially mediating effect on self-handicapping behaviour. Strube and Roemmele (1985) proposed that tasks differed on the extent to which they would diagnose individuals as possessing high or low competence on the task. A task on which failure clearly demonstrates possession of low competence is defined as high in ‘diagnosticity of failure’ and has negative implications for self-esteem. Tasks requiring minimum performance standards are likely to be considered high in diagnosticity of failure because although the statistical likelihood of failure may be low, with the majority of individuals regularly passing, failure clearly signifies incompetence on the relevant task components. Thus, a low failure rate increases both the expectancy to pass and the significance of not attaining the minimum standard. Conversely, such tasks have little diagnosticity of success since attaining a pass standard gives no further information to individuals about their ability or competence other than that they are above the minimum acceptable level. An analysis of the contextual factors associated with the use of self-handicapping strategies reveals the importance of the interaction between existing behavioural predisposition and the situation as precursors to self-handicapping behaviour. However, the success with which situational variables can be manipulated in experimental research is obviously limited. A basic task which has never been encountered previously and the results from which will not be disclosed, is unlikely to be highly salient or ego relevant to the participant. In addition the duration of the experiment is important, as many subjects required, or paid to participate, simply want to complete their duties as expediently as possible. Consequently an experiment that attempts to operationalise self-handicapping in terms of reduced practice time, may be confounded by this basic motivation of participants. Naturalistic research, whilst less rigorously controlled, provides situations where the motivation to engage in self-handicapping behaviour is genuine and the contextual parameters that may influence their behaviours are meaningful to the participants. 1.3. Present research The present study assessed the efficacy of trait self-handicapping and global self-esteem as predictors of behaviour in a real life setting namely, completion of, or voluntary withdrawal from, basic military training in the British Army. Military training provides a genuinely multi-faceted performance domain where the pressure to achieve is high and originates from both personal and social sources. Recruits undergo strenuous physical training and are required to pass practical and theoretical tests in weapon handling, shooting, first aid, map reading, field craft and nuclear and biological warfare drills. Within the training environment competitiveness between platoons is encouraged and at the same time recruits are in a totally new environment, possibly away from home and established relationships for the first time in their lives. The combination of the new demands of training and the absence of old resources, such as social support, contribute to the overall pressure experienced by recruits. The contextual factors previously outlined as relevant to self-handicapping, are present in the military training environment together with intense physical and mental challenge. Throughout the 12-week duration of training, trained and experienced staff conduct continual public evaluation, on different components of performance. Performance outcome is highly salient since individuals volunteer to join, have professional aspirations and are required to make significant personal and public commitments. The nature of military training and ultimate aim of attaining minimum performance standards identifies this as a task high on diagnosticity of failure. In addition feedback is often ambiguous and may be deliberately manipulated by instructors to disrupt feelings of self-esteem and confidence in order to ensure that maximum effort and attention are maintained throughout. Decisions as to whether an individual fails or passes training are made by experienced staff and encompass many different facets of performance. However, individuals wishing to self-handicap are very restricted in the options available to them. Alcohol and tobacco use and lack of sleep are not considered viable excuses and cannot be feasibly maintained throughout the duration of training. Lack of practice or effort is unavailable as a self-handicapping strategy because of the design of the training programme and the work of trained instructors. Finally, injury and illness are both thoroughly investigated and treated by medical professionals leaving little scope for an individual to strategically employ this behaviour without just cause. Thus, in the context of this study the decision of an individual to voluntarily withdraw from training represented the primary self-handicapping strategy available to them. Voluntarily withdrawing from training represents an extreme form of self-handicapping where all opportunity for success is forsaken in order to provide an ‘opportunity to externalise (or excuse) failure’ (Berglas & Jones, 1978, p. 406). Thus, in the context of military training individuals could rationalise their decision to voluntarily withdraw in terms of external factors such as recent parenthood, bereavement, marriage or a need to provide more immediate personal and financial support at home. Alternatively the decision to voluntarily withdraw could be excused because of claims by the individual that the army life did not suit them, the discipline was too harsh or that their individual needs were not met. In all of these examples the decision to self-handicap allows such externalised excuses to be made credibly, to both self and others, and thus protect self-esteem. The proposition that voluntary withdrawal is a form of self-handicapping needs to be considered in light of the different possible outcomes facing recruits in training. While passing would be the statistically most likely outcome, individuals with low self-esteem would be more likely to doubt their own potential for success. At the same time low self-esteem would exacerbate the self-relevant threat continually present in the challenging, evaluative environment of military training. Thus recruits with low self-esteem would be more likely to focus on voluntary withdrawal or being discharged as the alternative outcomes facing them. Being discharged would represent a categorical failure whereas the decision to voluntarily withdraw can be seen as a form of self-handicapping because it pre-empts the possibility of being discharged and at the same time permits external explanations or excuses to be made. This proposition is consistent with the association between behavioural disengagement and self-handicapping highlighted in previous research on academic performance (Zuckerman et al., 1998). The purpose of this study was to explore the ability of measures of self-handicapping and self-esteem to predict voluntary withdrawal in an ecologically valid and personally salient situation. In addition the characteristics of the situation demanded high levels of performance under public evaluation with feedback that was not always contingent on performance. The training outcome, with high pass rates produced a task that had a high diagnosticity of failure for recruits. Under these conditions it was predicted that the individuals most likely to voluntarily withdraw from training, would have a high propensity to self-handicap and low self-esteem

. Results 3.1. Descriptive statistics Of the 353 subjects, five failed to complete the questionnaires correctly and were therefore excluded from the analysis. A further nine recruits were discharged and scores from these recruits were not included in the main analysis. A brief summary of the discharged group is provided at the end of the results section. The descriptive statistics for the remaining 339 subjects showing mean and standard deviation values on the SHS and RSE are presented in Table 1. Table 1. Mean and standard deviation scores for self-handicapping and self-esteem, by group Group n SHS RSE PASS 242 36.71 (±8.70) 34.52 (±3.62) DAORa 97 49.85 (±10.09) 27.10 (±5.63) Total 339 40.47 (±10.87) 32.40 (±5.44) a DAOR, discharged army own right. Table options A significant negative correlation (r=−0.69, P<0.001) was found between self-handicapping and self-esteem scores that indicates that a total shared variance of approximately 48% exists between these two measures. The theoretical significance of this finding will be considered in the discussion. 3.2. Test selection The aim of this investigation was to examine whether the two scales, RSE and SHS, could accurately classify which recruits would pass or voluntarily discharge themselves (DAOR) by the end of basic training. Thus the design was one in which the numerical data from the two scales was used to predict membership to a nominal group. In accordance with this aim categorical modelling through a discriminant function analysis was selected to test the ability of the two scales (predictor variables) to correctly classify group membership (the criterion variable; Sportscience, 1997). The discriminant function analysis is a multivariate technique that analyses the contribution of the predictor variables simultaneously and thus includes information on their inter-relationships (Norušis, 1985). This is an important distinction with univariate tests that analyse variables one at a time. The discriminant function analysis also allows for unequal probability of distribution between groups to be accounted for (Manly, 1986) so that the naturally occurring biases in group membership (such as in this study where the majority of recruits pass) are not erroneously interpreted as accurate classifications. The discriminant function analysis was selected in preference to the conceptually similar calculation of multiple regression since the criterion variable data (pass or DAOR) was discrete rather than continuous. The stepwise method of discriminant function analysis automatically begins with the predictor variable that contributes most to accurate classification (determined by F values). In each subsequent step the predictor variable that results in the smallest Wilks lambda value is selected until inclusion of further predictor variables contributes no more to the overall classification accuracy. Any predictor variables that do not contribute to classification are omitted. Therefore, the stepwise discriminant function analysis is a test that assesses the combined predictive ability of a number of predictor variables in relation to a discrete criterion variable with two or more levels. The results of the test indicate which predictor variables contribute to accurate classification and the relative importance of the contribution of each variable. 3.3. Addressing assumptions for discriminant function analysis A consistent problem associated with applying discriminant function analysis to naturally occurring groups is the extent to which assumptions can be adequately met in order to minimise the risk of misclassification. Consequently, prior to conducting a discriminant function analysis the data was examined to establish the suitability of applying this test. Within this research all available recruits entering training in the ATR agreed to participate in the study and so the subject group was the full recruit population of the ATR. To this extent problems associated with sampling biases were limited. A consideration of the requirement for multivariate normality of the predictor variables is also important to avoid misclassification. Tabachnick and Fidell (1996) indicate that the discriminant function analysis is robust to failures of normality if violation is caused by skewness rather than outliers, and providing a minimum of 20 cases appear in the smallest group. Exploratory data analysis indicated a single outlier, which when removed from the analysis served to increase the overall accuracy of classification. Since classification coefficients are usually more accurate for the sample data from which they were derived than for a wider population, the outlier case was left in the analysis. This procedure provides a more conservative classification accuracy, which is more representative of the results that might be expected were the prediction to be applied to data from a new sample group. Exploratory analysis also demonstrated that the two groups (PASS, DAOR) had significantly unequal within-group variance–covariance (P<0.001). Large differences in group size typically produce this effect which creates a tendency for over-classification into the group with the larger variance. Thus the discriminant function analysis was calculated using separate covariance matrices to legislate against this factor ( Tabachnick & Fidell, 1996). 3.4. Discriminant function analysis Preliminary analysis confirmed significant differences (P<0.001) between the means of both groups, for each of the scales. Stepwise discriminant function analysis determined the ability of self-handicapping and self-esteem scales to predict outcome of military training. Given the expected unequal group size, a priori probabilities for each group, were calculated from the outcome group size. The result indicated a significant discriminant function (Wilks's lambda=0.59, χ2(2)=176.21, P<0.001) and a canonical correlation of 0.64 indicating that the two measures were capable of explaining about 41% of the variance between subjects either passing or voluntarily withdrawing from training. Self-esteem and self-handicapping both contributed significantly to the discriminant function, with self-esteem providing the largest coefficient. The classification accuracy of the variables in predicting group membership is presented in Table 2. Of these the highest accuracy of prediction was for individuals who passed training (92.6%). Prediction of those who voluntarily withdrew from training was achieved with 58.8% accuracy. Table 2. Accuracy of classification: predicting group membership against actual group membershipa No. of cases Predicted group membership PASS DAOR Actual group PASS 242 224 (92.6%) 18 (7.4%) DAOR 97 40 (41.2%) 57 (58.8%) a Overall total percentage of cases correctly classified: 82.9%. DAOR, discharged army own right. Table options Whilst Table 2 shows the number of cases correctly classified by self-handicapping and self-esteem scales, further analysis is required to determine the extent to which this classification exceeds that which would be expected by chance, based on probabilities calculated from the existing group size. Table 3 summarises the probability of group membership and expected accuracy of correct classification by chance. Table 3. Estimated accurate classification by chance Group Cases (n) Prior probability No. correctly classified cases expected by chance PASS 242 0.714 173 DAORa 97 0.286 28 a DAOR, discharged army own right. Table options Overall 201 cases (59.3%) would be expected to be correctly classified by chance. The total classification accuracy was substantially greater (23.6% more) using self-esteem and self-handicapping scales than classification based on chance. Prediction of PASS recruits was 22% more accurate and prediction of DAOR recruits was 30% more accurate, using the two psychological scales. A second discriminant function analysis, entering just self-handicapping scores as predictor variables, was calculated to determine the ability of the self-handicapping scale to predict training outcome on its own. Results confirmed a significant discriminant function (Wilks lambda=0.70, χ2(2)=119.60, P<0.001), although this Wilks lambda value was greater than that achieved for self-esteem alone as a predictor. Neither scale, on its own, produced a Wilks lambda smaller than when the scales were combined, indicating that the two scales together had the greatest predictive ability. 3.5. Discharged recruits The nine discharged recruits were discharged either for medical reasons (MD; n=3) or because they were assessed, by instructors, as unfit for army service (UFAS; n=6). Both of these subgroups had uneven distributions and so the descriptive statistics presented below ( Table 4) should be interpreted cautiously. No theoretical predictions were made about these groups since being discharged was not an action over which the individuals had complete control. Consequently individual differences such as self-esteem and self-handicapping could not be reliably associated with these groups. Table 4. Means and standard deviation of self-handicapping and self-esteem for discharged recruitsa Group n SHS RSE MD 3 61.33 (±6.66) 21.00 (±6.24) UFAS 6 61.00 (±9.34) 17.33 (±3.56) Total 9 61.11 (±8.10) 18.56 (±4.59) a SHS, Self-Handicapping Scale; RSE, Rosenberg Self-Esteem Scale; MD, medical discharge; UFAS, unfit for army service.