خود ناتوان سازی و بدبینی دفاعی: یک مدل از خود حمایتی از دیدگاه طولی

Abstract This research places self-handicapping and defensive pessimism (comprising defensive expectations and reflectivity) into a single conceptual and analytic framework that models the full self-protective process across time. Data on two occasions collected during students’ (n=328) first two years at university show: performance orientation positively predicts self-handicapping, defensive expectations, and reflectivity; task-orientation negatively predicts self-handicapping and defensive expectations and positively predicts reflectivity; uncertain personal control positively predicts defensive expectations and reflectivity; and an external attributional orientation positively predicts self-handicapping and defensive expectations. Although both self-handicapping and defensive expectations negatively affect academic outcomes, the negative effects of self-handicapping were more marked. In contrast to these counter-productive strategies, reflectivity had positive effects on academic engagement.

. Introduction Individuals can use a variety of strategies to deal with threats to their self-worth. Two such strategies—self-handicapping and defensive pessimism—have received little joint attention to date and their separate and combined effects are the focus of the present study. According to the self-worth theory of motivation (Covington, 1984, Covington, 1992 and Covington, 1997), the need to protect one’s self-worth arises primarily from a fear of failure and the implications this failure may have for one’s private and public sense of ability and subsequent self-worth. Using this theory as a basis to guide hypotheses, a longitudinal model of the processes giving rise to and following from self-handicapping and defensive pessimism is examined. In testing this model, the study seeks to (a) explore the processes of defensive manoeuvring as they relate to self-handicapping and defensive pessimism, (b) clarify the factors giving rise to self-handicapping and defensive pessimism and the consequences that follow from them, (c) draw together self-handicapping and defensive pessimism under a common conceptual and empirical framework, and (d) add to the small body of research that assesses defensive manoeuvring in a longitudinal fashion. A useful framework to guide the study of this process is one proposed by Buss and Cantor (1989). According to them, individuals’ dispositions or characteristic orientations influence the strategies they use to negotiate demands in their environment, and these strategies in turn influence their behaviour within this environment. In contrast to the bulk of research that typically studies such processes from personality and social psychological perspectives, the present study evaluates the generalisability of this model in an educational context.

7. Results 7.1. Preliminary confirmatory factor analysis Before examining the constructs in the hypothesised longitudinal model, correlations between these constructs were first tested in a CFA as well as the overall fit of the model to the data. The CFA yielded an acceptable fit to the data (χ2=3038.84, df=1736, RNI=.91, TLI=.89, RMSEA=.048). Mean factor loadings, Cronbach’s αs, and correlations are presented in Table 1. Generally the loadings are quite high as are the reliabilities for the factors. A number of correlations are noteworthy. The high correlation between self-handicapping and defensive expectations (r=.57 at Time 1 and r=.68 at Time 2) show support for the anticipated congruence of the two strategies. At Time 1, correlations between reflectivity and both self-handicapping and defensive expectations are lower than the correlation between self-handicapping and defensive expectations. It is noted, however, the sizeable correlation between reflectivity and defensive expectations at Time 2. Correlations amongst predictors and strategies are in broad agreement with hypotheses and the correlations amongst strategies and outcomes indicate the negative impact of both self-handicapping and defensive expectations and quite the opposite effect for reflectivity. Table 1. Mean target factor loadings, Cronbach’s αs, and correlations generated in the CFA of the Time 1–Time 2 data set ML α 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 1. T1POa 74 86 2. T1UCa 65 67 42 3. T1EAa 61 53 37 68 4. T1SEb 86 90 07 −38 −06 5. T1TOb 81 85 44 15 02 23 6. T1SHa 88 87 21 39 57 −14 −32 7. T1DEa 83 81 48 73 70 −15 02 57 8. T1REFb 70 75 61 56 41 05 52 07 40 9. T1SRb 78 82 11 −08 −33 18 46 −49 −18 42 10. T1Pb 73 78 −06 −27 −47 34 44 −58 −34 10 67 11. T1FPb 89 92 01 −15 −13 63 20 −14 −10 06 25 33 12. T2POa 76 87 64 30 30 07 14 28 38 38 −10 −18 −10 13. T2UCa 69 71 28 68 48 −30 −06 40 47 35 −20 −36 −26 50 14. T2EAa 73 69 17 38 66 −08 −13 46 42 19 −24 −43 −18 40 73 15. T2SEb 86 90 10 −33 −15 74 16 −11 −10 02 13 31 48 09 −38 −10 16. T2TOb 77 80 26 16 −08 15 59 −24 03 27 32 36 19 19 −02 −24 20 17. T2SHa 90 89 11 25 37 −05 −29 67 37 05 −34 −28 −11 29 44 56 −07 −44 18. T2DEa 84 82 31 50 51 −11 −15 49 67 29 −22 −32 −12 57 71 70 −15 −23 68 19. T2REFb 59 69 53 47 54 −01 16 28 50 63 03 −21 −10 72 58 51 02 26 26 69 20. T2SRb 76 80 17 03 −24 14 34 −37 −05 32 72 47 22 05 −14 −27 18 55 −37 −19 15 21. T2Pb 79 84 01 −09 −32 22 33 −40 −16 11 52 75 32 −15 −45 −52 37 50 −45 −41 −17 55 22. T2FPb 87 91 11 −08 −14 39 20 −10 04 04 15 30 57 −05 −27 −28 63 33 −18 −15 −06 22 50 23. 97GR − − 08 −16 −29 −01 15 −23 −06 04 16 12 −14 16 −21 −19 03 16 −20 −07 10 18 13 −07 Note. Decimal omitted; Test–retest correlations in bold; ML, mean factor loading; T1, Time 1; T2, Time 2; PO, performance orientation; UC, uncertain control; EA, external attributions; SE, self-esteem; TO, task-orientation; SH, self-handicapping; DE, defensive expectations; REF, reflectivity; SR, self-regulation; P, persistence; FP, future plans; 97GR, grades at end of 1997 (first year); coefficients >.13 are significant at p<.05. a Factor represented by composite scores. b Factor represented by item parcels. Table options 7.2. Testing the longitudinal structural equation model The central relationships in the hypothesised longitudinal model were then examined and these are presented in Fig. 1. This involved testing relationships at Time 1 between (a) the predictors and the three strategies (self-handicapping, defensive expectations, and reflectivity), (b) the three strategies and the four outcomes (end-of-year grades, self-regulation, persistence, and future plans), and (c) the three engagement outcomes (self-regulation, persistence, and future plans) and the performance outcome (grades). At Time 2, central analyses involved testing relationships between (a) Time 1 outcomes and Time 2 affective and motivational predictors, (b) the predictors and the three strategies, and (c) the three strategies and self-regulation, persistence, and future academic plans. In addition to these parameters, three other sets of parameters were freed to be estimated. These were (a) test–retest paths between parallel Time 1 and Time 2 constructs (e.g., between Time 1 task-orientation and Time 2 task-orientation), (b) correlations between parallel Time 1 and Time 2 uniquenesses (e.g., between the uniqueness of each Time 1 task-orientation item-parcel and its Time 2 counterpart), and (c) correlations among constructs at the same point in the model (e.g., among Time 1 predictors or among residuals of Time 1 strategies). To enhance parsimony, structural parameters not significant at the p<.1 level were dropped from the model. The hypothesised longitudinal model and modification indices were inspected for empirical information about additional paths that would result in a better model fit if freed. Using a conservative approach, the criteria for freeing additional parameters were that they (a) were conceptually feasible, (b) would yield substantially large changes in the model, (c) would result in a model that was nested under the initial CFA model (Table 1) in that only paths representing relations among the latent constructs were for considered for inclusion, and (d) were consistent with correlations among the corresponding factors in the CFA model (Table 1). Only three path coefficients were subsequently freed: in both Time 1 and Time 2, the private esteem-relevant competence valuation composite scores were freed to also load on task-orientation, while self-esteem at Time 1 was freed to predict persistence at Time 1. The parameters between private competence esteem-relevant competence valuation and task-orientation were considered feasible in that the former reflects a focus on personal standards and progress (rather than performance relative to others), quite consistent with a task-orientation. The path between self-concept and persistence is consistent with claims that efficacious individuals function better through elevated levels of effort and persistence (Bandura, 1986 and Bandura, 1997). Results based on this final model yielded an acceptable fit to the data (χ2=3294.50, df=1905, RNI=.90, TLI=.89, RMSEA=.047) and the structural parameters are presented in Fig. 2. Significant structural relations in longitudinal model. Note. Time 1–Time 2 ... Fig. 2. Significant structural relations in longitudinal model. Note. Time 1–Time 2 test–retest paths are indicated in brackets [ ] *p<.10. All other paths significant at p<.05. Paths bolded at Time 1 are also significant paths at Time 2. Figure options Results in Fig. 2 indicate that at Time 1, performance orientation positively predicts the three strategies. External attributional orientation positively predicts self-handicapping and defensive expectations. Uncertain personal control positively predicts defensive expectations and reflectivity. Task-orientation negatively predicts self-handicapping and defensive expectations, and positively predicts reflectivity. Self-esteem predicts reflectivity and persistence and to a large degree predicts future academic plans. Particular attention is drawn to the consistency across Times 1 and 2. Path coefficients between Time 2 predictors and Time 2 strategies are consistent with the parallel Time 1 relationships but are generally weaker (primarily because the bulk of the variance of Time 2 strategies has been explained by their Time 1 counterparts). Paths at Time 1 that are consistent at Time 2 are bolded. Pivotal aspects of the longitudinal model are the link-points between the two process models. In relation to this, academic grades positively predict Time 2 performance orientation and negatively predict uncertain personal control. Also, persistence negatively predicts Time 2 uncertain personal control and external attributional orientation and positively predicts Time 2 task-orientation. Particular attention is directed to the significant negative path between self-handicapping at Time 1 and later academic grades and the significant test–retest paths between Time 1 and Time 2 constructs. 7.3. Testing three submodels Relationships between strategies and outcomes, however, are problematic. For example, the generally non-significant predictive role of Time 1 and Time 2 defensive expectations—particularly in the context of the significant correlations with Time 1 and Time 2 outcomes (see Table 1)—is of some concern. Its non-significant role in the longitudinal model seems largely an artefact of collinearity with self-handicapping and this gives rise to the suppression of defensive expectations effects (for a more detailed discussion of this general problem, see Bollen, 1989). Indeed, as noted earlier, the collinearity between self-handicapping and defensive expectations is important in terms of the assessment of their empirical congruence. It was therefore considered important to examine three submodels designed to assess the separate effects of self-handicapping, defensive expectations, and reflectivity. In Submodel 1, self-handicapping and reflectivity at Time 1 and Time 2 were excluded from analyses. In Submodel 2, defensive expectations and reflectivity were dropped at Time 1 and Time 2. In Submodel 3, defensive expectations and self-handicapping were dropped. All submodels fit the data well (see Table 2). When defensive expectations, self-handicapping, and reflectivity are dealt with in this way, their respective roles—independent of their collinearity—are clarified. Because the relationships between predictors and strategies are much the same as those presented in Fig. 2, only the paths between strategies and outcomes are presented. Standardised path coefficients are presented in Table 2. Table 2. Fit of submodels and beta paths between strategies and educational outcomes estimated in each of the three submodels Time 1 outcomes Time 2 outcomes Grades Self-regulation Persistence Future plans Self-regulation Persistence Future plans Submodel 1. Defensive expectations ns −.21 −.31 ns ns −.19 ns χ2=2470.49, df=1349, RNI =.91, TLI=.90, RMSEA=.050 Submodel 2: Self-handicapping −.27 −.50 −.52 ns −.14 −.24 −.12 χ2=2364.42, df=1348, RNI =.91, TLI=.91, RMSEA=.048 Submodel 3: Reflectivity ns .39 .12 ns .20 ns ns χ2=2595.58, df=1457, RNI =.90, TLI=.89, RMSEA=.049 ns, not significant at p<.1. All other paths significant at p<.05. Table options In Submodel 1, Time 1 defensive expectations negatively predict Time 1 self-regulation and persistence. To a lesser extent, Time 2 defensive expectations negatively predict Time 2 persistence. In Submodel 2, Time 1 self-handicapping quite strongly negatively predicts Time 1 self-regulation and persistence and negatively predicts later academic grades. To a lesser extent, Time 2 self-handicapping negatively predicts Time 2 self-regulation, persistence, and future academic plans. In Submodel 3, Time 1 reflectivity positively predicts Time 1 self-regulation and to a lesser extent positively predicts persistence. At Time 2, reflectivity positively predicts Time 2 self-regulation.