International Journal of Education and Psychological Research (IJEPR) Volume 3, Issue 3, September 2014 13 Construct Validity of the EDS with Depression [1] Katrina M Travers [2] Peter A Creed [3] Shirley Morrissey [1] Australian Institute of Psychology, Brisbane, Australia, E-mail: katrinatravers@aipc.net.au [2] School of Applied Psychology and Griffith Health Institute, Griffith University, Gold Coast, Australia [3] School of Applied Psychology and Griffith Health Institute, Griffith University, Gold Coast, Australia Abstract: The Attributional Style Questionnaire [1] is a widely used scale that measures explanatory style when investigating the construct validity of the Reformulated Learned Helplessness Theory [2]. Despite its wide use, it has never been demonstrated to have satisfactory reliability or validity [3]. To address this, Travers, Creed, and Morrissey [4] successfully developed a more internally reliable and structurally valid measure of explanatory style, called the Explanatory Dimensions Scale (EDS), which achieved partial construct validity with the GHQ-12. It was recommended by Travers et al. that the construct validity of the EDS needs to be further investigated against established variables, including depression. Thus, we set out to investigate the construct validity of the EDS with depression in the community (termed depressive symptomatology), measured using the DASS and CES-D. Using two population samples, the structural validity of the EDS was again demonstrated, and the EDS retained improved alpha coefficients for all dimensions. Further, the EDS globality scale significantly but weakly predicted depressive symptomatology (5-6% of the variance) only when other known predictors of depressive symptomatology (non- productive coping, self-esteem, anxiety, and stress) were absent. Finally, we demonstrated that low self-esteem, anxiety and stress significantly mediated the relationship between globality and depressive symptomatology, and stress also moderated this relationship. Further recommendations and implications were discussed. Keywords: Explanatory Dimensions Scale, validity, depression I. INTRODUCTION The Attributional Style Questionnaire (ASQ; [1]) is a scale that has been used almost exclusively as a measure of explanatory style, when investigating the validity of the Reformulated Learned Helplessness (RLH) Theory by Abramson, Seligman, and Teasdale [2]. However, the ASQ has never been demonstrated to have adequate structural or construct validity, as outlined by Travers, Creed, and Morrissey [3], thereby rendering the RLH theory itself as lacking the tenets of a “good theory” as described by Epstein [5]. To address this, Travers, Creed, and Morrissey [4] set out to develop a more valid measure of explanatory style by: (a) rewriting the items measuring the existing internality dimension of explanatory style so that they reflect Lewin’s [6] life space theory (the precursor theory to the construct of internality), and (b) reformat the ASQ so structural validity studies of the measure are not confounded due to the embedded “testlets” [7]. They did so by ensuring each item stem was followed by only one item, not three items which was the format of the original ASQ. Results were successful in establishing a new scale (termed the Explanatory Dimensions Scale; EDS) that measures each dimension of explanatory style (globality, stability, and internality) with four items each. The EDS achieved structural validity using both EFA and CFA techniques, finding orthogonal components that were interpretable as the three dimensions of explanatory style. Further, Travers et al. achieved alpha coefficients for each of the dimensions of the EDS approaching acceptable to good (.68 to .78), reasoning that this was the first time a measure of explanatory style had achieved this. Interestingly, Travers et al. [4] were unable to find a second-order latent variable to justify the use of the composite scale score, adding that this was further evidence for the lack of validity of the ASQ. Finally, Travers et al. [4] found initial partial construct validity for the EDS with the GHQ-12 [8], finding weak significant correlations for the globality and stability dimensions (.15 and .13; p < .05). Construct validity was not established for the internality dimension, despite its improved internal reliability. It was recommended by Travers et al. [4] that further research into the construct validity of the EDS needs to be conducted, initially with established outcome variables theorised to be related to the construct of explanatory style, such as depression. 1.1 Explanatory Style and Depression Previous research on the construct validity of explanatory style with depression using the ASQ was summarised in Travers et al. [3]. Travers et al. detailed that, to date, the most that can be summarised from this research is that only the negative composite scale score of the ASQ is weakly related to depression (measured using the Beck Depression Inventory [BDI; 9]) in a young adult population, and that this relationship is strengthened to a moderate level if mediating variables such as brooding are considered. Further, Travers et al. [3] summarised that only one study reported the individual dimension results (see [10]), finding that the globality and stability subscales of the ASQ significantly correlated with the BDI (.30 and .25; p < .05), but the internality subscale did not significantly correlate with the BDI.