Self-discrepancy: comparisons of the psychometric properties of three instruments.

In 2 studies, the psychometric properties of 3 methods for measuring real-ideal and real-ought self-discrepancies were compared: the idiographic Self-Concept Questionnaire-Personal Constructs, the nonidiographic Self-Concept Questionnaire-Conventional Constructs, and the content-free Abstract Measures. In the 1st study, 125 students at a university clinic completed the 3 instruments and measures of anxiety and depression before individual therapy. In the 2nd study, 278 undergraduates completed the 3 instruments at 2 time points 4 weeks apart and completed multiple measures of anxiety and depression at the 2nd time point. Internal consistency alphas were consistently strong for the personal construct measures (.90 to .92) and moderate to strong for the conventional construct measures (.82 to .90). Test-retest reliability coefficients were above .70 for the personal construct and conventional construct measures, but the coefficients for the latter were inflated by the stability of their error terms. The 2 discrepancies were found to be factorially distinct even though they were highly correlated. Convergent and discriminant evidence of validity was found in both studies for all measures except the abstract real-ought discrepancy. Convergence was as strong or stronger for the personal construct measures in comparison to the other measures. Test-criterion evidence of validity, with multiple measures of anxiety and depression as criteria, was found in both studies for all measures except for the abstract real-ought discrepancy in relation to anxiety. Overall, the findings support the idiographic personal construct instrument most strongly for clinical assessment and for clinical, translational, and personality research.