The stability of structural measures derived from repertory grids

Although the use of repertory grids in psychological research has proliferated, studies of their psychometric properties are relatively rare. For this reason, we studied the reliability and convergence of several measures of cognitive structure derived from grids, including intensity, percentage of variance accounted for by the first component, cognitive complexity, ordination, extremity of ratings, self-ideal discrepancy, and self-other discrepancy, as well two measures of rating stability, construct consistency and factor loading consistency. Eighty-two Spanish (Catalan) and American participants completed small and large grids on each of four occasions, which enabled exploratory analyses of the impact of grid size and subject characteristics on the structural scores obtained. Results indicated that the majority of scores showed impressive test-relest reliability (modal r = .85 for periods up to 1 month), but also suggested a gradual tightening effect across subsequent administrations. In general, scores intercorrelated in theoretically anticipated ways, converging on the assessment of cognitive differentiation and discrimination of elements within construct dimensions. The results have implications for the reactivity of structural scores to variations in grid format, as well as for use of grids in the study of sex differences and cross-cultural comparisons.