How Should Attributions Be Measured? A Reanalysis of Data from Elig and Frieze

Abstract

Elig and Frieze used a multitrait, multimethod approach to contrast three methods for measuring attributions: unstructured/open-ended, structured/unidimensional, and structured/ipsative. The present paper reanalyzed their data using confirmatory factor analysis techniques. Six attribution dimensions, ability, stable effort, unstable effort, mood, task, and luck, were examined. The two structured methods were found to yield attribution dimensions that were similar to one another but quite unlike those defined by the unstructured/open-ended method; thus, findings from studies employing the unstructured method seem unlikely to be comparable to findings from studies using structured methods. The structured/unidimensional measures yielded a method factor with positive loadings on all attribution dimensions, demonstrating that that approach taps individual differences in the extremity of attributions made by subjects.