A discriminant function for diagnosing depressives with selected source trait factor measures from the O-A kit
Patrick, S. V.
Cattell, Raymond B.
Price, P. L.
Campbell, J. F.
MetadataShow full item record
Patrick, S. V., Cattell, R. B., Price, P. L., Campbell, J. F. (1981). A Discriminant Function for Diagnosing Depressives with Selected Source Trait Factor Measures from the O-A Kit. Multivariate Experimental Clinical Research, 5(2), 41-51.
Thirty-one depressives and 30 normals were administered the subtests of the O-A battery used for measuring the objective source traits of Independence (U.I. 19), Evasiveness (U.I. 20), Realism (U.I. 25), and Somindence (U.I. 30). The covariance matrices for the two respective groups were found to be significantly different, X = 22.048, p ? .015. The sample variances for the source traits were all greater in the depressives than in the normals, especially Evasiveness (U.I. 20), but none were significantly greater. A subsequent analysis of the correlation matrices, however, indicated that whereas all pairs of source traits were significantly correlated in the normals except Evasiveness (U.I. 20) and Realism (U.I. 25), only two pairs of source traits were significantly correlated in the depressives: Independence (U.I. 19) and Realism (U.I. 25) were positively correlated, and Realism (U.I. 25) and Somindence (U.I. 30) were negatively correlated. The data were then subjected to a discriminant analysis for the purpose of determining the direction and degree to which each of the four source traits contributes to the discrimination of depressives when the source traits are considered simultaneously. The obtained discriminant function was highly significant, X (4) = 26.296, p ? .001. Depressives were found to be lower in Independence (U.I. 19) and higher in Evasiveness (U.I. 20) and Somindence (U.I. 30). The contribution of Realism (U.I. 25) to the discriminant function was too small to be considered significant. Following the discriminant analysis, the overall ability of the four source traits to discriminant normals and depressives was assessed by the Mahalanobis (1936) generalized distance function to classify subjects as normal or depressed. The classification procedure, which did not assume equivalent covariance matrices for the two groups, correctly classified 78.68% of the subjects. The relatively high percentage of correct classifications attest to the diagnostic and theoretical relevance of objectively derived source traits, especially when one considers the three of the four measures chosen were more theoretically obscure and less related to the depression than other objective source traits which could have been selected for investigation. The discriminatory power of all the relevant objective source traits should be even higher.