Second stratum personality structure in joint rating and questionnaire measures: With new light on questionnaire distortion

Abstract

The study continues a test of the theory of instrument-transcending personality factors, which requires that the same personality factors load both the questionnaire and the observer-rating markers for a factor. It has recently been shown that this holds with primary factors, with 64 variables on an N of 164 undergraduates. It was argued that proofs of congruent structure are not entirely satisfactory unless the second order structure derived from the primaries also fits. Accordingly the correlation matrix of the 23 primary factors was factored, showing 9 second orders by the scree test. Communalities were iterated thereto and rotation to simple structure was pursued by Rotoplot to an unimprovable maximum at 52.5% for ± .10 loadings. At this point congruence were worked out with patterns of secondaries in Cattell's 1973 survey of results, which showed 8 out of 9 factors matching very significantly, but with Qv poorer than others. Inasmuch as desirability distortion has previously had to rest on data within questionnaires only, and this study was virtually the first factoring, of a Q and L (life observer) data together, it offers a definite contribution to motivation distortion. Two primaries, 19 and 20, appeared in Q-data only were hypothesized to be "popularity" and "respectability" distortions. The way they behaved here at the second order, loaded negatively by superego, Qviii, supports this interpretation.