A cross-validation of primary personality structure in the 16 P.F. by two parcelled factor analyses
Cattell, R. B. (1994). A Cross-Validation of Primary Personality Structure in the 16 P.F. by Two Parcelled Factor Analyses. Multivariate Experimental Clinical Research, 10(3), 181-190.
Since most recent factorings of primary personality structure as based on the 16 P.F. item domain have been by scales, they just conceivably could yield a simple structure perpetuating a past structure. Two new analyses have been made, one here, one prior, setting scales aside and using 46 'radial parcel' variables covering the 184 items. Radial parcels of four items each have built up quite regardless of scale groupings. The present analysis on 250 students and the prior one on 780 adults yielded 22 and 21 factors respectively, statistically significant simple structures in both and a substantial invariance (matching) of factors between them by congruence and salient variable similarity coefficients. Further analysis, beyond the present, shows that the factors found here (a) match when converted to loadings on items, the recognized 16 P.F. factor patterns, and (b) yield the patterns of second-order factors that would be expected from the identification by item content given them as primaries. Additionally, new primaries appear here beyond those in the 16 P.F., and they also match from the present to the prior parcelled analysis. However, one or two poor matches - E, H, M, Q2, and Q4 - also result, which could be due to the substantial differences of population in the studies used in the cross validation.