Radial parcel factoring-vs-item factoring in defining personality structure in questionnaires: Theory and experimental checks

Abstract

Allegations have been made that factoring by items is acceptable but by parcels not. Theoretical reasons are given for expecting that factors should actually be the same in number and nature from items and parcels, and that, moreover, the latter will give more accurately determined unique simple structure resolution. From 184 items found in several researches to cover some 19 to 20 primary factors, 46 parcels of 4 in each were made by radial parcelling. Factorings using blind rotation, employing both items and parcels on each of two groups (N = 246 and 780) independently showed close agreement on number of factors and in obtaining significant matching indices for each factor (a) among the four solutions and (b) with the ideal criterion pattern. As hypothesized, the advantages in precision of rotation turn out to be with the parcelling method.