ABSTRACT
Shortly after I arrived on the faculty of the University of Illinois in 1952, I joined Raymond B. Cattell in research on personality structure in children. Cattell was a pioneer in the use of factor analysis as a method for defining the “elements” of which the conglomerate human “personality” was composed. He was an internationally recognized expert in the conceptual and mathematical foundations of the method and taught me everything I knew about the techniques. We published a series of research re-ports with which I was, at the time, entirely content. Later, how-ever, as I followed another line of research on children’s behavior disorders, I found Cattell’s methodology to be less dependable than it had seemed during our collaboration. I discovered that I could obtain more decisively replicable results by a simple but radical modification of Cattell’s procedure. I reanalyzed the data Cattell and I had collected jointly, using my revised technique, and found the same improvement in structural definition for those data that I had found in analyzing my own. I published the findings in the Psychological Review in an article entitled “Scope and Generality of Verbally Defined Personality Factors.” The abstract read as follows:
134Factor analyses of verbal personality measures have typically generated highly complex multidimensional structural systems. Available evidence now suggests that the most dependable dimensions drawn from conventional factor analyses of ratings and questionnaires are simple, familiar dimensions of broad semantic scope. It also appears that most of the initially obscure, apparently more precise, more narrowly defined factors many investigators claim to have revealed are either trivial, artifactual, capricious, or all 3. Verbal descriptions of personality were reduced to 2 factors, and the 2 factors were reduced to 2 ratings, 1 concerning perceived adjustment and the other related to introversion-extraversion. Convergent and discriminant validities for the 2 simple ratings are evidently equal or superior to those for any of the more cumbersome and expensive questionnaires examined in the review. Implications for theory and method were discussed.
