Personality and factor analysis: A reply to Guilford.

Consideration is given to the problem of how we are to arrive at personality factors that possess some demonstrable degree of representativ eness. It is suggested that psychometric considerations play an important part and that factor analysis in particular can be of great value in this connection. It is also suggested that there are other considerations of at least equal importance. Factors emerging from such analyses must be replicable and reliable, and they must fulfill certain basic psychometric requirements. In addition, there should be evidence of their heritability, they should have theoretical backing leading to objective laboratory verification of deductions from such theories, and they should be socially relevant in the sense of correlating significantly with social parameters. The evidence suggests that there are three and only three such factors emerging from relevant research, that these are all higher order factors, and that primary factors lack some or all of the qualities required. It is suggested that the models presented by Guilford, Cattell, and Eysenck converge on some such model as is here presented. A recent paper by Guilford (197S) expressed a point of view regarding the determination of personality dimensions by factor analysis that will probably be shared by most psychometrists. In this reply I do not wish to dwell on the many points of agreement (e.g., distrust of purely psychometric considerations in rotation, preference for orthogonality where possible, use of factor analysis in a hypothetico-deductive manner, disbelief in the meaningfulness of specifying factors as second-, third-, or higher order, selection of variables for analysis), but wish rather to discuss briefly two points of disagreement, one factual and dealing with matters largely internal to factor-analyt ic methodology and findings, the other external and dealing with the relation between factor analysis and other parts of psychology. In his section on Eysenck's Factors E and N [Extraversion and Neuroticism], Guilford (197S) correctly stated that the early work on the Maudsley Personality Inventory (MPI) and the Eysenck Personality