Measurement equivalence of Watson and Clark's 18 surrogate measures of negative affectivity: applying the principle of tetrad differences to an incomplete correlation matrix

Watson and Clark concluded, based primarily on a subjective analysis of the positive manifold of 87 intercorrelations, that 18 personality measures were manifestations of the same stable and pervasive trait of negative affectivity. Confirmatory factor analysis to test a one-factor measurement model was not possible, since only an incomplete matrix of correlations (87 out of 153) was available. In this paper, the principle of tetrad differences was applied to an incomplete correlation matrix to test for a general factor across the 18 measures. This principle assesses the equivalence of different measures hypothesized to tap into the same construct. Of the 614 first-order partial correlations, only 44 (7.2 %) failed to satisfy the principle of tetrad differences, indicating support for a common underlying trait. In addition, confirmatory factor analysis and Herman's one-factor test provided support for a general factor across a subset of the 18 scales for which a complete matrix of correlations was available.