In two studies, the General Factor of Personality (GFP) remained intact after controlling for the Lie scale from the Eysenck Personality Questionnaire, despite the Lie Scale showing significant correlations with the GFP defining traits. In Study 1, a re-analysis of 29 self-ratings from 322 pairs of twins (644 individuals) yielded a GFP both before and after controlling for social desirability. In Study 2, four measures of affect in 133 university students loaded on a GFP both before and after controlling for social desirability such that those high on the GFP were high in self-esteem and positive affect and low in depression and negative affect. These results join those from other studies failing to find evidence that the GFP is merely an artifact of evaluative bias.

This commentary expands on issues raised by Delis, Jacobson, Bondi, Hamilton, and Salmon, in their paper on the use of shared variance techniques to establish construct validity. Significant discussion is focused on method variance, and how this can distort the results of factor analysis. Solutions are offered for the appropriate use of factor analysis in construct validation. Examples are also provided of construct validation procedures that do not rely on correlational or shared variance techniques. (JINS, 2003, 9, 947–953.)