A latent state-trait model (not only) for social desirability

Abstract

A latent state-trait model for social desirability is proposed, that—aside from the trait to be measured—takes into account method factors as well as systematic effects of the situation of measurement and the person-situation interaction. Parallel test halves of a German version of the Marlowe-Crowne Scale were administered to two independent samples: Sample 1, N 1 = 380, measured twice with a 6-month lag; and Sample 2, N 2 = 215, measured three times with 9-month lags. The traditional latent trait model and the latent state-trait model with method factors were tested against each other. As expected, the former fit the data poorly, the latter very well. The common trait explained 58% of the variance of the test halves, the test half-specific or method factors 17%, and the occasion-specific situation and interaction effects a small but significant 6%. The latent state-trait model enables a more adequate representation of the social desirability construct, and the model is applicable to a wide array of personality phenomena.