A comparison of an actuarial and a linear model for predicting organizational behavior

Abstract

Using an actuarial and a linear model for predicting organizational behavior, employee subgroups were identified through a hierarchical and convergent clustering of assessment variable profiles in a validation sample (N = 2,899) and cross-validated by assigning a holdout sample (N = 2,899) to the original subgroups on the basis of a minimum distance qualifier. Subgroup membership in both samples was significantly associated with current employment status and job performance. A linear discriminant function analysis of employment status and a linear regression analysis of job performance also yielded significant results. A comparison of the two models in terms of predictive accuracy indicated that the two models were essentially equivalent. However, it was concluded that the actuarial model was superior to the linear model, since a descriptive and behavioral taxonomy based on stable, homogeneous employee subgroups could be developed.