Head-to-head comparison of the predictive validity of personality types and dimensions

The utility of a configural type approach for predictions from personality is currently controversial. Configural types predict important personality correlates, but continuous dimensions based on the same data often fare much better in cross‐sectional head‐to‐head comparisons. However, many such comparisons can be considered unfair to the type approach, confound diverse differences between type and dimensional approaches, and rely only on cross‐sectional data. A sequence of analyses is reported that include fairer comparisons and deconfound differences due to the number of predictors, categorical versus dimensional predictors, dichotomized dimensions versus configural types, dimensional versus type criterion variables, and cross‐sectional versus longitudinal predictions. The results suggest incremental validity of configural types over dimensions only in a very few cases. Copyright © 2003 John Wiley & Sons, Ltd.