Estimating true short-term consistency in vocational interests: A longitudinal SEM approach

This study aimed at estimating the correlation between true scores (true consistency) of vocational interest over a short time span in a sample of 1089 adults. Participants were administered 54 items assessing vocational, family, and leisure interests twice over a 1-month period. Responses were analyzed with a multitrait (MT) model, which supposes no temporal change, and a latent change (LC) model with temporally stable method effects. The LC model fitted the data well in 75% of cases whereas the MT never held. Error measurement is not sufficient in explaining the imperfect test-retest manifest correlations. True consistency estimates were very high with an average of .87, suggesting that although true temporal change occurs even within short periods its magnitude is rather limited.