Comparing five alternative factor-models of the Student Jenkins Activity Survey: separating the wheat from the chaff.

The Student Jenkins Activity Survey (SJAS; Glass, 1977) is the most popular instrument for measuring Type A behavior in college undergraduates, and a variety of different unidimensional and multidimensional models for scoring the SJAS have been employed in the past. This study compared five alternative factor-models developed for the short (21-item) form of the SJAS in terms of their overall goodness of fit, factorial invariance, and factor reliabilities, using the data of two independent samples of undergraduates. Indices of relative fit for the five models ranged from .64 to .94 (median = .85), but only one unidimensional model and one multidimensional model achieved a minimum satisfactory goodness of fit (.90 or greater) for both samples. Reconceptualizing these models within a broader nomological net sheds light on their meaning and highlights existing gaps in the measurement of multifaceted Type A behavior.