DGPS-Berlin, 2002 abstract; Dr. J.M. Müller

The general meaning of Rasch variances as an indicator of variability of persons

Variances of raw scores of different measurements are yet not comparable, because of the lack of an test independent scale unit.

The recommended scale unit could be derived for metric constructs within the Rasch model. This opens the opportunity to compare Rasch variances of person parameters. Rasch variances have a unique psychological meaning, while they serve as a comparable measure for the variability of person parameters referring to a measured dimension. Starting with an introduction and explanation about test independent scale units for a scale of person parameters empirical results concerning different Rasch variances from several tests are given. Guided by an understanding of content validity Klauer (1984) defined of an item universe, leads to the hypothesis of comparable Rasch variances for measures that complain to measure the same psychological content. Beside this new opportunity to proof content validity, additional practical implications for diagnostic reasoning are given. Within an example, Rasch scale units can serve as a standardised distance unit from the mean of a population. The interpretation of a single test score could use not only the information about the relative position within a distribution, but also the new derived Rasch scale unit. Practical consideration about assumption and limitations conclude the topic.

Topic:  6.3 Test-, Measure-, Scaling Theory