Interpreting Rasch scale units as ‘levels of ability’
Jörg M. Müller, University of Tübingen (Germany)
Psychologisches Institut, Abteilung für Allgemeine und Angewandte Psychologie
Dr. Jörg M. Müller
Friedrichstrasse 21,
72072 Tübingen, Germany
Abstract:
Common standardization (i.e., sigma values, percent range, stanine, etc.) of raw scores transformation puts all distributions of different psychological dimensions into a comparable standard deviation. This leads to the (false) conclusion that persons differ on average more or less the same, independent of the dimension measured. This paper assumes there are differences in the variability of persons within a psychological dimension and shows how to display these differences. Such differences between personality dimensions are found in Rasch variances of the estimated person parameters. Rasch units can therefore be interpreted as ‘levels of ability’. An empirical reanalysis of norm data provides evidence for this approach. The foundation for the interpretation of Rasch variances is given and illustrated with examples. The Rasch variance is finally a characteristic of the trait (specifically hypothesis) and indifferent to the methodological approach (indifference hypothesis). These hypotheses are confirmed with data from different tests in various applications. The consequences and possibilities use in personality psychology and diagnostics are discussed.
Keywords: Personality, Item-Response-Theory, dispersion, Rasch
Type of contribution: paper