Dissertation - Testkennwerte psychologischer Verfahren

Abstract

The work "Kennwerte psychologischer Testverfahren" contains an introduction and three articles. The introduction clarifies the meaning and use of coefficients to evaluate the quality of a test, such as reliability or validity. The introduction closes with five new coefficients to describe different aspects of a test: "Ausdehnung", "Ausschöpfung", "Differenziertheit", "Mess-Effizienz" and "Personenunterscheidungsvermögen". The first article concerns criteria of usability, which help to distinguish a poorly understandable coefficient from a good one. The coefficient "Differenziertheit" is based on the critical difference of test scores and shows how many of these differences fit on the whole scale. The second article proposes the term "Personenunterscheidungsvermögen" (PUV). This index allows one to estimate the expectation (probability) of (statistically) different scores. One advantage of this coefficient is that it is independent of the test theoretical foundation of psychometric assessment, which makes it possible to use the same coefficient for categorical and metric constructs. This index as well fulfils most of the criteria of usability. The last article presumes differences in the variability of a psychological dimension and shows how to display these differences. For this purpose, a new kind of standardization based on the differences of probabilities is shown which can be derived from the application of a Rasch scale. This application, to measure the spreedness (Ausdehnung) of a psychological dimension, leads to a new interpretation of a Rasch scale. This article is in press in the journal "Zeitschrift für Differentielle und Persönlichkeitspsychologie". This new coefficient offers many new possibilities, such as testing construct validity or choosing those dimensions for prediction that show bigger differences between persons.

Vollversion