TARNAI, CH. & ROST, J. (1990). Identifying Aberrant Response Patterns in the Rasch Model: The Q Index. Sozialwissenschaftliche Forschungsdokumentationen 1. Münster: Institut für sozialwissenschaftliche Forschung. ISBN 3-926083-02-6. Preis --,-- €.
Aus der Einleitung
In recent years, there has been a growing
interest in methods and indices for detecting aberrant response behavior of single individuals in
a sample of examinees or respondents of a questionnaire. In fact in all applications of tests or
questionnaires there are some individuals showing an "unusual" or "unexpected" response pattern,
e.g. solving some difficult items while failing the easier ones, or in general, showing a
response pattern which does not fit with the profile of itemparameters at all.
However, some of those people must belong to each data set fitted by a probabilistic
latent trait model. In probabilistic models even the "most aberrant" response pattern has a (very low)
probability of occurence. But in a realistic setting, where a probabilistic model does not fit "perfectly",
the identification of unusual response patterns may be of great interest for basically two reasons:
First, identifying and selecting misfitting persons can be used as a mean for "reaching a better fit"
of the model to the data. This is the complementary strategy to the usual "item - selection" strategy and
may be regarded as even legitimate as the latter.
Second, the identification of unusual response patterns may be diagnostically relevant,
because it puts the finger on those persons, where the response pattern provides more information
than the "number of correct" responses.
Bezug der Studie direkt e-mail: email@example.com