The PCM2 Sampler - Part I

Beschreibung

Abstract: The Rasch sampler is a method (and a computer program) to sample uniformly binary matrices with given marginals. It forms the basis for constructing statistical models of goodness of fit for the Rasch model which can be applied to an arbitrary (univariate or multivariate) statistic. The PCM2
sampler is a generalization of this method to the partial credit model (PCM) with items with 0-1-2 scoring. In this model the sum score of the partial credits is sufficient for the person parameter and the frequencies of the 0-1-2 scores obtained per item are sufficient for the item category parameters. The PCM2 sampler is a method to sample uniformly 0-1-2 matrices with fixed sufficient statistics for all PCM-parameters.

In the first part of the contribution, the method - which is much more complicated than in the Rasch sampler - will be discussed. It could be formulated after a very illuminating working weekend the authors had with the late Reinhold Hatzinger in June 2012. It will also be shown how the PCM2 sampler can be used to generalize the Rasch sampler to incomplete designs of any kind.

Zeitraum	14 Feb. 2013 → 15 Feb. 2013
Ereignistitel	International Workshop on Psychometric Computing
Veranstaltungstyp	Keine Angaben
Bekanntheitsgrad	International