A. A. Maslak, S. I. Moiseyev, S. A. Osipov, “Comparative analysis of Rasch model parameters estimates obtained by maximum likelihood and least squares methods”, Probl. Upr., 2015, Issue 5,Pages <nobr>58

Control in the socio-economic systems

Comparative analysis of Rasch model parameters estimates obtained by maximum likelihood and least squares methods

A. A. Maslak^a, S. I. Moiseyev^b, S. A. Osipov^a

^a Kuban State University, Slavyansk-on-Kuban Branch, Slavyansk-on-Kuban
^b Institute of Management, Marketing and Finance, Voronezh

Abstract: Derived and investigated in a simulation experiment are statistical characteristics of Rasch model parameters' estimates, obtained by using maximum likelihood and least squares methods. Comparative analysis of these two sets of estimates was carried out, showing that estimates, obtained by using least squares method, are closer to data generated. Specificity of estimates obtained by both ways is considered.
Missing data influence on precision of Rasch dichotomous model parameters calculation is determined.

Keywords: latent variable, measurement, Rasch model, precision of estimates, maximum likelihood, least squares method.

UDC: 378:001.891