One Method to Check the Population Homogeneity of a Test
In this paper it is considered a method for checking the population homogeneity of a given psychological or didactical test with respect to the way of its accepting from different populations. For this purpose we look for a statistical association between the a priori known populations and posterior estimated clusters on the base of experimental data. The lack of such statistically significant association is considered as an evidence for the absence of a discrimination and therefore as a strong indicator for homogeneity. The posterior clusters are searched by means of the Generalized Partial Credit Model from the Item Response Theory.
Природни науки, математика и информатика
Natural sciences, mathematics and informatics