Abstract
A computer program for analysing test data according to a latent trait model which allows for dependencies among responses is described briefly. The model allows for dependencies among a subset of dichotomously scored items through a dispersion parameter associated with the subtest. The smaller the value for this parameter, the greater the dependence. In addition to the estimates of the difficulty and dependence parameters of subtests and the estimates of the abilities of persons, the program provides two indices of fit for each subtest and an index of fit for the response pattern of each person. It also furnishes an index of the degree of reliable separation among estimates of the abilities of persons in the sample analysed which is similar in value and interpretation to Cronbach's α.
Get full access to this article
View all access options for this article.