Abstract
Comparison of strategies of the data fit to the partial credit model. A comparative study of the results provided by two strategies for fitting data to Latent Trait Theory Models has been performed. The first, called Total-Subjects-Items (TSI), is structured in three phases: 1) assessment of item fit, 2) assessment of subject fit; and finally, 3) overall fit of data to the models (items and persons). The second strategy, the Total-Items-Subjects (TIS), changes the order of the phases: 1) assessment of subject fit, 2) assessment of item fit and, 3) overall fit of data to the models. To verify the results of these two strategies, a set of 30 items, designed to measure religious attitude, was administered to a sample of 821 persons. The Latent Trait Theory Model used were the Partial Credit Model. The results underline an important difference between the two procedures: the TSI maximizes the number of persons with good fit and the TIS maximizes the number of items with good fit. Moreover, a procedure for controlling the sensitivity of fit to sample size is proposed.