Invariability of Item Parameters in the Sample with Restricted Real Data Results

Abstract

There are two basic theories used in testing and item analysis in education and psychology, and they are called as the Classical Test Theory (CTT) and the Item Response Theory (IRT). Both theories have advantages and disadvantages changing according to the situation in which they are used. The most important advantage of the IRT, and its superiority to the CTT is that in the CTT the item parameters are based on sample and the ability parameters are based on the test. There have been many studies conducted to determine whether the IRT provides this advantage or not. However, in these studies, many different results have been reported. The item parameter invariance being multidirectional shows that continuing these types of studies is necessary. This study is based on this fact, and has been designed as a simulation study in which the distribution and parameter characteristics of the real data are used as restrictions. The number of the items have been kept stable, the sample size, the restrictions of distribution and item parameters, and the parameter prediction models have been taken as the simulation models. According to the results obtained, the restriction of the parameters and increasing the sample size have been determined as important factors in ensuring the parameter invariability.