Comparison of test equating methods based on classical test theory and Item Response Theory in polytomously scored tests

Test equating is a statistical process that is carried out to use two forms interchangeably which measure same latent trait and have similiar difficulty levels. This work is a stochastic simulation study aiming to compare test equating methods based on Classical Test Theory (CTT) and Item Response Theory (IRT) in tests consisting of polytomously scored items. It was aimed to compare the error, the sample sizes (50, 100, 300, 1000 and 3000 people) and the test length (10, 20, 30, 40 and 50 items) that were obtained from three CTT methods (linear, mean and equipercentile equating) and three IRT methods (mean-mean, mean-sigma and test chararacteristic curve). In this study, simulative data compatible with Generalized Partial Credit Model (GPCM) scored as 0-1-2 were used. In producting the data of the study, WinGen3 programme was used, to Equate this data sets, it was utilized from the R programme. At the end of the research, it was determined that the method which equates with the least error in equating test with different sample sizes and test lengths is equipercentile equating for test conformity to Generalized Partial Credit Model. This finding is also supported by the literature. The methods that give the highest equating errors in the research are the mean-sigma and test characteristic curve methods. When the test errors were analyzed based on sample size, it was found that as the sample sized increased, the error decreased. When the test equating methods were compared according to the test lenght variable, it was found that as the amount of item increased, the calculated error also increased.