Comparison of Validity, Accuracy, and Reliability of the Scores obtained from Thurstonian Item Response Theory in forced-choice scales

The main purpose of this study was to identify the optimum approach to analysis of epistemic data in measuring personality traits. For this purpose, MPLUS software and lavan software package simulate different formats of response scales, and two traditional scoring approaches and Thurston's IRT method were compared in some of the most important psychometric outcomes. Overall the findings of the study indicated that the Thurstonin IRT approach performs better than the traditional scoring approach in accurately estimating actual scores, criterion validity, locating a person on the attribute continuum. Also, the research findings showed that the average correlation between estimated scores and actual scores for all three correlation coefficients (0, 0.25 and 0.50) in Thurstonian Item Response Theory is higher than the scores obtained from the classical method; True for all three correlation coefficients, in Thurstonian Item Response Theory more than classical method scores and mean experimental reliability scores for correlation coefficients 0.25 and 0.5 with classical method are higher than scores obtained by Thurstonian IRT method, but for coefficient The mean correlation of the mean scores of the Turestonian IRT method is higher than the classical scores. The findings also showed that the difference between the two tests, both Classical and Trestonium IRT, was significant for both the 48 and 96 questions, and the correlation between estimated scores and actual scores in the Thursetonium IRT method was more than the classical method.