Accuracy of Person-Fit Statistics

Abstract

Using a Monte Carlo experimental design, this research examined the relationship between answer patterns’ aberrance rates and person-fit statistics (PFS) accuracy. It was observed that as the aberrance rate increased, the detection rates of PFS also increased until, in some situations, a peak was reached and then the detection rates of PFS decreased with increases in aberrance rates. Furthermore, the results suggest that ECI2Z was somewhat more robust to high levels of aberrance than lz, HT, and U3 when cheating was simulated. The results of this study shed light on a limitation of PFS analysis.