Nonparametric and Group-Based Person-Fit Statistics: a Validity Study and an Empirical Example

In person-fit analysis, the object is to investigate whether an item score pattern is improbable given the item score patterns of the other persons in the group or given what is expected on the basis of a test model. In this study, several existing group-based statistics to detect such improbable score patterns were investigated, along with the cut scores that have been proposed in the literature to classify an item score pattern as aberrant. Through a simulation study and an empirical study, the power of three person-fit statistics was compared, and the practical use of various cut scores was investigated. The empirical study involved 437 Dutch sophomores studying psychology and pedagogics taking an examination on test theory. It was also demonstrated that person-fit statistics can be used to detect persons with a deficiency of knowledge on an achievement test. While one of the statistics was less appropriate in the simulation, the power of the three approacnes was approximately the same in the empirical example. (Contains 9 tables and 31 references.) (SLD) *********************************************************************** Reproductions supplied by EDRS are the best that can be made from the original document. *********************************************************************** Nonparametric and Group-Based Person-Fit Research Statistics: A Validity Study Report and an Empirical Example 94-12 U.S. DEPARTMENT OF EDUCATION Office 01 EduCaloonal ReSeirCh and Irroprovemenl ECIUCA ONAL RESOURCES INFORMATION CENTER (ERIC; document nes been rebrottlucecl as febanved from the Person Or orgontlabon ongormtong It C Mono, Changes nave been made lo nnprove neoroduchon quauty Porms of vie* or optneons Stated on this deco rnnt 00 not necesSardy represent otficral OERI posmon or policy