An Assessment of Performance Differentials by Gender and Educational Level in ELI Placement Test

Abstract

This study attempts to examine test fairness in terms of sub-group bias (gender and educational level) and validity of ELI placement test. Four components of the test including academic listening, dictation, cloze test, reading comprehension were analyzed through item difficulty (ID), or item facility differential (IF), and item functioning (DIF). The results revealed no evidence of gender bias in the ELI placement test, whereas significant performance differentials between undergraduate and graduate examinees were observed. Nevertheless, it was thought that educational difference is not likely to be the results of testing bias but rather variations in constructs under measured. The overall reliability of the ELI placement test is respectably high.