The Diferential Functioning of Items and Tests (DFIT) framework is a widely used test of di erential functioning. DFIT uses the Item Parameter... Show moreThe Diferential Functioning of Items and Tests (DFIT) framework is a widely used test of di erential functioning. DFIT uses the Item Parameter Replication (IPR) algorithm to generate statistical cuto s, which relies on several simpli cations have not been formally tested. This study examines these assumptions by comparing the IPR algorithm with an extended, unsimpli ed procedure intended to represent results observed in practice. Monte Carlo simulation was used to compare cuto s between the two methods under a number of conditions. Di erences between the IPR and extended cuto s indicate that Type I error rates are conservative at the test level and in ated at the item level. Mean group di erences resulted in massive in ation. Sample size and test length had only minor inuences on error rates. M.S. in Psychology, December 2013 Show less
