Anomalous type I error rates for identifying one type of differential item functioning in the presence of the other

A number of statistical methods exist for the detection of differential item functioning (DIF). The performance of DIF methods has been widely studied and generally found to be effective in the detection of both uniform and nonuniform DIF. Anecdotal reports suggest that these techniques may too often incorrectly detect the presence of one type of DIF in the presence of the other type (Type I error). The purposes of this simulation study are to ascertain whether these observations are in fact accurate and, if so, to gain some understanding as to the cause of the inflated Type I error. Results do Support that the Type I error rates for detecting one type of DIF in the presence of the other are inflated for most common DIF detection techniques. Discussion focuses on potential causes of these results.