The MIMIC Method With Scale Purification for Detecting Differential Item Functioning

Abstract

This study implements a scale purification procedure onto the standard MIMIC method for differential item functioning (DIF) detection and assesses its performance through a series of simulations. It is found that the MIMIC method with scale purification (denoted as M-SP) outperforms the standard MIMIC method (denoted as M-ST) in controlling false-positive rates and yielding higher true-positive rates. Only when the DIF pattern is balanced between groups or when there is a small percentage of DIF items in the test does M-ST perform as appropriately as M-SP. Moreover, both methods yield a higher true-positive rate under the two-parameter logistic model than under the three-parameter model. M-SP is preferable to M-ST, because DIF patterns in real tests are unlikely to be perfectly balanced and the percentages of DIF items may not be small.

MIMIC Methods for Assessing Differential Item Functioning in Polytomous Items

“…As expected, the major findings are consistent with those found in dichotomous items Wang et al, 2009). The strategy is supported, as demonstrated in the three simulation studies.…”

“…To accomplish this, many scale purification procedures have been developed and widely incorporated in DIF assessment methods (Clauser, Mazor, & Hambleton, 1993;French & Maller, 2007;Wang & Su, 2004a, 2004b. Following the principle of scale purification, Wang et al (2009) implemented a scale purification procedure on M-ST, which was called the MIMIC method with scale purification (denoted as M-SP; detailed steps are shown below), and in a series of simulations found that both M-ST and M-SP maintain a well-controlled Type I error rate when tests do not contain DIF items; M-SP outperforms M-ST in controlling the Type I error rate and yielding a higher power of DIF detection when tests contain DIF items, but unfortunately, M-SP begins to yield an inflated Type I error rate and a deflated power when there are 20% or more DIF items in the test. That is, even M-SP cannot guarantee an expected Type I error rate and a high power.…”

Detecting Multidimensional Differential Item Functioning with the Multiple Indicators Multiple Causes Model, the Item Response Theory Likelihood Ratio Test, and Logistic Regression

“…However, a carefully designed purification procedure needs to be the first step for identifying potential DIF items when conducting DIF analyses with real data. In the literature, different anchor purification methods have been suggested to select DIF-free items for different DIF detection approaches (e.g., French and Maller, 2007;Wang et al, 2009;Woods, 2009b;Gonzalez-Betanzos and Abad, 2012). Depending on the selection of DIF-free items (i.e., purification), the DIF detection methods may provide different results regarding the number and type of detected DIF items.…”

DIF Analysis with Multilevel Data: A Simulation Study Using the Latent Variable Approach

“…In practice, it is possible that multiple items can be DIF-present within a short test. Although item purification procedures can be conducted in advance (Wang, Shih, & Yang, 2009) and the purified covariate (i.e., total score computed after excluding DIF-present items) can be used in HLR-LC to match θ between groups, with tests being short, the purified covariate might be difficult to cover a wide range of θ. One possible solution is to use multiple indicators multiple causes (MIMIC) model, which is robust against DIF contamination (Finch, 2005).…”