Please use this identifier to cite or link to this item:
An Examination of the MIMIC Method for Detecting DIF and Comparison to the IRT Likelihood Ratio and Wald Tests.
|Title:||An Examination of the MIMIC Method for Detecting DIF and Comparison to the IRT Likelihood Ratio and Wald Tests.|
|Authors:||Zimbra, Daniel J.|
|Contributors:||Educational Psychology (department)|
|Date Issued:||May 2018|
|Publisher:||University of Hawaiʻi at Mānoa|
|Abstract:||Differential item functioning (DIF) detection research has found the multiple indicators multiple causes (MIMIC) structural equation model (SEM) to be effective in detecting uniform DIF. Recent advances in the MIMIC method have also allowed for the detection of nonuniform DIF. However, few researchers have evaluated its performance, or compared it with the established DIF detection methods. The current study addresses this gap in the existing research by evaluating the MIMIC method in detecting uniform and nonuniform DIF, and comparing its performance to the established item response theory (IRT) based likelihood-ratio (IRTLR) and Wald tests. Monte Carlo simulations of tests and item responses were conducted, manipulating the number of examinees, type of DIF, magnitude of DIF, and proportion of contamination. The simulation results indicate that the MIMIC method outperformed the IRTLR and Wald tests based on Type I error and power rates when testing for a large magnitude of nonuniform DIF and contamination at 20%, regardless of sample size. When the proportion of DIF contamination rose to 40%, the Wald test outperformed IRTLR and MIMIC methods in all other experimental settings. IRTLR was the only method that was able to maintain well-controlled Type I error rates throughout the experimentation and adequate power when the magnitude of DIF was large. While the IRTLR method generally outperformed the others, the MIMIC method was particularly strong at detecting nonuniform DIF, and the Wald test performed well when the proportion of DIF contamination was high. The findings of this study inform future research and practice in the appropriate selection of DIF method.|
|Description:||Ph.D. Thesis. University of Hawaiʻi at Mānoa 2018.|
|Rights:||All UHM dissertations and theses are protected by copyright. They may be viewed from this source for any purpose, but reproduction or distribution in any format is prohibited without written permission from the copyright owner.|
|Appears in Collections:||
Ph.D. - Educational Psychology|
Please email email@example.com if you need this content in ADA-compliant format.