Search results

(1 - 2 of 2)
EVALUATION OF A MODIFIED ITEM PARAMETER REPLICATION METHOD FOR DIFFERENTIAL FUNCTIONING OF ITEMS AND TESTS ANALYSIS WITH UNEQUAL SAMPLE SIZES
Title
EVALUATION OF A MODIFIED ITEM PARAMETER REPLICATION METHOD FOR DIFFERENTIAL FUNCTIONING OF ITEMS AND TESTS ANALYSIS WITH UNEQUAL SAMPLE SIZES
Creator
Blitz, David L.
Date
2016, 2016-05
Description
In 1995, Raju, van der Linden, and Fleer introduced the Differential Functioning of Items and Tests (DFIT) framework. However, some concerns...
Show moreIn 1995, Raju, van der Linden, and Fleer introduced the Differential Functioning of Items and Tests (DFIT) framework. However, some concerns have been raised regarding the accuracy of DFIT (e.g., Meade & Lautenschlager, 2004, 2005). More recently, it was suggested that large differences in sample sizes might affect the sampling variance of the NCDIF statistic (e.g., Raju et al., 2009). The purpose of this study was to confirm if differing subgroup sample sizes affect the accuracy of the NCDIF statistic and to propose and evaluate a modification to solve this problem. Monte Carlo results indicated that the old method generally maintained fairly stable power, but tended to be overly conservative when the focal group was smaller than the reference group and exhibit inflated Type I error when the focal group was larger than the reference group. The new method generally maintained reasonable Type I error regardless of subgroup sample size and demonstrated comparable or better power except for conditions where the old method exhibited inflated Type I error rates. When impact was present Type I error rates were slightly higher and power was slightly lower but results otherwise conformed to the general pattern.
Ph.D. in Psychology, May 2016
Show less
Comparison of an Ideal Point and Dominance IRT Model on the Detection of Differential Item Functioning with DFIT
Title
Comparison of an Ideal Point and Dominance IRT Model on the Detection of Differential Item Functioning with DFIT
Creator
Spizzuco Jr, Daniel
Date
2019
Description
Item response theory (IRT) models can assume a variety of forms including,notably, dominance and ideal point-based probability distributions....
Show moreItem response theory (IRT) models can assume a variety of forms including,notably, dominance and ideal point-based probability distributions. But researchers haveonly recently begun to explore issues related to the above distinction. The current studytherefore examines whether model-data fit and rates of differential item functioning (DIF)detection remain comparable when data are analyzed via the ideal point-based generalizedgraded unfolding model (GGUM) vs. the dominance-based graded response model (GRM).To address these issues, item response data were simulated to contain dominance,ideal point and mixed response processes, and DIF and impact scenarios. Results indicatedthat model-data fit and DIF detection accuracy were not as closely aligned as anticipated.Overall, the GGUM fit data better than the GRM to the extent that any ideal point processeswere present, while the GRM was slightly better at fitting dominance-only data. With noimpact, however, the GGUM fit all embedded response data types better than the GRM.Results were mixed among impact scenarios. This pattern was found in both no DIF and DIFscenarios.Several points were made with respect to the DIF portion of the study. First, Type 1error rates were in most cases quite conservative for both models. Second, study-wide,more power emerged with dominance as compared to ideal point data for both models.Moreover, in no impact conditions, slightly more power accrued via the GGUM fordominance and ideal point data. With impact, however, the GRM produced somewhat morepower across data types. Third, in terms of DIF patterns/sources, power was high for bothmodels when DIF was embedded on the full set of location/threshold parameters, andlower with fewer differentially functioning (DF) location/threshold parameters. Notably,the GGUM was slightly more powerful in the fewest DF location/threshold scenarios, andthe GRM was more powerful in the most DF location/threshold scenarios. Fourth, neithermodel performed well in the complex within-item cancelling DIF scenarios. These patternsgenerally occurred in both uniform and non-uniform scenarios. The paper concludes with apresentation of recommendations, study limitations and issues for future research.
Show less