Detection of the effect of the Percentage of Items with DIF on the strength and efficacy of the singular characteristic curve test in the three-parameter model

Abstract

The current research is following the simulation studies in IRT to experiment with the effect and power of the ICC's method on DIF items. To achieve the experiment, data were generated for three tests consisting of 100 items to each one containing (10% -20%-30%) items with DIF. These three tests were applied through the simulation program to 3000 individuals with normal abilities are distributed in average (0) and standard deviation (1). The results of the study indicated the effectiveness of the ICC's test by using the Rajo equation (1988), which measuring the area below the curve between the Reference groups and the Focal groups in detection items of (10% -20% -30%) amount of DIF and counting them in each of the three tests. The results showed that the strength of the test was (0, 619) for the DIF of 10% and the test strength of the DIF of 20% is amount (0.703) and the strength of the test was (0786) for the DIF of 30%, all of Power Values are higher than the average rating by Cohen, D test to derive the size of the impact. And we recommend taking the results in instruments containing differential items.