دراسات: العلوم التربوية (Sep 2021)
The Implication of Wald Test Anchor-All-Test-All Procedure for Anchor Selection
Abstract
This study aims to provide information on whether the more practical and time saver anchor-all-test-all procedure (AATA) strategy for anchor selection can demonstrate similar performance or outperform the well known all-others-as-anchors (AOAA) procedure under certain conditions. FlexMIRT-3 and IRTLRDIF-2 was utilized for Wald’s x 2 AATA and IRT-LR AOAA. All the parameters were constrained between the groups equally for the estimation of focal group distribution. These parameters were estimated through a full model that confined the focal group mean and SD values from the baseline model. The X2 statistics have been used to evaluate the differences between two sets of item parameters of two different groups. DIF tests results were dependent on the anchor selection strategies employed. The results have revealed that appropriate anchor-selection not only depends on the sample size, but it has an association with the proportion of DIF items and the direction of DIF along with the length of the anchor. Test depicts the efficacy of the Wald method for the DIF. This study suggests future researchers compare the different modeling approaches for the detection of DIF methods such as multiple indicators multiple causes (MIMIC) modeling following variant simulation techniques. In addition, future studies are recommended to analyze the efficiency of the anchor selection strategies concerning different groups and times
