الملخص الإنجليزي
The current study aimed to investigate the effect of using person fit indices (WSR,
USR, UT, WT, and LZ) to exclude individuals with misfit response pattern on estimating item
and abilitis parameters in Item Response Theory. The study used data of the numerical ability
test in the Gulf Scale for Multiple Mental Abilities. The study sample was 4206 students from
all Gulf countries. Appropriate statisical methods were used to verify the assumption of the
1PLM, 2PLM, and 3PLM. The study used three samples of different size (1000, 2000, and
4206). Items and ability parameters were compare before and after exclusion of misfit response
pattern according to the indices used in this study.
The results of the study indicated the limited change in the relative status of estimates
of item parameters and the abilities across the different conditions of the study. They also
showed the effect of deleting misfit responses patterns on the estimates of items parameters
according to the item response models. The effect was more evident the estimates of difficulty
parameter in 1PLM, and in discrimination parameter in 2PLM, and in guessing parameter in
3PLM. The results also showed that the effect of deleting misfit response patterns according to
the all person fit indices was negligible to ability parameters.
In light of the findings of the study, the study recommends the need for studying the changes
in estimates of items parameters in different item response models. It also recommends
conducting more studies on the effect of deleting misfit response patterns on the accuracy of
the estimates of item and ability parameters in other test conditions such a test length.
