أثر معالجة القيم المتطرفة في اختبارات تيمس الدولية (TIMSS, 2011)‎ على تقديرات المعالم للفقرات و دقتها وفقا لنظرية الاستجابة للفقرة

Abstract EN

This study aimed to identify the percent of Outlier values in TIMSS international tests (TIMSS, 2011) and the impact of the method of treatment in the students' scores at TIMSS test in accuracy of the estimation Items Parameters according to item response theory according used three–parameter logistic model, The sample consists of (531) students', and pointed out the Outlier values by the Box-plot method, it used the trimmed mean in replacing the Outlier values, and it used the standard error for accuracy of estimating, which showed there are Outlier value at Maths data percent of %6.40, Outlier values in Science percent of %8.29, The results showed that substantial differences between the means of standard errors for the Items parameters estimates, the method of deleting the Outlier values has a positive effect comparing with the method of (replacing and keeping) Outlier values, and it is adviced that it is the best and the most successful method of Treatment with the Outlier values in TIMSS 2011 in Maths on the mean estimated Items parameters (discrimination and guessing) and its accuracy, and in Science on the mean estimated discrimination parameter and its accuracy.

As the results showed that the method of keeping Outlier values has a positive effect comparing with the method of (replacing and deleting), it is advised that it is the best and the most successful method of keeping with the Outlier values in TIMSS at Science on the mean estimated Items parameters (guessing and difficulties) and its accuracy, at Maths on the mean estimated difficulties parameter and its accuracy.