أثر حجم العينة على دقة تقدير خصائص المفردة و القدرة في اختبار التنمية المعرفية في مادة العلوم لطلبة الصف السابع بسلطنة عمان

الملخص EN

The study aims at investigating the effect of sample size on the accuracy of item parameter and ability parameter estimation for cognitive development test in science for seventh grade in Sultanate of Oman using Rasch model.

The sample consists of 8484 students from seventh grade who are taken the Cognitive Development Test in the academic year of 2016/2017.

From this sample 12 subsamples were randomly selected with replacement with 4 different sample sizes (200, 500, 1000, 1500).

In each of these four sample size, three replicated samples were randomly selected consisting the 12 subsamples.

All data sets were caliberated with Multilog7.0 software using Rasch model to estimate item difficulty, person ability, and test information using Marginal Maximum Likelihhod estimation method.

To evaluate the estimation accuracy of these three types of parameters, root mean square error (RMSE) was used to compare the estimation of the three types of parameters from each subsamples to the corresponding estimation from the large sample (treated here as a true value).

Results showed that sample size influenced the estimation accuracy of item and ability parameters, and test information.

The accuracy improved with increasing sample size.

In addition, a sample size of 500 was found as sufficient sample size to obtain accepatable estimation accuracy for the three types of parameters with Rasch model.