أثر حجم العينة و طرائق التقدير في دقة تقدير معالم نموذج راش

الملخص EN

The aim of this study was to investigate the effect of estimation methods (maximum likelihood, Beyes estimation, Proxy method) at different sample sizes (2000, 1500, 1000, 500) on the estimation of Item parameters and Ability Parameter and their accuracy in the light of the item response theory (Rush model).

To achieve the objectives of the study, a random sample of (2000) responses were generated on (40) single-response (0 or1) using Wingen software with difficulty and standard normal distribution, and then verified the assumptions of item response theory (Unidimensionalty, local independence, item characteristic curve) and to check the suitability of items for the Rush model, As a result, we got (31) items fits for the Rush model.

The results showed that: 1.

There are no statistically significant differences in the estimation of the item parameter of the Rush model (difficult) and Ability Parameter attributable to the sample size factor and the method of estimation and interaction between them.

2.

There are statistically significant differences in the accuracy of estimating the item parameters (difficult) due to the factor of sample size and the method of estimation and interaction between them.

3.

There are statistically significant differences in the accuracy of estimating the parameters of individuals (power) attributable to the sample size factor and the method of estimation and their interaction.

4.

The accuracy of the item parameter (difficult) and the accuracy of the ability parameter factor is more accurate as the sample size increases.

5.

There are no statistically significant differences between the maximum odds method and the Bayes method on the one hand and the Bayes method and Proxy method on the other.

6.

There are statistically significant differences in the accuracy of estimating the ability of individuals between the three methods.