أثر اختلاف طريقة المعادلة و طرق تقدير الدرجات و قواعد صياغة الفقرات على دقة تقدير معالم الفقرات و قدرات الأفراد في ضوء القياس الكلاسيكي و النموذج اللوجستي ثلاثي البارامتر

Abstract EN

The study aimed at investigating the effect of different functioning Method (Mean & Mean Method, Mean & Sigma Method), Methods of Scoring (the conventional method, the experimental Method and the method of probability assigned to the correct answer), and the rules of crafting items (the well-structured test, the ill structured test) on the accuracy of estimating the parameters of items and the abilities of individuals in the light of Classical Measurement and the threeParameter Logistic Model.

The sample of the study consisted of (1500) male and female students aging from (20.4-21.7) years, from the faculty of Education, at Umm Al-Qura University, who have been chosen stratified randomly.

In order to achieve the aims of this study and to answer its questions, the researcher prepared two test modules for the course of tests and measurements of multiple choices type with four alternatives.

Data were analyzed through using (SPSS 22, XCalibre 4.1.7, IRTEQ).

The results indicated differences between classical theory and three parameters logistic model.

The Classical perspective: The difficulty and discrimination mean of well-structured test items was higher than the difficulty and discrimination mean of ill structured test items.

Three parameters logistic model perspective: The well-structured test is more efficient and effective than the violated test at different ability levels.

The well-structured test is more accurate in estimating the parameters of individuals than the violated test, and the item analysis in the light of item response theory is better than classical theory of the test regarding parameter difficulty, discrimination and guessing.

The Conventional method was the most related method to the three parameters logistic model among the other classical methods in estimating the abilities of the students, the difficulty and discrimination of items, then the experimental method, followed by the method of proposed probability of the answer.

The results showed that bias values and root mean square errors decreased with the increase of sample size and test duration.

The bigger sample size and the longer test, the more accurate the equation becomes.

for the effect of three parameter model, in light of bias simulation and Root Mean Square Error, Mean & Mean Method is considered better than Mean & Sigma method in equating test score.