تقصي أثر عدد خطوات الأسئلة متعددة التدريج و شكل التوزيع لصعوبتها على تقديرات القدرة للأفراد، و الصعوبة للأسئلة، و دالة المعلومات للاختبار وفق نموذج التقدير الجزئي

Abstract EN

The study aims to investigate the effect of number of steps and difficulty distribution of test questions on the accuracy of persons, ability, questions difficulty parameters estimation, and Information function of the test according to Partial Credit Model (PCM).

To achieve the objectives of the study, data from three 30-question tests with varying number of steps for every question, (three steps, four steps, and five steps) were developed using WINGEN program.

The degree of difficulty between first and last steps in those tests was equal for all tests according to three different distributions related to difficulty of questions : normal, positively skewed, and negatively skewed.

Also, data related to responses from 1000 persons were generated.

Data were analyzed according to Partial Credit Model and SPSS program.

The results showed statistically significant differences in the standard errors of means for persons’ ability estimations attributed to the number of test steps and the interaction between the number of test steps and the shape of distribution, and the results also showed no statistically significant differences attributed to the shape of distribution.

Moreover, the results showed statistically significant differences in the standard errors of means for difficulty parameters estimation attributed to the number of test steps and the interaction between the number of test steps and the shape of distribution.

No statistically significant differences, attributed to the shape of distribution, was shown.

In addition, no statistically significant differences were found among estimated persons reliability coefficients, attributed to the number of test steps or at different shapes of distributions.

The results indicated that the five-step test provided the largest amount of information under different shapes of distributions (normal, positively skewed, and negatively skewed), then the four-step test.

Finally, the relative efficiency of the five-step test was higher than that of the three-step test.

In other words, the five-step test was more accurate in estimating persons abilities.