اللاتغير في تقديرات معالم فقرات الاختيار من متعدد عندما تطبق في سياقات مختلفة باستخدام نماذج النظرية الحديثة في القياس

الملخص EN

The study aims to study the Invariance of multiple-choice item parameters estimates when applied in different contexts.

To achieve the goals of the study, the researcher used a test consisting of (113) items multiple-choice type in mathematics in addition to (15) common items applied on a sample of 2511 students.

Using the suitable software to analyze data, the results of the analysis showed that the items calibration within all items of the test, when analyzed alone, leads to invariance in the items parameters estimate and this result achieved for the flowing items parameters : discrimination and difficulty, and guessing depending on the different model and contexts, except estimating a discrimination parameter depending on the different model, has been found.

There were significant differences in favor of the Three-parameter model compared with two-parameter model.

As shown by the values the correlation coefficient between the estimates of items parameters : discrimination and the difficulty and guessing that has been calibrated independently and within the all items of the test they were high and statistically significant as shown by the values correlation coefficients were high when it was calibrated item sample independently (alone).

This value can be reached from that we get better accuracy when calibrating a sample of items independently compared within calibrating within the all items of the test.

Based on the results of the study, it was recommended to use a representative sample of items with a few items in the tests of adaptive and further studies using a information function also recommends using the Three-parameter Model, when we calibrate the items and when we design and development the item banking.