The application of IRT Polytomous models in scoring high-stakes tests (Case of study: lawyer's license test)
The aim of this study was to compare the accuracy and measurement error of dichotomous and Polytomous IRT models in scoring high-stakes, large-scale ability tests.
The statistical population of this study was included all the participants of the lawyer's license external tests in 2016 and 2018, from which 5000 persons and 5000 persons respectively were selected by random sampling. In addition, data collection was done using the responses of the participants of the above exam. Accordingly, the research method is experimental.
The analysis of the findings showed that among the dichotomous IRT logistic models, the 3-parameter model, and among the nominal Polytomous models studied, the 3-parameter model are a better fits and information compared with other models on the data under study.
Considering the more favorable fit and the level of information of the 3-parameter dichotomous model and the 3-parameter Polytomous model compared with other models, the use of these models in scoring can increase the accuracy of measurement and reduce the error. In addition, the use of these models also helps the fairness of the selection process of the applicants for the lawyer's license exam.
- حق عضویت دریافتی صرف حمایت از نشریات عضو و نگهداری، تکمیل و توسعه مگیران میشود.
- پرداخت حق اشتراک و دانلود مقالات اجازه بازنشر آن در سایر رسانههای چاپی و دیجیتال را به کاربر نمیدهد.