НАН РА. Математические вопросы кибернетики и вычислительной техники=Mathematical problems of computer science

New Approach for Test Quality Evaluation Based on Shannon Information Measures

Haroutunian, Mariam E. and Avetisyan, Varazdat K. (2015) New Approach for Test Quality Evaluation Based on Shannon Information Measures. Математические вопросы кибернетики и вычислительной техники, № 44. pp. 7-21. ISSN 0131-4645

[img]
Preview
PDF
Download (675Kb) | Preview

    Abstract

    There are two currently popular statistical frameworks for addressing test data analyzing: Classical Test Theory (CTT) and Item Response Theory (IRT). Each of these approaches has its advantages and disadvantages. The detailed description of CTT was given in the previous paper of V. K. Avetisyan [1]. In this paper the description of IRT models are provided to show the complex mathematical apparatus used. In this investigation we suggest a new model of test quality evaluation based on information measures such as Shannon entropy, conditional entropy and average mutual information.We show that this approach is easier and more informative.

    Item Type: Article
    Additional Information: Շեննոնի ինֆորմացիայի չափի մեծությունների վրա հիմնված թեստերի որակի գնահատման նոր մոտեցում / Մ. Հարությունյան, Վ. Ավետիսյան: Новый подход к оценке качества тестов на основе мер информации Шеннона / М. Арутюнян, В. Аветисян
    Uncontrolled Keywords: Test quality, IRT models, Test information function, Shannon entropy, Average mutual information.
    Subjects: L Education > LT Textbooks
    Q Science > QA Mathematics > Computer software
    Divisions: UNSPECIFIED
    Depositing User: FSL Bibl. Dept.
    Date Deposited: 24 Sep 2018 13:02
    Last Modified: 26 Sep 2018 12:02
    URI: http://compsci.asj-oa.am/id/eprint/804

    Actions (login required)

    View Item