Fuzzy Classification of Dichotomous Test Items and Social Indicators Differentiation Property

Abstract

In many fields of human activities such as economics, sustainable development, construction, human resources management etc., dichotomous tests are employed to measure some observed property, for example knowledge level in a specific field or applicant’s eligibility for a job position. Fuzzy classification method for dichotomous test items is proposed in this paper. Depending on the observed property, each test item may well differentiate all testees or only the testees who are strong or weak at that property. Also, the test item may badly differentiate all testees and be inappropriate for that purpose. The method presented in the paper may be applied for small groups of testees with known estimates of the investigated property, for example raw test scores. The proposed method for dichotomous test item classification is based on the fuzzy set theory. Though the tests were originally constructed for knowledge measurement, their mathematical models can be applied for social indicators and wide range of other areas.