ITEM RESPONSE THEORY ANALYSIS ON STUDENT STATISTICAL LITERACY TESTS

Abstract

This study aims to evaluate the quality of statistical literacy instruments. With the ex post facto approach, the results of the statistical literacy tests of 150 students were used. Data were analyzed using PARSCALE, based on the Item Response Theory with a two-parameter GRM (2-PL) logistic model, namely item discriminant and item difficulty. The results showed that of the 12 items analyzed, all were good, and the GRM-IRT 2-PL model was the right model for statistical literacy instruments with polytomous answer types. The elements in the statistical literacy instrument have been constructed with accurate information about statistical literacy abilities ranging from -2.5 to +1.2. The least SEM occurs when the maximum test information function is assigned to the test set; therefore, it can be recommended for item analysis on other tests. The statistical literacy test items examined in this study can be used to assess student literacy statistics in various schools and regions.