This study aims to test the reliability, validity and wright person-items of the test instrument. It used to measure students' cognitive system thinking in genetics lectures. This cognitive system dimension (New Marzano Taxonomy) consists of 4 levels: retrieval, comprehension, analysis, and knowledge utilization. The research method used was descriptive quantitative, with respondents as many as 116 undergraduate students who had attended genetics lectures. The instrument used consisted of 20 multiple-choice questions developed from 4 levels of the cognitive dimension of the system. Data analysis using the Rasch Model-Dichotom to generate more accurate estimates based on individual abilities and difficulty levels of question items. The results showed that the Alpha Cronbach and Item Reliability value on the instrument was 0.81 (Very Good) to 0.92 (Best). Furthermore, person reliability is 0.63 (Moderate). It can be concluded the interaction between the question items and the quality of the question items on the test instrument indicates that the instrument effectively assesses constructs and is relevant for measuring the level of cognitive system thinking. Revisions are made to items that are too easy or too difficult, as well as a review of items with similar difficulties. So it can be assessed that this test has the potential to be widely used in measuring cognitive ability, but some improvements are needed to further improve its reliability for other population.
Read full abstract