Evaluation of Psychometric Characteristics of A Logical Reasoning Tool With The Classic Test Theory Approach

Abstract

Research on logical reasoning began to be carried out in Indonesia, it is necessary to conduct an evaluation of the psychometric characteristics possessed by logical reasoning measurement tools made by researchers. Logical reasoning measurement tools made by researchers need to be re-evaluated considering there are weaknesses in terms of using a relatively small number of subjects, so that it can affect the standardization of logical reasoning measuring devices. This study aims to obtain an overview of item difficulty index, item discrimination index, effectiveness of the distractors, validity, reliability, and error measurement from the logical reasoning measurement tool. Respondents in this study amounted to 7.730 students, consisting of 3.897 men and 3.833 women with an age range of 15-19 years. The research data were analyzed with the help of Microsoft Excel 2007, SPSS version 22.00, and ITEMAN version 3.6. The results of this study indicate that the item difficulty index moves from 0.35 to 0.80 which belongs to the easy and medium category. This study also shows that the item discrimination index moves from 0.10 to 0.71 with four items rejected and need to be aborted. This study also shows that the effectiveness of the distractors contained in each item in the logical reasoning measure is in the effective category. This study also shows that each item in the logical reasoning measure is considered valid in terms of factorial validity through exploratory factor analysis (EFA) procedures. This study also shows that the measurement of logical reasoning is relatively reliable with a Cronbach Alpha coefficient of 0.906. This study also shows that the measurement error obtained in this study is ± 3.92 of the total score obtained using a logical reasoning measure with a confidence level of 95%, with the actual score estimate obtained by respondents moving from 12.08 to 19.92, if the scores obtained by respondents sixteen.