Investigating the Quality of a Popular Classroom Assessment Instrument in Indonesia

摘要

This paper reports a preliminary study on the quality of a multiple-choice test as a popular classroom assessment in Indonesia. An item response analysis was conducted to scrutinize four major kinds of quality: reliability, construct validity, item discrimination ability, and distractor plausibility. The data were a form of test developed by a local teacher with strong credentials and administered to roughly eighty students, the test takers' responses to each test item and the data processed by ConQuest, a program developed by Australian Council for Educational Research. The study uncovered that the test requires some reconstruction to establish adequate construct validity. At least 13.5% of all items failed to discriminate students on their ability and roughly half of all items had deficient distractors, not to mention some which distracted no students. Additionally, the test was also found to have failed to meet the acceptable cut-off score of reliability. Finally, this preliminary research highlights two important implications: first, the importance of classroom teachers to receive a proper training on assessment to be able to accurately monitor and reveal the learners' genuine learning progress, and second, the importance of conducting wider scale research to investigate teachers' capacity in developing their classroom assessment instruments.