DEVELOPMENT OF AN ITEM BANK SCIENCE AND TECHNOLOGY FOR MATHAYOMSUKSA 3 STUDENTS

Abstract

This research aims to develop a science and technology item bank for Grade 9 students using the 2-parameter item response theory (IRT). The sample used in this study comprised 2,167 Grade 9 students in the Northeastern region during the second semester of the 2023 academic year, selected through multi-stage sampling. The research instrument was a multiple-choice test with four options. The parameters of the test items were analyzed using the item response theory model, which assesses individual item scores using the R programming language and the MIRT package. The research results revealed that a total of 543 science exam items were constructed for Grade 9 students. Content validity was assessed by experts, with 485 items meeting the criteria, showing a content validity index ranging from 0.80 to 1.00. The exam is divided into 10 editions, 50 questions each, with 10 shared questions in the set. These were tested with the sample group, and the test results were calibrated using the R programming language and the Equate package. Calibration was performed using the Haebera Method's item characteristic curve, resulting in 289 quality items. The item parameters included difficulty levels ranging from -0.84 to +2.50, with an average of 0.40 and a standard deviation of 0.21. The discrimination values ranged from 0.50 to 2.52, indicating that the items varied from moderate to high difficulty. The constructed item bank utilized PhpMyAdmin to create a database containing images, numbers, and text. This database will be used to develop a computer adaptive testing program in the future.