ประสิทธิภาพของวิธีการให้คะแนนความรู้บางส่วนด้วยวิธีการประยุกต์วิธีของคูมบ์และ วิธีตัดตัวลวงเลือกตัวถูก

Abstract

The purposes of research entitled efficiency of partial scoring methods using modified Coomb’s and number right elimination methods were to (1) compare reliability coefficient of test using modified Coomb’s scoring method and number right elimination method, (2) compare correlation between true ability and estimated ability of students were taken modified Coomb’s scoring method and number right elimination method, and (3) compare test information between test using modified Coomb’s scoring method and number right elimination method. The data was generated using a Monte Carlo simulation with multiple choice test with four choices had three different number of items; 20 items, 30 items and 40 items and had 1,000 test takers, each of which was repeated 10 times. Determined efficiency of partial scoring methods from reliability coefficient, correlation between true ability and estimated ability, and test information. Data analyzed by two-way MANOVA.

The results showed that (1) The number right elimination method is higher reliability coefficient than modified Coomb’s scoring method significantly at .05 which 40 items were higher than the 30 items and 20 items respectively. (2) The number right elimination method is higher correlation between true ability and estimated ability of students than modified Coomb’s scoring method significantly at .05 which 40 items were higher than the 30 items and 20 items respectively. (3) The number right elimination method is higher test information than modified Coomb’s scoring method which 40 items were higher than the 30 items and 20 items respectively. Especially in the range of intermediate and high ability.