Using Classical Test and Item Response Theories to Evaluate Psychometric Quality of Teacher-Made Test in Ghana

Teaching, learning, and assessment are key concepts in education and their relationship can be seen as participants of a three-legged race. In this regard, classroom assessment practices such as teacher-made tests are important and meaningful when they support students’ learning. The purpose of this study was to establish the psychometric quality of a teacher-made mathematics test item used in one of the Senior High Schools in Ghana. This study employed quantitative descriptive design where 400 selected students' responses to a teacher-made Mathematics test were collected and analyzed through various psychometric techniques. The results showed that the Mathematics test had low but acceptable reliability coefficient of 0.61. Also out of the 40 multiple-choice items, 26 were of satisfactory difficulty levels with only one test item found to be too difficult and three test items being too easy. The findings of the discrimination indices suggest that 25 test items had bad or weak discrimination indices and four items showed negative discrimination indices. The study further indicated that 30.8 percent of the options were functioning distractors whereas the majority of the options (69.2%) were non-functioning distractors.It is therefore recommended that inservice training on effective ways of developing test items should be organized regularly for in-service teachers to help improve of the quality of teacher-made tests across Senior High Schools in Ghana