Abstract

• The main purpose of this study was to develop two parallel forms of test for Grade (10) Chemistry. This thesis was mainly concerned with planning, constructing, and analysis of test items. And then, two parallel forms of test were administered to 524 Grade (10) students from Ten Basic Education High Schools in Yangon Region. Descriptive survey method and quantitative research design were used in this study. From 80 items of both forms, 12 pairs of items were selected as parallel items according to their similar levels and ranges of difficulties and discriminations. Since reliability coefficient of equivalence of Form A and Form B is 0.703, both forms have strong positive relationship. So, students who performed well in Form A would also perform well in Form B. Form A can be used to get more information for measuring chemistry achievement of students who have θ = -1.4. Form B had smaller standard error across the ability scale from -2.5 to +1.5 and larger standard error had at the low and high ends of the scale. The maximum amount of information was I(θ) = 6.7 at θ = +0.2. Ability estimates were more precise across the ability scale from -2.5 to +1.5 than at the high and low ends of the scale. Therefore, Form B could be suitable to measure for students whose ability is θ = +0.2. The expected ability distributions of the students (Form A and Form B were applied) were normally distributed across the ability scale.