This analysis was intended to compare the level of difficulty and discriminating power on multiple choices and essay items on math test. The research was done by design treatments by subject's method. The research object was 22 items of multiple choices items and 22 essay items. The essay items were generated from multiple choices item that modified into essay items by eliminating the options. The essay items were divided into two clusters and each of them consist of 11 items. The population in

more »

... this research was public and private Junior High Schools in DKI Jakarta Province, while the sample was defined by two-staged random sampling and stratified random sampling technique. There were 452 students act as the sample, taken from 6 private and public schools. Multiple choices item was scored by following rule; score 1 for correct respond and score 0 for incorrect respond, in meantime essay items were scored by using 3 different scoring rubrics, namely; Rubric Model 1 (0, 1), Rubric Model 2 (0, 1, 2) and Rubric Model 3 (key words). The difficulty level and the different power of students' response data that has been scored are calculated from the multiple choice and essay items, and then analysed by using descriptive statistical methods and inferential statistics and Rasch models. The results of the analysis with the Classic method obtained an average level of difficulty multiple choice questions greater than the essay items either on Model-1, Model-2, or Model-3. While the analysis using the Rash Model, the average level of difficulty multiple choice items is relatively the same as the essay items. For the average discriminating power, both using the Classic method and the Rasch Model, the average discriminating power of multiple choice items is greater than the essay items. In other words, based on research the form of multiple choice items is easier than the essay items, whereas the discrimating power of multiple choice items is greater than the essay items.