The Development Of Mathematics Achievement Test For Low Grade Of Elementary School

Evaluation reform is an important part of curriculum reform. The basis of the subject study assessment is the subject' s curriculum criterion now. The research develops Mathematics Achievement Test for Low Grade of Elementary School based on the compulsory educations' mathematics curriculum criterion and psychometrics theory, trying to afford a scientific evaluation method for curriculum reform. It' s a math subject achievement test for Grade One and Two of elementary school. It' s a group test and must be read to the student by tester. There is only one correct answer of four choices of all items. It' s composed of two alternative copies for every grade: A and B. The test carried out at four city elementary schools and two rural elementary schools and got 1157 valid answer sheets. After analyzed the result, row scores of all tests were obtained. The results of items analysis showed that average difficulty of all tests were from 0. 47 to 0. 54, the range of index of items discrimination were over 0.20. The distraction analysis pointed out that incorrect choices of 63. 9%~88. 8% items had good distraction. The reliability study bespoke that the range of the homogeneity reliability of the test was from 0.80 to0.88, standardized errors of measurement (SEM) were from 2.61 to 2.92. The validity study indicated that it was regarded qualified by expert' s method and experience method. There were two main factors by exploratory factor analysis of all tests. They decided 76. 40% for 1A and 72. 99% for IB and 64. 79% for 2A and 62. 28% for 2B of whole variance. The simultaneity validity analysis explained that there were remarkable differences among different grade, city and rural school, but no differences among sex. The correlations were 0. 703~0.732 between the test and the final math examination' s marks. All results illustrated that there were little discrepancies between A and B. The quality of the items is fine. It has higher reliability and validity by several evidences, and meets the need of psychometrics.