The Preliminary Establishment Of Reading Comprehension Test For 3～4 Grades

Objective: The preliminary establishment of reading comprehension test for 3～4 grades gives an analysis on its items, and has an initial test on its reliability and validity. The aim of the study is to test the ability of reading comprehension for the students who are in 3～4 grades in the primary school, to improve the teaching methods of reading for school and also to provide tools for the clinic diagnosis and curriculum reform.Methods: Combining with theories and experience, this paper adapts the classical theory of measurement theory (CTT) of psychological measurement. Taking crystal intelligence as the basic theoretical framework, this paper initially establishes reading comprehension test for 3～4 grades according to the characteristics of primary students' reading ability, the content of which is made up by words comprehension, phrases comprehension, sentences comprehension, paragraphs comprehension and fast reading for the 3～4 Grades. After preparation and calculation, it comes out two set of test paper A and B, which has same examining form, same examining items and same degree of difficulty. The formal examining samples all are assumed from Changsha, of which 1062 effective samples are adopted (533 in test paper A and 528 in test paper B). Then, 103 students did the same tests after one week (53 in A and 50 in B), 92 students did parallel test the another day. Moreover, 49 effective samples are adopted. The reading comprehension test gets item analysis, reliabilities and validities test by different measuring methods.Results: The statistcal analysis of sample shows that, besides the gender presents notable differences, the sample ditribution of different schools is in the conformity with the sampling plan, the firsthand score of the test increases progressively with grade. The item analysis shows that the project, the average difficulty of paper A is 0. 62, and that of B is 0. 63, item of which difficulty is between 0. 30 and 0. 69 reaches 60%. The average discrimination index of paper A is from 0.33 to 0.44, Paper B from 0.35 to 0.44, both achieves outstanding level. The result of reliability analysis shows that, the retest reliability is 0.88 and 0.82, the retest-parallel reliability is 0.85, the cronbach' sαcoefficients of Paper A and B is 0. 80 and 0. 84. The result shows that the test has good reliability. The content validity investigated by expert judgment, alternative-form method and empirical method is relatively high. Analyzing by concurrent validity and predictive validity, the concurrent validity' correlation of paper A and B is 0.73 and 0.85, predictive validity' correlation is 0. 65 and 0. 72. The result shows that the test have good concurrent validity and predictive validity. The construct validity shows that, there is notable statistcal difference in diffferent grades and different schools, and there is a phenomenon of gender differences in some parts of the test. The exploratory and confirmatory factor analysis shows that, the test can be divided into 5 factors which has same structure, they are words comprehension, phrases comprehension, sentences comprehension, paragraphs comprehension and fast reading. It meets the expected standards.Conclusion: The difficulty of the test is appropriate and discriminations of it is very well. The items reliability and validity of the test has basically meet the requirements of survey. The test has good concurrent validity, predictive validity and construct validity. Paper A and B are parallel on the whole.