Development Of Diagnostic Dyslexia Test Of Phonology And Font Style For 3-4 Grades

Objective: Based on the Classical Test Theory (CTT) and rule space model (RSM), preliminarily established a diagnostic dyslexia test of phonology and font style for 3-4 grades.Methods: The framework of the test was preliminarily established by drawing forefathers' research products and interviewing with experts. After selecting items for three times, Two parallel formal tests which separately contained 32 items came out. The formal samples all were assumed from YongShun,a small town in the west of HuNan, and 247 effective samples were adopted, including 88 normal samples of test A ,87 normal samples of test B,37 RD (reading disabilities) samples of test A and 35 RD samples of test B. Then , 28 pupils did the retest ,and the same number of pupils did the parallel test and the concurrent validity test. Moreover, 54 predictive validity samples of test A and 55 of test B were collacted. Various measuremental methods are used to test items, reliabilities and validities. The diagnostic report for test scores of pupils with RD could be obtained by RSM.Results:The result of X~2 test demonstrated that the formal samples were in conformity with the sampling plan. The Firsthand score of the test increased progressively with grade, and there was a significant difference between normal group and RD group in every subtest and test scores. The item analysis showed that the average item difficulty of test A was 0.74, and test B was 0.78. The average item discrimination of test A was 0.53, and test B was 0.51.The reliability analysis showed that, test A and B's parallel reliability was 0.77; The homogeneity reliabilitiy of test A was 0.91,and test B was 0.92; The retest reliability of test A was 0.88,and test B was 0.96. These results indicated that the reliabilities of the tests were well. Empirical analysis showed that the test had good content validity. Criterion validity was also well, as shown by concurrent validity (test A was 0.64,test B was 0.74) , predictive validity(test A was o.80,test B was 0.76).Besides, taking clinical diagnosis as criterion, taking 21 as cutoff score, the accuracy probability of test A was 0.87,and that of test B was 0.83; the detection rate of test A was 0.81,and test B was 0.60.The construct validity analysis showed that significant differences in the test and subtest scores were found beween grade three and grade five, but the differences were not significant in some test and subtest scores between grade three and grade four, so did grade four and grade five.The confirmatory factor analysis showed that the fit indexes were well. It indicated that the structure of the test was reasonable.The result of Analysis by RSM showed that more than 80% pupils with RD were classified to typical attribute-mastery patterns 1,2,3,4,5,7,10.Conclusion: Both test A and test B's difficulties and discrimations were well. The reliabilities and validities of test A and test B had basically met the requirements of survey. Test A and test B were parallel on the whole.Typical attribute-mastery patterns and attribute-mastery probability could provide embody diagnostic report for teachers and the children with RD.