| 摘要: | 本研究第一年度工作旨在編製一份評量國小四至六年級學生閱讀理解錯誤類型之測驗工具,並考驗其信度及效度。此外,為了後續研究,篩選出國小四至六年級之閱讀障礙學生以做為研究之樣本。茲將本研究之重點摘要說明如下:一、研究樣本:採分層叢集之取樣方式,選取中部地區台中縣、台中市、彰化縣及南投縣等四縣市四至六年級學生,預試樣本共1797名,信效度樣本117名。二、閱讀理解測驗:在測驗內容方面,測驗五種錯誤類型,字句增刪、字句修改、回憶基本事實、回憶序列、回憶主題等,每一種錯誤類型編製15題題目,全測驗共計75題。在試題的難度方面,在字句增刪部分介於0.36-0.87之間,平均值為0.71。字句修改部分介於0.39-0.92之間,平均值為0.69。回憶基本事實部份介於0.30-0.89之間,平均值為0.72。回憶序列部分介於0.35-0.88之間,平均值為0.69。回憶主題部分介於0.27-0.90之間,平均值為0.67。本試題的鑑別度方面,在字句增刪部分係數值介於0.18-0.66之間,平均值為0.40。字句修改部分介於0.16-0.84之間,平均值為0.45。回憶基本事實部分介於0.23-0.85之間,平均值為0.45。回憶序列部分介於0.18-0.72之間,平均值為0.43。回憶主題部分介於0.19-0.79之間,平均值為0.46。整體而言,試題的難度、鑑別度尚稱適中。三、測驗的信度:內部一致性.alpha.係數,三個年級的係數值介於0.90-0.96之間,全測驗的信度值為0.94;折半信度介於0.87-0.92之間;重測信度值介於0.81-0.92之間。四、測驗的效度:內容效度,本測驗商請三位心理測驗編製專家就測驗試題進行審核,認為本測驗具有內容效度及表面效度。效標關連效度,本測驗與羅桑二氏非語文智力測驗之相關值介於0.43-0.59之間;與國民小學國語文成就測驗之相關值介於0.40-0.79之間。構念效度,本測驗與學校之國語科學業成績之相關介於0.57-0.76之間。五、篩選閱讀障礙學生:閱讀障礙學生之篩選,以羅桑二氏非語文智力測驗及國民小學國語文成就測驗兩測驗之得分採已建立之直線迴歸公式,實際得分與預測能力得分之間相差在1.5個標準誤以上,並加入排除因素,每年級篩選出五十四名閱讀障礙學生,共計一百六十二名閱讀障礙學生。
The purpose of this study for the first year was to develop the "Reading Achievement Test for the Elementary Students" in order to select the students with reading disabilities from fourth to sixth grades. A stratification method was used to select 1,797 elementary students for collecting the data of item analysis and to select 117 subjects for estimating the reliability and validity of the test. The results were as follow: 1. The values of item difficulty of the test were ranged from 0.27 to 0.92, and the values of item discrimination of the test were ranged from 0.16 to 0.85. 2. Reliability data for the test consisted of test-retest, split-half, and .alpha. coefficient estimates. The coefficients of reliability for split-half were ranged from 0.87 to 0.92 and .alpha. coefficients were ranged from 0.90 to 0.96. 3. Test validities included: content validity, concurrent validity, and construct validity. The validity of this test was found very statis-factory. 4. 162 students with reading disabilities and 162 regular students were selected as the sample for analyzing the error patterns of reading for the future study. |
