檢査의 信賴度를 높이기 위한 問項選定基準에 관한 硏究 = Some criteria of item selection to raise up test reliability

다국어 초록 (Multilingual Abstract)

The study was intended to find some criteria of item selection to raise up a test reliability (mainly cronbach' α and KR20). The previous studies were examined for the properties of reliability coefficients in terms of a test. Based on the literature review, relative effects of item variables such as item mean, item variance, inter-item correlation, test-item correlation and the number of items on test reliability coefficients was confirmed by computer simulated data.
In conclusion, coefficient α or KR20 can reasonably be regarded as a degree of item homogeneity in a test. The relative weights of the item variables for the reliability coefficient can be listed in order of item-test correlation, or inter-item correlation, the number of items, item variance, and item mean. However, the criteria of item selection to raise up reliability coefficient can be determined by the interaction of these item variables. Finally, the efficient procedure of item selection was presented for obtaining maximum reliability coefficient of a test.

목차 (Table of Contents)

• Ⅰ. 緖 論
• Ⅱ. 信賴度의 統計的 理論과 그 推進
• 1. 信賴度 測定의 두 理論的 模型
• 2. 信賴度 算出을 위한 統計的 過程
• 3. 信賴度 推定과 그 意味
• 4. 檢査의 信賴度 向上을 위한 問項選定基準
• Ⅲ. 模擬資料에 의한 問項變人이 信賴度係數에 미치는 效果分析
• 1. 模擬資料에 의한 硏究의 目的
• 2. 分析을 위한 模擬資料의 性格
• 3. 分析結果 및 解釋
• 4. 要約
• Ⅳ. 要約 및 結論
• 參考文獻