受試特徵曲線在能力測量之應用研究

A Study on the Application of the Person Characteristic Curve to Ability Measurements

范德鑫(Der-Hsin Fan)

《教育心理學報》 20期 (1987/05) Pp. 107-129

https://doi.org/10.6251/BEP.19870601.7

摘要

本研究旨在探討試題邏輯分析與受試特徵曲線(person characteristic curve)在能力測量上可能之應用情形。受試者取自國中一二年級男、女生前、後段班四二九人，以實施普通分頷測驗語文理解與方塊計算兩類試題的結果作為分析與研究的資料。具體研究項目包括：(1)利用試題邏輯分析方法，找出造成方塊計算試題之難度差異的因素。(2)探究受試特徵曲線在鑑定試題測量單一能力上運用之適當性，(3)男、女限制草體(restricted group)是否影響到受試特徵曲線之運用。(4)前後段限制軍體是否影響到受試特微曲線之運用。(5)試題鑑別力指數是否影響受試特徵曲線之功能。
研究結果發現：(1)試題圖形之是否清楚，和方塊之排列方式影響方塊計算試題的心理計量的難度。(2)受試特徵曲線具有鑑定測驗是否測量單一能力之功能。(3)限制軍體不影響受試特徵曲線之運用。(4)刪除鑑別度。30以下之試題後，所剩優良試題仍能維持該能力測驗之特徵。
筆者的建議：(1)重視使用各種方法找出造成試題困難差異之因素，以作為能力界定之基礎。(2)使用受試特徵曲線以鑑定試題是否測量單一能力，以保障測驗之構念效度(construct validity)。(3)測驗編製者應進一步利用受試特徵曲線之功能以改進測驗之實施。

並列摘要

The present study was conducted to investigate the capacity of the application of the item content logic analysis and the person characteristic curve to ability measurements. A total of 429 seventh and eighth graders served as examinees. The verbal comprehension test and the block counting test, two subtests of the Army General Classification Test in the Chinese revision, were administered to all subjects.
The major findings are as follows:
(1) The degree of difficulty in the block counting test item depends upon the way in which the figures of blocks are arranged or printed.
(2) The person characteristic curve gives information whether a test measures a single ability.
(3) The restricted group has no influences on the application of the person characteristic curve.
(4) A few items with high indices of discrimination (larger than .30) still maintain the characteristics of a test.
The author suggests that test users and writers should:
(1) Adopt a variety of methods to find those factors that cause difficulty in test items with a view to defining exactly the ability which a test is claimed to measure.
(2) Make use of the person characteristic curve to determine whether a test masures a single abiliy for the sake of construct validity.
(3) Employ the person characteristic curve to further improve the administration of a test.