應用群聚分析法檢核國中教育會考標準設定結果之效度-以歷史科及國文科資料為例

Validating Standard Setting Results of the Comprehensive Assessment Program for Junior High School Students with Cluster Analysis: Using History and Chinese Test Data as Examples

本研究以2013年國中教育會考歷史科及國文科的部份測驗資 料為例，造行群聚分析法取向的標準設定，包含以臨界組法及對照 組法來定義切點分數，並以跨樣本比較法來撿核標準設定結果的效 度。研究發現，不論是歷史科或國文科，進行群聚分析之後以臨界 組法或對照組法找出的切點分數，在不同子樣本中的數據都相當接 近。這些結果顯示，應用群聚分析法取向找出之切點分數，具有適 當的效度，所以從某個子群體所獲得之切點分數也適用於具有相似 統計特性的另一個子群體。就教育決策的觀點而言，訂定切點分數 的適當性會影響到學生的受教權，而群聚分析法取向的標準設定方 法，在常用的以專家主觀意見為依據的標準設定法之外，增加了相 對較客觀的切點分數的訊息，可提供給教育主管機關不同觀點的決 策參考。文末則針對樣本數及資料型式，提出對未來研究的建議。

This study analyzed part of the History and Chinese test data of the 2013 Comprehensive Assessment Program for Junior High School Students in Taiwan， to demonstrate using the cluster analysis standard-setting approach， so as to validate standard setting results. Precisely speaking， the cutscores are defined separately by the Borderline Group Method and the Contrasting Groups Method after cluster analysis， and the Cross-Sample Comparison Method is used to validate these cutscores. The results show that these cutscores defined by the Borderline Group Method or the Contrasting Groups Method in different groups are very close or the same， which means the validity of cutscores provided by the cluster analysis standard-setting approach is appropriate because these cutscores are interchangeable in different groups. From the viewpoint of educational policy decisions， the appro-priateness of cutscores would influence students? rights to education. In addition to the commonly used experts-based standard setting method， the cluster analysis standard-setting approach provides educational authorities with some relatively objective information about cutscores for consideration. Suggestions about sample size and data type are also provided for future studies.