90-93年度國中基測量尺及等化程序之個案研究

基测實施六年來，公眾隱約可感受到基測計算方式可模糊前段及後段學生差距：一般大眾在艱澀學理如同黑箱作業下，無法判斷中段學生差距是否也被模糊了。由分發結果而言，中段學生差距之模糊可造成公、私立學校分發落差，連帶地使家長承擔不同對等的經濟負擔。鑑於一般大眾無法針砭基测量尺分數計算方式是否影響考生及家長權益，本文檢驗及分析基測量尺分數是否滿足程序正義的計算方式。此程序正義包括三個參照基準點：一爲正確的學理應用：二爲內部計分流程應與公佈於大眾的「計分遊戲規則」相符相合；三爲擇優政策隱含的名次排擠效應的資訊應該透明化。本文主要的發現：師大心理與測驗發展中心（心測中心）在計算量尺分數的過程中，存在著一些不尋常的地方：一是各個考科在計分過程中的最高分並非設定在60分，且各個科目在各個年度都不相同：二是第二次測驗量尺並無等化(equating)步驟，只是單純進行分數連結(linking)。本文依據此二不尋常的地方質疑心測中心計算量尺分數的缺失。

關鍵字

量尺分數；測驗量化； IRT模式；測驗等化；連結；強真分數理論；均等測量標準誤量尺分數形態； beta-binomial貝氏模式； Rasch模式

並列摘要

The Basic Competence Test (BCT) for junior high school students was developed and administered in 2001 by the research center for psychological and educational testing of the National Taiwan Normal University. The BCT is a test battery consisting of tests in five areas: Chinese, English, math, natural science, and social science. The BCT is administered twice annually. In the admissions to the post secondary high schools, students are ranked from high to low based on the composite scores of the BCT. The research center adopted the ACT scaling procedure for converting test raw scores to scale scores (1-60) and used IRT-Rasch method for equating scores on two forms of each test area. This study assesses the fairness of the scaling and equating procedures by the research center. The main findings of this study are: 1) the range used for the raw-to-scale conversions is not consistent across five test areas; 2) it is a linking procedure rather than an equating procedure that is used for transforming scores on two test forms. This study showed the negative impacts resulting from the unjust procedures carried out by the research center. Suggestions are presented as well.

並列關鍵字

Test scaling and equating ； IRT-Rasch model ； the BCT ； score scale ； beta-binomial model ； arcsin transformation ； compound binomial error model ； strong-true score theory