- 學位論文
仿國中社會科基本學力測驗之線上評量系統建置
Develop the Online Assessment System to Social Studies Mock Exam In Basic Competence Test for Junior High School Students
摘要
本研究旨在發展一個仿基測社會科的線上適性測驗系統，除可提供評估受試者能力的參考，也可作為預測基測表現的另一種工具。研究進行分為兩個階段，第一階段進行兩種組卷方法的預試。其中組卷方法一是對試題進行質的分析，即利用雙向細目表、九年一貫社會領域能力指標歸納出基測社會科的命題原則來組卷；組卷方法二是對試題進行量的分析，即挑選出不同難度區塊中的高鑑別度試題來組卷。透過施測後所得之量尺分數、受試者能力值與受試者基測量尺分數的相關決定建置題庫的依據，並發展成線上適性測驗系統。第二階段為適性測驗系統的成效評估。從預試對象中抽取受試者，實際進行線上施測，以驗證系統功效。 本研究歸納出以下結論： 一、從基測試題的雙向細目表分析可知，基測試題的能力層次分布具有規律性，其中以「了解」層次所佔比例最高；在內容方面，各細目大致皆有試題分布，而三分科的出題比例均等，但各科各年級的試題比例不相同。 二、基測試題皆能找到相對應的能力指標，其中歷史、地理科指標集中分布於第一及第二主題軸，約各佔三分之一，公民科則以第六及第七主題軸分布較多。 三、預試卷中以組卷方法二（量的分析）成果較好，其量尺分數與基測量尺分數的相關為 .898、受試者能力值與基測量尺分數的相關分別為 .857，顯示以量的分析組卷能建置成對基測成績有良好預測度的題庫。 四、本研究的適性測驗系統經實際的線上施測，可發現當測驗長度為13題時，全體受試者的能力估計值與全體基測量尺分數的相關已達 .8以上，與最大測驗長度63題相比，可節省約79% 的試題數；當測驗長度為23題時，全體的能力估計值與全體基測量尺分數的相關已達 .85以上，約可節省約63% 的試題數，故本研究的適性測驗系統可達到節省施測試題的功效。
並列摘要
The purpose of this study is to develop the computerized adaptive testing system to social studies simulated test for junior high school students. The system except for providing the reference of the examinee’s ability scale, it can be used as a tool of forecasting scores in Basic Competence Test. There are two stages in this research. The first stage is to do the pre-testing with two kinds of methods to generate the test paper. One is analyzing the qualitative data of the Basic Competence Test of Social Studies by the two-way specification table and the Competence Indicators of the Grade 1-9 Social Studies Competence Indicators to generalize the test principles. The other is analyzing the quantitative data of social studies simulated test by selecting the items of excellent discrimination from different difficulty phases. Through the pre-testing, the on-line adaptive testing system is developed based on the relation of the examinee’s ability scale and the score of the Basic Competence Test. The second stage is to evaluate the effect of the on-line adaptive testing.Choosing the examinees from the pre-testing subjects to conduct the on-line testing for testify the effectiveness of this system. The conclusions of this study are below： 1. From the analysis of the two-way specification table in the Basic Competence Test, we can find that, in the aspect of the competence level, there is regularity and the understanding phase takes the most part of it. As for the aspect of the content, each specification has items. The ratios of the three subjects in social studies are even but the ratios of the items from different subjects and different grades are not equal. 2.Each item can find the corresponding Competence Indicators. History and Geography gather on the first and second main topic, around 1/3, while Civic is on the sixth and seventh main topic. 3. According to the result of the method of analyzing quantitative data of social studies, the correlation coefficients of the examinee’s ability scale and the score of the Basic Competence Test are .857. It shows the analyzing quantitative method have great predictive validity. 4. Through the computerized adaptive testing, when there are 13 questions, the correlation coefficient of the examinee’s ability scale and the scale score in Basic Competence Test reaches to .8. Comparing to the actual 66 questions, about 79% can be retrenched. When there are 23 questions, the correlation coefficient is over .85 and around 63% can be retrenched. Therefore, the adaptive testing in this study can reduce the number of the questions in the test.