以試題差異功能（DIF）探討群體表現差異之原因Modeling differential item functioning using mixture model for multiple groups
| 計畫屬性 | 院內研究計畫 |
|---|---|
| 計畫編號 | NAER-107-12-B-1-03-02-1-04 |
| GRB編號 | PG10703-0021 |
| 計畫名稱 | 以試題差異功能（DIF）探討群體表現差異之原因 |
| 計畫名稱(外文) | Modeling differential item functioning using mixture model for multiple groups |
| 計畫類型 | 整合型計畫 |
| 整合型計畫名稱 | 十二年國民基本教育實施成效評估-臺灣學生成就長期追蹤評量計畫（第一期） |
| 整合型計畫名稱(外文) | Effectiveness of 12-year basic education program:A longitudinal study on Taiwan assessment of student achievement (TASA-L) (I) |
| 所屬計畫 | 子計畫二 |
| 計畫主持人 | 黃馨瑩 Huang, Hsin-Ying |
| 經費來源 | 國家教育研究院 |
| 執行方式 | 自行研究(本院經費-本院人員) |
| 年度 | 2018 |
| 執行期程(起) | 2018-01-01 |
| 執行期程(訖) | 2018-12-31 |
| 執行狀態 | 已結案 |
| 計畫經費 | 184200 |
| 摘要 | 為瞭解新、舊課綱學生的表現情形，並找出影響學生學力成長變化的因素，對學生進行長期追蹤研究，建立學生認知和非認知能力的量尺，並賦予教育上的意義實為重要。然而這種大型教育調查，其評量工具試題往往存在試題差異功能情形（differential item functioning, DIF），若題目具有DIF，可能會影響量尺的精準度，故本研究將處理學科試題和問卷題目的DIF情形。本正式調查的追蹤對象第一群為2018年9月入學的七年級學生（舊課綱最後一屆），第二群為2019年9月入學的七年級學生（新課綱第一屆），抽到該批學生後，分別連續追蹤這兩批學生在國中三年的表現，抽樣方式為兩階段分層群集抽樣（two stage stratified cluster sampling )，根據國中教育會考學校成績分層後，先抽出學校再抽出班級，抽到的班級整班填寫學生和家長問卷，該班導師和學科教師填寫教師問卷，該校校長填寫學校問卷。本計畫將分析國、英、數、自、社五個學科，以及學生、家長、教師、學校問卷的題目，檢視由外顯變項（manifest variable）或潛在變項（latent variable）造成DIF的情形，並檢查整份工具DIF的變異，評估學科和問卷的量尺穩定性，並根據問卷的題目，試圖找出造成試題DIF的因素。 |
| 摘要(外文) | In order to understand (compare) the effects of the new and the old versions of curriculum guidelines on the junior high school students' performances, and to identify the factors that affect the growth of students' learning ability, it is urgent to establish a scale, which measures the students' cognitive abilities and also provides the educational interpretation for stakeholders in the related fields. However, differential item functioning (DIF) of measurement tools is inevitable in the large scale major educational surveys. DIF of the test items affects the precision of the measurement tools. This study therefore addresses the DIF in the test items of academic disciplines and survey questionnaires. The investigation will be carried out in September 2018 (the last year using the old version of curriculum guidelines) and in September 2019 (the first year using the new version of curriculum guidelines) on the seventh graders. These two clusters will be continuously tracked on their academic performances through junior high school. Two stage stratified cluster sampling will go as follows: the population is stratified based on the results of the Comprehensive Assessment Program for Junior High School Students. Schools and classes are then randomly selected. The selected classes complete the student and the parent questionnaires; the homeroom teachers and the discipline teachers complete teacher questionnaires; and the principal of the school fill out school questionnaire. This study will look into the effects caused by DIF from manifest variable and latent variable through the following disciplines, namely Chinese, English, Mathematics, Science, and Social Studies, as well as the survey questionnaires for students, parents, teachers and schools. The variations of DIF in the measurement tools, the stability of academic disciplines tests and survey questionnaires, as well as the factors that cause DIF will also be investigated. |
| 關鍵字 | 試題差異功能大型教育調查測量誤差混合模式 |
| 關鍵字(外文) | differential item functioninglarge scale assessmentmeasurement errormixture model |