頁籤選單縮合
| 題 名 | 比較雙因素模式與題組效果模式對偵測試題題組效果之效益=The Study of Bi-factor Model and Testlet-effects Model in Detecting Item Testlet-effects |
|---|---|
| 作 者 | 盧思丞; 凃柏原; | 書刊名 | 教育研究論壇 |
| 卷 期 | 5:1 2013.12[民102.12] |
| 頁 次 | 頁19-42 |
| 分類號 | 521.3 |
| 關鍵詞 | 題組效果; 雙因素模式; 題組反應模式; Testlet effects; Bifactor model; Testlet-effects model; |
| 語 文 | 中文(Chinese) |
| 中文摘要 | 早年,Holzinger & Swineford(1937)提出雙因素模式(bi-factor model),僅被應用在驗證性因素分析中,而實際應用於心理計量的問題不多。直到Gibbons & Hederker(1992)提出full-information item bi-factor analysis,且應用TESTFACT軟體(Bock, Gibbons, Schilling, Muraki, Wilson, & Wood, 2003)進行分析,才逐漸引起研究討論熱潮。目前應用bi-factor模式進行研究探討的議題有題組效果、向度檢測、分量表分數與評分者效果,最多的是與題組效果有關的研究。本研究以題組效果的相關議題為主,利用模擬方式,1000位受試者,產生題組試題(20、40題;在bi-factor model中的是四種次要向度斜率;在testlet-effects model中的是不同題組效果),目的是探討題組效果模式(testlet-effects model)(Wainer, Bradlow, & Wang, 2007;Wang & Wilson, 2005)、雙因素模式(DeMars, 2006;Gibbons & Hedeker, 1992)在模擬資料的分析結果,並同時探討雙因素模式是否真的將題組效果反應在次要向度的因素負荷量上。 |
| 英文摘要 | The bifactor model ( Holzinger & Swineford, 1937) had been indicated for a long while. This model was always used in confirmatory factor analysis and was seldom used for solving the real psychometric problems. When Gibbons & Hederker (1992) described the full-information item bifactor analysis in the IRT context and had it been implemented in the TESTFACT program, the bifactor model became popular and receiving more attention. Researchers were able to use this method to investigate the issues related to testlet effects, dimensionality assessment, subscale score and rater effects. Among them, the testlet effect and dimensionality assessment seem to be the most studied topics. In this study, we focused on the application of bifactor model to the related issues of testlet effect. Three different data sets will be simulated full-testlet data. Two test lengths (20, 40 items) and three testlet effects (0, 0.5, 1.0) will be manipulated and the sample size will be 1000 for each data set. The purpose of this study is to investigate the performance of bifactor model when it was apllied to model the testlet effect among items. At the same time, the Q3 statistics proposed by Yen (1984) will also be calculated for each data set to see if this statistic was able to indicate the dependency among items. |
本系統中英文摘要資訊取自各篇刊載內容。