查詢結果分析
相關文獻
- 脈絡變數的多層次潛在變數模式分析:口試評分者效應的多層次結構方程模式應用
- 公共行政領域碩士論文主題與方法的研究趨勢--多層次分析觀點
- 課室目標與成就目標的多層次關係之檢驗
- 組內相關係數與樣本數對於脈絡效果估計的影響:貝氏估計與最大概似估計法的比較
- 如何成為一位好的臨床技能考官
- 大專生騎乘機車偏差駕駛行為之研究
- 線性混合效果模式在柔性鋪面道路試驗資料之初步分析
- 國中校長教學領導、學校知識管理與教師教學效能之多層次分析
- 性別、自我效能、工作價值、科學素養及學校層次因素對臺灣青少年學習情緒之影響:個人與情境交互作用之多層次分析
- 性別、自我效能及所知覺的學習環境對學習情緒之影響:線性混合模式在叢集資料之應用
頁籤選單縮合
題 名 | 脈絡變數的多層次潛在變數模式分析:口試評分者效應的多層次結構方程模式應用=Multilevel Latent Variable Modeling of Contextual Variables: Application of Multilevel Structural Equation Modeling on Rater Effects |
---|---|
作 者 | 邱皓政; | 書刊名 | 中華心理學刊 |
卷 期 | 49:4 2007.12[民96.12] |
頁 次 | 頁383-405 |
分類號 | 179 |
關鍵詞 | 脈絡變數; 潛在脈絡變數; 多層次資料; 多層次結構方程模式; 評分者效應; Contextual variables; Latent contextual variables; Multilevel data; Multilevel structural equation modeling; Rater effects; |
語 文 | 中文(Chinese) |
中文摘要 | 潛在構念的估計是社會與行為科學研究的重要命題,結構方程模式取向是當代據以進行潛在構念估計與關係探究重要典範,但是,如果研究樣本涉及叢集抽樣而呈現階層結構,研究者即必須進行多層次結構方程模式分析(MSEM)。階層資料的重要特性是低階層次變數在高階會形成脈絡變數,因此MSEM分析必需能夠處理潛在脈絡變數模型的萃取與模型評估。本文的目的,即在說明如何利用MSEM程序來進行潛在脈絡變數的因素模型與結構模型分析,並以一項口試實證資料(76位教師甄試口試委員針對841位考生的口試評分資料),以Mplus4進行單層次與多層次結構方程模式分析。分析結果證實了MSEM程序可以有效分離考生共變矩陣與評審共變矩陣,並證實了個體層次與總體層次的潛在因素模型並不相同,為一非等值模型(non-equivalent model)。個體層次測量變數在評審層次是一個單一完整的潛在構念,亦即方法效應或月暈效果的概念,但在考生層次則仍保有原先的二因素架構。本文的結論指出,MSEM比傳統的MLM取向在處理脈絡變數上更具有彈性,在方法學上具有重要的價值,但是在技術層次仍有諸多有待突破的瓶頸,值得進一步深究。 |
英文摘要 | The estimation of latent construct is the core issue of social and behavioral science. Structural equation modeling is the most important paradigm for estimation and for exploring the relationships among constructs. If the data for analysis involving multilevel or clustered structure, however, the multilevel SEM have to be applied for the special form of dataset. One of the features of multilevel data is that the observed variables in the individual level could be aggregated as a contextual variable into higher level. In terms of latent construct, contextual variable may behave as a latent form and should be estimated in a way of factorial analysis instead of manifest procedure. In a circumstance that the latent contextual variables have significant effects on dependent variable is called latent contextual effects. The purpose of present paper is to clarify the significance of the latent contextual effects in the social research. Rating data of oral examination examined in the present study contains a total of 76 raters and 841 interviewees. A series of single and multiple level SEM models was proposed for examining the latent contextual effects. Results of Mplus indicated that the MSEM modeling can effectively extract the latent contextual variables and furthermore used to explain the dependent variables. Different factorial structure on latent contextual variables in both macro and micro level reveals that the contextual model is not equivalent across levels. The present study proved that the MSEM approach to examine contextual variables is more flexibility than traditional MLM paradigm. However, technical as well as methodological issues have to be solved. |
本系統中英文摘要資訊取自各篇刊載內容。