教育研究應用所有子集迴歸分析

Using All Subset Regression in Educational Research

教育研究常應用逐步迴歸分析，卻少見採用所有子集迴歸分析，此肇因於論述少介紹，SPSS未提供分析功能，而所有子集迴歸分析能提供有關自變項對依變項影響關係的豐富資訊，實有待改善。據此，本研究比較四類多元迴歸分析方法，討論所有子集迴歸分析的優勢和限制，說明所有子集迴歸分析的分析流程、評選最佳子集模式和重要變項的規準和方法、及可資採用的分析軟體。再者，設計適用SPSS的所有子集迴歸分析程式，簡述其如何運作和解讀分析結果。最後以教師圖像調查資料進行實證考驗，發現子集迴歸比逐步迴歸提供更豐富的適配模式資訊，使用評選指標比較待選子集迴歸模式，可確認最佳子集模式，顯然所有子集迴歸和分析程式具有應用效益。

Educational research always conducted stepwise regression but less focus on all-subsets regression that caused by little introductions toward it and lack analytic function in SPSS. All-subsets regression can pro¬vide rich information about the influential relations between indepen¬dent variables and dependent variables that must be improved. The study compared four multiple regression analysis, discussed the strengths and weakness about all-subsets regression, illustrated analytic procedure of all-subsets regression, reported selecting criteria and methods toward the best-subset model and important variables as well as suitable analytic software. Moreover we developed two programs for all-subsets regres¬sion in SPSS, and discussed briefly how it works as well as how to inter¬pret the analytic outputs. Finally survey data of teacher image was used to test all-subsets regression, it showed that all-subsets regression provides rich information of fit model more than stepwise regression. The best-subset regression model can be confirmed by using assessment indices to compare candidate subset regression models. Therefore it showed that all-subsets regression and these programs have several applied benefits for educational research.