二階驗證性因素分析之理論與應用

Abstract

本研究的主要目的是：（１）簡介二階驗證性因素分析的理論與統計原則； （２）以一百名九歲半兒童在斯比量表上實際所得的觀察資料， 演示如何以 LISREL 7.16 電腦程式來解二階驗證性因素分析的問題。 本研究以 LISREL 7.16 電腦程式的「最大可能性法」（ ML ）進行分析， 以驗證斯比量表 的二階因素結構模式，並從模式的基本適配度、整體模式適配度（模式的外在品質）及模式 內在適配度（模式的內在品質）三方面來評鑑本研究所得結果。評鑑結果發現，在整體模式 適配度的幾項評鑑指標中，幾乎一致指出理論模式與實際觀察資料可以適配；但是在模式的 基本適配度及模式內在適配度兩方面，卻發現理論模式的內在品質不佳。因此，研究者建議 擴大樣本數，以進一步驗證斯比量表的二階因素結構模式。

The purposes of this study were (1) to introduce the algorithm of the second-order confirmatory factor analysis, and (2) to demonstrate the practical application of second-order confirmatory factor analysis by using the computer program LISREL 7.16. The Chinese version of the Stanford-Binet Intelligence Scale scores from 100 children of nine and a half years old were analyzed by a second-order confirmatory factor analysis to test the second-order factors model of the Stanford-Binet Intelligence Scale. The evaluation of model fit were analyzed by the preliminary fit criteria, overall model fit, and fit of the internal structure model. The results indicated that the second-order factor model of the Stanford-Binet Intelligence Scale fit the oberved data in overall model fit criteria but did not fit the observed data well in preliminary fit criteria and fit of the internal structure model. Results suggest that future research should increase sample size to retest the second-order factors model of the Stanford-Binet Intelligence Scals.