教育、經濟、人口、健康對平均餘命指標因果關係探索

Exploring the Causal Relations among the Educational, Economic, Demographical, Health

本研究主要在探索影響國民平均餘命因素之間的因果關係。本研究運用結構方程模式檢定影響國民平均餘命之間因果關係。本研究以1998年「世界發展指標報告」中1990年與1995年資料，選定10項影響國民平均餘命因素進行研究。本研究建構2個結構方程模式，提出2個虛無假設，以考驗潛在變項之間的因果關係。結果顯示，在結構方程模式上，如以潛在變項因果模式整體檢定指標，即卡方考驗未達顯著，且AGFI、GFI、CFI、IFI值都接近1等判定標準。表示二個結構方程模式的教育、健康、經濟、人口對國民平均餘命潛在變項之間適配度良好。也就是說，在1990年與1995年影響國民平均餘命之潛在變項因果關係存在，但以模式內在評估指標衡量2個結構方程模式，發現仍有若干變項有誤差存在。此外，本研究亦檢定潛在變項之間多共線性的問題，結果顯示，2個結構方程都有多共線問題。據研究結論提出相關建議。

The main purpose of this study is to explore the causal relationships among the latent variables of education, economic, demographic, health, and life expectancy at birth. The research employed the LISREL to test the causal relationships of those variables. The research used twelve variables to construct two causal models. Those data are collected from World Development Indicators Report (1998), and two panels （1990 and 1995） data are selected. And it provided two null hypotheses to test the casual relationships among those variables. The results showed that the x2 values of two causal models in structural equations are not significant. Besides the criteria values of AGFI, GFI, NFI, CFI, and IFI are also near 1.0 in structural equations. That is, two models are well fitted and rejected two null hypotheses in our study. It meant that the casual relationships among the latent variables of education, economic, economic, demographic, and life expectancy at birth are well fitted. But according to other criteria, there are some errors in research models. In addition, in two structural equations are found collinearity. From that finding, it provided some suggestions.