結構方程模型的基本原則及符號

The Basic Principles and Notation of Structural Equation Models

姚開屏(Kai-Ping Grace Yao)

《職能治療學會雜誌》 11卷 (1993/12) Pp. 79-93

https://doi.org/10.6594/JTOTA.1993.11.08

摘要

本篇簡介一種目前被廣泛慶用於社會行爲科學、經濟、教育等學界的研究方法—結構方程模型法(structural equation models)。結構方程模型是迴歸分析(regres-sion analysis)的一種延伸，它同時分析一個對閉理論模型中一組線性迴歸方程式，以瞭解模型中變項間的因果關係。這種方法可提供給職能治研究一個新的資料分析法，提昇職能治療的研究水準。

並列摘要

This paper introduces the basic theory and notation of a data analysis method, ”structural equation modeling”, which is an extension of regression analysis. This method is being widely used in behavior science, economic, and educational research. Structural equation models estimate the parameters of a set of linear structural equations representing the cause-and-effect relationship hypothesized by the investigator, and they test whether a relevant set of observed data is consistent with the proposed theoretical process hypothesis. This method can be used for the analysis of causal models with multiple indicators of latent variables, reciprocal causation, measurement errors, and correlated errors. This method provides a new method for analysis in occupational therapy research and, therefore, elevate the level of research.

並列關鍵字

causality ； regression analysis ； structural equation models ； LISREL ； latent variables.