Chapter 1 Concept and Principle of Structural Equation Modeling
1.1 Basic concepts of structural equation modeling
Structural equation modeling (SEM) is a commonly used empirical research method in sociological research and is widely used in data analysis and model testing.
Structural equation modeling was called linear structural equations (Linear Structural Relationships, LISREL) or covariance structure analysis (Covariance Structure Analysis) in the early days.
In principle, the structural equation model is a multivariate statistical method based on the covariance matrix (covariance matrix), which is used to detect and verify the relationship between the observed variable (manifest variable) and the latent variable (latent variable) in the theoretical model, as well as the relationship between the latent variable and the latent variable. Hypothesized relationships between latent variables.
Therefore, structural equation modeling is a multivariate statistical technique that combines factor analysis and path analysis, and is suitable for quantitative research on the interactive relationship between multiple variables.
The concepts of explicit variables and latent variables mentioned here, there are some abstract indicators in real life, such as trust level, customer satisfaction, etc. Strictly speaking, they cannot be directly and accurately observed, we call them latent variables; and latent variables It needs to be measured indirectly through explicit variables.