Covariance Based SEM (CB-SEM)

Covariance-based structural equation modeling (CB-SEM) estimates a system of measurement and structural relationships among latent variables by fitting the model-implied covariance matrix to the observed covariance matrix, typically with maximum likelihood.

공분산 기반 구조방정식모형(CB-SEM)은 모형이 함축하는 공분산행렬을 관측 공분산행렬에 적합시켜, 보통 최대우도법으로 잠재변수 간 측정관계와 구조관계를 추정하는 방법이다.

How do I test a theoretical model with several latent variables and paths between them?

여러 잠재변수와 그 사이 경로로 이루어진 이론모형을 어떻게 검증하나요?

Covariance-based SEM tests a theory-driven model by estimating latent constructs from survey items and the directional paths among those constructs in one model. It fits confirmatory research with adequate sample sizes and reports global fit indices such as CFI, TLI, RMSEA and SRMR.

When to use Covariance Based SEM (CB-SEM)

• Multi-item survey scales measuring several latent constructs
• Testing hypothesized paths among constructs, including mediation
• Confirmatory research where overall model fit must be evaluated

When to use something else

• Prediction-oriented models or composite (formative) constructs; use PLS-SEM, noting that it does not solve small-sample problems
• Only one latent construct; use confirmatory factor analysis

Assumptions and requirements

• Correctly specified model grounded in theory
• Multivariate normality for maximum likelihood, or a robust estimator
• Adequate sample size
• Model identification

What the analysis reports

• Chi-square test of model fit
• CFI, TLI, RMSEA and SRMR
• Standardized factor loadings
• Standardized path coefficients with p-values
• R-squared for endogenous constructs

How to report it

The structural model fit the data well, χ²([df]) = [value], CFI = [value], TLI = [value], RMSEA = [value], SRMR = [value].

Related methods

• Confirmatory Factor Analysis
• PLS-SEM
• Path Analysis

Run Covariance Based SEM (CB-SEM) on Quorant