Partial Least Squares SEM (PLS-SEM)

Partial least squares structural equation modeling (PLS-SEM) is a variance-based method that estimates relationships among composite constructs and their indicators, focusing on explaining variance in the outcome constructs.

부분최소제곱 구조방정식모형(PLS-SEM)은 합성 구성개념과 측정지표 사이의 관계를 추정하는 분산 기반 방법으로, 결과 구성개념의 설명된 분산에 초점을 둡니다.

How do I test a path model with multiple latent constructs measured by survey items?

설문 문항으로 측정한 여러 잠재 구성개념의 경로모형은 어떻게 검증하나요?

PLS-SEM estimates a measurement model linking items to constructs and a structural model linking constructs to one another, using composites and bootstrapping for inference. It fits prediction-oriented research, complex models with many constructs, and models with formative indicators, whereas covariance-based SEM is preferred for testing the overall fit of a common factor model.

When to use Partial Least Squares SEM (PLS-SEM)

• Several constructs, each measured by multiple items, linked in a path model
• Prediction-oriented research focused on explained variance in key outcomes
• Models that include formatively measured constructs

When to use something else

• Testing the overall fit of a theory-based common factor model: use covariance-based SEM
• All variables are single observed measures: use multiple linear regression or path analysis

Assumptions and requirements

• Measurement and structural models specified from theory
• Indicators specified as reflective or formative
• Adequate sample size for the most complex part of the model
• Inference based on bootstrapping rather than distributional assumptions

What the analysis reports

• Outer loadings or weights
• Cronbach's alpha, composite reliability, and AVE
• HTMT ratios for discriminant validity
• Path coefficients with bootstrap confidence intervals
• R², f², and Q² for endogenous constructs

How to report it

The path from [construct A] to [construct B] was significant, β = [value], t = [value], p = [value], 95% bootstrap CI [[value], [value]], and the model explained [value]% of the variance in [construct B].

Related methods

• Covariance-Based SEM
• Confirmatory Factor Analysis (CFA)
• Path Analysis

Run Partial Least Squares SEM (PLS-SEM) on Quorant