PLS-SEM vs CB-SEM: Which Method Should You Use?

June 2026 · 8 min read · ScholarAI

Decision rule: Use PLS-SEM when your sample is small (<200), your model is complex, or your goal is prediction. Use CB-SEM when your sample is large (200+), your constructs are well-established, and your goal is theory confirmation. For management and business research, PLS-SEM (SmartPLS) has become dominant at Q1 journals since 2015.

Choosing between PLS-SEM and CB-SEM is one of the most consequential methodological decisions in quantitative management research. Reviewers at Q1 journals will scrutinize your justification — and an unjustified choice is a common reason for major revision requests.

Key Differences

Criterion	PLS-SEM	CB-SEM
Algorithm	Partial Least Squares (variance-based)	Covariance-based (LISREL, AMOS, lavaan)
Goal	Prediction and exploration	Theory confirmation
Minimum sample	~100–150 (10× largest construct)	200–400+ recommended
Distributional assumptions	Non-parametric; no normality required	Multivariate normality assumed
Model complexity	Handles complex models well	Struggles with highly complex models
Formative constructs	Handles well	Problematic
Common software	SmartPLS, WarpPLS	AMOS, LISREL, R lavaan, Mplus
Fit indices	Limited (SRMR, NFI)	Full suite (CFI, RMSEA, TLI, χ²)

When to Use PLS-SEM

Sample size below 200
Exploratory research with new constructs
Model includes formative measurement
Goal is prediction (R² maximization)
Non-normal data distribution
Complex models with many constructs and indicators

When to Use CB-SEM

Sample size 300+
Testing a well-established theoretical model
All constructs are reflective
Goal is theory confirmation, not prediction
Data is approximately multivariate normal
You need full model fit indices (CFI, RMSEA, TLI)

Justifying Your Choice in the Methodology Section

Reviewers expect an explicit justification. Do not simply say "we used PLS-SEM." Cite Hair et al. (2019) for PLS-SEM or Kline (2023) for CB-SEM, and explain why your research objective, sample size, and construct type align with your chosen approach. A two-sentence justification referencing these foundational methodological texts satisfies most reviewers.

Required Reporting for PLS-SEM

Q1 journals now expect full PLS-SEM reporting including: outer loadings and AVE for convergent validity; HTMT ratios for discriminant validity; bootstrapped path coefficients with t-values and 95% confidence intervals; R² and Q² values; and effect sizes (f²). Missing any of these is a common cause of major revision.

Get Your Methodology Reviewed

ScholarAI's manuscript analysis evaluates your methodology section against Q1 journal standards and identifies structural weaknesses before submission.

Analyze My Methodology Free →