SEM method guides

Process flows, commonly used cut-offs, and key references for PLS-SEM, CB-SEM, regression, and multi-group analysis — aligned with how WAnalytic reports results.

Partial least squares SEM

Composite-based SEM for prediction and theory exploration when constructs are measured with multiple indicators. Suited to complex models, smaller samples (with care), and non-normal data relative to covariance SEM.

Process flow

  1. Specify measurement model — reflective or formative constructs, indicators, and higher-order structure if needed.
  2. Specify structural model — paths among constructs (mediation, moderation, controls).
  3. Estimate PLS algorithm — outer loadings/weights and path coefficients.
  4. Evaluate measurement model — reliability, convergent and discriminant validity.
  5. Evaluate structural model — collinearity, R², f², Q² / PLSpredict, significance via bootstrapping.
  6. Report and interpret — effect sizes, mediation, and managerial/theoretical implications.

Acceptable values (measurement)

CriterionTypical rule of thumbNotes / references
Indicator loading (reflective) ≥ 0.708 preferred; ≥ 0.40–0.70 may be retained if content-valid and AVE/CR hold Hair et al. (2019, 2022)
Cronbach’s α ≥ 0.70 (exploratory sometimes ≥ 0.60) Nunnally & Bernstein (1994); Hair et al.
Composite reliability (ρc / CR) ≈ 0.70–0.95 (very high values may signal redundancy) Hair et al. (2019)
AVE ≥ 0.50 Fornell & Larcker (1981); Hair et al.
Fornell–Larcker √AVE of construct > its correlations with other constructs Fornell & Larcker (1981)
HTMT < 0.85 (stricter) or < 0.90 (conceptually similar constructs) Henseler et al. (2015); Hair et al. (2019)

Acceptable values (structural)

CriterionTypical rule of thumbNotes / references
Inner VIF < 5 (often < 3.3 preferred) Hair et al. (2019)
Path significance Bootstrap CI excludes 0 (e.g. 5,000 samples); report p / t Hair et al. (2022)
R² Context-dependent; ~0.25 / 0.50 / 0.75 as weak / moderate / substantial (rough guide) Hair et al. (2011, 2019)
f² effect size ≈ 0.02 / 0.15 / 0.35 small / medium / large Cohen (1988); Hair et al.
Q² / PLSpredict Q² > 0 indicates predictive relevance; compare MAE/RMSE to naive benchmarks Stone–Geisser; Shmueli et al. (2016, 2019)

Key references

  • Hair, J. F., Hult, G. T. M., Ringle, C. M., & Sarstedt, M. (2022). A Primer on Partial Least Squares Structural Equation Modeling (PLS-SEM) (3rd ed.). Sage.
  • Hair, J. F., Risher, J. J., Sarstedt, M., & Ringle, C. M. (2019). When to use and how to report the results of PLS-SEM. European Business Review, 31(1), 2–24.
  • Henseler, J., Ringle, C. M., & Sarstedt, M. (2015). A new criterion for assessing discriminant validity in variance-based SEM. Journal of the Academy of Marketing Science, 43, 115–135.
  • Fornell, C., & Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18(1), 39–50.

Cut-offs are widely used heuristics, not universal laws. Always justify choices for your discipline, sample, and model complexity, and follow the latest journal guidelines.