Report PLS-SEM results for publication

PLS-SEM papers follow a predictable reporting structure, and reviewers check it point by point. A complete, well-ordered results section is one of the easiest ways to improve your chances. This lesson brings together everything from the course into a reporting template.

The methods section

• Why PLS-SEM: the research goal (prediction, theory development), model complexity, formative constructs or the need for construct scores, with citations.
• Sample size: justification with the inverse square root method or a power analysis.
• Measures: source of each scale, number of items, response format, and reverse-coded items.
• Software and settings: PLS-SEM Studio (cite it), weighting scheme, stop criterion, bootstrap subsamples, CI type and test type.
• Common method bias: procedural remedies and statistical checks.

The results section

1. Measurement model. Table: loadings (and weights for formative constructs), α, ρA, ρC and AVE. Text: thresholds and any deleted items.
2. Discriminant validity. Table: HTMT matrix with confidence intervals (and Fornell–Larcker if expected).
3. Structural model. Inner VIF; a hypothesis table with β, t, p, 95% CI, f² and the decision; R², adjusted R², Q² and PLSpredict.
4. Mediation and moderation. Indirect effects with CIs and mediation types; interaction terms with simple slopes and a slope plot.
5. Additional analyses. MGA, IPMA, NCA or robustness checks.

The figure

Every PLS-SEM paper has a path diagram with the results. In the Diagram tab, choose the label format (β with stars, or β with the CI), show loadings on the item arrows, show R² inside the endogenous constructs, and save as SVG (vector, for journals) or PNG at 300 dpi. Solid arrows are significant, dashed arrows are not. Keep the layout identical to your conceptual model, so readers can compare.

Exporting everything

The Export tab saves all tables at once: an Excel workbook (every table on its own sheet, with numbers kept numeric), a Word report with formatted tables and the diagram, a self-contained HTML report, CSV files for R, SPSS or Stata, LaTeX tables, the full results as JSON and the latent variable scores. It also writes a methods paragraph that describes your actual settings; adapt it rather than copying it verbatim. Keep the exported workbook and the saved project file with your thesis data, so you can answer reviewer questions months later.

Ten mistakes reviewers flag

1. Justifying PLS-SEM only with a small or non-normal sample.
2. Reporting only Fornell–Larcker for discriminant validity.
3. Deleting items to push AVE or reliability over a threshold without reporting it.
4. Treating formative constructs like reflective ones (reporting alpha and AVE).
5. Reporting p-values without confidence intervals or effect sizes.
6. Interpreting R² against the marketing guidelines without context.
7. Claiming “prediction” without PLSpredict or other out-of-sample evidence.
8. Claiming full mediation without including the direct path.
9. Comparing groups by comparing p-values instead of running MGA after MICOM.
10. Using goodness-of-fit (GoF) indices, which are not valid for PLS-SEM.

A compact results write-up of the course model

The measurement model met all criteria: loadings ranged from 0.814 to 0.932, ρA from 0.852 to 0.909, ρC from 0.903 to 0.941 and AVE from 0.723 to 0.843, and all HTMT values were below 0.85 (maximum 0.544, upper CI bound 0.632). In the structural model (all inner VIFs < 1.17), performance expectancy (β = 0.351), effort expectancy (β = 0.224), social influence (β = 0.185) and trust (β = 0.181) affected behavioural intention (R² = 0.413, Q² = 0.332), which affected use behaviour (β = 0.440; R² = 0.241, Q² = 0.179); all p < .001. Effort expectancy also had a small direct effect on use (β = 0.107, p = .043). Intention mediated the effects of all four drivers on use, and trust strengthened the effect of performance expectancy on intention (β = 0.211, p < .001; index of moderated mediation = 0.093, 95% CI [0.049, 0.140]).

That completes the course. The final assessment covers all nine lessons; pass it to download your certificate.