How to Report Mediation Analysis Results from SmartPLS and AMOS

Mediation results should be reported around four core ingredients: whether the indirect effect is significant, the total effect (IV → DV without the mediator), the direct effect (IV → DV with the mediator included), and whether the pattern reflects full or partial mediation. The practical payoff is straightforward: once those pieces are known, the mediation conclusion and the exact wording for a research paper become largely mechanical.

Indirect effect significance is the first checkpoint. It captures the impact of the independent variable (IV) on the dependent variable (DV) through the mediating variable. Next comes the total effect—how strongly the IV predicts the DV when the mediator is not in the model. Then researchers report the direct effect, which measures the IV’s influence on the DV while the mediator is present (often denoted as the “C complement” idea in the transcript). Finally, the mediation type is determined by comparing direct and indirect paths.

Full mediation occurs when the IV’s total influence runs through the mediator and the direct effect becomes insignificant. Partial mediation occurs when both the indirect path and the direct path remain significant—meaning the IV affects the DV both through the mediator and directly. In the example used to demonstrate reporting, “Team identity” mediates the relationship between “CSR” (corporate social responsibility) and “organizational performance” (OP). The write-up begins by stating the hypothesis: Team identity mediates the CSR → OP relationship.

Results are then organized around the mediation ingredients. The transcript emphasizes that the narrative should explicitly reference the results table (e.g., “Table 1 revealed a significant indirect effect”). For H1, the reporting format includes the indirect effect’s beta value, T statistic, and P value, taken from SmartPLS “specific indirect effects.” It also requires the total effect of CSR on OP to be significant, since total effects represent the sum of direct and indirect influence.

The mediation type hinges on whether the direct effect stays significant after including the mediator. In the example, the direct effect remains significant alongside the significant indirect effect, which indicates partial mediation: some of CSR’s influence on OP passes through Team identity, while some continues through a direct CSR → OP path. The transcript also notes how to present the direct effect path coefficients from the “direct effects” section.

For SmartPLS reporting, the transcript outlines a table structure: total effects (CSR → OP without the mediator), direct effects (CSR → OP with the mediator), and indirect effects (CSR → mediator → OP). For indirect effects, researchers should include coefficient (beta), standard error, T value, P value, and the percentile bootstrap 95% confidence interval (bias-corrected). For AMOS, the same logic applies: use the estimates output to extract total effects, direct effects, and indirect effects. P values can be obtained from the “tailed bias corrected” option, and T values can be computed as estimate divided by standard error when needed. The end result is a clear, defensible mediation conclusion tied directly to significance tests and effect decomposition.