How to Report #Moderation Analysis Results from #AMOS and #SmartPLS Output

Moderation reporting hinges on showing how much an interaction term improves model fit (via R² change), whether the interaction is statistically significant (via t/critical ratio and p-values), and what the moderation looks like in practice (via slope analysis). The transcript lays out a step-by-step workflow for reporting moderation results from both AMOS and SmartPLS, with a particular focus on calculating and interpreting F² effect sizes—an often-missed piece when writing up moderation.

For AMOS, the key technical requirement is to report R² included versus R² excluded, where “included” means the interaction term (IV × moderator) is part of the model and “excluded” means it is removed. Using squared multiple correlations in AMOS, the example values are R² included = 0.413 and R² excluded = 0.350. Plugging these into the F² formula—(R² included − R² excluded) / (1 − R² included)—yields F² ≈ 0.063/0.587 ≈ 0.107. Interpreting that effect size using Kenny’s thresholds (0.02–0.05 small/medium and 0.15+ large as commonly cited; the transcript uses 0.051 and 0.25 as cutoffs), the moderation effect is treated as large, meaning the moderator contributes meaningfully to explaining the endogenous construct.

SmartPLS follows the same logic but avoids manual F² calculation. After running bootstrapping, SmartPLS outputs F² directly. In the example, SmartPLS reports F² = 0.024, which the transcript interprets as a medium effect size under Kenny’s guidance. The practical takeaway is that both software outputs should be reported consistently: use R² change (included vs excluded) for variance explained, use the interaction’s significance for hypothesis testing, and use F² for effect size.

The write-up structure is then made concrete. Start by stating the moderation hypothesis in a form that specifies direction and mechanism—for instance: role ambiguity (RA) moderates the relationship between collaborative culture (CC) and organizational performance (OP) such that increased role ambiguity weakens the positive CC→OP link. Next, report the R² change: without the interaction, R² = 0.350 (35% variance explained), and with the interaction, R² rises to 0.413 (41.3%), an increase of 6.3% attributable to moderation.

Then comes the significance test. The interaction term’s estimate is negative and significant, with a critical ratio (t-value) exceeding 1.96 and a corresponding p-value, supporting H1. The transcript ties this directly to interpretation: a negative beta means higher role ambiguity weakens the CC→OP relationship.

Finally, slope analysis (Figure 1) clarifies the pattern. The line is steeper for low role ambiguity (blue), indicating CC has a stronger positive association with OP when role ambiguity is low. For high role ambiguity (red), the line flattens, showing that increasing CC no longer boosts OP as strongly. The conclusion summarizes the moderation: higher role ambiguity weakens the impact of collaborative culture on organizational performance, with F² reported as evidence of moderation magnitude and a table capturing estimates, standard errors, t-values, and p-values for the interaction and related paths.