Table 1
Conversion between some effect sizes.
| small | medium | large | |
|---|---|---|---|
| d | 0.20 | 0.50 | 0.80 |
| r | 0.10 | 0.24 | 0.37 |
| f | 0.10 | 0.25 | 0.40 |
| η2 | 0.01 | 0.06 | 0.14 |
| AUC | 0.56 | 0.64 | 0.71 |
Figure 1
Main window of G*Power calculating the power of a two independent samples t-test.
Figure 2
Sensitivity Plot of G*Power calculating the power of a two independent samples t-test: Lowest detectable effect size as a function of required N.
Figure 3
Sensitivity Plot of G*Power calculating the power of a two independent samples t-test: Power as a function of required N for fixed effect size.
Figure 4
Example of derivation of the effect size based on its constituent parameters: paired t-test case.
Figure 5
One-way ANOVA computation of effect size indexes and power in G*Power.
Table 3
Example of 2 × 2 design expected results.
| Case 1 | Case 2 | ||||
|---|---|---|---|---|---|
| A1 | A2 | A1 | A2 | ||
| B | replicated | 5 | 2 | 5 | 2 |
| moderated | 0 | 0 | 2 | 5 | |
[i] Note: Pooled standard deviation is equal to 1.
Figure 6
Example of a 4 × 2 expected interaction based on a one-way observed pattern of means.
Figure 7
Geometrical interpretation of the interaction beta coefficient, with a dichotomous moderator (a) and a continuous moderator (b).
Figure 8
Example of power analysis for interaction in moderated regression.
Figure 9
Mediation model.
Figure 10
Shiny app for Monte Carlo power analysis (Schoemann et al., 2017).