Sample size sanity check
Rules of thumb by analysis type — regression, factor analysis, SEM, group comparison — to check whether your planned sample is in the right range.
- Regression: at least 10–15 cases per predictor, more for small expected effects.
- Exploratory factor analysis: 5–10 cases per item, and rarely fewer than 150 in total.
- PLS-SEM: use an inverse square root power calculation, not the old ten-times rule.
- Group comparison: run a power analysis in G*Power against your expected effect size.
- Qualitative interviews: patterns usually stabilise between 15 and 25 within one segment.