Decide whether a difference is real with a shuffle test, fit a line and know what it means, and catch the confounder behind a seductive correlation.
3 units, about 3 hours. Free.
Run a permutation test and interpret a p-value as a statement about chance under no effect.
2Derive a least-squares fit from scratch and read slope, intercept and residuals critically.
3Manufacture a spurious correlation from a confounder and dissolve it by stratifying.