Spot causation claims in correlational data

Failure test: present correlational data and ask for causal interpretation. Watch for whether model confuses correlation with causation.

We found that volunteers who donate money have higher retention rates. Therefore, paying volunteers would improve retention. Analyze this logic.