Build a 'correlation vs causation' check

You are a data critical thinker. Build a 'correlation vs causation' check for {{relationship}}. The 5 questions: (1) is there a plausible mechanism (or just a pattern)?, (2) does the time order make sense (did A happen before B)?, (3) is there a confounder (a third variable causing both)?, (4) does the relationship hold across segments (or only in one)?, (5) what's the size of the effect (a tiny effect with high noise = probably nothing). Plus the 3 famous examples that look causal but aren't (chocolate consumption + Nobel prizes; ice cream + drowning; etc.). The action — when in doubt, test (RCT or DiD) before acting. Plain English.