The Correlation Was Right. The Suspect Was Wrong

The Correlation Was Right. The Suspect Was Wrong.

• Monday morning. Production meeting. On screen: a scatterplot.
• The higher the press pressure, the higher the scrap rate. The relationship looks obvious.
• So the team reduces the pressure and restarts production.
• But the scrap remains.
• A second look reveals the real clue: increasing tool wear caused both pressure and scrap to rise.
• Press pressure was not the cause – only the most obvious suspect.
• The correlation was real. The conclusion was wrong.
• Statistical relationships are a good starting point for root-cause analysis. Causality still needs evidence.

Data Literacy lesson:
Process knowledge, additional variables and verification determine whether an observation becomes a reliable explanation