What is the difference between causation and correlation?

Instruction: Explain the distinction between causation and correlation, and the importance of this difference in data analysis.

Context: This question assesses the candidate's ability to distinguish between correlation and causation, a fundamental concept in statistics and data interpretation.

Updated

Official answer

Correlation describes association. Pearson correlation summarizes linear association on a scale from −1 to 1 when defined; it can miss nonlinear dependence. Causation asks how an outcome would change under a specified intervention, compared with an alternative.

For example, customer count can increase both revenue and support calls. Their correlation does not show that generating more support calls would raise revenue. Reverse causation, common causes and selection can all create misleading associations.

A well-designed randomized experiment can estimate an intervention effect for its population and period. Observational causal analysis needs a credible identification strategy and explicit assumptions about confounding, selection and measurement. Adding a long list of regression controls is not sufficient by itself and can introduce bias if the controls are consequences of treatment.

State whether a result supports prediction, association or a causal decision. A strong predictive relationship can be useful without establishing an intervention effect.

Related Questions