Data Analyst interview question
How do you distinguish correlation from causation?
Correlation measures association; causation means changing one variable changes the outcome. Establish causality with randomization where possible or a defensible quasi-experimental design, temporal order, and control of confounders.
Direct answer
What a strong answer covers
How it works
Two variables move together for four possible reasons: one causes the other, the other causes the one, a third factor drives both, or chance. Observational data alone cannot tell these apart. A randomised experiment breaks the link with confounders by assigning the treatment independently of everything else. When randomisation is impossible, quasi-experimental designs such as difference-in-differences, regression discontinuity, or instrumental variables can support a causal claim if their assumptions are argued explicitly. Temporal order, dose-response, and a plausible mechanism strengthen the case but do not settle it.
Example
Users who enable notifications retain twice as well, but users who care more about the product are the ones who enable them; an A/B test that randomly prompted half of new users to enable notifications showed a real but much smaller 8 percent lift. A pricing change rolled out by region was evaluated with difference-in-differences against regions that kept the old price.
Common mistakes
- Presenting a correlation from a dashboard as proof that a feature works.
- Ignoring that the users who opt in differ from those who do not.
- Running an experiment and then reading results only for the subgroup that looks good.
A useful answer structure
Answer from your own work.
Do not memorise a script: use only examples you can explain honestly in follow-up questions.
Open the desktop workspace and rehearse a clear answer in your own words.