Statistics, experimentation, causal reasoning · insight-1025
Do not read a market-level correlation as a statement about how individual customers behave
- Evidence type
- documented fact
- Difficulty
- easy
- Roles
- Data analyst, Data scientist
- Industries
- aviation, public-sector, retail
Summary: situation, action, reason, boundary
Situation
Across 60 markets, higher average ancillary spend goes with a higher business-traveller share, and a manager concludes that business travellers spend more on ancillaries.
Action
State the unit of the claim. If a statement about people is needed, use person-level data or within-market comparisons; otherwise word the conclusion about markets and avoid causal verbs for customers.
Reason
A correlation between group averages depends on how people are distributed within groups and need not match the correlation among individuals.
Boundary
Region, route, store and cohort-level dashboards used to explain individual behaviour.
Symptom or risk
Aggregate relationships can differ from, or reverse, the individual-level one because composition and group-level confounders drive them.
Exceptions
Decisions that are themselves made at the market level, where the aggregate relation is what matters.
Trade-offs
Person-level data may be costlier or restricted.
Pitfalls
Adding more markets, which narrows the interval without fixing the unit mismatch.
Lessons and concepts
Claims and their sources
Correlations computed from group-level (ecological) data can differ sharply from the corresponding individual-level correlations, as shown in the 1930 US census state-level foreign-born and literacy data.