Statistics, experimentation, causal reasoning · insight-0019
When the total and every segment move in opposite directions, decompose rate and mix before concluding
- Evidence type
- practitioner heuristic
- Difficulty
- medium
- Roles
- Data analyst, Data scientist
- Industries
- aviation, banking, retail
Summary: situation, action, reason, boundary
Situation
Network load factor rose year on year while load factor fell on every individual route.
Action
Split the total change into a within-segment rate effect and a between-segment mix effect, and report both. Then ask which question the decision needs: route performance (use within-route rates) or network outcome (use the total).
Reason
An aggregate is a weighted average; when weights change, the total can move opposite to every component. The causal question, not the arithmetic, decides which view is right.
Boundary
Load factor, conversion, default rates, average fares and any KPI that aggregates segments with changing weights.
Symptom or risk
Leadership is told performance improved when the change came entirely from shifting capacity toward high-load routes, or the reverse.
Exceptions
When segment weights are stable, the total and segment trends agree and decomposition adds little.
Trade-offs
Decomposition adds a chart and an explanation; skipping it risks a confident wrong story.
Pitfalls
Choosing segments after the fact until the paradox appears or disappears.
Example
Hypothetical: shifting seats toward high-load intl-short routes can raise the network load factor while domestic and intl-short load factors both decline.