Decide whether a boundary change permits an area trend

Fix the geography as well as the time interval

A name can stay the same when a boundary changes. Look up the geographic vintage for the source product, then inspect the relevant boundary or relationship files. Classify each source area as unchanged, split, merged or unresolved relative to the target boundary. A geographic identifier alone is not evidence of a stable population.

Choose a valid transformation

If unchanged smaller areas exactly partition the target and counts are additive, aggregate the counts and derive the rate afterward. If a source area is split, a population-based allocation weight may support an approximation for some variables. Land-area shares alone generally do not describe where residents live. Never apply a count-allocation rule directly to medians.

Hypothetical boundary bridge

Area A has 100 eligible residents; area B has 50. An independently supported crosswalk assigns 80% of A and all of B to the new area. The allocated denominator is 80 + 50 = 130. That is an estimate under the crosswalk assumption. If the event distribution within A differs from its population distribution, using 80% of A’s events can bias the resulting rate.

Keep the unallocated balance explicit: 20 residents from A belong outside the target. If the balances do not reconcile or weights lack a defensible basis, publish separate maps by vintage instead of reporting an apparent trend. Document the source and target definitions and any allocation uncertainty that the survey margin does not capture.

Primary source checked 7 November 2026: U.S. Census Bureau: ACS geographic boundaries by year.

Use this in your study: Catchment allocation reconciliation · Download: Catchment count concordance.