In a sparsely populated area, an aggregated count can still point to exactly one person

What it is

Area aggregation hides people only when a cell contains enough indistinguishable records. In sparsely populated places, rare-event data, or narrow time windows, an aggregate can still correspond to one person or household. This small-cell disclosure means “aggregated” is not synonymous with “non-identifying.”

Why it happens

The same spatial resolution yields radically different candidate counts in a city and a rural area. Public locations, local news, and acquaintance knowledge can link “one event in this town on this day” to a person. Multiple seemingly safe tables can also be differenced to recover suppressed values. Privacy follows the smallest candidate set, not the dataset's average density.

Where it stops holding

Large cells are not automatically safe: a homogeneous sensitive attribute may be inferred once membership is known. Sparse data does not inevitably cause harm; sensitivity and an attacker's background knowledge matter. Uniform administrative boundaries aid communication but need not provide uniform protection.

Applying it

• Calculate the minimum count for every area, time window, and allowed filter combination; do not substitute a map-wide average.
• Merge, suppress, or delay low-count cells, and prevent totals minus subgroups from restoring hidden values.
• Verify on the sparsest areas by simulating linkage from known public events. If it narrows to very few people, increase granularity or restrict access.

Related

• Same group: U12.06.1 Precise coordinates can reveal homes and daily routines · U12.06.2 Aggregating to areas can reduce re-identification risk · U12.06.4 Location jitter introduces false conclusions at small scales · U12.06.5 Aggregation granularity must be set by the sparsest region, not average density
• Nearby: U8.05.2 Small counts should not be published as precise values · U12.06.2 Aggregation also loses local analytical detail
• Search terms: small-cell disclosure · sparse data privacy · spatial uniqueness