Contact

GlossaryPublic records, privacy, and data quality

Disaggregation

Disaggregation reveals the groups inside a total. Learn how to read breakdowns by place, time, or category without overstating what they show.

Disaggregation means breaking a summary into smaller groups so their differences can be examined. A citywide incident total might be separated by month, neighborhood, or call type. The breakdown must come from supporting data; a total alone does not reveal how its records are distributed among groups.

What a breakdown can reveal

A single figure can conceal different experiences within the same area. The National Forum on Education Statistics explains how examining smaller groups can reveal patterns hidden by combined figures. That general principle is useful when reading public-safety summaries as well. Forum guide presentation, pages 6 and 10.

A breakdown answers a more specific question. “How many calls were recorded citywide?” and “How did call volume change in each service area?” require related but different views. Neither view alone explains why the change occurred.

Example

Hypothetical: A city records 200 service calls in each of two months. In the first month, the north area has 80 and the south area has 120. In the second, the north has 110 and the south has 90.

The total is unchanged, but the distribution shifts. A citywide chart would not show that movement. The breakdown does. Before drawing a trend conclusion, the reader would still need to know whether the area boundaries and reporting coverage remained the same.

Common confusion

Disaggregation is not evidence of cause. A difference between areas could involve population, activity, reporting behavior, service responsibilities, or other factors. The displayed counts do not isolate those explanations.

It also differs from estimating an unavailable breakdown. Dividing a city total among neighborhoods according to population produces an allocation based on an assumption. It does not reveal the actual locations of incidents unless the source data supports that distribution.

Detail has practical limits

More detail requires suitable source fields. Records with unknown locations may need a separate category so the area totals still reconcile with the overall count. Dropping them silently can make a breakdown appear more complete than it is.

Small groups also need care. A change from two incidents to four doubles the count, but represents two additional incidents. Show the counts alongside percentages. Where a publisher withholds a small group to protect privacy, a blank or suppressed value should not be read as zero.

Useful breakdowns retain enough context to reconnect the parts with the whole: the same period, counting unit, and inclusion rules, plus an explanation of missing or withheld groups.

Sources