Contact

GlossaryCrime reporting systems

Estimated crime

Estimated crime uses a documented statistical method to infer totals or rates. Learn why estimates differ from submitted counts and forecasts.

Estimated crime is a crime total, rate, or other measure produced through a documented statistical method rather than only a direct count of available submissions. An estimate can represent a wider population or compensate for missing information. Its meaning depends on what the method is designed to estimate.

Identify the target

BJS and the FBI use NIBRS imputation and weighting procedures to account for nonresponse and missing data. Imputation supplies values through a statistical method; weighting changes how much observed records contribute to the estimate. The resulting products can describe incidents, offenses, victims, or arrestees. BJS program explanation

Those estimates concern crime known to law enforcement. They do not automatically recover crimes never reported to police. A separate victimization survey can estimate covered experiences from respondents, including unreported events, but it has different definitions and population limits. BJS/FBI estimation briefing

An estimated figure should therefore be labeled with its target, geography, period, counting unit, and methodology. “Estimated crime” alone leaves too much unspecified to support a meaningful comparison.

Example

Hypothetical example: A national program receives qualifying incident data from many agencies but lacks complete submissions from others. It applies its published weighting and missing-data procedures to estimate offenses known to law enforcement across the target population.

The estimate can exceed the raw sum of received records without inventing individual police reports. It is an inference about a defined total. It should not be presented as the number of actual case files in the downloaded submission dataset, nor as a complete estimate of all crime experienced by residents.

Common confusion

An estimate is not necessarily a forecast. Estimation can describe a completed period with incomplete observations; forecasting concerns a future period. A provisional direct count is also not automatically an estimate merely because later revisions are expected.

An estimate is not guaranteed to be exact. Sampling, nonresponse, model assumptions, and revisions can affect it. Appropriate uncertainty measures and methodological notes help readers assess how precisely it supports a conclusion.

Comparing results

Use comparable estimate products and their stated confidence intervals or significance guidance when evaluating change. Avoid comparing an adjusted national estimate with an unadjusted agency count as though they share coverage. If methods changed between releases, determine whether earlier values were recalculated. Record the release version so a later revision can be distinguished from a new event or a changed interpretation of the same period.

Sources