Sources of Crime Data (UGC-NET Unit VI)

The first classification NET tests is primary versus secondary data. Primary data is collected directly by the researcher for a specific question, such as a fresh victimisation survey fielded for one study. Secondary data already exists, built by an agency or an earlier study for a different purpose, and the researcher reuses it. Most criminological data is secondary.

The most common secondary source is administrative data: police incident and offence records, court case files, and correctional intake and release logs. Each agency records what it needs for its own operational work, not what a later researcher would ideally want, so these records describe the justice system's decisions (who got reported, charged, convicted, released) as much as they describe underlying offending.

A candidate should be able to name police, court and correctional records as the three core administrative streams and state that each is filtered by discretion and attrition at its own stage.

Above individual agency records sits a layer of aggregated national series built for public reporting and cross-time comparison. National statistical offices and law-enforcement agencies publish these as periodic reports or as open data portals that anyone can query. Such series are convenient for trend questions but inherit every limit of the administrative records they aggregate, plus new ones from how categories are defined and revised over time.

A newer source category is digital trace data: records generated automatically as a by-product of online or networked activity, such as social media posts, mobile location logs, or server logs from a platform used to plan or commit an offence.

Nobody designed a survey instrument to produce this data, which is both its appeal (scale, no recall bias) and its limit (no sampling frame, uneven access across platforms, and consent and privacy questions that administrative and survey data do not raise in the same way).

A cohort study follows a defined group over time rather than drawing a fresh sample at each wave. In criminology this usually means a birth cohort (everyone born in a given place and year) or an offender cohort (everyone who came into contact with the justice system in a given period). Because the same individuals are tracked repeatedly, a cohort study can show onset, persistence and desistance directly, which a single cross-sectional snapshot cannot.

Victimisation and self-report surveys sit alongside cohort designs as another secondary source once collected: researchers other than the original survey team reuse the data to ask their own questions of it. Surveys ask people directly about crimes experienced or committed, independent of whether an incident was ever reported to police, which is why they surface offences that never appear in administrative records at all.

Record linkage combines two or more separate administrative datasets by matching the records that belong to the same individual, using either a shared unique identifier or probabilistic matching on fields such as name, date of birth and address. It lets a researcher trace one case across stages, for example from a police contact through to a correctional outcome, that no single agency's data covers on its own.

Every source above shares one vulnerability: the quality of a secondary dataset is fixed by decisions the researcher had no part in. Missing fields, inconsistent classification over time, and gaps where an agency simply did not record a variable are the norm, not the exception, and comparability across jurisdictions is weaker than the headline numbers suggest because categories and recording practices differ. A NET answer that names a source without flagging its quality limit is only half the answer this bullet expects.