03 Section B Option 1, Key question 1 - Measuring crime (part 2)

What crime figures count

A crime figure depends on what is counted, whose experiences are included and how information is collected.

Official statistics are statistics produced by public bodies. This describes their institutional source, rather than one particular research method. Police recorded crime and the Crime Survey for England and Wales (CSEW) are both official statistical outputs. The first uses administrative records made through policing; the second estimates victimisation from a sample survey. It is therefore misleading to divide all crime evidence into “official statistics versus surveys”. Some surveys are official.

Always identify the unit of measurement. An incident count describes events; a victim count describes people or households affected; an offender count describes people committing offences. These cannot simply be substituted for one another. One person may experience several offences, and several people may commit one offence together. Prevalence means the proportion experiencing something at least once during a stated period; incidence concerns the number of incidents, sometimes expressed per a given number of people or households. These distinctions are used in the ONS explanation of crime measurement.

For an original illustration, imagine ten residents each experiencing one bicycle theft and one further resident experiencing three. There are thirteen incidents and eleven victims. A measure counting whether each person was a victim counts the repeatedly victimised person once. An incident measure captures the additional events. Neither number is a mistake: they answer different questions.

The useful methodological question is validity: how well does a measure capture the phenomenon it claims to describe? A record may accurately describe an offence recorded by a force while being a poor basis for claiming to count every offence committed. Reliability, or consistency of measurement, is different: applying the same categories repeatedly does not guarantee that hidden incidents become visible.

Police recorded crime

Police recorded crime measures offences that enter the police recording system under its recording rules.

Police recorded crime is a count of recorded notifiable offences, rather than a count of convictions or all rule-breaking. “Notifiable” refers to the offences included in the statistical collection. An offence can be recorded without anyone being caught or convicted. The ONS quality and methodology report, revised July 2026, explains the collection and its coverage.

Two processes matter. Reporting brings an incident to police attention, although officers also discover offences themselves. A victim may not report because they fear retaliation, feel embarrassed, consider the loss too small or expect little help. Another may report a theft because an insurer requires a crime reference. Reporting can therefore differ between offences and social circumstances. Recording is the subsequent classification and entry of the information as a crime. An incident coming to police attention does not automatically produce an equivalent statistical record.

The advantage is extensive administrative detail. Recorded data can identify local concentrations and help services plan responses. They cover offences beyond household victim surveys, including offences against businesses and homicide. For rare, serious offences, a national sample might contain too few cases for a useful estimate; the police collection can therefore be particularly valuable. These are advantages for particular purposes, not proof that every category is equally complete.

A disadvantage is that changes in records can arise from changes in behaviour or changes in detection, reporting and recording. For example, targeted patrols might detect more drug possession without an equivalent rise in possession itself. New recording guidance can also change consistency across time. This makes a simple before-and-after comparison difficult. The ONS account of diverging survey and police trends explains why recorded trends require care.

Consider an original worked example in which the actual number of assaults stays constant. A support campaign makes victims more willing to report, and improved training makes staff more likely to record qualifying reports. The recorded count can rise because a larger share of existing assaults enters the series. It would be wrong to infer that crime must have risen by the same proportion. It would also be wrong to infer that none of the rise could reflect more offending: the records alone cannot separate these possibilities.

This is a sociological link between measurement and social construction: institutions make decisions that turn information into categories and records. It does not mean that victims' experiences are imaginary or that every police record is false.

Victim surveys and the CSEW

Victim surveys ask a sample about experiences of crime, including experiences that never reached the police.

The CSEW's adult survey asks people aged 16 and over living in households in England and Wales about the previous twelve months. It produces weighted estimates rather than interviewing everybody. Weighting adjusts the contribution of different respondents to improve the sample's match to the target population. Its adult household coverage is important: it does not represent all people, places or offences. The current design and exclusions are set out in the ONS user guide, revised 23 July 2026.

The survey can reveal part of the dark figure by asking about incidents regardless of whether they were reported. Its consistent approach is especially useful for national trends in the offences and population it covers, because the estimates do not depend on police recording practices. It can also collect information about experiences and reasons for not reporting, allowing researchers to understand the process behind a recorded total.

Nevertheless, a victim survey is not a complete crime census. Undercoverage occurs when relevant groups or offences lie outside the design. The CSEW does not cover crime against businesses or people outside its household population, such as those living in institutions. It cannot interview a homicide victim, and victimisation questions do not measure drug possession in the same way as a survey of people's own offending. A survey of household theft therefore cannot establish the total volume of all crime.

Survey answers also depend on memory and disclosure. Recall error can mean forgetting an incident or its timing; bringing an older incident into the reference period is called telescoping. Social desirability means giving an answer that seems acceptable, rather than an accurate answer. Someone interviewed at home may avoid disclosing abuse, especially without privacy. Confidential self-completion can help, although it does not guarantee disclosure. ONS therefore uses separate self-completion measures for sensitive experiences such as domestic abuse rather than treating the headline interview count as complete. Completing a victim questionnaire privately does not turn it into a survey of the respondent's offending. ONS discusses these response errors in its property-crime methods guidance.