Measuring Crime: Official Statistics and the Dark Figure

Self-report studies. Ask people, usually under guarantees of anonymity, what they have done. Useful for juvenile and drug offending; unreliable for serious crime.

Regulatory and audit data. For the offences of the powerful, the useful sources are not police records at all: they are the reports of the Comptroller and Auditor General, of regulators, of the food and drug authorities and of commissions of inquiry. This is one reason Module IV of this paper is largely about bodies that are not part of the criminal justice system.

Why the dark figure is patterned, not random

This is the paragraph that connects this chapter to the rest of the paper.

Where the victim does not know. Adulteration, a rigged tender, a false certificate, a diverted subsidy and a mis-sold policy leave a victim who feels nothing at the time.

Where the victim is diffuse. If a company evades a crore of tax, the loss falls on everybody and on nobody in particular, and no individual has the incentive to pursue it.

Where the offender is respectable. A complaint against a doctor, a lawyer, a police officer or a company is harder to make, more likely to be disbelieved, and more likely to be deflected into an internal process.

Where reporting is costly. Sexual offences, offences within a family, and offences by an employer.

And where the police prefer not to record. Registration lowers the station's detection rate if the case is unlikely to be solved, and the incentive to avoid recording is well documented.

Put those together and you get the finding that governs Module III: the dark figure is largest exactly for the offences committed by those with power, so official statistics systematically understate privileged class deviance and overstate ordinary crime.

A worked example

A district records 400 thefts and 2 cases of criminal breach of trust in a year. A student concludes that theft is 200 times as common as breach of trust in that district.

Filter one. A theft victim knows immediately. A person defrauded by an agent may discover it years later or never.

Filter two. A theft is reported because an insurance claim or a lost document requires it. A breach of trust by a known professional is often taken to a family settlement or a professional body instead.

Filter three. Registration of a theft is routine. Registration of a breach of trust invites an argument that the matter is civil, and that argument is frequently accepted at the station.

Filter four. Where a scheme involves fifty investors, it may be recorded as one case.

The honest conclusion. The figures show that theft is reported and recorded 200 times as often as criminal breach of trust. They do not show that it is committed 200 times as often, and the correction is in an unknown but certainly large direction.

What beginners get wrong

"Crime rate" needs a denominator and a definition. Per lakh of population, of what offences, recorded by whom, on what counting rule.

A fall in recorded crime is not proof of success, and a rise is not proof of failure. Both are consistent with a change in reporting or in recording practice, and a police force rewarded for low figures will produce low figures.

The dark figure is not an argument for ignoring statistics. It is an argument for knowing which direction the error runs. Recorded crime is a reliable guide to serious violence, where reporting is high, and an unreliable guide to fraud, corruption and offences within families.

Quick revision

• Five filters between crime and crime statistics: perception, reporting, recording, classification and aggregation.
• The dark figure is crime committed but never recorded, and it is patterned rather than random.
• Sources: police records, court and prison figures, victimisation surveys, self-report studies, and regulatory and audit data. India has no continuous national victimisation survey.
• Prison figures measure the output of the whole system, not offending.
• The dark figure is largest where the victim does not know, where the loss is diffuse, where the offender is respectable, where reporting is costly, and where the police prefer not to record. That is exactly the profile of privileged class deviance.
• Recorded crime is a fair guide to serious violence and a poor one to fraud, corruption and offences within families.

Test yourself

1. What is the dark figure, and why does it matter that it is not random? It is the volume of crime committed but never recorded. It matters that it is patterned because a random error would leave comparisons between offence types valid; a patterned one does not. Since the figure is largest for the offences of the powerful, official statistics systematically understate those offences relative to ordinary crime, which distorts both public perception and the allocation of enforcement.

2. Name the five filters between an offence and its appearance in a statistic. Perception by someone that an offence occurred; willingness to report; recording by the police; classification as an offence and as which offence; and survival through aggregation, where a counting rule that records only the most serious offence in an incident removes the rest.

3. Why are prison statistics a poor measure of offending? Because they are the output of every earlier filter combined: who was noticed, reported, recorded, arrested, refused bail, charged, convicted and not released. Each stage selects, and the selection favours the poor and the unrepresented, so the prison population describes the operation of the system rather than the distribution of offending.