The crime report¶

Somebody telephones the police. A call handler listens, decides what has happened, and decides whether it amounts to a notifiable crime. If it does, a record is opened, and that record is one unit of the national crime statistics. If it does not, nothing is opened, and the call leaves no mark on any figure anybody will later publish.

The test is written down. Under the Home Office Counting Rules an incident is recorded as a crime if, on the balance of probability, the circumstances as reported amount to a crime defined by law and there is no credible evidence to the contrary. It is a judgement, made quickly, by one person, at the point where the public meets the institution.

The under-recording¶

In 2014 Her Majesty’s Inspectorate of Constabulary reviewed 10,267 reports of crime and published the result. Its summary opens without hedging: “Victims of crime are being let down. The police are failing to record a large proportion of the crimes reported to them. Over 800,000 crimes reported to the police have gone unrecorded each year. This represents an under-recording of 19 percent.”

The rates were worst where the offences were gravest. Violence against the person, 33 per cent unrecorded. Sexual offences, 26 per cent. Of 3,246 decisions to cancel a record as a no-crime, 664 were wrong, including over two hundred rapes. In more than eight hundred of those decisions the inspectors could find no evidence that the victim had been told.

Correctly classified, never recorded¶

One finding in that report reads like a reprieve and is the most interesting thing in it. “We were reassured to find little evidence of the misclassification of crime. Our audit showed that 96 percent of crime records reviewed were classified correctly.”

So the machinery downstream of the decision worked. Once something was a record, it was put in the right category ninety-six times in a hundred. The rules were understood, the categories were applied, the paperwork was sound.

What failed was upstream of all of that, at the moment somebody decided whether there was anything to classify. A system can be scrupulous about sorting what enters it and still lose a fifth of what should have entered, and the two facts do not contradict each other, because they are facts about different steps.

The sentence the inspectorate wrote carefully¶

Somewhere in the introduction the inspectorate addresses why the public does not believe the figures. A probable cause for scepticism about national crime figures, it says, is the culture in the police, “as in other parts of the public sector”, of “pursuing targets and being under pressure to demonstrate good performance”.

The sentence offers the target culture as a cause of the scepticism, not as the cause of the under-recording. The inspectorate had audited a nationally representative sample of 6,880 reported crimes and could say precisely how many crimes went unrecorded. It could not say, and did not claim to be able to say, how many went unrecorded because somebody was thinking about a number.

That restraint is the correct scientific posture and it also describes the difficulty. The decision not to record is made once, quickly, by one person, and leaves behind exactly nothing to audit. An unrecorded crime has no record. Motive is not recoverable from an absence.

The label¶

The statistics regulator did the one thing available to it. It removed the National Statistics accreditation in 2014, “following an assessment which found evidence that the quality and consistency of the underlying data may not be reliable”.

The figures continued to be collected, published, quoted in Parliament and reported in newspapers. What changed was a designation. A decade later it has not been restored.

The clerk’s brief¶

2024: The accreditation has not come back¶

The Office for Statistics Regulation reviewed the quality of police recorded crime statistics, setting ten recommendations across three areas and describing the review as a first step towards the statistics being considered for reaccreditation. The clerks note the interval: ten years without the label, the figures published throughout, and reaccreditation still described as something to be considered rather than scheduled.

2014: The accreditation is withdrawn¶

The regulator removed National Statistics accreditation from police recorded crime, “following an assessment which found evidence that the quality and consistency of the underlying data may not be reliable”. The clerks observe that this is the whole of a statistics regulator’s power over a number produced by somebody else: it may decline to call it a National Statistic, and it may say why.

2014: One in five¶

Her Majesty’s Inspectorate of Constabulary published Crime-recording: making the victim count, reviewing 10,267 reports of crime, auditing a nationally representative sample of 6,880 reported crimes, and finding an under-recording rate of 19 per cent, over 800,000 crimes a year, with violence against the person at 33 per cent and sexual offences at 26 per cent. Of 3,246 no-crime decisions reviewed, 664 were incorrect, including more than two hundred rapes. Ninety-six per cent of records that did exist were classified correctly. The clerks note that the inspection found the recording defective and the bookkeeping sound, and that these are separate findings about separate steps.

2002: The standard arrives¶

The National Crime Recording Standard was introduced in 2002 to bring consistency to the decision about whether a reported incident becomes a recorded crime, and the 2014 inspection sets out what it aims to achieve. The clerks record that a standard governing the decision had existed for twelve years by the time anybody measured how often it was followed.

The absence that has no record¶

Every number in this file describes something that did not happen: a report that did not become a record, a record that was cancelled, a victim who was not told. The inspectorate reached them by auditing the reports rather than the records, which is the only route by which a missing thing can be counted at all. The clerks’ standing assessment is that a statistic produced by discretionary decisions at the point of first contact cannot be checked against itself, that checking it required somebody to go and look at the raw material, and that nobody does this annually.