Thursday, September 25, 2008

How is crime down?

Over the past few days, the political editor of the Spectator, Fraser Nelson, has been doing a sterling job pointing out how Gordon Brown has basically been lying (Brownies) about the level of debt because the official statistics from the Government say the complete opposite to what he does.

Now, I've never been a brillaint mathmetician, I get by, nor was I particular good (or more correctly "inspired by") statistics. However, I am, I think at least, able to see trends and/or spot bizarre anomolies in sets of data that do not chime with what politicians might say is true, and I'm thinking I've spotted one.

Take the following two tables of data on "Police Manpower" from Hansard. The first lists the number of police officers per 100,000 population in each region of England and Wales between 1997 and 2008. The bottom line is that there are moe police officers now than then against an also growing population.

The second lot of data (split into two tables) shows the total offences per officer from 1997 to 2008. The bottom line of this dataset is that there are now more crimes per officers now than there were back then. So here's where I get confused.

If, according to the Government, there are more police officers per 100,000 of us since 1997, and there are also more offences committed per police officer since 1997. How can it be that crime is, also according Government, down? That doesn't compute. In fact it must mean the upward trend in crime is even greater than it appears because there are more coppers.

I simply cannot see how crime can have gone down whilst the number of police officers and the number of offences they have dealt with each has gone up. It doesn't logically work does it? Or am I missing something?


martini said...

Please tell me you worked this out on the back of a fag packet.

Edward said...
This comment has been removed by the author.
John Scott said...

You are missing something.

The statistics on offences per officer probably use recorded crime.

Recorded crime is no longer the government's preferred measure for determining the actual level of crime. Instead they use the British Crime Survey (BCS). The BCS undertakes widespread representative sampling of people to see who have experienced crime (whether or not they reported it to the police). At present, the BCS shows a fall in crime.

The choice of the BCS was very sensible back in the late-90s when the decision was taken - there was good evidence of under-recording of crime by the police so as to massage the figures for each of their forces (which also explains why offences per officer have risen even if crime has reduced).

Evidently this means the two statistics can move in opposite directions.

These days, there are criticisms of the BCS:
(a) it tends to underrate white collar crime (as victims may never know that the crime has occurred, and even if they do, it can be difficult to extrapolate quite how wide the impact of certain such crimes are);
(b) it entirely omits "victimless crimes" (eg possession of drugs/firearms as well as driving offences) which the police do record;
(c) finally, the BCS does not survey people under-16. That means it inevitably under-estimates crime against young people, when smaller scale surveys show that young people (especially in some areas) tend to have a higher than average risk of being victims of certain offences, particularly street crimes such as robbery. Given that such crimes now result in the theft of relatively high-value items which can be claimed for on insurance policies, the incidence of recording such crimes with the police (so as to obtain a crime number for the insurance claim) has also risen.

Rachel Joyce said...

Dizzy - this does not surprise me at all. It is against public perception to say crime is going down - and I meet victims of crime all the time - many don't report, but many complain that the police do nothing.

"Brownies" is going to become a commonly used term for lying soon everywhere!

Not a sheep said...

Under New Labour; black is white, white is black and all statistics prove whatever they want them to prove. If you don't understand then you will sent to a re-education centre for 42 days of explaining...

Anonymous said...

Its obvious: Iain had a burglary at his home and the rozzers took hours to get there. Then there was his `source` who was in a cop station and heard a mad man confess to murder when the real murderer was already arrested and in a cell in another (County) station. He "stands by every word." He also called the police "institutionally incopmpetent" This is how the crime rate and the increase in police officers has happened at the same time.

I wonder if the 300 + murdered RUC GC. Officers were institutionally stupid before they were murdered, or indeed any copper in GB.

Anonymous said...

Well you could be missing the note on Table 2 that says: "Recorded crime statistics were collected on a calendar year basis up to 1997 and thereafter on a financial year basis. Therefore, these figures are not comparable with those for later years." (My emphasis.)

Or perhaps Note 1 on Table 3 which state: "The coverage was extended and counting rules revised from 1989-99. Figures from that date are not directly comparable with those for 1997 and earlier." (My emphasis.)

Also if you look at the previous page it says in the question that the tables only relate to "notifiable offence[s]" while crime figures used by the Government also include data from the British Crime Survey.

This is because "A large proportion of crime is unrecorded, as many offences are not reported to the police. The propensity of the public to report offences to
the police also changes over time. Thus, statistics recorded by the police may not accurately reflect the underlying trend in all crime."

Anonymous said...

Only explanation I can think of is that more police officers are now involved in each case and so each crime is counted multiple times.

Oliver Mantell said...

It does seem odd. But if the figures the Government uses to argue that total crime is down are from the National Crime Survey (or whatever it's called), which is based on a survey of the experiences of the general population not on crimes recorded by the police, then those crime figures could be down whilst the recorded offences by the police could go up. Having more police officers should, presumably, result in more recorded offences, whilst reducing the amount of crime actually taking place. That'd be what would happened in the best-of-all-possible-worlds, in fact.

Whether they're telling porkies or not, this would be a possible consistent explanation, despite the apparent anomaly that you've spotted...

kinglear said...

Ah yes, lies,damned lies and statistics... or in this case, lies damned lies and spin

John Wilkie said...

You have to check the dates carefully - the table on Police manpower, Hansard gives all the data from 97-2008, but the crimes per officer data go up to 2002 in the first table, you have to go over the page for the 2008 data.

Broadly speaking, police numbers fell 97-02 (as crimes per office rose), from 02-08, police numbers increase and CpO falls. Overall suggestion is that total crimes per 100000 population stays fairly constant

TheFatBigot said...

Ah, Mr Dizzy, I think it works like this:
(1) More police does not mean more police doing policing, it means more police doing forms.

(2) More frustration at police doing forms rather than policing means more petty things being reported and appearing on the statistics as crimes when previously they were just part of the rough-and-tumble of life.

(3) More on-the spot penalties for breathing in a public place means not just more tax revenues but more crimes being recorded by the yellow-bibbed tax collectors / plastic plod.

(4) Legislation creating a new crime every day means no change in behaviour results in more crime.

But, and this is the clever bit, because the Home Office issues edicts that crime must be reduced, any crime which could have been dealt with by a fixed-penalty notice is counted not as a crime but as a breach of civil law.

Then, because the numbers are still too high, there is further fudging by discounting break-ins where nothing was stolen, discounting complaints where the victim is unwilling to go to court to testify, discounting "domestics" unless there was visible physical injury ... and so the list goes on.

dizzy said...

Anon, I did actually see those notes and the problem is the same anomoly exists if you only takes the figures from 1998. Regarding the reference to the BCS, again that still imply the figures are up and not down.

Mark said...

John Scott-

The BCS also doesn't include thefts from commercial premises (ie shoplifting, warehouse burglaries etc.). We were sold a pup when claims about the superiority of BCS stats were accepted.
The other problem with crime stats, particularly on police clear up rates, stems from the fact that our system no longer differentiates between felonies and misdemeanours. The former should always be prioritised, but the relative ease with which the latter can be 'cleared up', combined with the 'target culture', often means petty offences get a disproportionate amount of attention.

Bill H said...

there were two changes one in 1998-99 (which meant that far more offences were recorded by the Home Office - such as common assault for the first time)

There were also changes in 2002-03 where the emphasis was moved to the victim rather than the crime. So if there were three people injured in a fight before 2002-03 it would have counted for one crime, after 2002-03 it counted for three.

So figures aren't really comparable. In theory. But Home Office stats show that the net effect was about 28%. But violent crime rose by 80% b/w 2003 - 2008 so it has increased (even with the changes)

BCS is bollocks, doesn't include under 16's, students, those not living in owner occupied houses, loads of people in high crime areas etc etc.