« When mental algebra goes wrong.... | Gene Expression Front Page | Written Language »
February 27, 2005

Crime and Punishment

I linked recently to a Sunday Times article by Rod Liddle. His main point was that

There are some things that you can say and there are some things that you can’t say. Paradoxically, they are sometimes the same things.

For example, you can’t say that black people commit more crimes, but, using the same facts, you can say that they are victimised by the criminal justice system. And you can’t say that young black males are, for whatever reason, not academically able, but, using the same facts, you can say that the school system is failing them.

The article wasn’t great science, or even great journalism, but I thought it amusingly illustrated the absurdities of current bien-pensant discourse. So I was interested to see that a recent Crooked Timber post contained a diatribe against Liddle. It wasn’t specifically aimed at this article, but it linked to a blog by Matthew Turner, which ’ably exposed’ Liddle’s article. So I naturally checked out Turner’s blog, partly to see what he says about Liddle, but also to find out what the guys at Crooked Timber reckon is an ‘able exposure’….

So here’s what Turner says about the article:

His [Liddle’s] Sunday Times column yesterday… is probably best left to fester in its own idiocy. Nevertheless it's amusing to note how it begins,

For example, did you know that black and Asian women commit far more crime than their white counterparts? Almost one third of the total female British prison population is drawn from black and Asian communities.

Because one-third is obviously far more than two-thirds. No, hang on, it's not is it?


And that’s the ‘able exposure’, in full. Well, there’s glory for you - in Humpty Dumpty’s sense.

Needless to say, Liddle wasn’t suggesting that black and Asian women commit the majority of crimes, just that they are disproportionately represented among criminals. In his own words, they have ‘an apparently greater propensity… to commit crime‘. And he went on to give some figures:

Black and Asian women make up 8% of the general population and 29% of the female prison population…

Did Turner not read the rest of the column? Or did he read it and decide to misrepresent it? Dishonest, or just an idiot?

The source of the 29% statistic, as Liddle makes clear, is the impeccably liberal organisation the Fawcett Society, which campaigns for ’women’s rights’. But that doesn’t mean it is correct, so I did some checking. As so often happens, the facts turn out to be more complicated.

First, the Fawcett Society probably got the statistic from a Home Office summary of prison statistics for 2002, which states that ’ethnic minorities make up 22% of the male prison population and 29% of the female population‘. So far so good.

But on looking at the full report on Prison Statistics in England and Wales for 2002 (apparently the latest available on the Home Office website - see here for a copy as a 2Mb pdf file) things get murkier. The report gives an ethnic breakdown of women prisoners into four categories: White, Black, South Asian, and ‘Chinese and other’. In the text (para 6.1) the proportions of women prisoners are given as Black 24%, South Asian 1%, and ‘Chinese and other’ 5%, leaving by implication 70% for Whites. However, more accurate figures are given in Table 6.1, from which I calculate the proportions (to one decimal place) as White 70.6%, Black 23.9%, South Asian 0.9%, and ‘Chinese and other’ 4.7%.

One complication is that a very high proportion of women in prison are there for drug offences. Among Black women it is especially high, at 75% (para 6.12). The Courts are reluctant to jail women, especially mothers, for crimes like petty theft and assault. The figures for women prisoners therefore give an unrepresentative picture of women’s crime in general.

A second major point is that a substantial proportion of women prisoners are visitors to the UK, notably Caribbean drug mules arrested on entering the country. If the aim is to examine the criminal propensities of ethnic groups living in Britain, these ought to be excluded. Unfortunately the statistics do not provide a reliable way of doing so. The closest we can get is to exclude foreign nationals, who are separately identified for each ethnic group (Table 6.3). This undoubtedly excludes too many, as there are numerous foreign nationals living permanently in the UK. However, for what it is worth, I have calculated the number of British nationals in the different ethnic groups as a proportion of all British nationals among female prisoners (that is, excluding foreign nationals from both numerator and denominator). This gives the proportions (to one decimal place) as White 84.4%, Black 11.8%, S. Asian 0.7%, and ‘Chinese and other’ 3.3%. Note that the ‘Black’ share is dramatically reduced - in fact, halved - though as already noted, the adjusted figure will underestimate the true proportion among those living permanently in Britain.

Of course the raw proportions mean little unless they can be compared with proportions of different ethnic groups in the general population. Rod Liddle quotes a figure of 8% for Black and Asian women in the general population, which is about right for the overall proportion of non-whites, of all ages, in the population of Britain in the 2001 Census. However, this is not the most appropriate comparative figure, first because it is a figure for Britain (England, Wales and Scotland), and we want a figure confined to England and Wales to compare with the prison figures. More important, we really need a figure for the age groups most likely to be in prison. The ’all ages’ figures are likely to be substantially misleading, because the proportions of various ethnic groups in the population vary with age.

I suggest that the most appropriate age group for comparison is roughly age 16 to 35. I have extracted figures for women of ages 16 to 34 inclusive from the 2001 Census report (Table S101). On this basis I calculate the percentages of different ethnic groups in England and Wales as:

White___All non-white___Black___Mixed____S. Asian___Chinese or other

87.9_______12.1________3.0_____1.5______6.1_______1.5

It will be seen that the Census, unlike the prison statistics, has a ‘Mixed’ category, which is broken down into various sub-categories (white-Caribbean, white-Asian, etc.) It is not clear how mixed-race women are covered in the prison stats, but I would guess, from the size of the ‘Chinese or other’ category, that some have gone into that group, though possibly also the prison service have lumped Arabs, Kurds, etc, into ‘Chinese or other’ rather than ’white’. This creates yet another difficulty of interpretation, and I suggest it is best to ignore both the Mixed and ‘Chinese or other’ group.

Comparing the two sources - and omitting the doubtful categories - we can summarise them as follows:


___________White___All non-white___Black____S. Asian

In Prison_____84.4_____15.6________11.8_____0.7

In general
Population____87.9_____12.1_________3.0_____6.1


What is clear, even after all allowance is made for complicating factors, is that Black women are still heavily over-represented in the prison population, but that South Asian women are heavily under-represented. I think most of us would have guessed this anyway, but it’s nice to have confirmation.

So it seems that the real crime of Rod Liddle - or rather of the Fawcett Society - is to lump together two groups whose characteristics are radically different. In fairness to Liddle, much of his article is concerned precisely with this point.

It may still fairly be said that the proportions of people in prison are not a perfect guide to the number of criminals. Not all criminals are prisoners, and not all prisoners are criminals (some are detained on remand, and some convicted prisoners are innocent). And it is possible that the police and justice system operate differentially in relation to different ethnic groups. Numerous studies have been carried out in Britain to investigate possible discrimination in the justice system, with complicated and conflicting results. Some of the issues and evidence are discussed in this Home Office report. There is some evidence that black people are more likely than other groups to be convicted, and to receive longer sentences, for similar offences. This is a difficult point, as the behaviour of suspects affects their treatment, and different groups show different behaviour in police custody. As the HO report notes, when interrogated ’white suspects were far more likely to provide admissions [of guilt] (in 58% of cases) than either Asians (48%) or black people (44%)… The relatively low admission rate among black people in particular increases the chances of them entering the formal criminal justice system because those who deny their guilt are ineligible for a caution’ (page 73). Blacks were also more likely to exercise their right to remain silent when questioned (page 78). There is also evidence that in trials for serious offences blacks are less likely to plead guilty than whites (page 167), which would tend to increase their sentence if convicted, since there is a ’discount’ for pleading guilty.

It would be nice, of course, to have an accurate source of information, independent of the criminal justice system, on the numbers of crimes actually committed by different ethnic groups. Naturally, no such source exists. However, there is some evidence in the British Crime Survey, based on interviews with the general public on their experience as victims of crime. This report based on data from the 2000 Survey indicates that of those respondents who could describe the ethnic group of the offender, 5% of offenders were black, 3% were Asian, and 2% were ‘other‘ non-white. As compared with the proportions of youngish (16-34) people in the 2001 Census (see above) these show an overrepresentation of blacks and under-representation of Asians. Of course these figures mainly relate to personal assaults and robberies where the victim is able to see the offender.

Need I say - yes, I suppose I do - none of the above expresses or implies any opinion on whether differences in criminal behaviour among ethnic groups have any genetic basis.

Posted by David B at 03:50 AM