Friday, June 15, 2007

Trial By Jury: Not Black and White?   posted by DavidB @ 6/15/2007 04:32:00 AM
Share/Bookmark

There have been reports in the British Press this week about the influence of ethnicity on the decisions of juries. This is based on a study published by the newly formed Ministry of Justice (formerly part of the Home Office).

The press reports have generally been along the lines that 'juries are not affected by ethnic bias'. The truth, as usual, is more complicated. The research itself is available here (2Mb pdf file) The key part of the report for the present purpose is Chapter 6, on Jury Decision-Making.

Those who read the report will be able to make up their own minds, but I will offer a few comments below the fold.


It is important to understand the methods of the study. It is based on 27 simulated jury deliberations on a single case. The facts and evidence in the case were identical except that in 10 simulations the 'defendant' was white, in 9 he was Asian (which in Britain usually means South Asian), and in 8 he was Black (page 162). (There is also variation in the 'victim' of the crime.)

The 'jurors' themselves were taken from people summoned for jury service at Blackfriars Crown Court in London. The jurors watched the evidence on videotape then, without discussion, individually recorded their provisional verdict. The jury then discussed the case and recorded their final verdicts (individually and collectively). The 'pool' of potential jurors is ethnically mixed, and except for one (all-white) jury, the juries contained a mixture of white and non-white participants, but always with whites in a majority. As the authors stress, it would be hazardous to extend the findings to other parts of Britain or to unmixed juries.

Despite the optimistic headline reporting of the findings, there is in fact evidence of substantial ethnic or racial bias on the part of some jurors. As the authors put it (page 164), 'jurors of different ethnic backgrounds reached significantly different verdicts depending on the race of the defendant. BME [black and minority ethnic] defendants were less likely to be found guilty than white defendants, while the White defendant was much more likely to be found guilty by BME jurors than White jurors.'

The best quantitative indicator of the extent of juror bias is probably the table on page 180, which gives a breakdown of 'guilty' votes by the ethnicity of jurors and defendants. Since the facts of the case were the same, if jurors were unbiased we would expect the distribution of guilty votes with respect to the ethnicity of defendants simply to reflect the proportions of Whites, Blacks and Asians among them, which it will be recalled were 10: 8: 9, or in percentages 37: 30: 33. If then we express the figures in the table as percentages of all guilty verdicts by the jurors in question (using the 'first vote' figures), we get the following:

................................White defendant.....Black defendant....Asian defendant
Unbiased expectation.......37................................30..........................33
White jurors....................45.................................21..........................33
Black jurors.....................55.................................18..........................26
Asian jurors.....................72...................................0..........................28

It will be seen that both Black and Asian jurors appear more lenient towards Black defendants, and more severe towards White defendants, than would be expected by chance. There is a slight indication of leniency towards Asian defendants by Black and Asian jurors, but this may not be statistically significant in view of the small numbers of jurors involved. White jurors seem slightly biased in favour of Black defendants and against White defendants.

These figures cannot simply be explained by jurors being biased in favour of their own group, since Asians are more lenient towards Blacks than towards Asians, and Whites are more lenient towards both Black and Asians than towards Whites. (I do not understand the assertion at the bottom of page 180 that White jurors show some leniency towards Whites in comparison to all non-whites. The authors seem to have forgotten that non-whites make up a substantial majority of defendants in the simulations, so that the combined numbers of 'guilty' votes for non-white defendants from White jurors are still less than would be expected. If anyone thinks I have misunderstood this, please comment.)

The relative percentages of guilty votes might reflect either leniency towards blacks, severity towards whites, or both. It is difficult to disentangle these motives, but this may be partly remedied by looking again at Figure 6.12. If a particular group of jurors were severe towards one group of defendants, without being lenient to others, this would show up in a higher overall conviction rate. There is in fact some indication in the table that Black, and especially Asian, jurors are positively severe towards Whites, as well as lenient towards Blacks. The authors recognise the leniency towards Blacks, which they attribute to a perception by jurors that the justice system itself is biased against Blacks, and a consequent need to compensate for the bias. They do not (I think) recognise any bias of jurors against Whites, and therefore do not seek to explain it.

In view of all these findings the generally optimistic tone of the press reports (and the conclusions of the research report itself) may be surprising. The optimism is based on the fact that the overall final verdicts of the juries do not seem to have been affected by the ethnic composition of the jurors. The authors appear to take the view that the jury system is robust enough to cope with a bit of bias. Perhaps it is, but I suggest two reasons for caution. First, I suspect that the tone of the conclusions would have been less optimistic if the bias of jurors had been against non-whites. Much wailing, gnashing of teeth, and ritual self-flagellation would have ensued. Second, and more important, the lack of practical impact of the jurors' bias on the verdicts of juries may be largely due to the particular features of the real-life case on which all the simulations were based. Although few details were given, it seems to have been a case where there was a serious conflict of evidence and any reasonable jury would have been unlikely to reach a guilty verdict. In the real-life case there was a hung jury (page 164), while in the simulations only one of the 27 juries reached a guilty verdict, and a majority were 'hung'. It is impossible to say, based only on this case, whether the bias of some jurors would have a serious impact on other cases. It only takes 3 jurors to block a guilty verdict in English Courts (see note 257 of the report for details), so there is a serious possibility that in some cases a bias in favour of Black and Asian defendants would lead to people who would otherwise be properly convicted going free.