December 14, 2020

Are juries fair? Ten years ago, at the end of the court day in the Bar Mess at the Old Bailey, Professor Cheryl Thomas delivered the results of a mega-study into jurors’ decision-making. When she came to the question of conviction rates at different court centres, laughter erupted: every barrister knows that juries at Snaresbrook Crown Court — drawn from Hackney, Dagenham, Chigwell — love to let the villains off, whereas a Kingston-upon-Thames tribunal — from places like Barnes, Richmond and Epsom — will rarely award a defendant anything other than a silver medal. Awareness of this phenomenon goes back at least to the 1980s, when it was raised in parliament by the MP for Ilford South.

Except, of course, it turned out to be nonsense: for 2006-2008, the conviction rates for Snaresbrook and Kingston were within 1% of each other, either side of the national average of 63.9%. I suspect that this did little to dent the audience’s confidence in one of the profession’s most dearly held beliefs: arch references to Kingston juries still abound in London’s robing rooms.

But this was not the only surprise Professor Thomas delivered. The evidence, which came from analysis of over half a million criminal charges as well as a randomised trial with real jurors, strongly supported the proposition that white juries do not discriminate against black defendants. Indeed, it even appeared that a white defendant accused of a racially motivated crime is more likely to be convicted by a white jury than by a racially mixed one.

Those findings do not, of course, call into question the unfairnesses suffered by the non-white population upstream of trial — if anything they could be said to emphasise them, given that non-white people are over-represented in prisons. But they did allay the fears expressed in the Auld report of 2001, which made a recommendation (not subsequently implemented) that “Provision should be made to enable ethnic minority representation on juries where race is likely to be relevant to an important issue in the case”. Whether the findings have entrenched themselves in the public consciousness more generally, however, is another matter.

Last month, Professor Thomas — now an honorary QC — published the results of a new study, commissioned by the senior judiciary to investigate “rape myths”. Her conclusions are thoroughly at odds with received wisdom.

The idea that many people — and therefore many jurors — hold seriously wayward beliefs about this uniquely enraging crime is not new, and has solidified in recent years. It first became common in the early 2000s for judges to warn juries to approach with caution defence counsel’s arguments about, for instance, a lack of physical violence, or a delay in making a report. Such warnings were swiftly approved by the Court of Appeal and found their way into the “Bench Book” — the judges’ bible of approved jury directions.

The need for these warnings arose from a developing understanding that inaccurate assumptions about how rape allegations should be analysed are widespread among the public. The 2010 Bench Book pointed to “research by those who are expert in the subject”, much of which derived its data from surveys. One cited author concluded that “the crucial determinant of judgments about rape cases is the extent to which people subscribe to rape myths. Thus it tends to be attitudes rather than an unbiased evaluation of the facts which determine judgments in these cases.

And that message has grown louder in the decade since. In 2018 End Violence Against Women (EVAW) commissioned a YouGov poll and reported that “33% think that sex which happens without consent, but where there is no physical violence, isn’t really rape.” Horrifying to think that any jurors have that attitude, let alone a full third of them. No wonder, then, that judicial warnings have become more robust.

For example, on the question of inconsistencies (which is of course a way to get purchase on the evidence in any trial) 2010’s bench book includes the following direction for sexual offences: “The experience of the courts is that it is unwise to approach the issue of inconsistency with the assumption that a true account is always consistent, or that an inconsistent account is always untrue. It depends on the circumstances and the individual.” And concludes with, “If…you are sure that the essential parts of the complainant’s account are true, you will no doubt act on that conclusion. But, if you are left in doubt about the truthfulness of the complainant’s account, because the inconsistencies cannot be satisfactorily explained, you must find the defendant Not Guilty.

Fair enough. But the equivalent part of the 2020 version runs as follows: “Experience has shown that inconsistencies in accounts can happen whether a person is telling the truth or not”. And the line about doubt arising from inconsistencies that can’t be satisfactorily explained has disappeared.

Now, these are only short extracts, and much of the thrust is between-the-lines and therefore open to debate. But these warnings are not there primarily to help the defence. Some of them are felt to amount to a nudge in the other direction. Which might be fair — desperately needed even — if jurors believe rape myths. But it now looks like they don’t.

The UCL Jury Project surveyed almost 800 recently discharged jurors and found that “hardly any jurors believe what are often referred to as widespread myths and stereotypes about rape and sexual assault”. They don’t believe that rape must leave bruises, that a victim will always fight back, that dressing “provocatively” is inviting an assault, or that rapes will always be swiftly reported. But haven’t previous studies repeatedly found the precise opposite?

The reason for the disparity, the Professor suggests, is that her study involved actual jurors rather than members of the public. The jurors had heard trials of non-sexual as well as sexual offences, and of those approached 99% agreed to participate, which means a very low risk of selection bias. Something may be happening, she says, when a member of the public actually goes through the solemn process of sitting on a jury.

Another possible reason for the disparity, it seems to me, is that conclusions from other research don’t always accurately reflect the questions asked. Take the 2018 EVAW headline I mentioned. The small print reveals that the question, with which 33% of respondents disagreed, was, “Do you personally consider it to be rape if they don’t want to have sex but feel pressured to even though there was no other physical violence?” Leaving aside the confusing and redundant “other” (what was the first bit of violence again?) this is absolutely not identical to “Do you think that sex which happens without consent, but where there is no physical violence, isn’t really rape?” Indeed, the 2020 Bench Book even says, “reluctant but free agreement is not the same thing as submission and is still consent even if reluctantly given.”

But why does it matter? Aren’t judicial warnings a good idea just in case? Well, sometimes the defence’s arguments — about delays and injuries and intoxication and inconsistencies — are good arguments. And if the jurors aren’t prejudiced, then a warning about rape myths could easily be taken as a veiled encouragement to ignore a good point. And bad defence arguments? They’ll surely be counterproductive with a myth-free jury.

Shortly before Professor Thomas published her results the Crown Prosecution Service updated and expanded its list of rape myths, with a consultation on their accuracy and usefulness ongoing. Shortly after that, EVAW published another report, entitled “The Decriminalisation of Rape”. Under the heading “Obstacles to Conviction” is listed the very fact of trial by jury, and it goes on to state, “it has been widely accepted by criminal justice bodies that many members of the public continue to believe in long-standing myths and stereotypes relating to rape”. A lively public conversation seems inevitable — and desirable.

Also listed as an obstacle to conviction, interestingly, is the English “adversarial system” of criminal trials. Adversarial systems are having a bit of a moment: the software that creates those amazingly realistic faces of non-existent people, for instance, uses a type of Artificial Intelligence called a General Adversarial Network: progress is made by one algorithm trying to create something convincing, and another that tries to find the holes. This is very much not the tech equivalent of a continental inquisitorial magistrate. A bad defence point is best dealt with by a good prosecution point. It may turn out that the medieval jurists who designed our legal system were onto something.

And a similar process — a dialectic approach to enquiry more generally — is arguably a better route to the truth than any other. There has until now been very little questioning of the prevalence of rape myths among juries, and it’s not hard to see why: who wants to be “that guy”? But Professor Thomas’s research should be welcomed. Her methodology, and her conclusions, might turn out to be wrong. But they should not be ignored.