Juries are often thought of as a quintessential institution of English law, just as the investigating magistrate is a quintessential institution of French law. Most members of the public, even those who have never come into contact with the legal system, have a basic idea of what a jury is.

And the institution of the jury has often been romanticised as part of our national mythos. The right to trial by a “jury of one’s peers”, “twelve good men and true”, has historically been held up as a safeguard of the Englishman’s liberty. Many defenders of jury trial quote the famous words of clause 39 of Magna Carta: “No free man shall be seized or imprisoned, or stripped of his rights or possessions, or outlawed or exiled, or deprived of his standing in any other way, nor will we proceed with force against him, or send others to do so, except by the lawful judgement of his equals or by the law of the land.”

In fact, however, the use of jury trials in England and Wales today is rather limited. They are principally used in criminal cases, but juries decide only a small proportion of these. Many indictable offences are not in fact dealt with by a jury, for example, because the defendant pleads guilty prior to trial. In short, thousands of people every year are arrested, charged and convicted without ever seeing a jury.

Although juries were historically used in civil trials in England, the use of civil juries today is extremely limited. Many English civil lawyers go their entire careers without ever appearing before a jury. However, as a large proportion of my practice has involved actions against the police, I have more experience of civil jury trials than most lawyers do.

Coroners in England and Wales can also summon juries for inquests. Most inquests take place without a jury, but a jury is required in certain cases, such as deaths in custody and deaths caused by the police. The task of a coroner’s jury is of course very different from that of a jury in a criminal or civil trial and in many ways even more challenging.

When we look beyond the UK to other parts of the Commonwealth, we see that the use of jury trials is inconsistent. Some Commonwealth jurisdictions, such as India, Singapore and South Africa, have abolished juries altogether. However, in the Commonwealth Caribbean, where I practise, trial by jury continues to be used for serious criminal cases. The country most associated with jury trials today is the United States, where jury trials are used for both criminal and civil cases, and in both federal and state courts.

Prejudice and bias 

Many people are deeply attached to juries as a safeguard of liberty and a bulwark against state power. However, others are sceptical. In recent decades, much of the criticism of juries has concerned questions of bias.

Many people are understandably concerned that when guilt or innocence is decided by 12 members of the public who bring their own prejudices to the table, they will treat some defendants less favourably than others. For example, you might reasonably assume an all-white jury might harbour some biases against a black defendant.

In fact, however, the evidence paints a mixed picture as to whether jury verdicts are racially biased or not. There is some evidence that they are not, and this was a key point in David Lammy’s review of ethnic minorities in the criminal justice system. But the point is debated.

However, bias is not the whole story. We don’t just want to know whether juries are biased against particular groups. We also want to know whether they’re getting it right. After all, a legal system that decided guilt or innocence by tossing a coin would also show no racial bias, but that doesn’t mean it would be getting the right answers.

And this is a much more difficult question to answer. We can measure empirically whether juries are more likely to convict members of particular groups. But we can’t measure empirically whether juries are making the right decisions. To state the obvious, we don’t have any objective measure of whether the people they are convicting are actually guilty or whether the people they are acquitting are actually innocent. And unlike judges, juries don’t give reasons for their decisions so we can’t assess the quality of their reasoning.

In deciding whether we need juries, we also have to look at the alternatives to jury trials. Even if juries are getting it wrong, it doesn’t necessarily follow that professional judges or lay magistrates who also have their own biases, prejudices and failings, would be more likely to get it right.

Jury selection 

The first thing we want to look at is racial bias in jury selection. It’s interesting to contrast the systems in England and Wales with the very different systems in the United States.

The practice of allowing peremptory challenges to jurors, where either party can reject a certain number of jurors without providing cause, was abolished in England and Wales by the Criminal Justice Act 1988. This contrasts with the United States, where peremptory challenges to jurors are still allowed in the federal courts and in the courts of most states. Although the Crown retains a traditional right to “stand by” a juror, the Attorney-General’s guidelines on the use of the stand-by power make clear that it is to be used only exceptionally, and only in national security or terrorism cases.

So, the parties in an English or Welsh jury trial have less ability to influence the composition of the jury than those in an American jury trial. However, both the prosecution and the defendant do have the ability to challenge jurors for cause, and the judge does have discretion to stand down a juror (for example, because there is an appearance of bias or because they are not competent to serve).

Up until the late 1980s, English judges did occasionally use their discretion to discharge jurors in such a way as to ensure that juries were multi-racial. However, the Court of Appeal held in 1989 that a judge has no power to do this (R v Ford [1989] QB 868). So, a defendant of colour has no right to be tried by a racially representative jury.

Let’s look at bias in jury selection. In the United States this is a familiar subject. In its 2021 report Race and the Jury: Illegal Discrimination in Jury Selection, the Equal Justice Initiative highlights data from a range of American federal and state courts showing evidence of racial disparities at all stages of jury selection.

Many American jury pools under-represent people of colour. And both peremptory challenges and challenges for cause are often used disproportionately to exclude black jurors and jurors of colour. Although racial discrimination in jury selection is theoretically illegal, the Equal Justice Initiative argues that the legal tests laid down by the appellate courts have made it difficult to prove racial bias, and there continues to be a massive race disparity at every stage of the process.

It also reviews evidence from American studies suggesting that all-white juries are biased against black defendants. But as we have seen, we can’t just read those findings across to England and Wales and assume that they apply here, because our juries are very different.

Diversity study 

So, let’s turn to the English data. We’re going to start with Cheryl Thomas’ 2007 study Diversity and Fairness in the Jury System. For the purposes of discussing this study, I’m going to adopt Thomas’ own terminology when discussing race, including the use of the umbrella term “black and minority ethnic”, to accurately represent what the study shows. This is not necessarily the language that I would have chosen myself. This study approached jury diversity and fairness in several ways. I’m just going to highlight a few headline points that are important for our purposes.

Thomas studied all stages of the jury selection process. In her survey of 84 Crown Courts, in all but two there was no statistically significant difference between the proportion of black and minority ethnic jurors summoned and the black and minority ethnic population in the court catchment area. There was also no evidence of disparity when broken down by ethnic group.

However, she did highlight that for the majority of Crown Courts in the country, the black and minority ethnic population in the court catchment area is below 10%, which means in practice there is little likelihood of black or minority ethnic jurors serving on a jury at these courts. In other words, all-white juries tend to happen at these courts not because the selection process is biased, but simply because of the demographics of the local population.

She also highlighted, however, that some courts with an overall low population of black and minority ethnic people did have high concentrations of ethnic minorities in some parts of their catchment area. So, in these areas a person of colour might well end up being tried by an all-white jury, not because the summoning process was biased but because of the overall demographics of the court catchment area.

Consistent findings 

Thomas also looked at those who were actually selected to serve on a jury, as opposed to being disqualified or excused. She found in most of the 84 Crown Courts, the proportion of black and minority ethnic people serving on juries was generally consistent with the proportion of black and minority ethnic people in the court catchment area. There were only three Crown Courts where black and minority ethnic people were significantly under-represented on juries. In addition to analysing the jury selection process, she also carried out a case simulation study using real jurors summoned to Blackfriars Crown Court, selected in the same manner that a real jury would be selected.

The case, which was based on a real case, involved a male defendant accused of punching a male victim in the face after a confrontation outside a bar. In real life, the case had resulted in a hung jury. The facts of the case were kept the same in each simulation, but the race of the defendant and the victim was varied to black, white or Asian. In some simulations the defendant was charged simply with assault occasioning actual bodily harm, while in others he was charged with racially aggravated actual bodily harm.

Thomas found that in the 54 separate jury decisions in the study, outcomes for the defendants were remarkably similar regardless of race. Whether Asian, black or white, the defendants were almost always either found not guilty by a majority verdict, or the outcome was a hung jury. Some racial disparities emerged in relation to how individual jurors voted, but these did not result in any racial disparities in the overall jury verdict.

Thomas carried out a second study in 2010, Are juries fair? In this study, she ran the same case simulation, but this time at two courts with a mostly white catchment area, Nottingham and Winchester. She found that the all-white juries in this study were not more likely to convict a black or minority ethnic defendant than a white defendant. This held true for both black and Asian defendants.

This methodology has two obvious limitations. First, it’s possible that jurors who know they are participating in a study might be less inclined to display racial bias than jurors in a real case, whose deliberations are secret. Second, Thomas’ case simulation concerned a case that resulted in a hung jury in real life, meaning it may not be representative of most cases that come before the courts.

That said, Thomas’ 2010 study also carried out a large-scale analysis of jury verdicts. She looked at all cases in all Crown Courts in England and Wales from 1 October 2006 to 31 March 2008, including over half a million charges. This data showed white and Asian defendants both had a 63% jury conviction rate, while black defendants had a 67% jury conviction rate. Thomas described this as a small difference, although it is notable that black defendants were slightly more likely to be convicted.

For the Lammy Review, Thomas’ research was updated, with analysis of over 390,000 jury decisions between 2006 and 2014. Again, this found very similar conviction rates across ethnic groups. White, black, Asian and mixed-race defendants were all convicted at rates of between 66% and 68%. Lammy was very enthusiastic about this. His report called juries a “success story of our justice system”.

Judges and magistrates 

The Lammy review contrasted juries with judges and magistrates. Lammy cited a 2016 study by Hopkins et al which looked at racial disparities in Crown Court judges’ sentencing decisions. Hopkins found that under similar criminal circumstances the odds of imprisonment for black, Asian, and Chinese or other offenders were higher than for white offenders. The disparity varied a lot between different offence types. It was particularly high for drugs offences, where black and minority ethnic offenders were 240% more likely to be sent to prison than white offenders. Although the data was limited, the Lammy Review also found some racial disparities in magistrates’ courts verdicts. In particular, black and minority ethnic women were more likely to be found guilty than white women.

So, Lammy’s conclusions about jury trial were overwhelmingly positive. However, his approach to this issue has been criticised. One powerful criticism is this. We know that there is massive racial disproportionality in terms of who ends up in the criminal justice system to start with. Black people are therefore more likely to end up before a jury than white people.

As Thomas herself highlighted in her 2007 study, black people from 2006-08 made up 14% of all jury verdicts, compared with 3% of the population. We could reasonably infer from this that black people are more likely to be falsely accused of crimes than white people, and therefore we might expect a genuinely fair trial process to have a lower-than-average conviction rate for black defendants, rather than a slightly higher one.

Lee Bridges argues that the Lammy Review’s approach “carries an implication that those making decisions at later stages in the process have no role or responsibility for seeking to redress unfair treatment of particular groups at earlier stages.” There is also evidence that black and minority ethnic defendants are more likely to plead not guilty than white defendants, and are also more likely to be committed to the Crown Court for trial if charged with an offence triable either way. This, again, complicates the picture, because arguably Thomas and Lammy are not comparing like with like.

Other bias 

So, more research is needed on this subject. At this stage, we can’t be certain to what extent racial bias affects jury verdicts. However, we can probably at least say that there is more evidence of racial bias on the part of judges and magistrates than on the part of juries. And it’s also important to note that racial bias isn’t the only kind of bias. For example, it’s often been alleged that juries’ gender biases affect the outcomes of rape trials.

In her 2010 study, Thomas argued that there were misconceptions in this area, pointing to the fact that other serious offences have lower jury conviction rates than rape. But her study certainly isn’t conclusive on this question. Nor do we know to what extent juries are biased against LGBT people: for instance, we don’t know how the current wave of hostility towards transgender people in the UK might affect jury behaviour in cases with transgender defendants or victims. But again, we shouldn’t assume without evidence that professional judges are any less biased than juries.

The right answers? 

Whether a tribunal is biased on the basis of race is not the same question as whether it is getting the answers right. The law reports reveal an instance in 1995 where a jury attempted to communicate with one of the victims of the offence by means of a Ouija board, and another in 1736 where the jury decided their verdict by “hustling half-pence in a hat”. To state the obvious, neither of these methods is racist, but both of them are unreliable.

As explained earlier, there is no straightforward way to measure empirically whether juries are getting the right answers. We have data on how many people are acquitted and convicted. But, again to state the obvious, we don’t have data on how many people are actually guilty or actually innocent.

And unlike judges, juries don’t give reasons for their decisions. They don’t explain what they made of the evidence, or how they reached their conclusions. So, when faced with a jury verdict, we don’t know whether they understood the law or the evidence properly, or whether their reasons made sense, or whether they based their decision on false assumptions.

Understanding instructions 

Nor do we know whether they fully understood the judge’s legal directions.

Thomas’ 2010 study, which we have looked at earlier, also investigated whether the jurors in her case simulations understood the legal directions they were given. She found that while most jurors thought the judge’s legal instructions were easy to understand, a majority in fact did not completely understand them in the terms used by the judge in his instructions. For that matter, we don’t know whether jurors based their decision on whose counsel they found more attractive, or whether they just wanted to get the deliberations over with and go to the pub.

In her 2017 Blackstone lecture, Lady Justice Hallett recounted a 2003 fraud case at Southwark Crown Court where “a female juror sent to prosecuting counsel a bottle of champagne and an invitation to a dinner date with the question ‘what does a lady need to do to attract your attention?’” Another concern is whether juries are influenced by the unattractive facts of a particular case. For example, a jury might be thought to be more likely than a judge to be influenced by adverse publicity in a case involving a high-profile case or defendant.

In a case involving graphic sexual abuse, the jurors’ feelings of disgust may influence their decision-making. Or a jury might be unimpressed by a defendant who is relying on a technical defence to escape liability for acts of which the jury disapproves. All these arguments are raised by Penny Darbyshire, who in a 2014 article makes a powerful argument that defendants ought to be able to opt for a bench trial instead of a jury trial, as they can in some other common law jurisdictions.

Assessing credibility

In many trials, the trier of fact, whether it’s a jury, a judge or a bench of magistrates, has to assess the credibility of witnesses and decide between competing versions of events.

Where witnesses give different accounts, the trier of fact has to decide who is mistaken or who is lying. This is an inherently very difficult task. And the way that juries are currently expected to perform this task is not particularly effective. For example, lawyers are trained to look for inconsistencies between a person’s evidence in court and their previous statements, and to use those inconsistencies in cross-examination to show that the person is lying. The assumption is that an inconsistent account is more likely to be a lie.

But in fact, we know that this is not true. There is an excellent article by Hilary Evans Cameron, Refugee Status Determinations and the Limits of Memory, which summarises a large amount of empirical research on this topic. She shows that human memory for temporal information, such as the exact date something happened, how often it happened, the order it happened or how long it took, is extremely poor. For the most part, we reconstruct that kind of information by inference, estimation and guesswork, rather than actually remembering it.

These issues are exacerbated in the case of mental health conditions. Many people who come before the courts have experienced traumatic events in their lives and have conditions such as post-traumatic stress disorder (PTSD) and depression, which significantly affect memory and concentration. Likewise, there is evidence that autistic people may be more likely to be wrongly judged as deceptive or lacking in credibility. A person’s demeanour may also be affected by their cultural background. So, it’s very risky to assume that you can tell whether someone is lying by the way they behave when being questioned.

Professional judges are certainly not immune to making false assumptions about memory and demeanour. Judges are not trained in psychology or psychiatry, but in law and a legal education by itself does not give you any special insight into human nature. So, we shouldn’t assume that judges are any better at getting the right answers than juries.

Robot jurors

Finally, we should consider one of the more outlandish ideas that has been floated: replacing jurors with artificial intelligence. I’m not an AI expert, but I would have very grave concerns about whether such a system could ever be genuinely fair or just. An AI system is only as good as the data we human beings feed into it. And the datasets on which AI systems are trained can often inadvertently impart biases which have profound effects on decision-making. For example, a 2021 New York Times article highlighted an incident where Google Photos automatically sorted photographs of a black man into a folder marked “gorillas”. Similarly, the article quotes a black computer expert, Deborah Raji, who was working on a content moderation system designed to remove pornography from social networks.

She noticed the system was being trained to distinguish pornography from non-pornography by comparing anodyne stock photos with images from online pornography sites. Because the people in the stock photos were mainly white and the people in the pornographic images were not, the system was unconsciously being trained to identify images of black people as pornography. And another black computer expert, Joy Buolamwini, found that the facial recognition system at her workplace would not recognise her face – but when she wore a white mask, it did.

None of this is to say that AI isn’t useful. But it doesn’t always remove human biases from decision-making. Sometimes it replicates and exacerbates them. And given what we have already learned about how people may make false assumptions about a witness’s credibility, we can see that the same false assumptions could well affect an AI-based courtroom.

For instance, an AI trained to analyse a person’s body language might well misidentify a person with autism or PTSD as a liar because their body language might not match the programmers’ expectation of how a truthful witness behaves. In this scenario we wouldn’t have eliminated our human biases. We’d simply have automated them.

Conclusion

I want to wrap up by saying this: to an extent, the jury’s still out on juries. We don’t really know with any certainty whether racial bias is a significant factor in jury decision-making. Although Thomas’ research is often held up as proving that juries are race-blind, there are good reasons to question this conclusion. Nor do we know with any certainty whether juries are getting the answers right, a question which is very difficult to test empirically. But before throwing out the institution of trial by jury, we should keep in mind that trial by professional judges is not necessarily a better alternative. A legal education does not give a person the wisdom of Solomon.

And while we don’t know to what extent juries are racially biased, there is good evidence that judges are. Nor should we subscribe to gimmicks such as replacing juries with artificial intelligence. It may well be that the key to reducing bias and improving rigour in the justice system lies in changes to the trial process and the rules of evidence, rather than in replacing trial by jury. ■

Professor Leslie Thomas KC is a lecturer at Gresham College in central London. The above is an extract from a recent public lecture ■