Not proven: getting the numbers right



Professors James Chalmers, Fiona Leverick and Vanessa Munro take issue with recent claims about how often and in what sort of case the ‘not proven’ verdict is used.

In a recent piece for Scottish Legal News, Alistair Bonnington criticised calls to scrap the not proven verdict, making a remarkable claim which deserves further attention. He said the following:

“The most recent available figures shows that sheriffs use the ‘not proven’ verdict ten times as often as jurors. This gives the lie to the argument advanced by those against ‘not proven’. They wish to make out that the verdict is a cop-out for indecisive and inexpert jurors. They pretend that the verdict is most often used in sex crime cases by jurors. The facts show that argument to be false. ‘Not proven’ is used much more by sheriffs in road traffic cases, for example, than in sex crimes.”

Leaving aside whether this actually is the argument for abolishing ‘not proven’ (at best it is a caricature of one argument for its abolition), is this claim about the figures correct? Mr Bonnington’s article does not identify the “most recent available figures”, but he may have in mind the Scottish government’s statistical bulletin Criminal Proceedings in Scotland, the most recent version of which gives figures for 2018-19. The problem here is that these figures say something rather different.

As table 2(b) of the bulletin shows, 10 per cent of acquittals in “motor vehicle offences” – the category used in these statistics – are ‘not proven’ verdicts. That compares with a figure of 32 per cent in the category of “sexual crimes” and an even higher figure of 40 per cent for rape and attempted rape.

It is difficult to see how anyone could look at these figures and claim that sheriffs use ‘not proven’ ten times as often as jurors. It might be, of course, that what Mr Bonnington has in mind is the raw numbers – there are, after all, far more summary prosecutions than there are solemn ones. But even there, the numbers do not stack up, although explaining why requires digging into the data in a bit more detail.

The bulletin does not break down the use of ‘not proven’ between solemn (jury) and summary (non-jury) cases. The Scottish government did, however, release jury-specific figures for several years in response to a parliamentary question in 2018. In the most recent year for which data was given (2016-17), there were 359 ‘not proven’ verdicts and 818 not guilty verdicts in solemn procedure. From the overall figures in the Criminal Proceedings bulletin for the same year (table 2(a)), we can deduce that this left 875 ‘not proven’ verdicts and 5080 not guilty verdicts in the summary courts (the published data does not allow us to break this down between sheriffs and JPs).

So it is certainly not true – or even close to true – that sheriffs use ‘not proven’ ten times as often as juries. On the raw numbers, we can say that the summary courts (sheriffs and JPs taken together) use ‘not proven’ almost two and a half times as often as juries. But this is a misleading comparison, because they decide far more cases: we might equally point out that they use not guilty over six times as often as juries do. A clearer comparison can be demonstrated with percentages, which show that in choosing between verdicts of acquittal, juries are significantly more likely to opt for ‘not proven’ than sheriffs and JPs. Presented that way, the figures given here show that in 2016-17 31 per cent of jury acquittals were ‘not proven’ verdicts, whereas only 15 per cent of summary acquittals were.

What about Mr Bonnington’s more specific claim that the ‘not proven’ verdict “is used much more by sheriffs in road traffic cases, for example, than [by juries] in sex crimes”? Here, we have raw numbers in the most recent Criminal Proceedings bulletin (table 2(a)). In 2018-19 there were 152 ‘not proven’ acquittals in the “sexual crimes” category, as compared to 72 in the “motor vehicle offences” one. Of course, not all these prosecutions for sexual crimes will have been in solemn cases, just as some of the motor vehicle cases will have been in the JP court rather than the sheriff court. But at the very minimum all of the 68 ‘not proven’ verdicts in prosecutions for rape and attempted rape must have been jury verdicts. Mr Bonnington’s claim is, on the published data, impossible.

Mr Bonnington is right to highlight the fact that the debate over ‘not proven’ focuses particularly on jury trials, but it is unsurprising – and surely right – that a debate such as this should focus on the most serious of criminal cases. He is also absolutely right that attention should be paid to the question of whether removing ‘not proven’ should lead to changes to the majority requirement. And he is correct, as he says in his final paragraph, that we should “keep to the facts”.

James Chalmers, Fiona Leverick and Vanessa Munro are professors at the Universities of Glasgow and Warwick, and were co-investigators on the Scottish Jury Research project.