An experienced risk consultant once told me that reporting on uncertain statistical analysis could end up being 'a game of smoke and mirrors'. During another project, a highly qualified computer science researcher advised me to always do 'a pencil and paper exercise' on expected inputs and outputs before embarking upon any expensive program of development work. How do these words of experience relate to the newly published Jury Research findings?
As an example of a simple ‘pencil and paper exercise’, we can take one aspect of the recent Scottish Jury Research work which was carried out by Ipsos MORI, namely the question of ‘what difference does the not proven verdict make?’. Putting the question in broad terms, we have a group of people (the jury) who are asked a question. There are three options for the answer; affirmative, negative and uncertain (Guilty, Not Guilty and Not Proven). The question relates to an assessment (trial) which is ‘finely balanced’, so if we ask any number of people the question, we may get a ‘finely balanced’ response which could perhaps be, on average, one third affirmatives, one third negatives and one third uncertain. Just how definitive of ‘finely balanced’ this suggested split may be is questionable, as there is every possibility that in such cases there will be more uncertainty and thus more Not Proven votes, however this is discussed further below. If we then pose the same question with exactly the same set of circumstances but remove the ‘uncertain’ option, the outcome of a such a ‘finely balanced’ question is likely to be that half of the people will be in favour of each answer (the very definition of ‘finely balanced’).
So, in the first instance we have one third of the people voting affirmative, (Guilty) and in the second instance, half of the people are voting affirmative. A rise in the number of affirmative or Guilty votes thus appears to be an inevitability when the Not Proven vote is removed in a ‘finely balanced’ scenario such as the Ipsos MORI trials were said to be.
This analysis is reinforced by uncertainties over what Not Proven actually means. Gordon Jackson Q.C. stated on STV’s Scotland Tonight program during discussion of the findings that nobody knows what Not Proven means but logically it can only mean one of three things to any particular juror, namely that the juror:
A) believes the accused is guilty but the case has not been proven,
B) believes that the accused is innocent, and the case has not been proven,
C) believes that the accused may be either guilty or innocent but either way the case has not been proven.
We can debate how likely each of these three variations may be, but they nevertheless represent the full range of possibilities of what a juror could believe Not Proven means. In any case, and in accordance with the Ipsos MORI report, they collectively mean one thing – uncertainty.
Looking at how the Jury Research findings were represented in the Executive Summary, in answer to the question:
‘What difference does the not proven verdict make to verdict choice?’
Ipsos MORI’s Key findings (page viii of the report, enumerated here for reference) were:
1. ‘Where the not proven verdict was available, acquitting juries tended to choose not proven rather than not guilty as the means to acquit the accused.’
2. ‘Individual jurors were also less likely to favour a guilty verdict when the not proven verdict was available.’
Looking at these findings in reverse and simplifying and contextualising the text in order to clarify:
2. Where Not Proven is not an option, more jurors are likely to vote Guilty.
All things considered as discussed above, in a ‘finely balanced’ trial, a three-way split typically means less people voting each way than in a two-way split, this is a statement of the obvious. This is particularly true if the option being removed is the one signifying uncertainty in meeting the standard of proof, because those voters will then split two ways.
1. In ‘finely balanced’ trials, more jurors voted Not Proven than Not Guilty when acquitting.
I would argue that a simple ‘pencil and paper exercise’ considering the uncertainty aspects and the full range of meanings of Not Proven, as described above, predicts this. Further, more detailed ‘pencil and paper exercises’ developing on this kind logic have been presented previously in Scottish Legal News and other e-publications (references below). It is interesting to note that in one of those articles (featured in the 2018 SLN Annual Review entitled ‘Hung Jury – The De Facto Third Verdict?) I referred to the Oscar Slater trial, noting this most infamous miscarriage of justice ‘…turned on a narrow majority of nine guilty votes, with one not guilty vote and no less than five for not proven…’ indicating that Not Proven can also be the preferred option of those jurors seeking to acquit in a ‘finely balanced’ Guilty verdict. This speaks to Not Proven as being a vote for uncertainty, although ‘prejudicial effects’ will no doubt bring other smoke and mirrors into play in this case.
In concluding their findings on how the Not Proven verdict is regarded and used by jury members, Ipsos MORI state (on page 57):
‘Removing the not proven verdict might incline more jurors towards a guilty verdict – and might, therefore, lead to more guilty verdicts over a larger number of trials.’
I would go further and make the following, more unequivocal statement, based on my pencil and paper exercises here and in previous articles:
Removing an option that indicates uncertainty at meeting the legal standard of proof, i.e. Not Proven, will lead to more guilty verdicts. Furthermore, a hung jury indicates uncertainty at meeting the legal standard of proof as does anything short of a unanimous verdict. Where there is no provision for such expression or demonstration of uncertainty, there is a related increase in likelihood of miscarriages of justice.
(Elaborations of Lorimer’s logic on the above issues can be found in his previous SLN articles and at: https://www.abdn.ac.uk/law/blog/posts-by/david-lorimer/)