The Scottish Jury System in 2020

The Scottish Government’s summary of mock jury research stakeholder events together with SLN’s Year in review round-up prompted a reconsideration of a number of this year’s prominent legal issues in relation to jury trials. In particular, drivers underpinning the idea that the current jury system may not be the best forum to assess evidence in rape trials are not limited by empirical evaluations of whether or not jury members buy into rape myths; the fact that the law prescribes to withhold or present evidential information through fairly elaborate rape shield legislation means that we already have a system which recognizes and attempts to control the influence of bias in lay fact-finders. It is no more than an empirical observation that the existence and development of elaborate codes for screening and, in effect, spoon-feeding evidential information to the jury interferes with fact-finding in a purely scientific sense.

Nor is it the case that mock jury research should be dismissed because of the potentially zealous nature of the participants. There are intrinsic levels of comparison within the mock jury work which may not have been examined to their full potential. Take for instance how the SJR findings may be viewed in the light of the proposed ‘Level of Certainty’ model[1] of jury fact-finding.

The model is summarized in Table 1 below. It describes a simple ‘level of certainty’ comparison for unanimous and simple majority jury systems, where the minimum level of certainty required for a verdict is described as a simple percentage of overall juror verdict opinion at jury level:

Focussing on Scottish versus English style jury systems, this ‘level of certainty’ model is clearly reflected in the fairly symmetrical distributions of verdict outcomes found in the Scottish Jury Research for the Scottish and English equivalent systems shown in Tables 2A and 2B below:

Whether these results are recognized as significant in some statistical sense is a point of discussion, however, given that both the rape and assault trials were said to be similarly balanced[2] and the same trials were viewed by different juries in each case, these results have a comparative significance. By comparing the ‘level of certainty’ required for conviction under the English and Scottish style systems, the higher certainty level required for the English style system appears to equate and translate to a shift in verdict severity for all but one outcome – in other words, under Scotland’s simple majority system, 2 hung verdicts in the same trials viewed under the English system move to become 2 guilty verdicts with the Scottish system and 6 not guilty outcomes shift up to become 5 hung or not proven verdicts (the theoretical equivalence of the not proven verdict and the hung jury as a third verdict has been discussed elsewhere, and given that the numerical pressures on unanimity/2 verdict jurors polarize them away from third verdict outcomes, it is arguable that the drivers for not proven relate directly to the drivers for a hung jury).

The characterization of not proven presented throughout the Scottish Jury Research publications as being something vague and ill-defined in Scots law is not one that entirely squares with case law (Larkin v HMA 1993), the Jury Manual[3] or the writings of David Hume, however its impact on the finely balanced cases of the Scottish Jury Research does appear to be overshadowed by the relatively low level of certainty required for a simple majority verdict when one focusses on comparison between the English and Scottish systems. That does not mean that not proven is disposable. The prevalence of a not proven vote in any given trial outcome may be an indicator of a risk of miscarriage of justice or the existence of some deficient evidence; either way it does not detract from the possibility of a guilty verdict – if anything the existence of the not proven option appears to allow jurors more freedom to opt for a guilty verdict according to the results of the SJR work.

The overshadowing of the third verdict by the simple majority (and the relevance of the ‘level of certainty’ model) is further evidenced by the fact that 6 out of the 7 guilty verdicts found in the SJR work were found in simple majority systems whereas the 7 convictions were split 4 to 3 between three verdict and two verdict systems. As alluded to above, the surprise here is that there were more convictions in the three verdict systems and the three verdict systems produced most of the convictions for rape - 3 out of the 4 guilty verdicts found in the rape trials were in three verdict systems. Furthermore, one third of the simple majority convictions were found in the Scottish equivalent system (2 out of 6) which represented only 1 out of the 8 different jury system configurations trialled in the research, and no other system had more convictions (the English equivalent system had no convictions).

The lingering single not guilty verdict found in the Scottish system may be an indicator that there is not enough discrepancy between the Scottish level of certainty requirement and the English one to cause a full shift of all the finely balanced verdicts in moving from one system to another – it may also be an indicator of the Hollywood Effect at work. Further detailed research on how the other various SJR mock jury system variables compare when examined in the light of the certainty model would be interesting and examining how hung juries and not proven verdicts may be categorized together as ‘third verdicts’ would be of interest also. On general overview however, the Scottish and English systems, when the detailed SJR findings[4] are considered as a whole, appear to represent two pragmatic systems at, or at least near, opposing ends of a spectrum and they have a unique relationship among the 8 jury systems trialled.

I offer Table 3 below as a more general summation of the SJR’s finely balanced trial outcomes[5], grouping, as it does, the verdict outcomes together in the form of an answer to a question, where both rape and assault trials are regarded as equally representative, where polarization (and possibly leniency bias) in two verdict systems is apparent, where the powerful impact of the simple majority is highlighted, where not proven appears to provide more freedom to choose rather than offering a ‘cop-out’ and where third verdicts (both hung and not proven) are categorized together as not proven:

A final note on professional attitudes to fundamental reform. Any experienced professional bases his/her outlook (and role modelling) on many years of dealing with a framework of given constants against which day to day variations of specific circumstances play and contrast. This is the basis of professional intuition and insight. It would be unusual if any body of established professionals would welcome, with open arms, fundamental root and branch step-changes to such a framework of constants which has been built up, carefully evolved and supported over many generations and upon which their very livelihood and professionalism has been founded through time and meticulous labour. We should not be too surprised therefore if experienced courtroom practitioners do not want to change the ground rules of the jury system. Any change that could be so welcomed would have to be greatly desired.

[1] This was first described in a presentation given over a year ago at Aberdeen University and was due to be given again at Dundee University this year but was cancelled due to the pandemic lockdown.

[2] It is worth noting that the variation in verdict outcomes between rape and assault for each the 8 categories of jury system trialled never varied by more than a single verdict difference over each of the 4 trial groups.

[3] Per the Judicial Institute for Scotland: ‘…it has been held not to be a misdirection that a jury might return a not proven verdict where the Crown have not established guilt beyond reasonable doubt but where there are still “lingering doubts” as to the accused’s guilt, or where the words “not guilty” would “stick in their throats” (ref Larkin v HMA 1993).

[4] My thanks to Fiona Leverick for putting me out of my misery in trying to calculate the detailed trial outcomes from the categorized figures published in the Scottish Jury Research Findings.

[5] Source Table 3.1 on page 20 of the Scottish Jury Research: Findings from a Large Scale Mock Jury Study.