Scottish jury research: findings from a mock jury study

Footnotes

1. If the number of jurors is reduced from 15 due (for example) to a jury member becoming ill during the trial, the accused cannot be convicted unless eight jurors support conviction.

2. A 12-person jury which split evenly between conviction and acquittal would be deemed to have returned a verdict of not guilty: see section 1.1 below.

3. Assault trial: https://youtu.be/gxeU-sFzOxQ; Rape trial: https://youtu.be/kDAGaSedje8; Judge's opening and closing directions: https://youtu.be/ecemRns-gDk

4. An additional 105 'spare' jurors watched the trial videos and completed a questionnaire, but were sent home prior to deliberations to ensure that each jury only included either 12 or 15 people at this stage. Ipsos MORI recruiters deliberately recruited more jurors than were needed, to allow for non-attendance on the day. One further juror left immediately prior to deliberations, as they became unwell.

5. Post-Corroboration Safeguards Review: Final Report (2015) para 12.24.

6. See e.g. the detailed paper prepared by researchers for Lord Justice Auld's Review of the Criminal Courts of England and Wales: P Darbyshire, A Maughan and A Stewart, What Can the English Legal System Learn from Jury Research Published up to 2001? (2002). More recently, see DJ Devine, Jury Decision Making: The State of the Science (2012), who states (at 8) that "1,500 is a plausible estimate of the number of published jury studies by the end of 2011".

7. Although the trend over time - and here practice varies from jurisdiction to jurisdiction - has been to permit the jury to return a verdict with either one or two jurors dissenting from that outcome. See J Chalmers, "Jury majority, size and verdicts", in J Chalmers, F Leverick and A Shaw (eds), Post-Corroboration Safeguards Review: Report of the Academic Expert Group (2014) 140 at 143-151.

8. While the three possible verdicts open to the jury mean that there may be no verdict which commands the support of eight or more jurors, such cases will always be regarded as an acquittal because not guilty and not proven have the same effect in law. Where jurors are split between the two acquittal verdicts, the rules differ depending on whether the jury deliberated with its full 15 members or not. Where the jury had 15 members, the verdict will be recorded as one of not guilty or not proven where there is a majority for that verdict amongst the jurors voting for acquittal, but an equal split between acquittal verdicts will be recorded as not proven: Kerr v HM Advocate 1992 SLT 1031. Where, exceptionally, the jury had an even number of members (if one or more of the initial 15 were unable to continue e.g. due to illness), such a jury, if split evenly between conviction and acquittal, will always be deemed to have returned a verdict of not guilty: Criminal Procedure (Scotland) Act 1995 s 90(2).

9. These were supplemented by a number of more detailed sub-questions, expanding on different aspects to be addressed in answering these over-arching questions (see Annex A for a full list of the research questions).

10. J Chalmers and F Leverick, Methods of Conveying Information to Juries: An Evidence Review (2018); V Munro, The Impact of the Use of Pre-Recorded Evidence on Juror Decision-Making: An Evidence Review (2018).

11. This is not to assert that the rest of the extensive body of existing research focused on the traditional common law jury is irrelevant to Scotland. It would be relevant to any policy decision to introduce a jury system more akin to the model found in most major English language jurisdictions, and might also be relied upon more generally if research suggested that the differences made by the distinctive features of the Scottish jury were limited in their effect.

12. This does not imply any criticism of the researchers involved - highly realistic mock jury studies require significant resources, and variations in research methods are generally a result of resource constraints.

13. This research arose out of decisions of the US Supreme Court which concluded that the size of a jury could not be reduced below six members without violating an accused's Sixth Amendment right to jury trial. See Williams v Florida 399 US 78 (1970); Ballew v Georgia 435 US 223 (1978).

14. MJ Saks and MW Marti, "A meta-analysis of the effects of jury size" (1997) 21 Law and Human Behavior 451.

15. See J Chalmers, "Jury majority, size and verdicts", in J Chalmers, F Leverick and A Shaw (eds), Post-Corroboration Safeguards Review: Report of the Academic Expert Group (2014) 140 at 156.

16. R Hastie, SD Penrod and N Pennington, Inside the Jury (1983) 60.

17. See e.g. the research surveyed in R Hastie, SD Penrod and N Pennington, Inside the Jury (1983) 29-32.

18. R Hastie, SD Penrod and N Pennington, Inside the Jury (1983) 60.

19. BH Bornstein and E Greene, "Jury decision making: implications for and from psychology" (2011) 20 Current Directions in Psychological Science 63 at 65.

20. The one study ('Study 2', as reported in L Hope and others, "A third verdict option: exploring the impact of the not proven verdict on mock juror decision making" (2008) 32 Law and Human Behaviour 241) which involved deliberation used groups of four to eight jurors, much smaller than the Scottish jury of 15 members. In addition, the existing studies discussed here (except Hope and others, Study 2) relied heavily on the use of students as mock jurors, rather than a broader community sample.

21. See Annex C for more detail on existing studies of the not proven verdict.

22. Judicial Institute for Scotland, Jury Manual (2019) 107.1. The mock juries in the present study were directed by the judge in accordance with this practice.

23. K Bruce, "Not proven verdict leaves 60% of Scots in dark, claims poll", The Herald 13 May 1993, p 2. A further 11% said that they did not know. Following the Double Jeopardy (Scotland) Act 2011 it is now, in some circumstances, possible to prosecute an acquitted person again if new evidence becomes available, but this does not depend in any way on the initial verdict having been not proven rather than not guilty. See further section 5.4.2.

24. L Hope and others, "A third verdict option: exploring the impact of the not proven verdict on mock juror decision making" (2008) 32 Law and Human Behaviour 241.

25. At 246 (35% of participants who had the option of three verdicts).

26. At 249.

27. LJ Curley and others, "Threshold point utilisation in juror decision-making" (2019) 26 Psychiatry, Psychology and Law 110.

28. At 118.

29. L Hope and others, "A third verdict option: exploring the impact of the not proven verdict on mock juror decision making" (2008) 32 Law and Human Behaviour 241 at 245.

30. One 12-person jury proceeded with 11 jurors, as a juror went home ill immediately before the start of deliberations. A real trial can similarly proceed where a juror falls ill and can no longer participate: Criminal Procedure (Scotland) Act 1995 s 90.

31. C Thomas, "Exposing the myths of jury service" [2008] Criminal Law Review 415 at 422.

32. Where more than 12/15 jurors attended, researchers tried to ensure gender and age balance in the juries when selecting 'spare' jurors.

33. G Reyes, "Understanding non response rates: insights from 600,000 opinion surveys" (World Bank, 2016).

34. Heehaw, based in Edinburgh.

35. Available at http://www.scotland-judiciary.org.uk/Upload/Documents/JuryManualFebruary2019.pdf

36. The judge's directions on the legal tests that should be applied were not truncated from those included in the Jury Manual, although the overall length of directions in any given trial will vary depending on the complexity of the case and the number of legal issues on which the jury requires to be directed.

37. Jurors in an actual criminal case may either swear an oath or affirm: that is, "swear by Almighty God" or "solemnly, sincerely and truly declare and affirm" that (in either case) they will "well and truly try the accused and give a true verdict according to the evidence". A decision was taken to use the affirmation throughout to heighten the realism of the experience while avoiding undue complexity.

38. See Annex E.

39. See Questionnaire C in Annex E.

40. See Annex E.

41. See Annex F.

42. See Annex G.

43. Any discrepancies between the two were discussed and resolved, involving additional members of the research team where necessary to ensure consistency of approach.

44. That is, significance tests on differences reported in the text produced P-values of <=0.05, indicating that the probability of such a difference occurring in our sample when there is no actual difference in reality is less than 5%.

45. Section 2.3.

46. In the unanimity juries, none of the 32 juries would have been able to return a verdict on the basis of the jurors' initial preferences as stated in their pre-deliberation questionnaires. There were no instances in which those initial preferences were unanimous, nor any in which there were only one or two jurors with a different view from the majority position. However, in every instance where a unanimity jury returned a verdict, the decision to convict or acquit (but not necessarily the choice of acquittal verdict) was identical to the preference of a majority of jurors prior to deliberation, aside from one acquittal where the jurors' initial preferences had been split 50:50.

47. Some support for the predictive value of individual pre-deliberation verdicts can also be found in previous research: Hastie and others, for example, conducted a study involving 69 juries comprised of 12 members. The jurors observed a simulated trial and then completed pre-deliberation questionnaires. The researchers concluded that the most accurate model for predicting final jury level verdicts was a weighted average drawn from individual jurors' initial preferences (which correctly predicted the result in 75% of cases). R Hastie, SD Penrod and N Pennington, Inside the Jury (1983) at 63-65. Note that none of these juries applied a simple majority rule, however, and so extrapolation to the Scottish context is unclear. Three different decision rules were applied, each by one-third of the mock juries: 12 votes out of 12, 10 votes out of 12 and eight votes out of 12.

48. The table also shows the split by the number of verdicts available within trial type, jury size and majority required (since in each case, half the juries were asked to choose between guilty and not guilty, and half between guilty, not guilty and not proven).

49. As described in Chapter 2, if juries were unable to reach a completely unanimous decision, they were allowed to return a verdict if either 10 out of 12, or 13 out of 15, of them could agree on it.

50. As noted in Section 1.1 above, one distinctive feature of the Scottish jury system as it stands is that hung juries are impossible: the simple majority decision rule means that a verdict will always be reached.

51. Juries must attempt to reach unanimity but ultimately can return a verdict that 10 out of 12 of their members agree on.

52. J Chalmers, "Jury majority, size and verdicts", in J Chalmers, F Leverick and A Shaw (eds), Post-Corroboration Safeguards Review: Report of the Academic Expert Group (2014) 140 at 151-152.

53. Scottish Government, Criminal Proceedings in Scotland, 2017-18 (2019) 53. Figures on the use of the not proven verdict in all cases prosecuted as assault (the other trial type used in this research) are difficult to discern as assault is included in three different categories used in the data.

54. In answer to a question asked in the Scottish Parliament, it was stated that, in 2016-17, 31% of acquittal verdicts in solemn cases (heard by a judge and a jury) were not proven. The figures for the two previous years were 28% (2015-16) and 30% (2014-15). See Scottish Parliament Question S5W-18904.

55. See e.g. the meta-analysis in RJ MacCoun and NL Kerr, "Asymmetric influence in mock jury deliberation: Jurors' bias for leniency" (1988) 54 Journal of Personality and Social Psychology 21.

56. In 12-person juries, 69% of jurors favoured acquittal pre-deliberation, compared to 79% post-deliberation (a shift towards acquittal of ten percentage points). In 15-person juries 66% of jurors favoured acquittal pre-deliberation compared to 70% post-deliberation (a shift of only four percentage points) (Tables 3.2 and 3.3).

57. In simple majority juries, the percentage of jurors favouring acquittal was exactly the same pre- and post-deliberation (67%), whereas in unanimity juries 66% of jurors favoured acquittal pre-deliberation compared to 80% post-deliberation (a shift of 12 percentage points) (Tables 3.2 and 3.3).

58. One unanimity jury started off with an even split between jurors favouring guilty (six) and acquittal (six) - this jury acquitted at the end of deliberations.

59. See section 2.3.

60. The proportions who thought the verdict should be guilty, after deliberating, were: Rape trial - 30% (three-verdict) vs 37% (two-verdict); Assault trial -13% (three-verdict) vs 25% (two-verdict).

61. The proportions who thought the verdict should be guilty, after deliberating, were: Rape trial - 36% (15-person) vs 29% (12-person); Assault trial - 23% (15-person) vs 14% (12-person).

62. This was confirmed via logistic regression, which examines whether different features are significantly associated with an outcome - in this case verdict - even after controlling for possible inter-relationships between these features. Each of the three unique features (number of verdicts, jury size and majority required) were included in the model, alongside trial type (rape vs assault) and individual jurors' demographic characteristics (gender, age group, education level). This confirms that the number of verdicts is significantly and independently associated with the likelihood of individual jurors favouring a guilty verdict both pre- and post-deliberation, while jury size and majority required are significantly and independently associated with likelihood of favouring a guilty verdict post-deliberation (but not pre-deliberation).

63. See Annex E.

64. See Annex J, Table J.1.

65. MJ Saks and MW Marti, "A meta-analysis of the effects of jury size" (1997) 21 Law and Human Behavior 451 at 458-459. However, only two of the studies which Saks and Marti reviewed reported data on this point.

66. Section 2.5.3. See Annex G for further details.

67. Neither were there any significant differences in the mean number of evidential issues raised between two-verdict and three-verdict juries or between 12-person and 15-person juries.

68. All three of these circumstances must exist for the defence to succeed, but where the defence is raised the burden of proof is on the prosecution to disprove the defence. If the prosecution disproves any one of these three circumstances beyond reasonable doubt, the jury cannot hold that the accused acted in self-defence.

69. Judicial Institute for Scotland, Jury Manual (2019) 5.15.

70. For a summary of the content of the trial, see Annex D.

71. "Free agreement" is the legal definition of consent in sexual offence cases: see Sexual Offences (Scotland) Act 2009 s 12. The jurors were told this by the trial judge.

72. For a summary of the content of the trial, see Annex D.

73. Where, in a jury trial for a sexual offence, it is suggested that the sexual activity took place without physical resistance on the part of the complainer, the judge must normally advise the jury that there can be good reasons why someone against whom a sexual offence is committed might not physically resist and that the absence of such resistance does not necessarily indicate that an allegation is false: Criminal Procedure (Scotland) Act 1995 s 288DB, as inserted by the Abusive Behaviour and Sexual Harm (Scotland) Act 2016 s 6. This direction was included in the judge's directions to the jury in the rape trial simulation.

74. Though all three assault trial juries that convicted referred to all three tests.

75. This excludes misunderstandings relating to the not proven verdict. Juror understanding of the not proven verdict is considered separately in Chapter 5.

76. Note that it was not always clear whether particular statements reflect actual misunderstandings, or rather a poor articulation of a legitimate position. For example, with respect to statements that self-defence did not operate as a legitimate defence, it may be that in some cases jurors accepted the validity of the self-defence argument in principle, but viewed the degree of violence such that they felt the incident could not be considered self-defence - and were therefore attempting to short-circuit the debate since they were already convinced it was not self-defence.

77. J Chalmers and F Leverick, Methods of Conveying Information to Jurors: An Evidence Review (2018).

78. For more details on the how these aspects were recorded and analysed, see Annex G.

79. See tables J.2 to J.6 in Annex J for full data.

80. R Hastie, SD Penrod and N Pennington, Inside the Jury (1983) 91-92.

81. While not the main focus of our research questions, there were some statistically significant differences in observed and self-assessed juror participation between the rape and assault trials. It is not appropriate to generalise from these, since they reflect the particular nature and content of the two trials in question. However, in summary: there were more minimally contributing and dominant jurors in assault trial juries and jurors were more often observed apparently wanting to contribute but being unable to do so in assault trial juries. But there were no statistically significant differences in jurors' own perceptions of their influence over the verdict, or their perceptions of whether other jurors spoke too much, between the two types of trial.

82. See Annex J, Table J.7.

83. Based on comparing what they personally said the verdict should be in their post-deliberation questionnaire with the verdict returned.

84. See Annex J, Table J.8.

85. While not directly relevant to the research questions, it is worth noting that confidence in the verdict was significantly higher among assault trial jurors (89%) than among rape trial jurors (73%).

86. The precise legal position regarding re-prosecution is described in more detail in section 5.4.2 below.

87. I Willock, The Origins and Development of the Jury in Scotland (Stair Society, 1966) 217. The explanation of the history of the not proven verdict that follows is drawn from Willock's book.

88. GH Gordon, Renton and Brown: Criminal Procedure according to the Law of Scotland, 6^th edn (1996) para 18-79.41.

89. The direction was taken from the Judicial Institute for Scotland's Jury Manual (2019).

90. See Annex E for details of the different questionnaires. For an explanation of 'spare' jurors, see section 2.4.

91. See section 3.4.

92. Five of 15 in one of the juries and four of 15 in the other. As discussed in section 5.5.1, however, jurors rarely gave explicit reasons for switching verdict.

93. Four of the five were assault trial juries; one was a rape trial jury. Four were simple majority juries; one was a unanimity jury. Two of the juries involved very short deliberations (19 minutes and 24 minutes), indicating that discussion generally (i.e. not just of the not proven verdict) was limited. The other three had deliberation times of 35 minutes, 41 minutes and 50 minutes.

94. Research suggests that jurors think they understand legal concepts better than they actually do and that confidence does not necessarily equate to accuracy: see L Hope, N Eales and A Mirashi, "Assisting jurors: promoting recall of trial information through the use of a trial‐ordered notebook" (2014) 19 Legal and Criminological Psychology 316 at 326 and B Saxton, "How well do jurors understand jury instructions? A field test using real juries and real trials in Wyoming" (1998) 33 Land and Water Law Review 59 at 92.

95. This mirrors the findings of Hope and others' research (cited in section 1.2.2), where self-assessed understanding of the not proven verdict was also significantly higher among jurors who had been directed on the verdict than those who had not - see study 1 (at 246) and study 2 (at 249).

96. A point that is returned to later - see section 5.4.5.

97. See Annex J, Table J.10.

98. There were also 10 statements to this effect across six of the 32 two-verdict juries.

99. See Double Jeopardy (Scotland) Act 2011 sections 3 and 4.

100. Hope and others (see citation in section 1.2.2) at 249. The difference was even more stark in her study, with 37% of three-verdict participants believing the accused could be re-tried for the same offence following a verdict of not proven, compared to 87% of two-verdict participants.

101. Seven statements were also made in the two-verdict juries (six expressing the view that retrial was possible after a not proven verdict).

102. And in four of the two-verdict juries. In the two-verdict juries, this was always in the context of jurors wishing that the not proven verdict was available to them, so that they could more easily reach a verdict.

103. The Protecting Vulnerable Groups (PVG) check is a scheme intended to ensure that people whose behaviour makes them unsuitable to work with children and/or protected adults are prevented from doing 'regulated work' with these vulnerable groups.

104. Nothing was said about an impact upon criminal records in the two-verdict juries.

105. This is incorrect as previous acquittals (and convictions) are generally inadmissible evidence in a Scottish criminal trial.

106. Statements to this effect were also made in two of the 32 two-verdict juries.

107. See Judicial Institute for Scotland, Jury Manual (2019) 5.15.

108. Three statements to this effect were also made across three of the 32 two-verdict juries (all rape trial juries).

109. See Tables J.11 and J.12, based on jurors' explanations of their initial verdict preferences (prior to deliberating). Jurors were also asked in the post-deliberation questionnaire for their personal choice of verdict and the reasons for this. These reflected similar themes to the reasons given pre-deliberation but, as so few jurors in the three-verdict condition selected a not guilty verdict at this stage, there were insufficient numbers for meaningful analysis. The data is based on an open-ended question, with responses categorised by Ipsos MORI's specialist coding team, using a code frame developed by the research team.

110. Tables J.11 and J.12, Annex J.

111. This figure is based on comparison of the verdicts they favoured in their pre- and post-deliberation questionnaires.

112. This reflects the fact that a small number of jurors appeared to have changed verdicts based on the verdicts they recorded on their pre- and post-deliberation questionnaire, but appeared not to recall having done so when asked directly about this.

113. Although these are not the only methods that have been used. For a more detailed discussion, see J Chalmers and F Leverick, Methods of Conveying Information to Jurors: An Evidence Review (2018) Annex 1.

114. In Scotland, see Contempt of Court Act 1981 s 8.

115. Researchers refer to this as the 'external validity' of the study.

116. The steps we took in the present research to address these issues are outlined in chapter 2.

117. BH Bornstein and others, "Mock juror sampling issues in jury simulation research: a meta-analysis" (2017) 41 Law and Human Behavior 13 at 25.

118. RL Wiener, DA Krauss and JD Lieberman, "Mock jury research: where do we go from here?" (2011) 29 Behavioral Sciences and the Law 467 at 472.

119. F Leverick, "Jury instructions on eyewitness identification evidence: a re-evaluation" (2016) 49 Creighton Law Review 555 at 566-567.

120. E Finch and V Munro, "Lifting the veil: the use of focus groups and trial simulations in legal research" (2008) 35 Journal of Law and Society 30 at 45; L Ellison and V Munro, "Getting to (not) guilty: examining jurors' deliberative processes in, and beyond, the context of a mock rape trial" (2010) 30 Legal Studies 74 at 84; P Ellsworth, "Are twelve heads better than one?" (1989) 52 Law and Contemporary Problems 205 at 223.

121. M Smithson, S Deady and L Gracik, "Guilty, not guilty, or…? Multiple options in jury verdict choices" (2007) 20 Journal of Behavioral Decision Making 481.

122. A civil jury would not in fact be asked to return a verdict in the forms "guilty", "not guilty" or "not proven". The judicial instructions did, however, instruct the jurors to apply a "balance of probabilities" standard in the civil case.

123. M Smithson, S Deady and L Gracik, "Guilty, not guilty, or…? Multiple options in jury verdict choices" (2007) 20 Journal of Behavioral Decision Making 481 at 486.

124. Both reported in L Hope and others, "A third verdict option: exploring the impact of the not proven verdict on mock juror decision making" (2008) 32 Law and Human Behaviour 241.

125. LJ Curley and others, "The bastard verdict and its influence on jurors", (2019) 59 Medicine, Science and the Law 26.