During debates on the Victims, Witnesses and Justice Reform (Scotland) Bill, Justice Secretary Angela Constance cited research indicating that the odds of a jury convicting were “40 per cent lower” in a three‑verdict system, as compared with a two‑verdict system.

While correct, in context, this figure can be misleading when interpreted as a 40 per cent drop in actual convictions. The research referenced (Leverick, Curley, Lages, Jackson, 2024 Open University meta-analysis) reported an odds reduction of just over 0.4 (the ‘40 per cent’) in individual jurors voting guilty. In probability terms – the metric most laypeople intuitively understand – this corresponds to a 10 percentage point difference in juror voting (and not jury verdicts), as observed in mock trials.

Put simply: the difference between odds and probability matters. As does the distinction between juror votes and jury verdicts. A 40 per cent reduction in odds does not mean a 40 per cent reduction in the number of convictions, and a 10 per cent reduction in juror guilty votes does not mean a 10 per cent reduction in jury guilty verdicts. Misunderstanding these issues can exaggerate the perceived effect of jury reforms.

For example, the research indicates that if in a two-verdict system 65 per cent of mock jurors vote to convict, a three-verdict system reduces this to about 54 per cent – far from a 40 per cent drop in convictions.

Also, research consistently shows that verdicts are generally driven by initial, pre-deliberation majority votes. As indicated by Grofman and others, pre-deliberation voting is independent of different majority regimes/systems for verdict determination – rather, it is a function of the evidence presented in the courtroom and individual jurors perception of that evidence.

These subtle nuances are important for public understanding and for policymakers interpreting research evidence. Reporting odds ratios as percentages without clarification can unintentionally mislead, even when the underlying research is robust.

This is why medical journals require that odds are translated to, and presented as, percentage probabilities. The Scottish government ought to do likewise.

David Lorimer