In the UK it is currently in the news about the fallout from the Post Office Horizon case. My understanding is that all those accused are now considered innocent. This is a relatively rare example where we know the true outcome of a large enough number of trials after they have happened that we can examine the data and perhaps produce meaningful statistics. In particular it would be illuminating to for example examine the false positive rate of decisions made by judges compared to that made by jury's as a way of comparing these trial methods, particularly in light of the moves to try more cases in front of a judge.

For this we would need the numbers. For all the cases relating to the Post Office Horizon case how many were decided by judge, and what proportion of those where found guilty, and how many were decided by jury and what the outcomes of those trials were, including the actual number of jurors who found guilty vs. not guilty.

Is this information available publicly? If not, is it available to researchers such as we may get an answer in the future?

Since some of the comments seem to miss the point, the idea would be to use the data to help determine what features of a trial can affect the false positive rate. If for example we found that in these trials the jury decided ones had the same conviction rate as all trials, but the judge decided ones had a much lower rate of conviction that would be evidence that judge decided trials have a lower false positive rate.

  • 3
    I'm not sure it will be particularly illustrative to find the error rate among one batch of highly related financial fraud cases stemming from the same root cause, and extrapolate that to the general class of criminal cases. Perhaps a useful data point, but be careful when generalizing based on non-representative data. Commented Jan 9 at 19:15
  • @NuclearHoagie This is certainly true for a general error rate, but in a comparison between judge and jury decision making it would be of considerable relevance, especially as it is postulated in the linked article that single judge trials have a lower false negative rate.
    – User65535
    Commented Jan 9 at 19:22
  • 1
    You do realise that, at present, of the 736 convictions, only 93 have been overturned. That means that 643 are not (yet) legally “false positives”.
    – Dale M
    Commented Jan 9 at 20:33
  • @DaleM I did not. That does not mean the relative rates in different trials will be illuminating though.
    – User65535
    Commented Jan 9 at 20:35
  • 2
    Notwithstanding, it’s not clear that these are false positives in any event. Based on the law that, absent evidence of error, computer records are correct, the trial decisions were likely “legally” correct. It is unlikely that expert evidence of computer error was introduced in most or any of those trials; therefor, under the law, the guilty verdicts were accurate. The legal system is a blunt tool for discovering bye time facts. Under English law, if “computer says ‘no’”, then the law says “no” too.
    – Dale M
    Commented Jan 9 at 20:41

1 Answer 1


Based on the Court of Appeal's decisions in Hamilton v Post Office [2021] EWCA Crim 577 (which reviewed the convictions of 42 people, and quashed 39), there may be limited data when it comes to the question at hand. Here, the majority of defendants (35) entered a guilty plea in the face of prosecution pressure: so these cases did not reach a judge or jury, as finder of fact, at all. This was part of the point of why the trials were unfair - the Post Office withheld evidence which could have let them mount a proper defence, and in some cases used the threat of further prosecution in order to extract payment. Thus, there are rather fewer cases where the judge-or-jury distinction would come into play than the headline figures might suggest. There is a further confounding effect in that whether a particular prosecution took place before a jury depends mainly on what charge was brought, so the juries and judges are seeing a different population of cases.

We also do not have good data on acquittals (as the public campaign has sought to find people who were convicted and seek vindication), or on people who the Post Office investigated but declined to prosecute. Some of the known cases are about people who were convicted on some charges and acquitted on others, but it would seem difficult to draw any firm conclusions on the false positive rate without any estimate of the number of people acquitted in general.

For cases which did go to full trial, instances like that of Butoy (paras. 333-340) illustrate another reason that the judge-jury distinction may not be relevant. In that case, the defendant was convicted by a jury on ten counts of theft and acquitted on one. In his initial appeal, R v Butoy [2018] EWCA Crim 2535, quoted at 338, the court said that

during the course of the trial there was an agreed fact before the jury that at all material times the Horizon system had been operating properly

The prosecution had failed to disclose important evidence, and on this basis the judge was misled, causing the jury to be misdirected. Had the judge been sitting alone, this chain of events would be shorter, but seemingly leading to the same conclusion. The entire process was tainted at an early stage.

Of course, a jury might not obey the judge's instructions. A solicitor's note (para. 93(i)) observed:

However, it is absolutely vital that we win as a failure could bring the whole of the Royal Mail system down. Counsel's concerns is that juries will still believe in conspiracies and there don't need to be many people on the jury who do believe in conspiracy for us to have a problem.

The Post Office's barrister in that 2010 case may have been right that a jury would be more inclined to believe that the Post Office was concealing problems with Horizon (as in fact it was). That said, this is a little different from the proposed "natural experiment" on Horizon-related convictions, since it comes from a period when issues with the system were more well known.

Some data on prosecution misconduct may arise from a comparison of English and Scottish prosecutions. The latter were undertaken by the public prosecutor rather than by the Post Office directly. Any differences in case management or outcomes would be relevant to proposed reforms of private prosecutions in England, where it's been suggested that the Crown Prosecution Service ought to take over from the Post Office.

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .