The Application was heard on Tuesday and the Court unanimously refused leave to appeal. After hearing from counsel French CJ said ([2010] HCATrans 120):
At trial in this matter the parties acquiesced in the statistical conclusions drawn from evidence relating to DNA profiles being expressed qualitatively rather than quantitatively. More particularly, they acquiesced in the expression of the statistical conclusions drawn from analysis of material taken from the complainant’s clothing being compared with the applicant’s DNA profile as comprising “strong” or “extremely strong” evidence in support of the contention that the applicant was the source of the material taken from the complainant’s clothing without the jury being told that the particular conclusions made by the witness in the case had yielded a figure of greater than one in 10 billion.
It was open to the jury to conclude from the evidence that was led at trial that the applicant was guilty beyond reasonable doubt. In light of the way the parties conducted the trial this is not, in our opinion, a suitable case to consider the larger question which the applicant seeks to agitate. It is the opinion of all of us that special leave should be refused.
Looking at the transcript of argument one matter jumped to my mind even before coming across the Court’s decision: the outcome may have been different had counsel approached the issue differently.
Counsel of Forbes, Hastings QC, seemed to put the argument (in some articulations) that DNA evidence, without more, cannot lead to a safe verdict of guilty. That is, says the Hastings, there must be a substratum of established fact which the DNA evidence supports, because it cannot establish a fact (or guilt) by itself.
The problem with this argument is that the Court is often loath to make such a general rule that DNA evidence can never convict. If for no other reason (and I’m sure its not), it may require revisitation to the principle whenever technology improves and may thereby attract a treacherous consideration of what level of statistical certainty is necessary for proof beyond reasonable doubt.
I think it is a good rule of appellate advocacy to try not to offer rules of general application, but to proffer articulations of rules that can be restricted to specific instances. If the Court wants to take it further it can, but that should be the fallback position for counsel, not the primary argument.
As an example (which was eluded to in arguendo), a better argument might be: given the sample which was the subject of the evidence, the ACT database of DNA samples against which the sample was tested was not sufficiently large to render the expert evidence sufficient to convict without other facts.
Or even better: even if it could be sufficient, the jury was not advised of this difficulty and ought to have been, in the circumstances, to weigh the evidence properly.
Ultimately though, this appeal was difficult for appellant counsel because of earlier concessions (or at least failures to complain) about the way evidence was to be presented and used, by counsel at trial. In my experience, I have come across a few trials where an accused has been convicted predominantly or exclusively on DNA evidence. In those cases too there could be no such ground of appeal because of the way defence counsel presented its case at trial.
Hopefully, defence lawyers will be wise to the issue in the future. Perhaps we will see the matter back before the Court before long.
Jeremy Gans says:
ReplyDeleteI think you're assuming that the ACT’s ‘small’ statistical database is somehow bad. Not sure why you're making that assumption, though. Statistical databases in the hundreds are commonplace. (That isn’t to say that there’s nothing controversial about DNA statistics.)
Personally, I don't think revisiting the 'DNA wars' of the 1990s, when these statistical issues were exhaustively worked through, is a promising line at all. The appeal will either fail because the trial court has the task of making these calls or (as here) because the parties acquiesced in the trial court being kept from having to make these calls.
Rather, the real issue for the courts is whether it's ever safe to convict solely on the basis of either trusting experts to get the statistics right or trusting courts to apply statistics themselves. The statistical problems fade away when there's other independent evidence, hence the call for a 'corroboration' rule. No, these aren't much liked these days, due to issues of pejorativeness and workability when applied to witness testimony, but applying such a rule to statistical evidence is arguably a different matter.
Frankly, it’s hard to see why Forbes’s acquiescence in allowing the jury to be told that the DNA match was ‘extremely strong’ means that this was somehow a bad vehicle to test the proposition that statistics alone can never found a conviction. ‘Extremely strong’ is just statistics by another (potentially highly misleading) name. That being said, I suspect there were other unspoken reasons why this case was considered a poor one.
Of course, there’s nothing wrong with DNA evidence per se (there are plenty of journal articles on the issue) and I agree that when the statistics in Forbes are scrutinised there is probably not going to be anything unreliable about them. But my point is that the Court is unlikely to create a general rule such as “we don’t trust experts to get the statistics right without corroborating evidence”. It is better to try to point to specific matters of the case at hand and identify why – in this case – the evidence was not sufficient to support a guilty verdict. Or why the verdict is unsafe because of the way the DNA was presented to the jury. I only use sample size as an illustration which, as you say, is an argument that would probably be destined for failure.
ReplyDelete(On a side note: My statistics professor was never able to convince me that n=100 is sufficient for statistically significant predictions about a population in the millions but I accept that that is my own hang up and doesn’t have any acceptance in ‘science’.)