Citation: R v Keir [2002] NSWCCA 30
Court: New South Wales Court of Criminal Appeal
Date: 28 February 2002
Judge(s): Giles JA; Greg James J; McClellan J
Background
The appellant was convicted of murdering his wife at Tregear in February 1988 following a trial before Adams J and a jury in the Supreme Court. The Crown case was that he killed her during the night of 8 February 1988 and buried her body beneath the foundations of the family home. The appellant maintained that his wife had left him voluntarily and had contacted him and been seen alive by multiple witnesses on several occasions after that date.
No body was recovered, but bone fragments excavated at the Tregear property in 1991 were subjected to DNA analysis in 1997. The forensic biologist who conducted the analysis expressed his finding as a likelihood ratio: the DNA profile found in the bones was approximately 660,000 times more likely to have come from a child of the wife's biological parents than from a child of a random mating in the Australian population. The wife was her parents' only child.
The jury convicted the appellant, and he was sentenced to 24 years imprisonment with an 18-year non-parole period. He appealed against both conviction and sentence on four grounds, including that the trial judge's directions on the DNA statistical evidence were erroneous.
Legal Issues
- Whether the trial judge's directions to the jury correctly explained the significance of the DNA likelihood ratio evidence
- Whether the directions gave rise to the "prosecutor's fallacy" by conflating the likelihood ratio with the probability that the bones were those of the deceased
- Whether the same erroneous reasoning was then used to undermine the reliability of the identification evidence called by the defence
- Whether, even if there was a misdirection, a substantial miscarriage of justice had nonetheless occurred
Decision
The Court of Criminal Appeal found that the trial judge's directions committed a textbook instance of the prosecutor's fallacy. Rather than explaining to the jury that the DNA evidence meant it was 660,000 times more likely to find that profile in a child of the deceased's parents than in a random member of the Australian population, the judge directed the jury that there was a "660,000 to one chance" that the bones were those of the wife. These two propositions are not the same thing. The first is a likelihood ratio that must be weighed alongside all other evidence; the second wrongly converts that ratio into a direct statement of probability of identity.
The misdirection was compounded by a second, equally significant error. The trial judge used the same "660,000 to one" formulation to characterise the defence's identification evidence, telling the jury in effect that there was a 660,000 to one chance that witnesses claiming to have seen or spoken to the wife after February 1988 were accurate and reliable. This was an impermissible use of the statistic. The DNA evidence could properly be used to evaluate whether the bones were those of the deceased; it could not mathematically neutralise independent witness accounts in the way the directions suggested.
Giles JA accepted that the Crown case was strong, but held that the identification evidence could not simply be swept aside. One independent witness, in particular, had given evidence that she believed she had seen the wife in October 1988 and had noted a change in the wife's hairstyle consistent with other evidence. Given the misdirection, the Court could not say the appellant would inevitably have been convicted had the jury properly understood the DNA evidence. The appeal was allowed and the conviction quashed.
Giles JA added a caveat noting that even without a prosecutor actively introducing the fallacy, juries may reason in that impermissible way themselves when presented with large likelihood ratios. He observed that in cases of this kind, outside the straightforward crime-stain scenario, it may not always be easy to give adequate corrective directions, and that if satisfactory directions cannot be given, the admissibility of DNA statistical evidence itself may warrant reconsideration in the exercise of discretion.
Orders Made
- Appeal allowed
- Conviction quashed
- New trial ordered
Key Takeaways
- The Court of Criminal Appeal confirmed that a likelihood ratio expressed in DNA evidence (such as "660,000 times more likely") is not the same as a statement that there is a "660,000 to one chance" the bones belong to the deceased. Treating them as equivalent is the prosecutor's fallacy and constitutes a material misdirection.
- A DNA likelihood ratio cannot be transposed directly onto unrelated witness evidence to express the probability that such witnesses are mistaken or unreliable. The two inquiries are distinct and require separate evaluation.
- Where a trial judge's directions positively endorse the prosecutor's fallacy rather than merely fail to correct it, the error is particularly serious and is likely to have infected the jury's reasoning across multiple issues.
- Even a strong Crown case will not automatically preclude a finding of substantial miscarriage of justice where credible, independent defence evidence has been undermined by an erroneous direction on DNA statistics.
- Following R v GK and R v Galli, the Court of Criminal Appeal reiterated that DNA statistical evidence of this kind should ordinarily be accompanied by directions warning the jury against impermissible reasoning, and flagged that admissibility itself may be affected if appropriate directions cannot satisfactorily be formulated.
Legislation and Cases Referenced
Cases:
- Doheny and Adams (1997) 1 Cr App R 369
- R v Galli [2001] NSWCCA 504
- R v GK [2001] NSWCCA 413
Legislation: No legislation was cited in the decision.