Saturday, April 26, 2008

Propensity evidence: admissibility and Bayes Theorem

The High Court of Australia has given detailed consideration to how evidence of an accused’s propensity should be handled, in three jointly heard appeals: HML v The Queen; SB v The Queen and OAE v The Queen [2008] HCA 16 (24 April 2008).

The seven Judges delivered seven judgments. Unfortunately, and a trace of regret about this is evident in the remarks of Kirby J at para 82, the issue of real general interest was not unanimously decided because of the different approaches taken. This issue is, what role, if any, the standard of proof beyond reasonable doubt has in connection with proof of the alleged propensity.

The admissibility rule in Australia
I say that was the issue of general interest, because much of the judgment is concerned with the nearly-extinct rule in Pfennig v The Queen [1995] HCA 7, (1995) 182 CLR 461. This is, in essence, a rule about the probative value that propensity evidence must have in the context of a particular case before it can be ruled admissible. Only if the judge finds that, assuming the propensity evidence is accepted as true, and assuming the other prosecution evidence in the case is also accepted, the effect of the propensity evidence would be to exclude any reasonable doubt that would otherwise exist about the accused’s guilt, can the evidence be said to have sufficient probative value to be admitted. This affects the more generally applicable requirement that evidence must be excluded if its probative value is outweighed by its illegitimately prejudicial effect, by replacing the discretion with a rule (Gleeson CJ at 15). Hayne J discussed the application of this rule at 112 – 118. Gummow J and Kirby J agreed (41, 51).

Standard of proof of propensity evidence
It is often claimed that the only thing that needs to be proved beyond reasonable doubt in a criminal trial is the guilt of the accused. Each element of the offence must be proved beyond reasonable doubt. Policy has supported an evidential rule that the voluntariness of a confession must be proved beyond reasonable doubt. It is by no means clear that evidence of the accused’s propensity has to be proved beyond reasonable doubt.

Propensity evidence is, after all, a form of circumstantial evidence, and circumstantial evidence does not carry a standard of proof. But the view favoured by Kirby J is that propensity must be proved beyond reasonable doubt. He summarised the approaches of the other members:

“82 … Heydon J considers that it is unnecessary to decide whether the criminal standard of proof has a wider application in cases such as the present, because whatever the case, the judges' summing up in each of the three appeals included a direction incorporating the criminal standard[Reasons of Heydon J at [339], [376], [395]-[396]]. This is so, notwithstanding that the ostensible purpose of these appeals was to settle that issue with an authoritative statement by this Court. Crennan J endorses a principle similar to that stated by Gleeson CJ[Reasons of Crennan J at [477]], although she ultimately relies on the conclusion of Heydon J that directions incorporating the criminal standard were in fact given in the trial of OAE[Reasons of Crennan J at [483]]. It is apparent from the analysis of Kiefel J[See reasons of Kiefel J at [512]-[513]] that her Honour considers that, because the relevant evidence was relied upon for a purpose other than "disclosing [OAE's] sexual interest" in the complainant[Reasons of Kiefel J at [517]], a direction as to the criminal standard of proof was not required.”

Gleeson CJ held at 31 that the standard of proof beyond reasonable doubt applies only to indispensable facts, and Crennan J at 477 agreed. Hayne J held at 196 that where propensity evidence is admitted as an essential step in the reasoning (and Pfennig indicates that, since it is the admissible propensity evidence that removes any reasonable doubt about the accused’s guilt, it is essential) it must be proved beyond reasonable doubt.

Kirby J held that propensity evidence must be proved beyond reasonable doubt:

“[83] I support the conclusion of Hayne J. It is necessary and desirable for this Court to resolve the issue concerning directions to be given on the standard of proof applicable to evidence of "uncharged acts" for the guidance of trial judges and intermediate courts still observing the common law in this respect. I would hold that wherever such evidence has been admitted under the Pfennig test and is propounded as relevant to a step in reasoning towards the accused's guilt of an offence charged, the jury must be told that they are to be satisfied beyond reasonable doubt that such evidence has been proved before they reason that the accused is guilty on the basis of it[Reasons of Hayne J at [132], [244]]. This is the essential quid pro quo for allowing such evidence to be placed before the jury at all. It is mandated by considerations of law but also of basic fairness, considered in the context of an accusatorial trial that still observes rules of particularity as to the offences charged.”

The Australian approach to propensity evidence was devised to avoid a perceived risk that judicial directions on the proper use of such evidence will not necessarily be effective. Hayne J put it this way at 116:

“… Judicial directions about use of such evidence have not hitherto been seen, and should not now be seen, as solving that problem. The possible uses to which evidence of other acts (which does not meet the Pfennig test) may be put are inevitably so intertwined that they cannot be sufficiently disentangled to give useful instructions to the jury. And even if the various uses of such evidence could be disentangled, that would leave unaddressed and unanswered the further difficulty that the jury may attach more significance to the evidence of other acts than they should. That is why the solution that has been adopted for so long by the common law, reflected in this Court's decision in Pfennig, is to limit the circumstances in which evidence of other discreditable acts of an accused will be received in evidence.”

He summarised the Court’s position on the standard of proof of propensity evidence that demonstrated that the accused had a sexual interest in the respective complainant in these appeals:

“[247] …It is important to recognise, however, that at least a majority of the Court[Gummow J at [41], Kirby J at [63], Kiefel J at [506] and these reasons at [132]] is of the opinion that "[i]n the ordinary course a jury would be instructed by the trial judge that they must only find that the accused has a sexual interest in the complainant if it is proved beyond reasonable doubt"….”

Logical reasoning
Since the choice of the standard beyond reasonable doubt for propensity evidence is a policy choice, it should not be criticised for being illogical. But it is illogical.

It imposes on the jury an “if and only if, then …” form of reasoning: if and only if the propensity evidence is true, then the accused is guilty on the present charge. It is more appropriate to apply a logic of conditional probabilities to this situation (and, indeed, to most decisions a jury has to make). In the absence of concessions or admissions, certainties don’t arise in trials. Even fingerprint evidence is given in the form of opinion, and testimony involving any form of measurement inevitably involves a range of error. The truth of testimony is appropriately thought of as a probability. Evidence of propensity will be assessed by a juror as being more or less likely to be true, that is, as having a probability of being true.

The majority approach in HML v The Queen, requiring this probability to equate with beyond reasonable doubt, is too restrictive. Conditional probability reasoning, as expressed in Bayes’ Theorem, involves considering the likelihood of getting the assessed probability of the propensity, given the accused is guilty on the present charge, compared to the likelihood of getting that assessed probability of the propensity, given that the accused is innocent on the present charge.

I have discussed the application of Bayes’ Theorem to propensity evidence in a draft paper (a perpetual draft so that it can be updated), available on this site here.

This question of whether some or any facts need to be proved beyond reasonable doubt before a verdict of guilty can be returned has given rise to controversy: see for example Chamberlain v The Queen (No 2) (1984) 153 CLR 521 (HCA), Shepherd v The Queen (1990) 170 CLR 573 (HCA), Thomas v R [1972] NZLR 34 (CA), R v Puttick (1985) 1 CRNZ 644 (CA), R v Holtz [2003] 1 NZLR 667; (2002) 20 CRNZ 14 (CA), R v Morin (1988) 44 CCC (3d) 193, and R v MacKenzie (1993) 78 CCC (3d) 193 (SCC).

A point that has led to confusion is the need for a juror to be sure (beyond reasonable doubt) of what fact is accepted; this is not to say that the fact itself establishes anything beyond reasonable doubt. For example, a witness may say that on examination of a bullet there is an 80% chance that it was fired from a specified gun. To require the juror to be sure, beyond reasonable doubt, that the result “80%” is correct, is not the same as requiring the juror to be sure, beyond reasonable doubt, that the bullet came from that gun. Once the testimony has been accepted, the juror can assess the likelihood of the result “80%” being obtained, given that the accused is guilty, compared with the likelihood of getting the “80%” result, given that the accused is innocent.

It is better, when trying to grapple with the role of proof beyond reasonable doubt, to think of the reasoning as following the logic of conditional probabilities, rather than to use the traditional metaphors of ropes and chains of reasoning. This is because they are too vague. In HML v The Queen the rope metaphor was not used by any of the Judges, but chains of reasoning were spoken of.

No comments: