CLI Seminar: Propensity Evidence Reform
Thank you very much for the invitation to contribute again to the Current Legal Issues seminar series. I am a strong believer in the benefits of a dialogue between the academy, the profession and the judiciary.
My focus today will be on reform in the regulation of prosecution propensity evidence in the criminal trial. (I use the term ‘propensity evidence’ as a general term to include similar fact evidence, tendency and coincidence evidence.) The typical scenario involves the prosecution adducing evidence of a defendant’s other misconduct with the intention of inviting the jury to reason that, the fact that the defendant did this kind of thing before suggests that the defendant has a tendency to do this kind of thing and makes it more likely that the defendant committed the charged offence. Actually this is not too distant from the previous times I talked in this venue, regarding Phillips  and HML  respectively. Despite (I won’t say ‘because of’) the High Court’s repeated efforts, the area raises a lot of difficult questions. This time, while I will mention several recent HCA decisions, I want to focus on the prospects for legislative reform. This topic is timely given that the Council of Attorneys General is currently working on propensity evidence reform in the wake of the Royal Commission into Institutional Responses to Child Sexual Abuse.  I have provided advice to both the Royal Commission and CAG.
I will talk about the direction of reform being pursued by the CAG, against the backdrop of the current treatment of propensity evidence at common law and in the uniform evidence law (UEL). The regulation of propensity evidence has already been the subject of considerable legislative intervention over the last few decades. These days the UEL applies in the Cth, NSW, Victoria, Tasmania, and the Territories. In the area of propensity evidence, WA and SA have their own statutory provisions which bear some resemblance to the UEL.  Only Queensland still has the common law, albeit slightly modified.  While my focus here is on child sexual assault cases, the discussion has implications for tendency evidence generally.
While a degree of legal technicality will be unavoidable, I want to address the fundamental underlying issue – the potential of propensity evidence to contribute to ‘the effectiveness of the criminal justice system … in securing the conviction of those guilty of criminal offences and the acquittal of those who are innocent’.  This broader focus requires us to stray beyond matters beyond legal doctrine into a consideration of criminology and the logic and psychology of proof. So there is quite a lot to cover.
Exclusionary rules and admission tests – common law and UEL
I’ll begin by briefly outlining the admissibility of propensity evidence under current common law and the uniform evidence law provisions. Where this kind of reasoning is relied upon the evidence will be subject to an exclusionary rule both at common law and under the UEL. However, despite the exclusionary rule, the evidence may gain admission if it satisfies certain requirements.
The common law has evolved a great deal over the last hundred or so years. At one point, on one view, propensity reasoning was absolutely ‘forbidden’ and evidence revealing a defendant’s misconduct could only admitted for non-propensity purposes.  US law still appears to be expressed this way.  However, as the law moved into the second half and fourth quarter of the previous century, it became increasingly recognised that many of the supposed non-propensity purposes were fictions. Propensity evidence could be admitted to support propensity reasoning. However, where this was its intended function the evidence must have sufficient probative value to gain admission. As I’ll mention below, there are still lingering traces of the old fiction that propensity reasoning is absolutely forbidden.  Sometimes evidence is admitted for a propensity purpose while, at the same time, juries are instructed not to reason that the defendant may be guilty because they did this kind of thing before and are the kind of person who does this kind of thing.
What level of probative value was required? The simple answer, at one point, appeared to be that probative value must outweigh the risk of prejudice.  This basic cost-benefit assessment made good sense as a matter of principle. Of course, application could be an issue. Probative value has a reasonably clear meaning – it is defined in the UEL as ‘the extent to which the evidence could rationally affect the assessment of the probability of the existence of a fact in issue’.  However, as I will discuss later, courts and commentators often have difficulty grasping this concept, conflating it with the related notion of proof. Further, even when probative value is properly understood, assessment can be anything but straightforward.
In some respects, figuring out the prejudicial risk appears even more difficult. The concept of prejudice may be more complex and less well understood than that of probative value. The dominant meaning risk of erroneous reasoning leading to factual error. Ultimately, the concern with prejudicial prosecution evidence is that it will cause an innocent defendant to be wrongfully conficted. Traditional jurisprudence has suggested that propensity evidence may lead the jury into error in two different respects.  First, the jury may apply propensity reasoning more strongly than is warranted. The jury may disregard the weaknesses with the propensity evidence and reason that since the defendant has committed this kind of offence previously, the defendant must be guilt. The second risk is that the propensity evidence may deflect the jury from carrying out their job properly. In particular, having learnt that the defendant has committed other misconduct, and is a bad person, the jury fails to give the defendant the benefit of a reasonable doubt. Whereas the first risk may be open for all propensity evidence, the second risk may depend upon the nature of the evidence, increasing where the other misconduct is particularly reprehensible – extreme physical and sexual violence.
The epistemic risk of erroneous reasoning appears to be viewed as the dominant form of prejudice. However, less commonly, policy-based non-epistemic objections to the use of propensity evidence have also been raised. It has been suggested that propensity reasoning – because the defendant has done this before, it is likely that the defendant committed the charged offence – is inconsistent with notions of rehabilitation  and individual autonomy.  Even if the evidence is probative and presents no risk of being misused, it should be prohibited on these policy grounds. 
Here, I will focus on the dominant form of prejudice – the epistemic risk that propensity reasoning poses, either through inflating the weight of the evidence, or discounting the criminal standard of proof. In both cases the probability of guilt flowing from the propensity evidence is wrongly treated as exceeding the criminal standard of proof. Further, it appears that expressions of concern about the risk of epistemic prejudice are premised on propensity evidence having limited probative value. If propensity evidence was viewed by the legal system as highly valuable, concerns about overvaluation would be less likely to arise, and the exclusionary rule may not have developed. In Pfennig McHugh J suggested that probative value and prejudicial risk are incommensurable.  In fact, in the case of epistemic prejudice, they are quite closely related. 
The risk of epistemic prejudice can be linked with the one of the ultimate concerns of the criminal trial – to avoid wrongful conviction. On this view, the concern is that the evidence will be misused resulting in conviction whereas, properly assessed, the evidence would leave a reasonable doubt. The scope for epistemic prejudice is related to the gap between the supposedly lowish probative value of the evidence and the probative value that may be required by the evidence for the prosecution to meet the elevated criminal standard of proof.
It is this relationship between probative value and prejudicial risk that led a majority of the HCA in Pfennig  to modify the admissibility test. The majority expressed concern with the flexible discretionary nature of the balancing test and sought to fix the level of the probative value required for admissibility.  Because they viewed propensity evidence as carrying a very high risk of prejudice they fixed the required level of probative value at a very high level. In fact, they drew the admissibility threshold from the criminal standard of proof – to gain admission, the evidence must be so strong that there is no reasonable view of it consistent with the innocence of the accused.  If evidence has this much probative value there is virtually no room left for prejudice to operate.
The Pfennig test is open to conceptual and practical criticisms. Conceptually, the Pfennig test conflates probative value and proof. It fashions an admissibility test for a single piece of evidence out of the criminal standard of proof which,  of course, ultimately has application to the entire prosecution case.  The conceptual conflation carries a related practical problem. It is difficult to make sense of the test. The most obvious interpretation – propensity evidence must be sufficiently strong by itself to prove the defendant’s guilt beyond reasonable doubt – appears so demanding as to be virtually impossible to satisfy.  As I’ll discuss in a moment, the HCA has provided a response to this concern, but this still fails to address the underlying concern – the conflation of probative value and proof.
I will return to the problems with Pfennig later, as they are symptomatic of some broader problems. But at this point I want to move on and introduce the admissibility tests in the UEL. The original UEL, the NSW and Cth Acts, became law in 1995, the same year as Pfennig. In some areas, the UEL deliberately and dramatically departed from the common law, particularly with regards to hearsay evidence. With regards to propensity evidence, my impression is that the common law was viewed as being in a state of flux and unclear, and the UEL just sought to bring greater clarity. 
Unfortunately, the admissibility tests in the UEL are scarcely less problematic than Pfennig. In some respects the UEL tests are more complex. Whereas Pfennig treats propensity evidence as a single category, the UEL unnecessarily and unhelpfully distinguishes tendency evidence and coincidence evidence in ss 97 and 98. These are very closely related variants of propensity evidence and the UEL applies the same admissibility test to both varieties, so it is difficult to see the point of the distinction. Also, whereas Pfennig requires propensity evidence to meet a single admissibility threshold, the UEL requires tendency and coincidence evidence to meet two different thresholds. It must meet the fixed threshold of ‘significant probative value’ (ss 97, 98), and further it must satisfy a variable threshold – probative value must substantially outweigh prejudicial risk (s 101). This is an asymmetrical balancing test which lacks the simple logic of the symmetrical balancing test. This test will require exclusion where the benefit of admission (probative value) exceeds the cost (prejudicial risk) but not by a substantial margin.
Despite identical wording being used for tendency and coincidence evidence, a line of authority applies a more demanding ‘significant probative value’ threshold to coincidence evidence. This differentiation is the result of some fairly weak statutory interpretation – the appearance of the word ‘similarities’ in the coincidence rule. As a matter of principle it is difficult to understand. Evidence that would invite tendency reasoning rather than coincidence reasoning is often more damning of the defendant on its face – eg, a prior conviction or uncontested evidence of another victim. The defendant did do this other misconduct and has demonstrated his bad tendency. Coincidence reasoning applies more naturally to a contested allegation or evidence merely linking the defendant to other harms. What is the chance that this evidence would mount against an innocent defendant? Coincidence evidence leaves open the possibility that the defendant has been falsely implicated. Tendency evidence then would appear to pose greater risks of prejudice. If anything, it should require a more stringent standard. Because tendency evidence now comes in more readily, prosecutors are increasingly fitting other misconduct evidence into the tendency category, even if it would support coincidence reasoning more naturally. This can only present problems the jury’s comprehension of the evidence and the arguments, and problems for the judge in trying to manage this.
The tests in operation
Common law – HCA
As discussed above, in Pfennig, a majority of the High Court sought to replace the flexible balancing admissibility test – probative value versus prejudicial risk – with a fixed admissibility threshold - the evidence must be so probative as to exclude innocence as a reasonable possibility. As I mentioned, this test, based upon the criminal standard of proof, conflates proof and probative value. The conceptual problem generates practical problems. On its face appears impossible for an individual item of evidence to satisfy a standard of proof that is demanding when applied to an entire body of prosecution evidence.
Various strategies were developed by state courts in an attempt to make the test workable. These varied in their details, but had a common theme. Essentially, because the admissibility threshold is a proof threshold, it should not be applied to the propensity evidence in isolation, but should be applied to the prosecution case as a whole. Dyson Heydon extrajudicially endorsed Hodgson JA’s proposal along these lines in NSWCCA in WRC and Joiner,  endorsement subsequently repeated by the High Court.  The QCA developed a similar strategy in O’Keefe  only to be criticised by the High Court in Phillips, an appeal from Queensland, for endorsing a test that is ‘expressed differently’ to,  ‘qualifying or ignoring a rule established by a decision of this court’, contrary to the rules of precedent. 
This criticism of the QCA by the HCA seems a little harsh in a couple of respects. First, it is not as though the HCA’s admissibility test has the fixed expression of legislation. In fact, in Phillips, the High Court provided more than half a dozen different formulations of
the need for similar fact evidence to possess some particular probative quality. The "admission of similar fact evidence ... is exceptional and requires a strong degree of probative force". It must have "a really material bearing on the issues to be decided". It is only admissible where its probative force "clearly transcends its merely prejudicial effect". "[I]ts probative value must be sufficiently high; it is not enough that the evidence merely has some probative value of the requisite kind." The criterion of admissibility for similar fact evidence is "the strength of its probative force". It is necessary to find "a sufficient nexus" between the primary evidence on a particular charge and the similar fact evidence. The probative force must be "sufficiently great to make it just to admit the evidence, notwithstanding that it is prejudicial to the accused". Admissible similar fact evidence must have "some specific connexion with or relation to the issues for decision in the subject case". 
Interestingly, the HCA itself, in Phillips de-emphasised the ‘no reasonable view’ version of the test from Pfennig. 
Second, the High Court developed its own contextual strategy along similar lines to that advanced in O’Keefe (and similar also to the one developed in NSW by Hodgson JA). ‘Pfennig v The Queen does not require the judge to conclude that the similar fact evidence, standing alone, would demonstrate the guilt of the accused of the offence or offences with which he or she is charged’; ‘due weight must be given to the necessity to view the similar fact evidence in the context of the prosecution case’. 
This contextual approach to testing probative value is understandable as a response to Pfennig’s conflation of probative value and proof. However, it is not a solution. If the assessment is fully contextual, then the demands placed on the propensity evidence will depend upon the strength of the prosecution’s other evidence. Weakly probative, highly prejudicial propensity evidence could gain admission on the back of other evidence. And vice versa; highly probative propensity evidence may be excluded because of other perceived failures in the prosecution case. This doesn’t seem to be a sensible approach to the risk of prejudice.
The contextual approach also appears at odds with other measures of probative value – ‘influence’,  and the ‘need principle’.  The stronger the other prosecution evidence the less the need for the propensity evidence, and the less the scope for propensity evidence to influence the result. These other perspectives can present something of a paradox for the contextual approach. As the probative value of the other evidence increases, the probative value of the propensity evidence assessed contextually increases. However, there comes a point where the other evidence proves guilt without the propensity evidence, at which point the propensity evidence becomes irrelevant. ‘ Just as the evidence is about to reach the probative value summit, it falls off a cliff.’ 
In Phillips itself, the High Court showed that, despite the potential for the contextual approach to temper the exclusionary rule, the Pfennig test can remain extremely stringent. The defendant was charged over a series of sexual offences against six young women. The trial judge had joined the charges with cross admissibility, noting the similarities between the allegations, including that each assault had taken place at a social gathering, the defendant had engineered an opportunity to be alone with her, he had first sought consent and failing that had used threats of violence. Further, the alleged offences all took place within the space of a couple of years.  There was no real suggestion of joint concoction. The trial judge considered the evidence highly probative – what would be the chance of such similar allegations being made if they weren’t true? The defendant’s appeal was dismissed by the QCA, but a further appeal to the High Court was upheld. The defendant, released on bail, promptly committed further similar offences.
There are actually two bases to the HCA’s decision, both deeply problematic. To the extent that commission or the defendant’s conduct was in issue, the HCA held that the allegations of other misconduct was insufficiently probative. The Court said ‘The similarities relied upon were not merely not “striking”, they were entirely unremarkable.’  This highlights a key connection long drawn between the probative value of propensity evidence and the degree to which it shares similarities with the charged offence. This focus on similarities is heightened further by a second point that the HCA made about assessing probative value. The assessment assumes that the evidence will be accepted.  Doubts about a propensity witness’s credibility should not diminish probative value at the admissibility stage. The relationship between probative value and similarity is key and I’ll return to it later.
The second basis of the HCA’s decision in Phillips is directed to offences where consent was in issue. Here, the HCA made an even stronger and more dubious finding. The other allegations were not only insufficiently probative. They were irrelevant! Evidence of other complainants’ lack of consent ‘proves only what mental state each of the other complainants had on a particular occasion affecting them, and that can say nothing about the mental state of the first complainant on a particular occasion affecting her’.  This argument is fundamentally flawed. Of course there is a connection between the different complainants having non-consensual sex with the defendant – the defendant’s conduct. He forced them. Why else would they do something that they didn’t consent to? The defendant’s conduct is the common thread. The fact that one woman (was forced to have) had non-consensual sex with the defendant lends support to another woman’s claim to have (been forced to have) had non-consensual sex with the defendant in similar circumstances.
Common law – QCA
I think that Phillips, on the consent/relevance point, should not be considered to establish a precedent. As a matter of principle, evidence is relevant if it impacts on the probability of a fact in issue. Whether or not evidence is relevant in a particular case is very much tied to the facts of that case and is a question for the trial judge. 
Phillips could be avoided on the basis that consent is never an isolated issue. The defendant’s conduct – how it came about that the complainants had non-consensual sex with the defendant – will also be in issue. The QCA adopted this approach in Little.  The Court noted that ‘the only real issue about the elements of these offences’ was consent.  However, there was another factual issue which ‘was important for the resolution’ of the consent issue – ‘whether the appellant’s sexual acts occurred in the circumstances and in the manner described by the complainant in her evidence, which the appellant challenged by defence counsel’s cross-examination’.  The prosecution case was that the defendant used force – including threats with a knife – while the defence claimed sex was entirely consensual. The propensity evidence in that case ‘was relevant and admissible at the trial because of its probative force as support for the complainant’s evidence that the appellant disguised himself, forced entry into the complainant’s room, and assaulted, threatened, and bound the complainant to facilitate his sexual conduct’.  Phillips was distinguished on the basis that ‘[i]n that case, unlike in this case, the similar fact evidence was not tendered for the purpose of proving that the defendant engaged in any conduct. It appears that the similar fact evidence was tendered to prove only that the complainant did not consent. … [T]he similar fact evidence is not rendered inadmissible merely because it indirectly proves that the complainant did not consent to the sexual acts permitted by the appellant.’ Actually, in Phillips, the same kind of argument could, and should, have been adopted by the HCA, but wasn’t.
Strangely, Little  sought support from Collins,  a case in which Phillips was followed uncritically. In Collins the prosecution ‘accepted that joinder was impermissible to prove lack of consent’.  The QCA accepted ‘evidence of each complainant as to lack of consent is irrelevant as to whether any other complainant consented’.  The force and other strategies of the defendant alleged in Collins were less extreme than in Little, but they still served as a common connection between different complainants’ claims of non-consent.
Putting consent to one side, there has also been variability in the extent to which the QCA has accepted the stringency of the admissibility test under Pfennig and Phillips. In Phillips the HCA criticised the QCA from straying from the HCA’s language. In subsequent decisions the QCA has been careful to adopt the HCA’s language but has applied the HCA test in some cases with less stringency than in other cases.
The claim that the Pfennig/Phillips test has been applied inconsistently is difficult to establish definitively. The test has been formulated in various different ways, and even with a single formulation, would not be mechanically applicable. Even with a fixed identifiable probative value threshold, there would be a great deal of judgment involved in determining whether the evidence meets that threshold. Each case is decided on its facts. But this is not entirely a case of comparing apples and oranges. While probative value assessments are based on a wide range of factors, they still end up on the same scale. And sometimes evidence that seems to lie lower on the scale has gained admission while evidence that seems to lie higher on the same scale has been excluded.
Gregory  provides a clear example of a more lenient application of the Pfennig/Phillips test. The defendant was charged with sexual offences against a 14-year-old boy relating to 2009 events. The prosecution adduced evidence of convictions for sexual offences against an 11-year-old boy from 16 years earlier. There was just one other victim with a considerable time gap. In addition, defence counsel, pointed out other differences between the offences, including that the complainant was a streetwise teenager who the defendant met on the street, whereas the defendant had a social connection with the other boy, through his parents. Also, the earlier victim was alone, whereas the offences against the complainant were committed with someone else present, Ms Smith (a slightly older friend of the complainant).  However, the court still found sufficient ‘striking similarities’, ‘pattern’ or ‘unusual features’.  In both cases the defendant struck up a friendship with a male child and engineered things so that the child very rapidly shared his bed in the evening.  Massage was used as a pretext, as part of the sexual grooming process. The court actually used one difference – the presence of Ms Smith – to explain another difference – the fact that coercion could not be used as directly with the complainant. 
Another example of relatively lenient application of Phillips is Little,  discussed above on the consent point. The defendant was charged with offences including a sexual assault. The complainant’s evidence was that he had forced his way in to her hotel room and raped her. The prosecution adduced evidence that the defendant had pleaded guilty to offences relating to three incidents 16, 17 and 21 years earlier similar to that related by the complainant. He had broken into women’s homes wearing a balaclava, threatened them with a knife, bound their hands and feet (except in one of the earlier incidents) and sexually assaulted them.  The defendant admitted to having had sex with the complainant but disputed the circumstances and said it was consensual. The two had ‘recently had many intimate encounters’.  While the shared features might be described as ‘striking’, there was the significant time gap following the other offences, and one major difference – in the earlier offences the defendant did not know the victims (and identity was in issue) whereas he knew the complainant well (and consent was in issue). The Court nevertheless upheld admissibility, commenting ‘[i]t is not ordinarily to be expected that a modus operandi will involve behaviour by an offender which is identical in every single respect upon each occasion of offending. Some changes are virtually inevitable as a result of differing circumstances’. 
Gregory and Little are cases where admissibility was upheld. However, given the nature of the evidence, had the admissibility test been applied as stringently as in Phillips, the Court might have been expected to go the other way. Recall that, in Phillips, there were half a dozen complainants with similar accounts of the defendants’ serial sexual assaults committed over a short period of time. In the QCA’s very recent decision in Davidson,  the evidence appears more probative than in Gregory and Little; there were nine alleged victims over a far more concentrated time-frame. Joinder and cross-admissibility was upheld but the court split.
Interestingly, McMurdo JA, delivering the leading judgment for the majority, quoted from Bauer,  a recent appeal HCA appeal regarding UEL admissibility from Victoria: ‘the logic of probability reasoning dictates that, for evidence of the offending against one complainant to be significantly probative of the offending against the other, there must ordinarily be some feature of or about the offending which links the two together’.  As I discuss below, Bauer may be viewed as adopting a stringent interpretation of the UEL test. However, McMurdo JA found the admissibility test satisfied. There were many common features across the numerous alleged offences. All women attended the defendant in his capacity as masseur. They were rendered vulnerable by this context – they were isolated and in a state of undress. The way in which the defendant offended was broadly similar in each case and his actions were accompanied by inappropriate or suggestive comments. In some cases he went quite a bit further than in other cases, but it was held ‘[a]ll were committed in relevantly identical circumstances’ and admissibility upheld.
Dissenting, Boddice J, like McMurdo JA, drew upon the HCA’s reaffirmed demand in Bauer for ‘sufficient nexus … common linkage … specific connection … [which] requires an assessment of the similarities in the evidence’.  However, unlike the majority, Boddice J placed weight on differences between the offences in terms of the parts of the body being touched, and the degree of sexual touching. He held that, by reference to these differences, some of the counts should be severed without cross-admissibility.
In other recent QCA cases the Court has been unanimous in applying the exclusionary rule with greater stringency. As in Phillips and like Boddice J in Davidson, the QCA has, in some cases, been unpersuaded by a catalogue of similarities among an entire series of alleged assaults, and has given weight to differences in detail. Collins is the case discussed earlier where the QCA uncritically applied the Phillips notion that propensity evidence is irrelevant to consent. In Collins the QCA also drew on the stringency of Phillips where commission was in issue. The defendant was charged with a series of sexual offences against seven young women while they were on his yacht. The prosecution relied upon the fact they’d all been lured there by him offering them employment. He then took advantage of their isolation and vulnerability. The Court, however, gave emphasis to a relatively fine distinction – while in all cases they had been offered a drink, only in some cases did the complainants testify that they felt stupefied by the drink, suggesting it was spiked. The QCA drew a distinction between the two classes of complainants on this basis.  Only those counts with the striking similarity of a stupefying drink could be joined. 
In Collins President McMurdo, giving the leading judgment, said she found Gregory unhelpful. These cases ‘necessarily turn on their own facts’.  Of course, assessing probative value does require a detailed consideration of the facts, and this makes it difficult to compare approaches between cases. Nevertheless, like Phillips, the approach in Collins is very much towards the stringent end of the spectrum.
A relatively strict approach was also taken in Nibigira  although, admittedly, the evidence in this case appears weaker than in Phillips and Collins. The defendant was convicted of CSOs against four complainants. The counts had been joined with cross-admissibility. On appeal Gotterson JA giving the leading judgment held the evidence was not cross-admissible and the charges should have been severed. The prosecution had relied on an ‘underlying pattern’.  In each case the defendant was alleged to have taken advantage of his position of trust and authority as a church leader; the offences occurred in connection with choir practice and other church related events; the victims were girls of similar age around 10, members of the choir; and the offences all occurred around the same time. However, the Court accepted the defence argument that there were differences in the seriousness of the acts, from indecent touching to penetration; locations, whether a car or at the defendant’s house; riskiness, in terms of the proximity of other people; some differences in the ages of the complainants; and wide variation in the extent to which the complainants were groomed.  The QCA quoted extensively from Phillips with emphasis on the stringency that the HCA gave the admissibility test in principle and in application. 
The QCA in Nibigira also relied upon MAP,  decided shortly after Phillips. That case involved two counts of digital rape of two 15-year-old females, known to him, in the same beachside holiday house a year apart, with others nearby (resulting in one conviction and one acquittal). The QCA in MAP held that, despite these similarities, there were also differences – whether the offences were accompanied by drinking and physical threats – and as a result there was no ‘underlying pattern’.  The QCA held that the same reasoning applied in Nibigira. Because of the differences there was no ‘underlying unity’  or ‘pattern’.  The similarities were only at ‘a rather generalised level’.  The counts in Nibigira were ordered to be split. However, it held the counts involving complainants A and D could be joined with cross admissibility, as could the counts involving complainants B and C. In Davidson just a year later, the majority upheld admissibility and Gotterson JA said Nigibira is ‘clearly distinguishable’. Perhaps that is correct – there were more complainants in Nigibira with more distinctive similarities among their allegations. However, it is more difficult to distinguish Davidson from Gregory and Little. There are points of difference, of course, however on balance, the evidence in Gregory in particular appears no stronger.
While considering the QCA decisions it is worth noting couple of further points. And this is with regard to the directions that trial judges give juries where propensity evidence is admitted. Trial judges continue to warn the jury against ‘pure propensity reasoning’.  The jury ‘must not reason that because the defendant committed those offences [on a prior occasion], he is therefore the kind of person who would or might commit the offences that have been charged’.  These directions were apparently not challenged by the prosecution – indeed, they are endorsed in the Benchbook.  However, they are problematic. They show signs of the lingering notion, mentioned earlier, that propensity reasoning is prohibited absolutely. Propensity reasoning is now recognised as acceptable, provided the evidence satisfies the admissibility tests. It is absurd to allow the evidence in for a propensity purpose and then warn the jury not to engage in propensity reasoning.
Standard Queensland jury directions are problematic in another respect. They embody the Pfennig admissibility test. Juries are being directed to only use propensity evidence if the evidence is so probative (or ‘strikingly similar’) that there is no reasonable explanation for it other than the defendant’s guilt.  Again, this direction was given in several of the cases discussed, apparently without objection, and it appears in the Benchbook.  This direction, like the Pfennig test conflates probative value and proof, applying the criminal standard to the propensity evidence in isolation. To suggest that the jury can only use the propensity evidence if it is sufficient to exclude innocence as a reasonable possibility is contrary to the cumulative operation of proof. As I’ll mention later, UEL jurisdictions have also experienced this conflation and slippage, long after having got rid of the Pfennig test. But there are recent signs that the problem is being remedied in the UEL jurisdictions.
As with the common law tests, the UEL admissibility tests are inherently flexible. This is obviously the case with the asymmetric balancing test in s 101 – the required level of probative value varies depending upon the assessed prejudicial risk which is variable. Moreover, while some courts have accepted that the risk of prejudice can be addressed through judicial direction, others have made the point that this simplistic assumption may, in effect, render s 101 inoperable.
The ‘significant probative value’ test of ss 97 and 98 has received more attention than the s 101 test. While ‘significant probative value’ is fixed rather than variable, the expression is a little indeterminate. The High Court in Hughes recently sought to provide clarification. Gageler J observed that ‘significant probative value … is lower than … “substantial” probative value; but, to meet the threshold of significant probative value, evidence must still be “important” or “of consequence” to the assessment of the probability of the existence of a fact in issue.’ Gordon J and the majority added a further synonym: ‘the evidence must be influential in the context of fact-finding’. Of course, much indeterminacy remains. And even if ‘significant probative value’ could be pinpointed on some kind of scale, its assessment would depend upon a range of factors and would be open to different approaches.
Within each UEL jurisdiction, as with the common law in Queensland, courts have interpreted the ‘significant probative value’ requirement with varying degrees of stringency.  As well as this intra-jurisdiction variation, a split recently emerged between Victorian courts, maintaining a fairly strict exclusion, and NSW courts, which liberalised admissibility. In Velkoski v The Queen  the Victorian Court of Appeal (VCA) suggested that ‘sufficient similarity or distinctiveness in the features of the proposed tendency evidence’ may require something ‘“remarkable”, “unusual”, “improbable” [or] “peculiar”’.  The Victorian Court criticised statements of the NSWCCA, that the other misconduct need not be ‘closely similar’  with the charged offence, for lowering the admissibility threshold ‘too far’. 
Shortly after the VCA decision in Velkoski, the NSWCCA in Hughes v The Queen indicated it did ‘not accept that the language used by the VCA represents the law in New South Wales’.  Robert Hughes, the 1980s ‘Hey Dad..!’ TV star was convicted for a series of child sex offences against five complainants. The prosecution had relied heavily on tendency evidence. It argued that the complainants’ allegations were cross-admissible – that each charge derived support from evidence of similar misconduct provided by the other four complainants. Further tendency evidence of uncharged misconduct from six other witnesses was also admitted.  The NSWCCA upheld the admissibility and cross-admissibility of the allegations of complainants and other tendency witnesses, notwithstanding that the complainants’ ages ranged from six to 15 with another alleged victim in her early twenties, that they were in a variety of social and professional relationships with the defendant, and they gave evidence of various sexual touching and exposure behaviours, and in various contexts.
The defendant appealed to the High Court arguing that the Court should adopt the Victorian demand for specificity in Velkoski and recognise that the alleged behaviours were too ‘dissimilar’  and the alleged tendency at too high a level of ‘generality’  for the evidence to acquire significant probative value and gain admissibility.  For example, the defendant argued
there was a ‘world of difference’ between the evidence concerning EE (count 10), who was 15 years old and whom the appellant encouraged to commit indecent acts in a park and in a driveway, and the evidence concerning SH (counts 3 to 6), which involved intrusive acts ‘in a darkened bedroom, in her bed, when she was only six, seven or eight’. 
The appeal was dismissed by a narrow 4:3 margin. The majority judgment of Kiefel CJ, Bell, Keane and Edelman JJ upheld the admissibility determinations of the trial judge and NSWCCA, expressly disapproving Velkoski as ‘unduly restrictive’  and inconsistent with the legislative scheme.  There were three dissenting judgments. Gageler J was not that far from the majority in approach; he interpreted the facts differently with regard to one count. Nettle J, in the longest judgment, appeared to diverge far more sharply. He defended the VCA’s approach as orthodox  and criticised the NSW ‘for so lowering the bar’ without ‘justification in principle or as a matter of statutory interpretation’.  Gordon J agreed with Nettle J.
It should be noted that, to the extent the majority did endorse of a more liberal approach to admissibility, this was limited to commission cases. The majority suggested that more would be required of tendency evidence ‘to prove the identity of the offender for a known offence [than] where the fact in issue is the occurrence of the offence’.  In relation to identity, ‘the probative value of tendency evidence will almost certainly depend upon close similarity between the conduct evidencing the tendency and the offence’.  Similar suggestions can be found in Gageler J’s judgment  and elsewhere,  including the Royal Commission’s work.  But, despite this widespread support, the distinction has not been properly substantiated. As I’ve argued elsewhere,  this distinction appears to reflect the heavily contextualised approach to probative value. In commission cases the propensity evidence operates in conjunction with the complainant’s evidence, whereas in identity cases it may have to do all the work itself. However, as I’ve mentioned this heavily contextualised approach seems to conflate probative value and proof. The better view is that ‘there is no special rule for identification cases’. 
In Hughes, a commission case, the majority upheld admissibility despite several dimensions of variability among the charged and uncharged sexual misconduct. It held significant probative value does not require ‘operative features of similarity with the conduct in issue’.  Having said that, it is unclear how liberal the majority decision is. The majority accepted that the probative value of a tendency is in proportion with the ‘particularity’  or ‘specificity’  with which it can be expressed. The majority doubted whether it would be sufficient if ‘the evidence does no more than prove a disposition to commit crimes of the kind in question’.  And, in Hughes, in finding significant probative value the majority emphasised two additional factors which it saw as common to all the offences. As well as displaying the defendant’s ‘sexual interest in … underage girls’, the evidence revealed that the defendant had ‘a tendency to act on that interest by engaging in sexual activity with underage girls opportunistically, notwithstanding the risk of detection’. 
Nettle J used stronger language in identifying the requirement for admissibility. He indicated that there must be ‘ some logically significant underlying unity or commonality’.  This ‘logically significant connection’  may be found in ‘similarity in the relationship of the accused to each complainant; … between the details of each offence or the circumstances in which each offence was committed; [or in the] modus operandi or system of offending’.  And his application of this requirement was quite stringent. He expressly held that the shared features of opportunism and riskiness did not amount to a sufficient connection between the offences. 
Hughes was decided only two years ago. But since then the High Court has handed down two further decisions regarding the admissibility of tendency evidence under the UEL, Bauer  and McPhillamy.  These are all but unanimous decisions (in McPhillamy Edelman J agreed with the majority in a short separate judgment) and while the composition of the HCA has not changed, these decisions seem more aligned with Nettle J’s dissenting decision in Hughes and the more restrictive Victorian approach than with the more liberal approach of the NSWCCA which the Hughes majority appeared to approve. Admittedly, the difference in language between the majority and Nettle J in Hughes is relatively subtle, and the difference in decision turned upon the significance to be attached to a couple of specific features of offending – opportunism and riskiness. Nevetheless, the language used in Bauer and McPhillamy is more in line with that of Nettle J in Hughes and these more recent cases may mark a more demanding approach to tendency evidence.
In Bauer, a prosecution appeal from the VCA, the High Court spoke of the need for a ‘special, particular or unusual feature’,  ‘some feature of or about the offending which links the two together’.  Without this, ‘evidence that an accused has committed a sexual offence against the first complainant proves no more about the alleged offence against the second complainant than that the accused has committed a sexual offence against the first complainant’.  In Bauer, the tendency evidence was held to be cross-admissible, and the counts correctly joined, but the Court placed emphasis on a special feature. All the offences were against the same complainant. This provided the necessary link. 
In McPhillamy, on appeal from the NSWCCA, the tendency evidence was ultimately held to be inadmissible. The defendant was charged with sexual offences against an 11-year-old altar boy in 1995-6 when the defendant was an acolyte. The tendency evidence related to admitted prior sexual misconduct against two 13 year old boys at a boarding school, ten years earlier when the defendant was a housemaster. A majority of the CCA had upheld admissibility on the basis that there was an ‘overriding similarity’ from which differences between the cases ‘did not detract’.  It was open to the jury to reason that the defendant’s sexual interest in boys had not attenuated over the 10 year time difference.
But the High Court preferred Meagher JA’s dissenting judgment in the CCA in which he referred to the ‘generality of the tendency’  and the ‘absence of sufficient similarity’.  They emphasised that there was ‘no evidence that the asserted tendency had manifested itself in the [intervening] decade’.  There was a need for ‘some feature of the other sexual misconduct and the alleged offending which serves to link the two together’.  The defendant’s supervisory position over the three boys was insufficient as were the similar ages of the boys and the similar sexual misconduct. The Court noted differences  – the boarders were homesick and vulnerable and the offending took place in the privacy of the defendant’s bedroom; the altar boy did not have that vulnerability and the offence allegedly took place in a public toilet. The evidence ‘rose no higher in effect than to insinuate that, because the appellant had sexually offended against B and C 10 years before, in different circumstances, and without any evidence other than A’s allegations that he had offended again, he was the kind of person who was more likely to have committed the offences that A alleged.’  It is difficult to compare between cases. However, while there were far fewer victims in McPhillamy than in Hughes and a time gap, the similarities appear greater and the differences slighter than in Hughes.
A further aspect of Bauer should be noted. While Bauer was a Victorian appeal concerned with the admissibility of tendency evidence, the HCA took the opportunity to disapprove of a common NSW judicial direction.
Contrary to the practice which has operated for some time in New South Wales, trial judges in that State should not ordinarily direct a jury that, before they may act on evidence of uncharged acts, they must be satisfied of the proof of the uncharged acts beyond reasonable doubt. 
This direction had already been prohibited in Victoria by ss 61, 62 of Jury Directions Act 2015. These provisions emphasise that the standard of proof is generally only applicable to the elements of the offence, a point made by the HCA many years earlier in Shepherd.  However, it appears to be an easy trap to fall into, and support for the direction was re-emerging, including in HCA decisions. 
The direction disapproved in Bauer resembles one of the problematic Queensland directions noted above – that a jury can only use propensity evidence if it finds there is no reasonable explanation for the evidence other than innocence. Like the direction disapproved in Bauer, the Queensland direction applies the criminal standard of proof to propensity evidence in isolation, conflating probative value and proof. A difference between the two is that the direction rejected in Bauer applies the criminal standard of proof to the circumstantial fact of the defendant’s other misconduct, while the Queensland direction focuses on the extent to which the propensity evidence proves the defendant’s guilt. Both are inappropriate. Propensity evidence can lend support to the prosecution case even though there is some doubt about whether the other misconduct occurred, and even though there may be a reasonable explanation for that evidence consistent with innocence. Proof operates cumulatively. The propensity evidence doesn’t have to do the whole job itself. It is a brick, not a wall. 
Child sex offence reform
The Royal Commission into Institutional Responses into Child Sexual Abuse identified the exclusion of propensity evidence as ‘one of the most significant issues affecting criminal justice in child sexual abuse cases’ and recommended admissibility be broadened.  The Royal Commission’s concerns related to both the common law  and the UEL, even in its most liberal manifestations. Reporting after the High Court decision in Hughes, the Royal Commission noted with approval that it supported the more liberal NSW approach to admissibility in the UEL, but suggested the decision did not go far enough, and also failed to provide courts with sufficient guidance.  (As just discussed, since Hughes, in Bauer and McPhillamy the High Court appears to have retreated from this relatively liberal approach.) The Royal Commission put forward draft legislation to further broaden admissibility in child sex proceedings.  The New South Wales government rejected the Royal Commission’s draft legislation,  but found force in the arguments for broadening admissibility and referred the matter to the Council of Attorneys-General (CAG). The CAG set up a working group to consider reforms extending beyond CSO cases to propensity evidence in criminal matters generally. 
In June the NSW Attorney General, Mark Speakman, issued a press release indicating the direction of reform flowing from the CAG process.  The press release refers to ‘proposed reforms to the Uniform Evidence Law’, however, given that the Royal Commission and the CAG are both bodies operating across Australia, the reforms may also be considered by non-UEL jurisdictions such as Queensland. The press release noted the following features.
“A new rebuttable presumption would ensure evidence that a defendant has, or has acted on, a tendency to have a sexual interest in children is presumed to have ‘significant probative value’.”
“Judges would be required to exclude tendency or coincidence evidence about a defendant if its probative value does not outweigh the danger of unfair prejudice to the defendant.”
“Targeted legislative guidance, based on the findings of the Royal Commission, would help dispel misconceptions that have minimised the perceived value of this evidence in the past.”
From this press release it appears that, unfortunately, the CAG did not take the opportunity to simplify propensity law. The UEL’s unnecessary and counterproductive coincidence/tendency distinction will be retained,  as will its double admissibility test incorporating both a fixed threshold and a balancing test. In effect, the law contains four tests instead of one.
Instead the focus is on opening up admissibility, particularly in CSO cases. To achieve this, the double admissibility test will be modified in its application in two respects. While the first test remains a requirement of significant probative value, for CSO cases this will be presumed. It would then be up to the defendant to demonstrate the evidence lacks significant probative value. The planned change to the second test appears to extend beyond CSO cases to tendency and coincidence evidence in general. The test remains a balancing test, but it will become symmetrical. The prosecution will only need to prove that probative value outweighs prejudicial risk – under the existing s 101, probative value needs to substantively outweigh prejudicial risk. The reform removes the illogic of excluding evidence where its benefits may outweigh its costs. These changes to the two admissibility tests are accompanied by a third feature of the reforms – legislative guidelines to assist the courts in assessing probative value. These guidelines will reflect the Royal Commission’s view that courts have traditionally undervalued propensity evidence.
The modification to the first admissibility test would presume that propensity evidence has significant probative value, but only in CSO cases. Should CSO cases receive this special treatment? As the Royal Commission recognised, propensity evidence can be particularly valuable in child sex offence cases. Given the nature of child sex offending, evidence can be hard to come by. The offences are conducted in private, the victim may delay reporting the offence for months, years or even decades, resulting in the loss of forensic science or medical evidence of the offence. The trial may turn in to a battle of credibility – the complainant’s allegation versus the defendant’s denial. The law now has some appreciation the many reasons why the victims of child sexual abuse do delay reporting;  generally these are variations on the power imbalance between adult offender and vulnerable child victim – the offender’s grooming, normalisation and threats; the victim’s embarrassment, confusion and fear. Nevertheless, the delay may still damage the complainant’s credibility,  and bring greater scrutiny to the complainant’s allegation.  And so the battle of credibility may be a difficult one for the prosecution to win, bearing in mind that it is not sufficient that the complainant be found more credible than the defendant. The criminal standard of proof requires that the complainant must be found so credible that there is no reasonable possibility that the defendant’s claim of innocence is true.
As is well known, the nature of CSO cases is such that, compared to other crimes of comparable seriousness, reporting rates are lower, prosecution rates are lower, rates of guilty pleas are lower, and, where they do go to trial, conviction rates are lower.  In CSO cases there is often a particular need for additional prosecution evidence. This consideration certainly influenced the Royal Commission’s thinking in making its recommendations. 
And, it seems, it is not uncommon for propensity evidence to be available to fulfil this need.  In a significant proportion of CSO cases, the prosecution has evidence of other alleged victims or a defendant’s prior convictions for similar offences. It is an interesting and important question whether evidence is more generally available for child sex offences than for other offences – whether CSO defendants are more likely to have a criminal history than other defendants. If so, this may suggest that, not only is there a greater need for propensity evidence, and not only is it readily available, but propensity evidence is also more probative in CSO cases. Child sex offenders are more likely to reoffend; they have stronger propensities. Many of the cases examined by the Royal Commission did involve offending against numerous victims,  and this may have influenced the Royal Commission’s view that the evidence carries significant probative value.
However, while the Royal Commission’s work did highlight the need, availability and probative value of propensity evidence in CSO cases, it was wary of making comparative assessments for the simple reason that non-CSO cases were beyond its remit. Indeed, its focus was on institutional child sexual abuse. It appears that there may be stronger policy reasons for freeing up admissibility of propensity evidence in CSO cases than in other types of cases, but the Royal Commission did not do this work. The Royal Commission recommended changes only for CSO cases simply because they were its focus. And, it seems, this narrow focus has restricted the subsequent law reform process.
Even if there were policy arguments that may justify giving CSO special treatment with regards to propensity evidence, this proposal may raise practical objections. The line between CSO and non-CSO cases is a further bifurcation: tendency/coincidence, fixed threshold/variable threshold, and now CSO/non-CSO. It introduces yet another complexity into the law. And this latest line may be difficult to draw. CSO cases may have significant non-CSO aspects – physical violence, physical and sexual violence against adults. It is unfortunate that the Royal Commission and the CAG Working Group did not give greater weight to the practical benefits of simplicity.
Of course, in terms of the scope of the reforms, the hands of the Royal Commission and the CAG were tied, to a degree, by the original limited terms of reference. There may be something to be said for postponing propensity evidence reform pending a broader enquiry. However, against this, the Royal Commission has highlighted urgent need for reform, and generated momentum for it. Now may be the moment. In view of this, it is to be hoped that at some point before the legislation is passed, someone will make a bold intervention to generate simpler, more workable legislation that draws more general lessons from the Royal Commission’s findings.
I have advocated – so far unsuccessfully – for a return to the symmetrical balancing test as a basis for admissibility.  Get rid of the fixed threshold requirement of ‘significant probative value’, but retain the balancing test in its symmetrical form. The requirement should be simply that probative value outweigh prejudicial risk. This should apply to CSO and other cases, and there should be no distinction between coincidence and tendency evidence. As compared with the current double admissibility test, this will liberalise admission, responding to the Royal Commission’s recognition of the traditional undervaluation of propensity evidence. At the same time, evidence will be excluded if its costs may outweigh its benefits. The test clearly operates at a high level of generality, and would leave much to individual judgement. However, this can be handled by legislative guidance. The Royal Commission’s work and the research it draws upon provides the foundation for the drafting of appropriate guidance. To maximise the value of the reforms, the aim should be to formulate guidance that will be useful broadly, not just in CSO cases.
Guidance on assessing probative value
The earlier discussion of the application of existing common law and UEL admissibility tests suggests that the precise details of the test don’t matter. The variation in QCA decisions shows that even with the apparently stringent Pfennig/Phillips test there is scope for a liberal approach to admissibility. And the divergence between Victoria and NSW together with the meanderings of the HCA indicate that the UEL tests may support approaches of varying degrees of stringency. This should not be surprising. The admissibility tests are open to interpretation. And judgment is also required in applying the tests – the determination of probative value and prejudicial risk. This observation adds to the desirability of having a simple admissibility test – the complicated details, even if they have a sound rationale, may achieve their purpose.
Signalling more liberal admissibility
This is not to say that trial judges, in determining admissibility, are wholly unconstrained by the terms of the admissibility test (whether common law or legislative). Some aspects of the test may be mechanically applicable – for example, that the trial judge at the admissibility stage take propensity evidence at its highest; which would include the assumption that other alleged victims are telling the truth. A further aspect of Bauer was to confirm that this position under the UEL extends to the risk of concoction – ‘unless the risk of … concoction … is so great that it would not be open to the jury to rationally accept the evidence … the risk … goes only to the credibility and reliability of the evidence, and … must be left to the jury’.  (Legislation adopting this approach, overriding the HCA authority Hoch,  has already been taken in many jurisdictions.  ) This principle is mechanically applicable precisely because it is absolute. The evidence, in this respect, is taken at its highest. Other aspects of the admissibility test require judgment and are inherently flexible. The way the test is expressed does exert some influence – it is not like pushing on a string – but it is like pulling on an elastic band.
Nevertheless, if the proposed legislation were introduced, it may exert greater influence because of the contrast with the existing laws. The proposed CAG reforms – presuming significant probative value for CSO cases and the more liberal symmetrical balancing test – would be clear signals to trial and appeal judges that this evidence should be admitted far more readily. Likewise my preference for the symmetrical balancing test by itself. And the judiciary would be receptive to these signals. 
These signals can be strengthened by the CAG-proposed legislative guidance on the application of the tests. The CAG press release may be construed as referring only to probative value – it suggests the guidelines will address ‘misconceptions that have minimised the perceived value of this evidence’. However, the guidelines could also address the assessment of prejudicial risk. As mentioned above, the considerations of probative value and prejudicial risk are linked – the more probative the evidence, the less room there is for prejudice to operate. Indeed, some of the traditional concern about jury overvaluation of propensity evidence may be the direct consequence of the law’s traditional undervaluation of propensity evidence. It should also be noted, the guidelines may also be of assistance to the trial judge, not only at admissibility, but in crafting jury directions. Here I will focus on the legislative guidance on assessing probative value – not prejudicial risk – and without reference to any complications that may arise in crafting jury directions.
Checklist of considerations
The question arises where this guidance might be derived from. One obvious source is past decisions. Notwithstanding differing levels of judicial stringency and the law’s traditional undervaluation of probative value, common factors do emerge and these may be worth considering. The Supreme Court of Canada in Handy suggested that courts have regard to: 
- the proximity in time and place of the similar acts
- the extent to which the other acts are similar in detail to the charged conduct
- the number of occurrences
- the circumstances surrounding or relating to the similar acts
- any distinctive features unifying the incidents
- any intervening events
- any additional factors tending to support or rebut the underlying unity of similar acts.
This kind of checklist may be useful in that, in some cases, key features may be missed. In Phillips, for example, the High Court seemed not to have appreciated the significance of the number of young women that had come forward with similar accounts of the defendant’s sexual misconduct.
However, this kind of list by itself would not be enough. While it certainly offers greater specificity than vague references to ‘underlying patterns’ and ‘logical connections’, it does not provide a proper standard as to how demanding courts should be in weighing up the various factors. In many cases the defence would be able to draw on the list to highlight respects in which the propensity evidence lacked probative value – low frequency, a lack of shared distinctive features, a time lag with intervening events, etc. This would work against the Royal Commission and CAG’s intention. The purpose of the guidelines is to address the traditional undervaluation of propensity evidence.
Similarity standards and criminal behaviour
In Hughes, the majority found the other misconduct evidence significantly probative on the basis of the shared features of riskiness and opportunism, while Nettle J viewed these features as unremarkable. Which approach is appropriate? In these cases the court is engaged in an enterprise of resolving uncertainty about criminal behaviour. As Gageler J recognized in Hughes, this is something that should ‘ be informed by social science data [and] scholarly work bearing on actual probabilities’.  He regretted that the parties had not made such material available in that case.  The Royal Commission, in its work drew upon a great deal of empirical data, and in determining how to use that data, it had reference to work on the logic of proof. It viewed the majority approach in Hughes as too strict. In CSO cases it is unnecessary to descend into detail. ‘T he two most important similarities are already present – sexual offending against a child’. 
To demand ‘striking similarities’, or something ‘remarkable’ or ‘peculiar’ seems to assume that CS offenders are highly specialized in their offending. In some cases it seems that courts demand that each offence carries the same distinctive hallmark. This assumption is not backed up by empirical data. The Royal Commission’s work revealed that while offenders may show particular preferences, they offend against ‘both girls and boys and children of quite different ages, … in a variety of ways [and] in different contexts – institutional, familial and others’.  Moving beyond the Royal Commission’s focus on child sex offenders, the same point can be made of other offenders. Recidivism analyses shows that criminals are often ‘“specialised generalists”, tending to commit a range of offences but … more likely to commit their specialization offence than any other particular offence’.  This data suggests that other misconduct may have probative value even without shared distinctive features.
Questions remain as to how recidivism data translates into probative value assessments. Quite a few commentators have pointed out that recidivism figures are often quite low. Annie Cossins (who favours more liberal admissibility) recently conducted a review of recidivism rates for extrafamilial child sex offenders. With follow-up periods ranging from four to 10 years and a variety of measures of recidivism, rates varied from 1.5 per cent to 26 per cent with about half the studies reporting figures of around 15 per cent.  These kinds of figures lead to suggestions that because an offender is unlikely to reoffend, prior offending is more consistent with innocence than with guilt.  But such views suffer the familiar problem of conflating proof and probative value. An item of evidence can be valuable even though, by itself, it does not prove guilt to any particular level. Consider motive evidence. This may form an important part of the prosecution case (where identity or intention is in issue). However, its value does not hinge upon the claim that a person with that motive would be likely to commit the offence. Many people have may have a motive to commit an offence without committing the offence. Motive evidence gets its value from the fact that someone with a motive is far more likely to commit the offence than someone without motive.
Probative value is a comparative assessment. Evidence will be probative of guilt if it is more consistent with guilt than to innocence. The Royal Commission quoted from British evidence theorist, Mike Redmayne.
[W]hat really matters is … a comparative judgment: whether the person with the murder conviction is more likely than other people to commit murder. … the question is whether the person has a comparative propensity to commit crime. 
The Royal Commission also drew upon my submission:
Prior conviction evidence is probative, not because it is highly probable that someone with a prior conviction is likely to reoffend, but because such a person is far more likely to commit an offence than someone without a prior conviction. Recidivism rates, while not incredibly high, are far higher than crime rates. Probative value is a ‘comparative judgment’, and its comparative nature suggests that tendency and coincidence evidence can have considerable probative value. 
There are methodological difficulties in settling upon precise figures for the relative likelihood of offending among prior offenders and the general public, but a rough indication can be provided by considering data gathered by Smallbone and Wortley of child sex offenders in custody. They identified 540 child sex offenders who were in custody or serving community correction orders in June 2000 in Queensland.  This is between one and two hundredths of one per cent of the population.  The researchers then obtained detailed information regarding the histories of 323 of these offenders.  They found that 21.3 per cent had prior convictions for sex offences (and 61.6 per cent had prior convictions of some kind, illustrating generalised patterns of offending).  Among the general population it is extremely unusual be in custody for a CSO. However, among those in custody for a CSO it is not at all uncommon to have prior convictions for sex offences (or other offences). Most people are very unlikely to commit a CSO. The likelihood of committing a CSO increases considerably for a person who is shown to have committed a CSO on another occasion. Evidence of other child sex offences is far more consistent with the defendant’s guilt on the current charges than with the defendant’s innocence. Evidence of other (more or less similar) offending is highly probative.
This analysis highlights a difficulty with Nettle J’s approach to probative value in Hughes, a difficulty suffered by other advocates of a stringent approach to admissibility. Nettle J suggested ‘[t]he commission of sexual offences by adults against children of either sex is depraved and deplorable, but, regrettably, it is anything but unusual’.  ‘[T]he bulk of the work of criminal courts in this country is devoted to dealing with sexual offences and the bulk of those offences are sexual offences against children’.  However, in assessing unusualness by reference to criminal defendants generally Nettle J is using an inappropriate conception of ‘unusualness’. At another point Nettle J conceded that child sex offending ‘ is unusual by the standards of ordinary decent people’.  This is a more appropriate conception of unusualness. The question is the consistency of other misconduct with guilt relative to its consistency with innocence. In determining this it is useful knowing how uncommon child sex offenders are among the general population, most of whom, by and large, are law-abiding. It is not surprising that the criminal courts will encounter a lot child sex offenders. But the frequency of child sex offenders among the criminal defendant population is not the question. 
Flowing from this analysis, guidelines for the assessment of the probative value should address the following related points. First, the guidelines should ward against the common problem of probative value being conflated with proof. Propensity evidence may be highly probative even though, by itself, it does not prove guilt to any particular standard. Second, the guidelines should emphasise the comparative nature of the probative value assessment. The question is not only how consistent the evidence is with guilt. The question is how more consistent the evidence is with guilt than it is with innocence. Some guidance on this point may be provided by considering the likelihood of that kind of offender reoffending, as compared with a member of the general population committing the offence.
Flowing from these guidelines, probative value assessments should not make strong demands in terms of striking similarities, logical connections or underlying patterns. It will be worth considering whether these features are present, and in this connection a checklist drawing on case law may be valuable. If present, these features may contribute to very high probative value – if the offence does display the same hallmark as a defendant’s prior offending, then it is very unlikely that this has appeared coincidentally. However, evidence of prior offending may acquire considerable probative value without a hallmark. Offenders generally do not specialise to that degree. The checklist should not obscure the important point that probative value turns on comparative propensity. Generally, it is far more likely that a person with a history of offending will commit an offence than someone without that history.
The regulation of propensity evidence has a long troubled history in Australia and other jurisdictions. The Royal Commission has given fresh impetus to the debate around propensity evidence, highlighting the exclusionary rule as one of the major obstacles to the enforcement of the prohibition on child sexual assault. The Royal Commission recommendations are in the process of being turned into legislative reform by the CAG. Unfortunately, so far it appears that the reforms will not take the opportunity to provide much needed simplification. The UEL currently contains two admissibility tests where one would suffice, and the UEL also unnecessarily distinguishes between two closely related varieties of propensity evidence – tendency evidence and coincidence evidence. It appears that the reforms will retain the two tests and the two types of propensity evidence and then further increase the complexity of the law by drawing a further distinction – between CSO and non-CSO cases. It will be regrettable if the opportunity to simplify the law is missed. There should be a single symmetrical balancing test for propensity evidence – the evidence should be excluded unless its probative value outweighs its prejudicial risk.
While it seems that the opportunity to simplify the law will be missed, the reforms do promise to relax admissibility. CSO evidence will be presumed to have significant probative, satisfying the first admissibility test. And the second admissibility test will be relaxed to a symmetrical balancing test, instead of requiring probative value to substantially outweigh prejudicial risk. Guidelines are also proposed to address the traditional undervaluation of propensity evidence. This direction of reform appears appropriate, not only for CSO cases, but across the board. The scepticism that the law has traditionally directed towards propensity evidence in large part is based on empirical misconceptions and logical errors. Evidence of other misconduct may acquire significant probative value even though the other misconduct does not share distinctive features with the charged offence. Criminals may reoffend without great specialisation or the adoption of a hallmark. Propensity evidence has been viewed as lacking probative value on the basis that it does not, by itself, prove guilt. However, this conflates probative value with proof. Probative value is a relative assessment. Most people don’t commit criminal misconduct, but having committed one offence, a person is far more likely to commit another. Propensity evidence – evidence of other misconduct – may not display a hallmark and may not prove guilt to any particular level of probability by itself, but is still often highly probative.
David Hamer * , University of Sydney Law School
 (2006) 225 CLR 303; David Hamer, ‘Similar Fact Reasoning in Phillips: Artificial, Disjointed and Pernicious (2007) 30 UNSW Law Journal 609-638.
 (2008) 235 CLR 334; David Hamer, ‘Admissibility and Use of Relationship Evidence in HML v The Queen: One Step Forward, Two Steps Back’ (2008) 32 Criminal Law Journal 351-368.
 See also David Hamer, ‘Propensity Evidence Reform after the Royal Commission into Child Sexual Abuse’ (2018) 42 Criminal Law Journal 234.
 Evidence Act 1929 (SA) s 34P; Evidence Act 1906 (WA) s 31A.
 Evidence Act 1977 (Qld) ss 132A, 132B.
 Royal Commission on Criminal Justice, Report, Cm 2263 (1993) i.
 Eg, ‘If the inadmissible chain of reasoning is the only purpose for which the evidence is adduced as a matter of law, the evidence itself is not admissible. If there is some other relevant, probative purpose than for the forbidden type of reasoning, the evidence is admitted, but should be made subject to a warning from the judge that the jury must eschew the forbidden reasoning.’ Boardman v DPP  AC 421, 453 (Lord Hailsham)
 Eg, Federal Rule of Evidence 404(b).
 See also David Hamer, ‘The structure and strength of the propensity inference: Singularity, linkage and the other evidence’ (2003) 29 Monash University Law Review 139, 145-6.
 DPP v P  2 AC 447, 460. Australian authority includes Sutton (1984) 152 CLR 528, 534 (Gibbs CJ); Harriman (1989) 167 CLR 590, 593-4 (Brennan J), 597-99 (Dawson J), 610 (Toohey J); B (1992) 175 CLR 599, 618-9 (Dawson and Gaudron JJ); Pfennig (1995) 182 CLR 461, 478 (Mason CJ, Deane and Dawson JJ); 515, 528 (McHugh J); BRS (1997) 191 CLR 275, 305 (McHugh J); Gipp (1998) 194 CLR 107  157 (Kirby J). See also Handy  2 SCR 908 ; Evidence Act 2006 (NZ) s 43(1).
 UEL Dictionary.
 Andrew Palmer, ‘The Scope of the Similar Fact Rule’ (1994) 16 Adel LR 161, 169 ; DP Leonard, ‘In defense of the character evidence prohibition: Foundations of the rule against trial by character’ (1998) 73 Indiana Law Journal 1161, 1184; Hughes (2017) 344 ALR 187  (Gageler J).
 People v Zackowitz 172 NE 466, 468 (NY 1930); Colin Tapper, Cross and Tapper on Evidence (London: Butterworths, 9th ed, 1999), 357; Adrian Zuckerman, Principles of Criminal Evidence (Oxford: Clarendon Press, 1989), 232
 Hock Lai Ho, A Philosophy of Evidence Law—Justice in the Search for Truth (2008) 337.
 DP Leonard, ‘In defense of the character evidence prohibition: Foundations of the rule against trial by character’ (1998) 73 Indiana Law Journal 1161, 1192.
 Pfennig (1995) 182 CLR 461, 528.
 See David Hamer, ‘Legal Structure of Propensity Reasoning’ (2016) 20 The International Journal of Evidence & Proof 131, 154-155.
 The test also appeared in earlier High Court decisions: Sutton v The Queen (1984) 152 CLR 528, 563–564 (Dawson J); Hoch v The Queen (1988) 165 CLR 292, 296; David Hamer, ‘Similar Fact Reasoning in Phillips: Artificial, Disjointed and Pernicious’ (2007) 30 University of New South Wales Law Journal 609, 612–613.
 (1995) 182 CLR 461, 483 (Mason CJ, Deane and Dawson JJ), 506 (Toohey J).
 (1995) 182 CLR 461, 482 (Mason CJ, Deane and Dawson JJ).
 Grant v The Queen (1975) 11 ALR 503, 505; Hodges Case (1838) 2 Lewin 228; 168 ER 1136.
 David Hamer, ‘Legal Structure of Propensity Reasoning’ (2016) 20 The International Journal of Evidence & Proof 131, 156.
 R v W  2 Qd R 531, 537, 533–4; R v Vinh Le  NSWCCA 49 ; see also Velkoski  VSCA 121 .
 The ALRC’s research confirmed the need for a relatively strong exclusionary rule: Australian Law Reform Commission, Evidence, Report No 26 (Interim), (1985), vol 1 at 456 . But the ALRC’s draft legislation was not adopted, so it is difficult to say how much this influenced Parliament: Ellis (2003) 58 NSWLR 700 at 714-715  (Spigelman CJ); Hughes (2017) 344 ALR 187  (Kiefel CJ, Bell, Keane and Edelman JJ).
 WRC (2002) 130 A Crim R 89; Joiner (2002) 133 A Crim R 90; Dyson Heydon ‘Similar Fact Evidence: The Provenance of and Justification for Modern Admissibility Tests’ in Aladin Rahemtula (ed), Justice According to Law: A Festschrift for the Honourable Mr Justice BH McPherson CBE (2006) 241, 251.
 BBH v The Queen (2012) 245 CLR 499 – (Crennan and Kiefel JJ); HML (2008) 235 CLR 334  (Gleeson CJ),  (Heydon J).
  1 Qd R 564, 573-574.
 (2006) 225 CLR 303 .
 Ibid .
 Ibid , with references in fnn:
 R v Boardman  AC 421 at 439 per Lord Morris of Borth-y-Gest, approved in Markby v The Queen  HCA 29; (1978) 140 CLR 108 at 117 per Gibbs ACJ, Stephen, Jacobs and Aickin JJ concurring.
 Perry v The Queen  HCA 75; (1982) 150 CLR 580 at 609 per Brennan J; Sutton v The Queen  HCA 5; (1984) 152 CLR 528 at 548-549 per Brennan J, 560 per Deane J, 565 per Dawson J; Harriman v The Queen  HCA 50; (1989) 167 CLR 590 at 633 per McHugh J; Pfennig v The Queen  HCA 7; (1995) 182 CLR 461 at 481 per Mason CJ, Deane and Dawson JJ.
 Sutton v The Queen  HCA 5; (1984) 152 CLR 528 at 534 per Gibbs CJ.
 Hoch v The Queen  HCA 50; (1988) 165 CLR 292 at 294-295 per Mason CJ, Wilson and Gaudron JJ.
 Hoch v The Queen  HCA 50; (1988) 165 CLR 292 at 301 per Brennan and Dawson JJ, approving words of Lord Hailsham of St Marylebone LC in R v Kilbourne  AC 729 at 749.
 Director of Public Prosecutions v P  2 AC 447 at 460 per Lord Mackay of Clashfern L Pfennig v The Queen  HCA 7; (1995) 182 CLR 461 at 483 per Mason CJ, Deane and Dawson JJ.
 Only mentioned in two paragraphs and then in the context of criticising the trial judge and O’Keefe  1 Qd R 564: (2006) 225 CLR 303 , .
 (2006) 225 CLR 303 .
 Hughes (2017) 344 ALR 187  (Gordon J), quoting from IMM v The Queen (2016) 257 CLR 300 ; see also (2017) 344 ALR 187 , (Kiefel CJ, Bell, Keane and Edelman JJ).
 Donald Piragoff, Similar Fact Evidence: Probative Value and Prejudice (Carswell, 1981), 146.
 David Hamer, ‘The Significant Probative Value of Tendency Evidence’ (2019) 42 Melbourne University Law Review 506, 527.
 All but one occurred over a 16 month period, with one occurring 18 months later.
 (2006) 225 CLR 303 .
 Ibid .
 Ibid .
 For this reason, it seems odd that the Hayne J raised it as an appeal point, and the court made it one of the bases for the decision: David Hamer, ‘Similar Fact Reasoning in Phillips: Artificial, Disjointed and Pernicious (2007) 30 UNSW Law Journal 609, 617 fn 62. See also Kirby J in Evans (2007) 235 CLR 521 -.
  QCA 113.
 Ibid .
 Ibid .
 Ibid .
  QCA 389.
 Ibid .
 Ibid .
  QCA 86.
 Ibid .
 Ibid .
 Ibid .
 Ibid .
  QCA 113.
 Ibid .
 Ibid 
 Ibid .
  QCA 120.
  HCA 40.
  QCA 120 , quoting from Bauer  HCA 40 .
  QCA 120 - citing Bauer  HCA 40 ; Phillips (2006) 225 CLR 303 ; Pfennig (1995) 182 CLR 461, 485.
  QCA 389 .
 Ibid .
 ibid .
  QCA 115.
 Ibid -.
 Ibid -.
 Ibid -.
  QCA 220.
 Ibid .
  QCA 115 .
 Ibid .
 Ibid .
 Ibid .
 Gregory  QCA 86 .
 Supreme and District Courts Criminal Directions Benchbook (March 2017 Amendments), ‘Similar Fact Evidence’ .
 Eg Little  QCA 113 ; Gregory  QCA 86 ; Nibigira  QCA 115 .
 Supreme and District Courts Criminal Directions Benchbook (March 2017 Amendments), ‘Similar Fact Evidence’ .
 Eg, DAO (2011) 81 NSWLR 568  (Simpson J).
 Eg, BC  NSWCCA 327  (Adams J, dissenting).
 (2017) 344 ALR 187  citing Lockyer (1996) 89 A Crim R 457, 459.
 (2017) 344 ALR 187  (Gordon J), quoting from IMM v The Queen (2016) 257 CLR 300 ; see also (2017) 344 ALR 187 , (Kiefel CJ, Bell, Keane and Edelman JJ).
 Eg, contrast the more liberal approach to incestuous CSO cases in RHB  VSCA 295  and DR  VSCA 440  with the stringency of Velkoski (2014) 45 VR 680 .
 (2014) 45 VR 680.
 Ibid  citing Reeves v The Queen (2013) 41 VR 275 .
 Velkoski (2014) 45 VR 680, ,  citing, eg R v Ford (2009) 273 ALR 286 ; R v PWD (2010) 205 A Crim R 75 .
 Velkoski (2014) 45 VR 680, 717 .
 (2015) 93 NSWLR 474 . This is a little ambiguous. It could mean that the NSWCCA considers that the VCA has misrepresented the NSWCCA’s approach. But in the context it appears to mean that the NSWCCA does not accept the VCA’s approach.
 Some was only held admissible in respect of some counts: Hughes v The Queen (2015) 93 NSWLR 474 .
 Hughes, ‘Appellant’s Submissions’, Submission in Hughes v The Queen, Case No S226/2016, 7 October 2016, .
 Ibid .
 Ibid .
 (2017) 344 ALR 187 .
 Ibid .
 Ibid .
 Ibid .
 Ibid .
 Ibid  (emphasis added).
 Ibid .
 DPP v P  2 AC 447, 462; Pearl Davidson, ‘A Tendency to Convict: Section 97 Evidence Act in Hughes v The Queen’ (2018) 22 International Journal of Evidence and Proof 144, 154-155. See discussion in David Hamer, ‘The Structure and Strength of the Propensity Inference: Singularity, Linkage and the Other Evidence’ (2003) 23 Monash University Law Review 137, 183-5.
 Royal Commission into Institutional Responses to Child Sexual Abuse (Criminal Justice Report, 2017) Parts III–VI, 594-595, 606.
 David Hamer, ‘The Significant Probative Value of Tendency Evidence’ (2019) 42 Melbourne University Law Review 506, 524-525; David Hamer, ‘The Structure and Strength of the Propensity Inference: Singularity, Linkage and the Other Evidence’ (2003) 29 Monash University Law Review 137, 175, 184.
 R v John W  2 Cr App R 289, 301; quoted with approval, English Law Commission, Evidence of Bad Character in Criminal Proceedings, Law Com No 273, Cm 5257 (October 2001), [2.23], [4.6]; see also David Hamer, ‘The Structure and Strength of the Propensity Inference: Singularity, Linkage and the Other Evidence’ (2003) 29 Monash University Law Review 137, 184-5. See also, R v Bromley  SASCFC 41  where the court notes the suggestion in Hughes that more is expected in identification case and adds, ‘Of course, the term “identification cases” may itself subsume cases of varying types so as to impact on the degree of similarity required.’
 (2017) 344 ALR 187 .
 Ibid  (Kiefel CJ, Bell, Keane and Edelman JJ).
 ibid  (Gageler J).
 Ibid ; see also  (Gageler J).
 Ibid  emphasis added; see also  (Gageler J).
 Ibid .
 Ibid. Nettle J used the term ‘logically significant connection’ in a conclusory fashion without unpacking it.
 Ibid , .
  HCA 40.
  HCA 52.
  HCA 40 .
 Ibid .
 Ibid , at  distinguishing IMM (2016) 257 CLR 300.
  HCA 52 .
 Ibid .
 Ibid .
 Ibid .
 Ibid .
 Ibid .
 Ibid .
  HCA 40  citing various cases at fn 88. See also David Hamer, ‘“Tendency Evidence” and “Coincidence Evidence” in the Criminal Trial: What’s the Difference?’ in Andrew Roberts and Jeremy Gans (eds), Critical Perspectives on the Uniform Evidence Law (Federation Press, 2017) 158, 164-168.
 (1990) 170 CLR 573 at 584-585.
 Eg, HML (2008) 235 CLR 334; David Hamer, ‘Admissibility and Use of Relationship Evidence in HML v The Queen: One Step Forward, Two Steps Back’ (2008) 32 Criminal Law Journal 351-368.
 The familiar image is attributed to Edward W Cleary (ed), McCormick’s Handbook of the Law of Evidence (West Publishing Co, 2nd ed, 1972) 436.
 Royal Commission, Criminal Justice Report (Commonwealth of Australia, 2017), Parts III-VI, 411; see at 591.
 In fact the Royal Commission expressed concern that the stringency of the common law had adversely affected the interpretation of the UEL: ibid 641-642; on the relationship between common law and the UEL see Hughes (2017) 344 ALR 187 - (Nettle J); David Hamer, ‘The Significant Probative Value of Tendency Evidence’ (2019) 42 Melbourne University Law Review 506, 520-521.
 Royal Commission, Criminal Justice Report (Commonwealth of Australia, 2017), Parts III-VI, 635.
 Royal Commission, Criminal Justice Report (Commonwealth of Australia, 2017), Parts VII to X, Appendix N. The legislation was drafted by New South Wales Parliamentary Counsel’s Office: above n 3, 649.
 Following consultation: NSW Department of Justice, ‘Strengthening child sexual abuse laws in NSW’, (Discussion Paper September 2017), 94.
 Council of Attorneys-General, ‘Communique, 1 December 2017.
 Mark Speakman, NSW Attorney General, ‘Evidence Law Reform’ (‘Media Release’, 28 June 2019) https://www.justice.nsw.gov.au/Pages/media-news/media-releases/2019/evidence-law-reform.aspx .
 The Royal Commission saw ‘considerable merit’ in eliminating the distinction, and ‘little merit in maintaining what seems to be an artificial distinction’: Royal Commission, Criminal Justice Report (Commonwealth of Australia, 2017), Parts III-VI, 643. Ultimately, however, the Royal Commission recommended maintaining the distinction and focused on broadening admissibility, ‘anticipat[ing] that in due course’ it would be abolished: ibid. Why wait?
 Eg Criminal Procedure Act 1986 (NSW) s 294.
 Crofts (1996) 186 CLR 427, 452.
 Longman (1989) 16 CLR 79, 90; UEL s 165B.
 Royal Commission, Criminal Justice Report (Commonwealth of Australia, 2017), Parts I-II, 164-168.
 Royal Commission, Criminal Justice Report (Commonwealth of Australia, 2017), Parts III-VI, 411.
 Eg, ‘In 58 per cent of cases, the offender had committed CSA offences against more than
one victim’: Karen Gelb, A statistical analysis of sentencing for child sexual abuse in an institutional context, (Royal Commission into Institutional Responses to Child Sexual Abuse, 2016) ix.
 Eg, Royal Commission, Criminal Justice Report (Commonwealth of Australia, 2017), Parts III-VI, Ch 24.
 David Hamer, ‘Propensity Evidence Reform after the Royal Commission into Child Sexual Abuse’ (2018) 42 Criminal Law Journal 234. The common law version of test continues to operate in Canada, and has legislative form in NZ: Handy  2 SCR 908 ; Evidence Act 2006 (NZ) s 43(1). See David Hamer, The Admissibility and Use of Tendency, Coincidence and Relationship Evidence in Child Sexual Assault Prosecutions in a Selection of Foreign Jurisdictions (Royal Commission into Institutional Responses to Child Sexual Abuse, 2016).
  HCA 40 .
 Hoch (1988) 165 CLR 292.
 Eg, Evidence Act 1977 (Qld) s 132A.
 As were the English courts in response to similar reforms in the Criminal Justice Act 2003 (UK) Part 1 Ch 11: Saleem  EWCA Crim 1923 ; see also Chopra  1 Cr App R 225 ; Manister  1 Cr App R 19 ; Edwards  EWCA Crim 3244, ; Mike Redmayne, Character in the Criminal Trial (2015), 145.
  2 SCR 908 ; A similar list appears in the New Zealand Evidence Act 2006 s 43(3). The New Zealand Act does not draw a distinction between the two types of reasoning. Stephen Odgers’ commentary on the Uniform Evidence Law provides a similar list under the ‘tendency’ heading, and notes that it has ‘some application’ to coincidence evidence: Stephen Odgers, Uniform Evidence Law (12th ed, 2016) [97.120], [98.120].
 Hughes (2017) 344 ALR 187 .
 Ibid; although the parties and the Court did draw on the work conducted by the ALRC in its evidence law reviews: David Hamer, ‘The Significant Probative Value of Tendency Evidence’ (2019) 42 Melbourne University Law Review 506, 535-540.
 Royal Commission, Criminal Justice Report (Commonwealth of Australia, 2017), Parts III-VI, 595.
 Mike Redmayne, Character in the Criminal Trial (2015), 30.
 Annie Cossins, ‘The Behaviour of Serial Child Sex Offenders: Implications for the Prosecution of Child Sex Offences in Joint Trials’ (2011) 35 Melbourne University Law Review 821, Table 1, 825–828.
 Peter M Robinson, ‘Prior convictions, conduct and disposition: a scientific perspective’ (2016) 25 Griffith Law Review 197, 205.
 M Redmayne, Character in the Criminal Trial, (2015), pp 21-2, quoted at Royal Commission, Criminal Justice Report (Commonwealth of Australia, 2017), Parts III-VI, 606.
 David Hamer, Submission to the Royal Commission into Institutional Responses to Child Sexual Abuse, Consultation paper: Criminal justice, 2016, p 6; quoted by Royal Commission, Criminal Justice Report (Commonwealth of Australia, 2017), Parts III-VI, p 606.
 Stephen Smallbone and Richard Wortley, ‘Child Sexual Abuse in Queensland: Offender Characteristics and Modus Operandi’, Trends & Issues in Crime and Criminal Justice No 193 (Australian Institute of Criminology, 2001) 6.
 Australian Bureau of Statistics, Population by Age and Sex, Queensland, 2000, Cat No 3235.3.
 Stephen Smallbone and Richard Wortley, ‘Child Sexual Abuse in Queensland: Offender Characteristics and Modus Operandi’, Trends & Issues in Crime and Criminal Justice No 193 (Australian Institute of Criminology, 2001), 13.
 Ibid 18.
 Hughes (2017) 344 ALR 187 .
 Ibid. Other courts have also made dubious claims about child sexual assault not being unusual. See discussion in David Hamer, ‘Proof of Serial Child Sexual Abuse: Case-law Developments and Recidivism Data’ in Thomas Crofts and Arlie Loughnan (eds), Criminalisation and Criminal Responsibility in Australia , 242, 249-250.
 Hughes (2017) 344 ALR 187 .
 See also Annie Cossins, ‘The Future of Joint Trials of Sex Offences after Hughes: Resolving Judicial Fears and Jurisdictional Tensions with Evidence-Based Decision-Making’ (2018) 41(3) Melbourne University Law Review (advance) 12, 25.