Friday, April 18, 2008

Straddling the point

By the slimmest of margins the Supreme Court of Canada has just avoided making an awful mistake.

Evidence can be admissible without it having to support only one side of a case. This elementary observation is plainly true. Many items of prosecution evidence may be consistent with both the guilt and the innocence of the accused, and only when the whole of its evidence is adduced may the prosecutor be able to say that this proves guilt beyond reasonable doubt.

It is often possible to express the probative value of an item of evidence as a probability, or a likelihood, of the proposition it is advanced to support. This can also be termed the weight of the evidence.

In Gibson v R [2008] SCC 16 (17 April 2008) the accused was charged with driving with excess alcohol in his system, and the issue was whether evidence that his real alcohol level could have been somewhere in a range extending from below, to above, the statutory limit, was admissible. The statutory scheme provided a presumption that the level was as indicated in a test result, but this was rebuttable by evidence that tended to show that the driver’s level was below the limit.

The sort of evidence the defence wanted to rely on was called “straddle evidence”, because it showed a range of possible levels that straddled the limit.

It should have been obvious that this evidence was no different from any other evidence that has a greater or lesser tendency to support the defence in its aim of raising a reasonable doubt. Plainly, where the evidence only has a slight tendency to support an inference of innocence, and is more a basis for an inference of guilt, it will weigh in favour of guilt. In such a case, it is likely to be insufficiently relevant to a defence proposition to be admissible as defence evidence (while being admissible – if the prosecution had obtained it, which of course here they hadn’t – for the prosecution).

In Gibson the defence expert indicated a range between 40 and 105 (the limit is 80). The majority held that this evidence was admissible: McLachlin CJ, LeBel and Fish JJ jointly, and Binnie and Deschamps JJ jointly. However, although admissible, these Judges differed on whether it was insufficient to raise a reasonable doubt. Binnie and Deschamps JJ held that it was sufficient to do so as the “prevailing direction” of the range was under the limit. But the other three Judges who agreed on the admissibility point held that the range of values and the extent to which the range exceeded the limit indicated that the evidence was insufficient to raise a reasonable doubt.

The awful mistake was made by the minority, Bastarache, Abella, Charron and Rothstein JJ (jointly, delivered by Charron J). They held that instead of being evidence in support of a reasonable doubt, straddle evidence was an attack on the presumption itself, and was therefore inadmissible. On this view, straddle evidence only tended to show that the accused fell into the range of people targeted by the legislation (para 33), and that it did not tend to show that the driver in question was not in this range. In order to raise a reasonable doubt, the driver could only adduce evidence of his actual alcohol consumption.

It is difficult to see why evidence that supports an inference of innocence (to the extend sufficient to raise a reasonable doubt) should be seen as an attack on the rebuttable presumption itself. The reasons are given in para 32 (and see also 19 – 20), and essentially come down to saying that Parliament must have taken into account the possible range of actual alcohol levels in drivers’ systems when it established the limit and the rebuttable presumption, and so the presumption could only be rebutted by evidence other than that based on metabolic rates of alcohol elimination in the population at large.

That, however, is to overlook the point that the evidence concerning the population at large must relate to the facts of the case: to the driver and the amount of his actual consumption (see McLachlin CJ, LeBel and Fish JJ at 58). Otherwise, the expert evidence has no weight. The minority’s mistake was to overlook the fact that it would be impossible for the defence to rebut the presumption by evidence of the driver’s consumption, without also adducing evidence of the level of alcohol that one would expect to find, and that could only be based on evidence from the population a large which would inevitably involve metabolism.

Imagine that the defence called evidence that satisfied the minority’s criterion: that the driver’s level fell within a range that was entirely below the limit. The prosecution might then seek to rebut this by calling evidence that his range straddled the limit. Why should straddling evidence be allowed to prove the driver was over, but not under, the limit? The alternative, that neither side should have recourse to straddling evidence, would be untenable because it would prevent the prosecution contesting the defence evidence.

But that’s not all. None of the judgments refer to the correct way in which the proposed evidence would be given. In accordance with Bayes’ Theorem, the expert should present the findings as a likelihood ratio: the probability of getting the results, given that the defendant was guilty, divided by the probability of getting the results, given that he was not guilty. The findings concerning straddling are only part of the data behind the result that the expert should report to the court. The minority’s restrictive approach would limit expert testimony to occasions where the testing indicated a very low posterior ratio of probability of guilt to probability of innocence, and this would be inconsistent with the presumption of innocence and the requirement that the defence only raise a reasonable doubt.

As an alternative to the rebuttable presumption, Canada may find more attractive the approach adopted in jurisdictions where there is a conclusive presumption. In para 76 LeBel J mentioned the issue of limitation on the presumption of innocence, but it is clear that such limitation would be demonstrably justified in a free and democratic society. That is the test in New Zealand under s 5 of the Bill of Rights, and we have a conclusive presumption in this context: s 77 Land Transport Act 1998; no one has as yet argued that this is unjustified.

And on a completely different matter, dicta in this case (para 50 – 51) illustrate a point that has been of some interest in another context: a requirement that the defence “show” something, in order to rebut a presumption, is not the same as a requirement that the defence “prove” that thing. The defence can “show” the thing (innocence) by raising a reasonable doubt, and this is not the same as “proving” it. See Hansen v R (blogged here 20 February 2007).

No comments: