Getting judges to obey the law on eyewitness identification
A rigorous application of statutory requirements for identification warnings is required in Fukofuka v R  NZSC 77 (16 August 2013). The Court requires greater vigilance to prevent erroneous convictions that was apparent in some Court of Appeal decisions [28, footnotes 21-23], .
The governing statutory provision is s 126 of the Evidence Act 2006 (quoted at ). It replaces s 344D of the Crimes Act 1961 (quoted at ), and it expands the original requirement of a judicial warning that it should "include the reason for the warning" by requiring the judge to warn the jury "that a mistaken identification can result in a serious miscarriage of justice".
This, however, requires more than just a recitation of those statutory words. The legislation carries a "statutorily required level of scepticism as to identification evidence" . So,
" ... trial judges should ... explain that the reason for the warning was that miscarriages of justice resulting from mistaken identifications had been known to occur."
This reflects the observations in R v Turnbull  QB 244 (CA) at 228, referred to in Fukofuka at -.
Where a warning under s 126 is required, it "must" be given, and failure to comply will have adverse consequences .
A digression by me: criteria for determining appeals against conviction
At the time relevant to this appeal the criteria for determining appeals against conviction were set out in s 385(1) of the Crimes Act 1961, which was a traditional formulation of what one might call the proviso approach: if the appellate court found there had been a miscarriage of justice the appeal had to be allowed, provided that the appeal could be dismissed if the appellate court was satisfied that no substantial miscarriage of justice had actually occurred.
In Fukofuka the Court of Appeal had applied this incorrectly -. The correct method was to firstly ask whether the error could have affected the result of the trial. If it could have, it was a miscarriage of justice. Then, if it was, the second question arose: was the appellate court independently satisfied of the appellant's guilt? Only if it was, could the appeal against conviction be dismissed. This reasoning was required by R v Matenga  NZSC 18,  3 NZLR 145 (discussed here on 9 July 2009, and again here on 20 July 2009).
The separation of these questions means that under s 385 there could have been a miscarriage of justice even if the appellant was obviously guilty. You do not answer the first question by going directly to the second.
From 1 July 2013 we have had new criteria for determining appeals against conviction: s 232 of the Criminal Procedure Act 2011. This is not structured with a proviso, and, among other grounds, the existence of a miscarriage of justice (as defined in subsection (4)) of itself requires an appeal to be allowed. Included in the definition of miscarriage of justice is any error in the trial that has created a real risk that the outcome of the trial was affected. (As to what real risk means, see Sarrazin, discussed here on 22 November 2011).
The question therefore arises, how does the court assess this "real risk"? Can it go straight to ask itself whether it is independently satisfied of the appellant's guilt, and if it is, conclude that the error at trial did not give rise to a real risk that the outcome was affected? Or must it treat the analysis as requiring two steps: first, ask whether the error might have created a real risk that the outcome of the trial was affected, and, if it might have, second, ask whether the court is independently satisfied of the appellant's guilt, and conclude that if it is, then the error could not have affected the result of the trial?
The latter suggestion requires an unwarranted reading-in to the legislation of "might have", whereas s 232(4)(a) goes straight to whether the error "has" created the real risk. So the question becomes whether the appellate court can assess the existence of a "real risk" by independently (that is, aside from the error) asking whether it is satisfied of the appellant's guilt.
That is certainly a possible interpretation of subsection (4)(a). It is consistent with convictions being quashed where appellants might reasonably possibly be innocent, or where, whether guilty or not, they had trials that were unfair or were nullities. But one should remember (1) why the proviso was abolished, (2) how difficult it can be for appellate courts to assess the strength of evidence (see also my discussion here on 16 November 2012), (3) the traditional practice of recognising risk of effect on result without the appellate court itself attempting to reach a verdict, and (4) the requirement of the rule of law that trials must be conducted according to law and not be dictated by what are thought to be facts. Those considerations support appellate courts refraining from determining guilt. Even so, it must be recognised that immaterial errors do not matter, and that is why the legislation requires there to be a real risk that the error affected the outcome of the trial.
But back to identification warnings: some issues
There is a possibility that not all defendants will be adequately funded at trial (shush ma sarcasm). Not all will be able to call expert evidence on the risks that can accompany eyewitness identification. Jurors are not allowed to research that for themselves. Should judges have a responsibility to keep up with research on the accuracy of eyewitness identification, and to pass information about that to juries? Or should they repeat expert evidence that may be on the record of trials in other cases? It is all very well to require judges to tell juries that there have been occasions when innocent people have been convicted because eyewitness identification was accepted when it shouldn't have been, but most people would know that. Should the Law Commission, which has looked into (in 1999, and as background to what is now s 126) the reliability of eyewitness identification, be asked to offer a standard direction that could helpfully elaborate s 126? The Commission had said in that background material (at ):
"Credible identification evidence is of particular importance to the criminal justice system. Research indicates that incorrect identifications are a major factor in miscarriages of justice. In its draft Evidence Code, the Law Commission has drawn upon the research discussed in this paper to develop a regime that will help to exclude unreliable identification evidence. In proceedings where the case against the defendant depends wholly or substantially on identification evidence, the judge will continue to warn the jury of the special need for caution before convicting on the basis of such evidence. Such procedural safeguards, and the use of expert witnesses where their evidence can provide substantial help to the jury, should help to ensure that identification evidence is both reliable and properly evaluated."
The Supreme Court has bolstered the Law Commission's – what some sarcastic people may say is a - rather limp reading of Turnbull, by strengthening s 126(2)(a).