R. v. A.P., 2013 ONCA 344 is a useful reminder that a judge in a criminal matter is not there to determine what happened but rather to see if proof beyond reasonable doubt is established:
 When the trial judge set out on his analysis of the evidence, he correctly listed the four elements of the offence of sexual assault that must be proven by the Crown. The trial judge then expressed the view that "the sole issue in this case is the credibility of the complainant and the credibility of the defendant", thereby, in the appellant's submission, viewing the case as a credibility contest.
 The appellant nevertheless concedes that the trial judge identified the correct approach to be taken in such cases, which is that outlined in R. v. W.(D.),  1 S.C.R. 742, at p. 758:
First, if you believe the evidence of the accused, obviously you must acquit.
Second, if you do not believe the testimony of the accused but you are left in reasonable doubt by it, you must acquit.
Third, even if you are not left in doubt by the evidence of the accused, you must ask yourself whether, on the basis of the evidence which you do accept, you are convinced beyond a reasonable doubt by that evidence of the guilt of the accused.
The appellant argues, however, that the reasons show that the trial judge did not properly apply that approach to the evidence in this case. I agree. Correctly setting out the W.(D.) approach is not determinative of the correctness of a decision – the critical issue is whether the reasons reflect its correct application: see e.g. R. v. Wadforth, 2009 ONCA 716, 247 C.C.C. (3d) 466, at paras. 50-51.
 The trial judge's approach to this case as a credibility contest is evident in his reasons when he is dealing with the assessment of witness credibility and the need for the Crown to prove the elements of the offence beyond a reasonable doubt. The trial judge indicated that in a case such as this, "the truth will be found to be the version of events that is in harmony with what a practical and well-informed person would recognize as most probable in all of the existing circumstances" (emphasis added). The trial judge referred to the "version of events" told by the appellant and the complainant a further five times in his reasons. These statements are problematic for two reasons.
 First, it suggests that the point of the trial and the trial judge's task is to determine which of the two recounted versions of the event is true. This approach is an error. As the Supreme Court of Canada held in R. v. C.L.Y., 2008 SCC 2,  1 S.C.R. 5, at para. 8:
[The purpose of the W.(D.) analysis] was to ensure that triers of fact — judges or juries — understand that the verdict should not be based on a choice between the accused's and Crown's evidence, but on whether, based on the whole of the evidence, they are left with a reasonable doubt as to the accused's guilt. [Citations omitted.]
See also R. v. Avetysan, 2000 SCC 56,  2 S.C.R. 745, at para. 21. An approach that emphasizes having to decide whether to believe either the Crown's evidence or the defence's evidence neglects the third alternative that while the trier of fact may not believe the accused, they may still be left with a reasonable doubt as to the guilt of the accused on the whole of the evidence: see R. v. S.(W.D.),  3 S.C.R. 521, at p. 532.
 Second, having determined that the sole issue at trial was credibility, the trial judge approached this issue as one to be decided on a balance of probabilities. I acknowledge that individual facts need not be proven beyond a reasonable doubt at trial; only the elements of the offence must meet this higher standard: see R. v. Bouvier (1984), 11 C.C.C. (3d) 257 (Ont. C.A.), at p. 264, aff'd  2 S.C.R. 485. However, by setting up the whole case as a choice between two competing versions of events and stating that the version that is "most probable in all of the existing circumstances" will be selected as "true", the trial judge came dangerously close to deciding the ultimate issue at trial on a balance of probabilities. See R. v. Quidley, 2008 ONCA 501, 232 C.C.C. (3d) 255.
 That the trial judge viewed his role in this case as choosing between the appellant's and the complainant's competing versions of events is further confirmed in his method of assessing their credibility. The trial judge acknowledged, in an exchange with counsel during closing submissions, that the appellant had testified in a forthright and honest manner. Since the complainant had testified in an equally honest manner, however, the trial judge set aside the fact of the appellant's honest presentation as neutral:
[B]oth the complainant and the accused, to my mind, presented honestly and well. Both gave their evidence in a forthright manner. But it seems to me that the issue of credibility is not to be decided by demeanour, it's not to be decided by the delivery of evidence, it's not to be decided by manner. It's to be decided by the internal and external consistency of the evidence given, with the evidence as a whole … .
So I say that only because I would almost rather have you focus on those issues of internal and external consistency or inconsistency, because to my mind that's where the truth emerges.
 The trial judge assigned relevant demeanour evidence no weight because it did not help him pick between the competing versions of events told by the complainant and the appellant at trial. This is demonstrative of the fact that the trial judge viewed his task as having to choose between the complainant's and the appellant's evidence. The trial judge's either/or approach was in error because it neglected the third option, namely, that even though the trial judge did not believe the appellant, he was not convinced of the appellant's guilt on the whole of the evidence.