Arriving at the truth is the aim of any adjudication. But that’s easier said than done. Decision makers (judges, arbitrators, statutory bodies such as the human rights tribunal or labour relations board) have to assess the credibility of witnesses when weighing accounts of disputed facts Employers also must assess credibility in cases where a conflict about “what happens” is present. For example, when investigating a complaint of employee misconduct.
How do decision makers decide who is telling the truth and who isn’t?
The Supreme Court of Canada in R. v. Gagnon 2006 SCC 17 (CanLII) put the challenge as follows:
Assessing credibility is not a science. It is very difficult for a trial judge to articulate with precision the complex intermingling of impressions that emerge after watching and listening to witnesses and attempting to reconcile the various versions of events.
A leading case is Faryna v. Chorny  2 D.L.R. 354 (B.C.C.A.) in which the Court of Appeal said:
If a trial Judge's finding of credibility is to depend on which person he thinks made the better appearance of sincerity in the witness box, we are left with a purely arbitrary finding and justice would then depend upon the best actors in the witness box. On reflection it becomes almost axiomatic that the appearance of telling the truth is but one of the elements that enter into the credibility of the evidence of a witness. Opportunities for knowledge, powers of observation, judgment and memory, ability to describe clearly what he has seen and heard, as well as other factors, combine to produce what is called credibility ... A witness by his manner may create a very unfavourable impression of his truthfulness upon the trial Judge, and yet the surrounding circumstances in the case may point decisively to the conclusion that he is actually telling the truth. I am not referring to the comparatively infrequent cases in which a witness is caught in a clumsy lie.
The credibility of interested witnesses, particularly in cases of conflict of evidence, cannot be gauged solely by the test of whether the personal demeanour of the particular witness carried the conviction of the truth. The test must reasonably subject his story to an examination of its consistency with the probabilities that surround the currently existing conditions. In short, the real test of the truth of the story of a witness in such a case must be its harmony with the preponderance of the probabilities which a practical and informed person would readily recognize as reasonable in that place and in those conditions. [Emphasis added]
Demeanour is a factor, but it can be a misleading factor. As an arbitrator stated in Quality Meat Packers Limited (2013) CanLII 15073 (ON LA):
…. although I have considered the demeanour of the witnesses when giving their evidence and the firmness of their professed recollections, I have placed greater emphasis on the clarity, consistency, and overall plausibility of every witness' testimony when compared to the testimony of other witnesses and tested by cross-examination; the ability of all witnesses to resist the influence of self-interest or self-justification when framing their answers; the consistency of witness testimony with the documentary material available to test recollection; whether the assertions of the witnesses were consistent with, or corroborated by, other objective evidence; and, what seems to me to be most likely in all the circumstances established by the reliable evidence
Assessing credibility is often very difficult. Obviously, it is non-scientific exercise (as the Supreme Court of Canada Gagnon).
The best we can do is ask how objectively speaking the witnesses story hangs together with “the preponderance of the probabilities”? Assessing credibility is a very human endeavour and like all human endeavours is uncertain and imperfect.
That said, appellate courts have shown great deference to trial judges findings of credibility. In R. v. W. (R.),  2 S.C.R. 122, McLachlin J. (as she then was) explained why courts of appeal must show particular deference to trial courts on issues of credibility. At the same time, however, she noted:
…. it remains open to an appellate court to overturn a verdict based on findings of credibility where, after considering all the evidence and having due regard to the advantages afforded to the trial judge, it concludes that the verdict is unreasonable.
The Supreme Court of Canada took up the point in H.L. v. Canada (Attorney General)  1 S.C.R. 401:
But as a matter of principle, it seems to me that unreasonable findings of fact -- relating to credibility, to primary or inferred "evidential" facts, or to facts in issue -- are reviewable on appeal because they are "palpably" or "clearly" wrong. The same is true of findings that are unsupported by the evidence. I need hardly repeat, however, that appellate intervention will only be warranted where the court can explain why or in what respect the impugned finding is unreasonable or unsupported by the evidence. And the reviewing court must of course be persuaded that the impugned factual finding is likely to have affected the result.
In Lensen v. Lensen,  2 SCR 672 Chief Justice Dickson stated:
It is a well-established principle that findings of fact made at trial based on the credibility of witnesses are not to be reversed on appeal unless it can be established that the trial judge made some "palpable and overriding error which affected his assessment of the facts”
In conclusion, the Supreme Court of Canada in H.L. v. Canada held:
With respect, I do not find persuasive any of the arguments advanced in support of the contention that the rules governing appellate intervention in Saskatchewan differ from those set out in Housen. On the contrary, I am satisfied for the reasons given that the standard of review for inferences of fact, in Saskatchewan as elsewhere in Canada, is that of palpable and overriding error and its functional equivalents, including "clearly wrong", "unreasonable" and "not reasonably supported by the evidence”.
But, assessing credibility is difficult. I read an article by James Fallows entitled The Brian Williams Story as Emblem of the Chickenhawk Era in The Atlantic about Mr. Williams' and his recent revelation about his experience in Iraq. Mr. Fallows makes the following comment:
…. narrative and recollection are strange. I think I clearly recall vivid or traumatic episodes in my life, starting with the time a pickup truck rammed the car in which I was riding with my mom as a pre-schooler in Jackson, Mississippi. I believe I'm sure that I was sitting in the front seat, in that era before seat belts or child safety-seats, and just missed hitting the windshield, being stopped by the padded dash. But maybe, this many years later, I'm fooling myself. There is no one else around who was there. Three or four times in the past 20 years, I've been in uncomfortable situations while flying an airplane. I think I could recount those episodes in second-by-second slo-mo detail. But I can't be absolutely sure.
Misremembering. This is a word that has been much in the news of late. When is someone lying and when is he or she simply not remembering something as it really happened? I recall reading an article in the Harvard Gazette several years ago that is still available online (Making sense of memory by Taylor Beck). Memory is a funny and complicated thing but these credibility principles are important in trying to get to the truth (whatever that is).
There’s some interesting studies and analysis (sociological, psychological and legal) that review some of the challenges including Serious Lies, Lying in Everyday Life, and The Many Faces of Lies and others that consider credibility including A Wipe of the Hands, A Lick of the Lips: The Validity of Demeanor Evidence in Assessing Witness Credibility, Demeanor Credibility and How do decision makers decide when witnesses are telling the truth and what can be done to improve their accuracy In making assessments of witness credibility?.