How do you gather and weigh evidence — and assess credibility?

An investigation stands or falls on how evidence is weighed. Two categories matter: documentary evidence (emails, texts, Teams/Slack messages, calendar entries, photos, swipe/CCTV records, HR files, contemporaneous notes) and testimonial evidence (what parties and witnesses say). Documentary evidence is generally the more durable, but rarely speaks for itself — a message that looks damning in isolation can read differently in a full thread, so context matters and authenticity should be checked.

Corroboration is how strong investigations are built. Very few cases turn on a single piece of evidence. Look for evidence that supports or contradicts each account: a contemporaneous text, a witness who heard a disclosure shortly after the event, a calendar entry that places someone in a room. The AWI Guiding Principle is to gather relevant evidence weighed for probative value against cost and disruption.

Hearsay is admissible but weighted, not banned. Workplace investigations are not courts; the strict rules of evidence do not apply. Second-hand information is not automatically excluded, but it cannot be tested the way direct evidence can, so findings should not rest on uncorroborated hearsay alone. There is a meaningful difference between bare hearsay and a contemporaneous disclosure (someone reporting an incident to a colleague immediately afterward, whose account of the timing and the person’s distress can itself be evidence). Assess where information came from, whether it can be tested, and how much weight it can bear.

Relevance vs. scope creep. Gather what is relevant to the allegations in your mandate — not everything that surfaces. When new issues emerge mid-investigation, the AWI principle is to notify the employer so they can decide whether to expand the mandate or open a separate file, rather than letting the investigation sprawl. Exceeding the mandate is itself a fairness error: in Shoan v. Canada (Attorney General), 2016 FC 1003, the Federal Court found an investigator breached procedural fairness by making findings beyond her mandate. Scope discipline belongs to the investigation planning and scope stage; see also the note on how investigations go wrong.

The credibility test — anchor to the case law. Where accounts conflict, credibility must be assessed against the probabilities, not the witness’s demeanour. The governing statement is Faryna v. Chorny, [1952] 2 D.L.R. 354 (BCCA), where O’Halloran J.A. held that “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.” Crucially, the court warned that credibility “cannot be gauged solely by the test of whether the personal demeanour of the particular witness carried conviction of the truth” — a witness may make an unfavourable impression yet be telling the truth, and a confident, polished witness may be “adept in the half-lie.”

The modern Canadian case that enumerates the working factors is Bradshaw v. Stenner, 2010 BCSC 1398 (per Dillon J., aff’d 2012 BCCA 296, leave to SCC refused). At para. 186 the court lists, as factors in assessing credibility, “the ability and opportunity to observe events, the firmness of his memory, the ability to resist the influence of interest to modify his recollection, whether the witness’ evidence harmonizes with independent evidence that has been accepted, whether the witness changes his testimony during direct and cross-examination, whether the witness’ testimony seems unreasonable, impossible, or unlikely, whether a witness has a motive to lie, and the demeanour of a witness generally” — expressly relying on Faryna. Restated for the investigator’s toolkit, the reliable factors are: internal consistency; external consistency (with documents and other witnesses); inherent plausibility; corroboration; motive to fabricate; prior inconsistent statements; and material omissions.

Demeanour-only findings are unreliable and judicially criticized. Body language is not a reliable lie detector. As the Queen’s University Industrial Relations Centre summarizes the research, “body language is not a reliable predictor of honesty/dishonesty and cannot be relied on to determine the credibility of a witness in a workplace investigation.” Canadian courts have repeatedly cautioned against over-reliance on demeanour, and the AWI advises extreme caution. Nervousness, flat affect, or poor eye contact can flow from stress, trauma, culture, or personality. Demeanour may be a minor cross-check; it must never be the load-bearing factor. Plausibility and “past record” are also weak factors easily distorted by bias — weight them lightly.

For an Ontario SMB, the discipline is to write down why you preferred one account: a credibility finding that says “I found the complainant more believable” without reasons is indefensible; one that says “the complainant’s account is corroborated by the timestamped texts and the respondent’s account is internally inconsistent on the key date” will hold up. The standard you are applying — balance of probabilities — is covered in the findings-and-report note. This is general information, not legal advice.