Just want to hop in and also point out the vastly different costs of being wrong in each case.
On one hand, we have a supervisor having to take a diversity course, and an employee getting a written warning about their performance. On the other hand, a person is losing their income and health insurance. If the evidence equally supported both sides and we had to guess, the detrimental effect of incorrectly supporting one side is vastly more significant than incorrectly supporting the other.
And that assumes a hypothetical where the evidence doesn’t support either side, something I do not think is the case. I think the article supplies enough information to support Alm’s case.