Let's think about conditional probabilities here. Let's take the example of a man at a regular day of work. Suppose that 50% of the men in sales wear suits, but only 5% of the men in engineering wear suits, at work. And suppose that 20% of the male engineers wear jeans and a t-shirt, but 0% of salesmen do that. Then, taking a random male employee from either sales or engineering, we can say: (1) if he's wearing jeans and a t-shirt, that is a maximally strong signal that he's an engineer, while (2) if he's wearing a suit, that is a moderate-strength signal that he's in sales.
If there are outfits where you can say "X% of engineers wear this, while Y% < X% of salespeople wear this", then someone wearing that outfit provides evidence (weak or strong depending on the values of X and Y) that they are an engineer. Contrariwise, if there's some other outfit where X% of engineers < Y% of salespeople wear it, then that's evidence for the wearer being in sales. Furthermore, since all the X's and Y's must add to 100% when taken across all outfits, if there are any outfits where X < Y, then there must be some outfits where X > Y.
If there is some pattern of visually obvious signs that, say, 30% of engineers show and only 1% of non-engineers show, then it follows that 99% of non-engineers and 70% of engineers show every other pattern of signs. Which means that, if you see someone showing some other pattern of signs, and you don't know anything except that it's not the abovementioned pattern that 30% of engineers show, then your knowledge logically implies that it's somewhat less likely (1.4 to 1 odds ratio) that they're an engineer. (Maybe someone else knows more than you, and could say that this pattern is actually also a strong signal of "engineer"—let's say it's even stronger, that 2% of engineers are like that and 0% of everyone else is. But the above statement about your knowledge remains accurate. Also, that would imply that the set of all other patterns is expressed by 68% of engineers and 99% of non-engineers, making the average of all other patterns a slightly stronger signal of "non-engineer".) That is a relatively diffuse signal, of course. If someone's very good at recognizing engineer-specific traits, and can see them in, say, 80% of engineers, then the average signal value of "all not-obviously-engineer traits" would be a strong "not an engineer". Or if someone can only recognize "definitely an engineer" traits in 2% of engineers, then that's only a 1.02 to 1 odds ratio for someone who doesn't show those "definitely an engineer" traits.
In conclusion, statements of the form "I can look at some people and conclude that they're very likely an engineer" logically imply statements of the form "There are other people I could look at and be less confident they're an engineer". The quantities—how much less confident—depend on the details.
I'm curious: (1) Do you think all statements of the form "Engineers are more likely to exhibit visible trait X" are worth calling out? (2) If 'mabbo had stated his criteria, and they were, for example (I haven't seen the videos), "Innsprucker is wearing very informal clothes while the others are obviously dressed up", would you think that was worth calling out? (3) If the answer to 2 is no, then would it have been better to ask 'mabbo, "I'm curious what makes Innsprucker so obviously a rocket scientist", before assuming it was absence of the traits "younger, or women, or of color" and calling him out for it?