> On what basis should men be treated by-default as if they are, "the type that won't take no for an answer?" Wouldn't it be parsimonious to try the "no" and see what happens, in that case?

It would be parsimonious, and on average, it would work out. The problem is that the cost of being mistaken is high enough that it's worth working to avoid the worst-case result rather than optimize for average-case results. Does that make sense?