I don't remember if I made this post already but it like, okay so suppose you've got a statistical trends about broad demographic groups. "Women are underrepresented in top neurosurgery programs", for instance.
Now let's say you're hiring for a neurosurgeon. Does this information help you?
No - you're not looking at "all men" and "all women", you never are. In this case, you're looking at the people who applied to this job.
When women are underrepresented within a speciality, female applicants have, on average, better CVs than their counterparts. (Another way of saying this is that men in a male-dominated speciality are more likely to apply to jobs they might be a little underqualified for.) And you've got other factors going into this too - can't say for certain in any particular case whether a disparity is primarily discrimination based or primarily inclination, but even if it's the latter, how inclined to be a neurosurgeon the average woman is tells you fuck-all about your actual interview candidates. If it's even a little bit the former, then the chances are that the women are actually a little better than their CVs would lead you to conclude.
That demographic information is much more likely to steer you wrong than it is to help you. And it is likely to steer you wrong - humans find it really really hard to grasp that a statement like that can simultaneously be true and provide them as an individual with no useful information.
This is true in other situations as well! Most demographic statistics about Men and Women are also not likely to provide useful information about, say, whether people you meet at a party will share your interests, because once you factor in that a) most of that science actually isn't great for various reasons and b) any party you go to is not a public survey sample, the predictive value of anything you read about Mars and Venus in a magazine is smaller than the lowest probability you can usefully casually think about.
Frankly most humans can only conceive of x different probabilities in casual thought: definitely not, probably not, coin flip, probably and definitely.
Maybe some rare people are not like this or have trained themselves not to be like this, but most people are.
Telling people that something isn't "coin flip" moves their thinking to at least "probably/probably not". People find it incredibly difficult to fathom that something can be technically true but not useful for them to make predictions off, and that certainly won't occur to them spontaneously. In this way, true information can actually make people more wrong, if "coin flip" is practically speaking, within the situations they encounter, a closer approximation of the truth than "probably/probably not".
This isn't to advocate for hiding true things necessarily, but it is a reminder that without a lot of care in how they are communicated even true things can effectively be misinformation.
