>That means the most talented and impressive young man who applied and didn’t get in is objectively more talented and impressive than the least talented and impressive (though still extremely talented and impressive) woman who applied and got in.
Unless the pool of female applicants was way more qualified than the male pool (i.e. by several standard deviations) this is true as a simple matter of statistics.
It seems quite likely that this is the case. We know a bunch more men than women are applying, and a reasonable prior is that an average woman is about as good a candidate as an average man. Therefore there seem to be a bunch of women who aren't applying who would have applied if they were equivalently qualified males.
Suppose there are 100 female applicants and 200 male applicants, and there is an even distribution of GPAs and ACT scores (or whatever criteria is used) in both the male and female pools. If Caltech takes the top 10 females and the top 10 males, the bottom 5 females of the 10 selected will be less qualified than all 10 of the males that were selected.
We would need more data from Caltech about total applicants and their qualifications etc. to say for certain, but barring some significant aberrations or selection effects what the OP said is going to be true.
You made my point. A lot of assumptions, no data. The female pool could very well skew higher in their stats. My assumption is that the pools aren’t any different as you are randomly selecting from a general population of people above a threshold level of qualification. Given that you can easily take in more females that have the same qualifications as the males. Given how stuff like the act is scored there will be a lot of people with the same score at every score level.
Prove it.