In other words, 20% of women end up changing their major. I'm guessing it was worded this way to trick those that skim read into thinking that 80% of women change their major.
But the "1/3 as many women go into the program" is a meaningful stat. The most meaningful stat is "how good are the graduates", and from what I've seen, the graduates of Hackbright have been quite good.
Actually, the problem is that some people use the phrase "all but" in its well established idiomatic sense which is conventionally used to modify a non-quantitative description (where it means "almost" -- e.g., "He all but went bankrupt) when it is modifying a quantitative description, where it has a very different meaning -- from the individual words -- as "all except".
Note that the former definition (but not the latter, which is just a fairly direct combination of "all" and "but") is in most good dictionaries, and has examples in print stretching back to the 16th century, so its hardly reasonable to say that people saying it don't mean "all but". Its very much a long-established part of the language.
Its confusing and should be avoided in quantitative contexts because of the way it conflicts with the normal use of the individual words, but likewise "all but" in the sense of "all except" should be avoided in preference to, e.g., "all except" in the same circumstances, because of the danger of confusion with the idiomatic sense of "all but".
> (where it means "almost" -- e.g., "He all but went bankrupt)
I see your point that because in a non-quantitative description, it effectively means "almost", people might confusedly use it to mean "almost" in quantitative descriptions, without thinking about what it means.
> Actually, the problem is ...
> In other words, 20% of women end up changing their major.
Or very nearly 80% [1]. Its generally just better to avoid "all but" in quantitative contexts; in qualitative contexts its clear, but in quantitative ones its clear-as-mud.
The corresponding stats for male majors would be 140 declare the major and 64 graduate, so they lose 55% of declared students. Also pretty high, but the difference is large enough to be worth looking at.
disclaimer: these numbers are from the internet, definitely not dealing with the same cohort at each step, and may be quite wrong. If anyone has better ones, do share.