You're making a lot of assumptions about those 100 women. We have no idea "how" these women were chosen. Given the strong opposition that civilians have to women in combat, it's realistic to believe that in the military, itself, it's much higher. These women could have possibly wanted to avoid the grief often thrust upon trailblazers (see Tuskeegee Airmen). Or perhaps these women simply did not want a combat role. 90%+ of military jobs are non-combat roles, so even if you do use the erroneous figure of "100," the numbers have more parity than opponents would want to give credit, with a 6/7% margin of error.
It's EPIC fail to imply that the 85 women who didnt participate, didnt because they couldn't do it.
You're speculating too much. The only assumptions I'm making about those 100 women is that they were chosen to participate. If they chose the women as a statistical test, I imagine they based it on the necessary sample parameters to get the answer they're looking for. But in any case taking 15 as the denominator only compares what they want to do and think they can do vs what they can actually do. That's meaningful in some contexts. But if its some kind of test to learn something about combat likelihood for women in the marines, probably 100 is the more telling denominator.
And I'm not saying that only 4% of female marines can pass the test. The best we could say is that 15% of those eligible tried; 4% passed. There's nothing wrong with 1775usmarine's math.