I am going to try and describe this in a math/programming way.
Say we have two functions randM() and randF() the first returns a Male Person, the second returns a Female Person. Person has an Ability and a Preference. We also have a function stemJob(Person p) that takes a Person and returns a Job with an Income, Prestige, and Satisfaction.
Equality of outcome would mean that after 10^9 calls of each the this is true average(stemJob(randM()) == average(stemJob(randF()). This could be either because 1a) randM() and randF() produce on average the same person modulo Sex and stemJob() does not consider Sex or 1b) The difference between the average randM() and randF() is corrected for by stemJob() considering Sex.
The other option is equality of chance where stemJob(randM()) == stemJob(randF()) where randM() and randF() have the same Ability and Preference.
Notice that in equality of chance and in equality of outcome stemJob() may or may not be the same. If randM() and randF() produce on average the same person modulo Sex then stemJob() is the same for both equality of chance and equality of outcome. If that assumption does not hold then stemJob() must be different.
Thus, if I support equality of chance and believe avg(randM()) != avg(randF()) I would not support equality of outcome because that would require stemJob to consider Sex in a way that invalids equality of chance.
Hope this give you a different way to look at why some people are against equality of outcome in STEM.