If one group on average has been brought in with lower success indicators, then they will have lower performance metrics throughout their career - on average when taken as a group - unless of course you argue that there is no such things as objective success indicators, which if you do - then why have them at all?
Especially in very technical areas like the Navy has in abundance (nuke power, aviation, etc) where academic skills are almost as important as other intangibles for a successful JO tour, you will have one group that will fail initial training more, will qualify later, and will have trouble grasping concepts quicker than their peers in other groups selected with more rigorous selection criteria.
As a result, if evaluated against each other - the group let in with lower average success indicators will be ranked lower, advanced slower, and given less demanding (and therefor rewarded) jobs than the group let in with higher average success indicators. Individual results may vary - but on average this is the result.