If you find the article and show me they tested actual players batspeed per and post steroid, I'll concede defeat.
As I said from the get go, I agree with premise that more muscle mass should equal more bat speed. My only point of contention is the 5% you posted, and the article you posted for support itself initially said 5%, but was then revised and only came up with 3.8%, and that final number came down to multiple assumptions, including estimates on how much kinetic energy was tranferred and how much presteroid weight is muscle mass. Those assumptions would have to be 100% accurate for 3.8% to be factually correct.
That's my point. It's a guess. Educated... but still a guess.