The analysis studies pre-crack and post-crack sales, and specifically observed the dip in sales after the crack. The dip was larger, the closer to release the game was cracked. A theoretical day 1 crack caused a 20% drop in sales.
I'm also not sure what you mean by games that are cracked almost immediately are a better sample. You can't measure sales before and after the crack was released because you only have the latter. Sure, if we could somehow measure how the game would have sold in an alternate universe where it wasn't cracked that would be a more robust finding. But obviously that's not possible.
The study focused on denuvo protected games because those are essentially the only games that go for extended periods of time without being cracked. They're the only games that actually offer any insight into how games sell without a crack available.