I mean the other way around. Sure, bully your users by default but let them opt out via a somewhat hidden setting. If they use the setting instead of installing the app you have your result.
It's still not quite what the experiment is designed to test. How many people who opt out when given the ability would stick around when not given the ability?
If they make opting out difficult (ie. You go to a page in settings that you have to directly link to because you’re “signed out” in the main app) then that wouldn’t skew the results that significantly, while still providing an escape hatch that isn’t “wait for this to be finished”.