I'm not sure how that would help. If I opt-out of data collection, I don't think I'd be particularly pleased if I get randomly selected to be one of the users in this "random sample" and the stats get sent anyway.
And if I opt-in to data collection, why would it matter to me whether the stats I'm sending are a result of me being selected as part of a random sample or not? Might as well just _always_ send those stats; it doesn't matter to me.
There is no mention that once validated, that the RAPPOR-based metrics when fully deployed would be take from a random population. Only that the initial study of the system will be done to a random population.
FTA:
What we plan to do now is run an opt-out SHIELD study [6] to validate our implementation of RAPPOR. This study will collect the value for users’ home page (eTLD+1) for a randomly selected group of our release population We are hoping to launch this in mid-September."
Notably:
"this study will collect ... for a randomly selected group"
I added the second part about the random sample later to the comment when I moved the proposal already out of my short term memory. I hope they use the data from their initial study to test whether the opt-out group actually is different from the group they already get data from.
To my amateur ear, that actually sounds like a good compromise to lessen the blow somewhat more. You should suggest it to Mozilla :)