That's not how statistics work, unless you assume there is a correlation between tab users salary and their willingness to respond to surveys. Which would also be an interesting find

I'm curious if you've actually worked with surveys before, because this is how statistics work. All self-reported data has a bias based on a user's willingness to reply, and considering that predisposition as a variable is required.

This is the exact reason that crime statistics are so crummy, even from anonymous surveys. Very few are compelled to admit to a crime that they haven't been convicted of (and some won't even to admit to that). Or why post-sale NPS scores have a negative bent - you're more likely to respond if you have a complaint.