IF you compare leavers with downvotes to those leavers without downvotes that may shed some light on their impact.
Remainers might also be informational.
How many people get downvotes and persist vs those that get them and leave?
The timing relative to dv's to leaving might shed some light, too.
The issue for me is that this may end up being quite a complex query. Either difficult for the database to process for me or difficult for me to create the right query in the first place. The queries that I wrote for user retention took upwards of an hour to run and I expect this would be more complex than those. But you are right that it should be possible to analyze it with what we have in @hivesql