3 ms·
That's my understanding too, that despite being heavily data driven, the Clinton campaign (and most of the polling agencies as well since they were using the sa
by monocasa 8y ago
That's my understanding too, that despite being heavily data driven, the Clinton campaign (and most of the polling agencies as well since they were using the same techniques) didn't have anywhere near a correct model for a "likely voter". That biased the results of their total model heavily towards Clinton.
- pmyteh 8y agoIt's very hard to come up with reliable likely voter models, not least because there are significant changes in people who vote every time. The opinion polls in the 2017 British general election were out, for example, because in the event unusual numbers of unlikely voters came out to vote Labour. Things were obviously going to be different that time (Labour had elected a deeply unlikey leader and repositioned itself) but no-one had any solid evidence how turnout patterns would change, or to what degree. So yes, inaccurate likely voter modeling is a serious problem. But it's really hard if things are changing, politically.
- monocasa 8y agoOh absolutely, it's a nearly impossible task. You can't even trust current polls asking people if they're likely to vote.
- luke0016 8y agoIf we had paper trails, this claim could be even more confidently asserted.
- monocasa 8y agoWell, hopefully paper trails don't have any identifiable information connected to the votes or else you end up with vote buying and coercion attacks.
- luke0016 8y agoOf course not. I go in, I vote, the vote is recorded electronically, and a piece of paper that only indicates who was voted for is spit out. No personally identifiable information. It's not like this is hard. We were doing paper secret ballots for literally centuries before electronic voting machines came along.