The Crimson Cyclone wrote:jerseyhoya wrote:pacino wrote:
really interesting read
In a real departure between public and private polling, they're talking in there about limiting people to being upweighted by 5x or 10x. I don't think we've had any polling where anyone would be weighed more than 2-3x. Targets are usually set before calling on age/race/gender/party, people get terminated past a certain point, and often you know these things about them before you call so you can target sample accordingly.
are campaign internal polls supposed to have better science? I keep seeing statements similar to "his/her internal polling must be showing a closer race in OH than the polls indicate"
in the media every now and then
Campaign polls are usually better than even the best public polls because they spend more money to know who they're calling and can do a lot more with the data. And so much of public polling isn't done well even taking into account their limitations.
Upshot has done good stuff looking at voter list surveys, their differences to normal public polling, and how even with very well done data gathering, there's a lot of art left on the decision maker side in addition to the science.
Introducing the NYT Upshot-Siena College Poll
The error the polling world rarely talks about
