Republican candidate 'body-slams' Guardian reporter in Montana

[Benton]

(05-26-2017, 05:10 PM)Belsnickel Wrote: A lot of people say this about polling, but there are pollsters that work to avoid the question bias and use models to compensate for low response rates. The issue with special elections is that they are tougher to model and so the MoE is often higher. Overall, polling has been fine. The polling wasn't wrong last year, the popular election results were within the MoE of most polls (and the statistical models of political scientists for that matter). The pundits were talking about the data wrong and the EC us hard to predict with models and polls.

I'd disagree with the bold.

Using Trump and swing states as an example, he outperformed in many of those states far above polling. I've seen his average margin in Utah at +10, but he ended up at +19. Ohio was +2, he was +9. It's less that overall polling was wrong (it was) and more that overall polling failed more noticeably in some aspects. It wasn't like a few of the pollsters missed here and there... the majority of pollsters using a variety of methods were in error. They couldn't account for nonresponse bias on the scale that it keeps growing as more people become join the ranks of uneducated and poor.