A Few Deep Breaths
ByJOSH MARSHALL
SEPTEMBER 16, 2016, 4:05 PM EDT
... I try to always remain a poll realist and resist partisan wishful thinking ... I frequently shoot people down who have some explanation of why the latest bad poll isn't really bad because something in the numbers is wrong ... I've repeatedly dismissed efforts to look at poll internals and show how they can't be right because they don't include enough of this or that group. Dems did that in 2004, Republicans did it in 2008 and 2012 and until about a week ago in 2016.
... What people often don't fully realize is that these 'weights' of different groups aren't set in stone or formulas pollsters do or not press onto the data. The data pollsters get from the phone calls can shift the models. Opinions and motivation can change in ways that make the crosstabs and composition of the electorate change too in ways that would surprise us and not seem 'right' ... In any case, your wishful thinking is going to be a very bad guide to untangling that skein of data. Don't try to unskew polls. Watch the averages ...
Clinton has come off an awful few weeks. She took a bunch of days off the campaign trail for fundraisers. Then she had a media storm with "deplorables", then she collapses on camera, was diagnosed with pneumonia and then spent three or four days totally off camera and off the campaign trail. These events may hurt Clinton. They may damage her campaign fatally. But a more likely explanation of the rapid shift in the polls is that they sharply demoralized her supporters and shook free her least committed supporters. That shows up in likely voter models; it leads to differential poll response. This does not mean the polls are wrong. It means they are accurately measuring the effect of those events. But there is good reason to think that it may be ephemeral because it is more as much a shift in enthusiasm and engagement than opinion ...
Trump is not expanding the Republican coalition. In fact, he's not holding a lot of voters who voted for Romney. Clinton, at the moment, is failing to unite the Obama coalition. That's very bad news for Clinton. But there's a flip side of that equation. By definition, everyone who is supporting Johnson or Stein doesn't like Clinton or Trump. But in extremis, the most of these voters look more like Hillary voters than Trump voters. We know this anecdotally from interviews and also statistically from polls: Clinton's margin drops when poll respondents are given the option of all four candidates rather than just the two. What does this mean? It may mean we're looking at a catastrophic replay of 2000 in which Trump is able to win the election by solidifying the GOP base while the Democratic coalition splinters across three candidates and pulls Clinton just below Trump. (Yikes!). But here's the thing to remember. All our historical experience shows that third party candidates tend to fade toward the end of the campaign. This was true in 1968, 1980, 1996, 2000, etc ...
http://talkingpointsmemo.com/edblog/a-few-deep-breaths
= new reply since forum marked as read
