OK, a slightly stream-of-consciousness, post-Super Bowl Iowa polling thread.
For starters, here is our current polling average, with comparison to 1 week ago / 1 month ago.
projects.fivethirtyeight.com/polls/presiden…
For starters, here is our current polling average, with comparison to 1 week ago / 1 month ago.
projects.fivethirtyeight.com/polls/presiden…
Sanders (along with Klobuchar) has made the biggest gains over the past month. But his trajectory over the past week has been flat, especially if you look at comparisons among the same pollsters. Not necessarily anything to worry about and I'd probably take the over...
Still, there's a fine line between "still peaking as we head into the caucus", "peaked at the right time" & "peaked a little bit too soon" and any of the 3 are at least plausible. Especially if you consider house effects; live polls are more mixed for Bernie than the online ones.
Biden seems like the hardest candidate to pin down, with fairly wide variation from poll to poll. I really have no idea, and his purportedly mediocre ground game makes for a downside case.
At the same time, there's this perceptual perception that Biden is declining when he isn't. He's actually up in Iowa month over month, and although there's been a bit of erosion over the past week, it's margin-of-error-ish stuff. A number of polls still have him ahead.
Buttigieg has declined both week-over-week and month-over-month. Nonetheless, he's not *that* far behind and is probably a bit unlucky that no poll as shown him with the lead. The mechanics of the caucus potentially also favor him. He does well in 2nd choice numbers...
...He should have a good ground game. And he reportedly does well in rural areas, where you get more bang for your buck in terms of delegates. Purely as an Iowa play—our model is very skeptical re: his post Iowa path—he may be a bit overlooked.
Warren is actually UP a bit month-over-month and week over week. Sanders's surge has really worked against her, though, in that he should be comfortably >15% in most precincts, denying her a chance to pick up many Sanders votes upon realignment, for whom she's often a 2nd choice.
So while a "progressive dream" of Sanders-Warren finishing 1-2 in Iowa is certainly plausible, it's *more likely* that her fate is inversely correlated with Sanders. Her support also seems to be concentrated in upscale areas, which isn't helpful to her re: the delegate math.
On the other hand, she has invested a LOT in Iowa, and should be helped by her ground game, having good precinct captains, etc. I am curious if Warren forms an alliance with other candidates (Klobuchar or Buttigieg) to try to direct votes away from the frontrunners.
Klobuchar would be fairly interesting if she were at 13%, and not at all interesting if she were at 7%. She's wound up right in the middle, at 10%. She has made some gains over the past month but we can now rule out a Santorum-like trajectory with a huge last-minute poll surge.
Empirically, the error on Iowa polls is high. It's entirely possible Klob could sneak in to 3rd place. Higher than that? Well, our model gives her a ~1 in 40 chance of winning. But that would be a case where the polls had missed something, not "normal margin of error" stuff.
p.s. It's not unheard of for polls to come out on caucus day. We'll normally freeze our model in the morning on days with big elections, but with Iowa not caucusing until the PM, we'll keep it open until the afternoon. With that said, pollsters shouldn't make a habit of this.
