A look at how the final polls fared, with all the vote in and... roughly counted. I can think of plenty of ways to measure pollster error in a multi-candidate race, and it's possible other measures are subtly different, but here taking the rmse of the Sanders - rest of top 4.
Conversation
A clear pattern, I suppose: online & live interview with full universe doing well; live phone with a constrained universe of voters (like last two primaries, in case of neighborhood research).
3
5
42
As an aside, I do think poll aggregators could have excluded some of these polls who didn't call the full universe of eligible voters, and we'd be better off. It's not like they'd add a poll of "White Wisconsin voters" to the average, even if they're >85% of the electorate
1
3
42
This is the only poll I saw with the correct order. With Pete on top and Biden slipping. Maybe they are a good source.
Quote Tweet
Caucus Day Poll:
Buttigieg 19%
Sanders 17%
Warren/Biden 15%
Klobuchar 11%
Poll by David Binder Research, January 28-30 of 300 likely Iowa caucusgoers statewide using landlines, cellphones, and text-to-web.
#IACaucus
Show this thread
9
2
Iowa's caucus is of course weird to poll/predict, but it's disturbing how unidirectional the errors were: every Biden error was too high, everyone else's were too low. Seems a lot of models might have weighing issues. Perhaps N.H.'s results will give us more insight into that.
3
4
36
So the best pollsters were:
1. Selzer/Des Moines/CNN
2.Civiqs
3. Siena College/NYT
Average error of 2.4
The worst:
1. Neighborhood Research
2. Suffolk University
3. Morning Side College
Average error of 11!
Emerson College was pretty bad too with an average error of 7 points.
1
6
8
We just not going to factor in Sanders error? DBR and Suffolk were catastrophically bad when it came to polling him
4