Selzer/DMR/Bloomberg FINAL IOWA POLL: Trump +5, Clinton +3

[Adam Griffin]

As a reminder, DMR's final pre-Iowa poll in 2012:

Romney 24
Paul 22
Santorum 15
Gingirch 12
Perry 11
Bachmann 7
Huntsman 2

"Gold standard"

They weren't far off. They just overestimated Bachmann/Huntsman and underestimated Frothy.

Their numbers were - statistically-speaking - pretty much dead on for everybody but Santorum. The difference between (Romney minus Santorum) in the poll and (Santorum minus Romney) in the final result = the undecided number.

[Adam Griffin]

Bodes well for Hillary, especially since it sounds like she has a built-in delegate advantage.

Yes, I think barring some major screw-up by Selzer or a huge event at the last minute, Clinton wins IA by a comparable number in voters and a slightly larger amount in delegates. Sanders is hurt by the fact that the caucus system penalizes his demographics - who are clustered on college campuses and in urban areas in general - while Clinton benefits from a larger than proportionate share of delegates being allocated to rural counties and caucus sites. Sanders would need to win by 3-5 points in order to eek out a plurality/majority of delegates in all likelihood.

[Adam Griffin]

Sanders is hurt by the fact that the caucus system penalizes his demographics - who are clustered on college campuses and in urban areas in general - while Clinton benefits from a larger than proportionate share of delegates being allocated to rural counties and caucus sites.

Is it true that, college towns aside, Sanders's support is more concentrated in urban areas than Clinton's support is?  I haven't seen evidence that it is.  See the discussion in this thread:

https://uselectionatlas.org/FORUM/index.php?topic=227529.0

Looking at the initial numbers in the OP, I'm not convinced what I said isn't true. Relative to population distribution, Sanders will have a larger than average share of his voters in actual urban areas, based largely on the fact that the generational gap appears to be the largest determining factor in the poll (older people tend to stay in rural areas; younger people leave them). Sanders may be doing better in the west than he is in the east, but there are fewer voters overall in the west than in the east.

I'd really need to see what specifically are their defining boundaries for each region, as there are probably suburban and rural areas that get counted as part of "Eastern Cities" and so forth, naturally inflating Clinton's dominance there when compared to my hypothesis. If the 64% of Iowa that is classified as "urban" was actually urban in a real sense, then I'd expect more than 64% of his vote to come from there. Based on that 64% urban definition, however, it's possible that less than 64% of his vote comes from "urban areas". However, in terms of what I'd consider to be urban, I imagine a larger percentage of his vote will come from those places than they comprise as a share of the state's population.

[Adam Griffin]

Sanders is hurt by the fact that the caucus system penalizes his demographics - who are clustered on college campuses and in urban areas in general - while Clinton benefits from a larger than proportionate share of delegates being allocated to rural counties and caucus sites.

Is it true that, college towns aside, Sanders's support is more concentrated in urban areas than Clinton's support is?  I haven't seen evidence that it is.  See the discussion in this thread:

https://uselectionatlas.org/FORUM/index.php?topic=227529.0

Looking at the initial numbers in the OP, I'm not convinced what I said isn't true. Relative to population distribution, Sanders will have a larger than average share of his voters in actual urban areas, based largely on the fact that the generational gap appears to be the largest determining factor in the poll (older people tend to stay in rural areas; younger people leave them). Sanders may be doing better in the west than he is in the east, but there are fewer voters overall in the west than in the east.

I'd really need to see what specifically are their defining boundaries for each region, as there are probably suburban and rural areas that get counted as part of "Eastern Cities" and so forth, naturally inflating Clinton's dominance there when compared to my hypothesis. If the 64% of Iowa that is classified as "urban" was actually urban in a real sense, then I'd expect more than 64% of his vote to come from there. Based on that 64% urban definition, however, it's possible that less than 64% of his vote comes from "urban areas". However, in terms of what I'd consider to be urban, I imagine a larger percentage of his vote will come from those places than they comprise as a share of the state's population.

In other words, I imagine that if the definition of "urban" is the area in gray (or even larger, god forbid), then my original statement wouldn't hold true. If the definition of "urban" is along the lines of what I would consider it to be (the areas in black), then I don't see how a disproportionate share of his vote doesn't come from inside of it.

[Adam Griffin]

40% first time caucusers seems kinda high for the caucuses considering the registration numbers are slit lot lower now than they were in 2012.

What are you talking about?

The number of registered republicans is 17000 lower than in February 2012, post caucus. 612k now vs 629k 2012.

People can flip their registration at the caucus. In reality, considering how many apathetic independent TRUMP supporters are likely to show up and do just that...it's not a terrible position number-wise to be just 17k off of 2012.

[Adam Griffin]

http://www.bloomberg.com/politics/articles/2016-01-30/bloomberg-politics-des-moines-register-iowa-poll-democrats

Only one in three likely Democratic voters in the survey are first-time caucus-goers, who break decidedly toward Sanders. That compares with 60 percent in the final pre-caucus survey of 2008

In addition, the survey finds Clinton’s support is deeper and sturdier than Sanders’ across many areas

“Most of the ways you look at it, she’s stronger than the three-point race would suggest,” said pollster J. Ann Selzer.

Sanders does not have as broad a reach as Obama did.
In the final pre-caucus survey of 2008, Obama led in many categories, with both definite and probable caucus-goers and decided as well as persuadable voters.

What I do find interesting in the poll, however, is that Hillary is within the margin of error of being toppled yet again in IA with only 34% of intended caucus-goers being first-timers, whereas it took 60% to dethrone her in 2008. This tells me that Clinton, in some ways, is even weaker than she was in 2008. However, it's worth noting that there is an eight-point difference between now (Clinton +3) and 2008 (Obama +5), and the difference between 34% and 60% isn't that large of one when you consider that it might be a 60/40 split in both years in terms of first-timers' preferences.

[Adam Griffin]

I'll ask again, since no one answered the first time: Have they released the full crosstabs for this poll?  Because all I can find is this:

http://media.bloomberg.com/bb/avfile/r1OvZ1NeDjnY

which includes results on a number of questions, but doesn't tell you how each demographic group is breaking.

The one I have seen was Hillary with a 12% lead among women or something? But... that doesn't seem like enough.

Considering Clinton was losing to Sanders among women under 45 by something like 15 points and was ahead with women over 45 by like 20 points (I saw these numbers mentioned in the live feed), that doesn't sound terribly far off.

[Adam Griffin]

http://www.bloomberg.com/politics/articles/2016-01-30/bloomberg-politics-des-moines-register-iowa-poll-democrats

Only one in three likely Democratic voters in the survey are first-time caucus-goers, who break decidedly toward Sanders. That compares with 60 percent in the final pre-caucus survey of 2008

In addition, the survey finds Clinton’s support is deeper and sturdier than Sanders’ across many areas

“Most of the ways you look at it, she’s stronger than the three-point race would suggest,” said pollster J. Ann Selzer.

Sanders does not have as broad a reach as Obama did.
In the final pre-caucus survey of 2008, Obama led in many categories, with both definite and probable caucus-goers and decided as well as persuadable voters.

What I do find interesting in the poll, however, is that Hillary is within the margin of error of being toppled yet again in IA with only 34% of intended caucus-goers being first-timers, whereas it took 60% to dethrone her in 2008. This tells me that Clinton, in some ways, is even weaker than she was in 2008.

It didn't necessarily take 60% to "dethrone" her. She got beat by 9 points. You're better with data than I am: what would the results have been if 2008 was 34%? Or 2016 being 60%?

Well, in 2008, she lost to Obama by 5 points as best I can tell. Today, she's up by 3. It's totally possible that if 60% of voters were first-timers in this caucus, that she'd be exactly where she was.

If we assume that it's 59/39 for Sanders among first-timers (34%), then it'd be about 57/40 Clinton for the rest (66%). This would give us a 51/48 Clinton result (Clinton +3, as in the poll).

If I've done it correctly, then a 60% first-timer crowd/40% repeat caucus-goers group would be:

52% Sanders, 45% Clinton

Which means she would lose by 7 (as opposed to losing by 5 in 2008). So, not very significant movement in one direction or another, but one could make the argument that she is possibly a bit weaker than in 2008, or at the very least, just as weak, when it comes to this one metric.

[Adam Griffin]

^^^ It's also worth noting that I edited my original post in between the time you quoted it and the time you posted it, explaining that the difference likely wouldn't be as monumental as I initially had thought.

[Adam Griffin]

Here are some data-points from the Democratic side:

18-35: 63-27 Sanders
65+: 65-27 Clinton

Those earning >$100k: 57-28 Clinton

"The system works reasonably well for those who work hard to get ahead": 60-29 Clinton
"The system is rigged against all but the very rich and powerful": 50-39 Sanders

First-time caucusgoers: 53-34 Sanders

Men: Sanders +5
Women: Clinton +10

No religious affiliation: 67-27 Sanders

Independents: 55-30 Sanders
Liberals: 51-41 Sanders

[Adam Griffin]

Well, in 2008, she lost to Obama by 5 points as best I can tell.

Um, she lost by 8 (37,6 to 29,5).

Ugh, why does IA have two caucuses and a primary? I was going off of the numbers Leip has listed and I'm assuming his are correct in some context. Are those numbers Leip has for the county delegates or what?

[Adam Griffin]

^^^ Interesting. There are two schools of thought in regards to which is more accurate, but I tend to believe that voter file sampling is the superior method in all but the most unorthodox and/or high-turnout scenarios (like 2008):