Pollster.com

Articles and Analysis

 

Caucus & Entrance Poll Results Thread

Topics: 2008 , Exit Polls , The 2008 Race

Since news organizations are free to release entrance poll tabulations as soon as they have them, we will likely see results soon and from multiple sources. Marc Ambinder has some early headlines:

CBS: CLINTON, OBAMA BATTLING FOR FIRST
EDWARDS TRAILING;
NOEDGE TO ROMNEY OR HUCKABEE
FROM ENTRANCE POLLS
RICHARDSON SUPPORTERS SAID
BE MOVING IN LOCKSTEP TO OBAMA

On MSNBC, Keith Olberman just read essentially the same broad description. All sorts of live blogging is occurring. Please feel free to post links to anything interesting in the comments.

8:15 - (all times Eastern): Tim Russert a few moments ago on MSNBC: "It seems to be, based on our early data, with Clinton, Obama in neck-and-neck, a lot more women, a lot more independents and a lot more young people than voted in '04. How they divide up in those three critical areas determines whether Obama or Clinton wins the caucuses."

8:20 - More from Ambinder: "On Dem side, half are first-timers, according to entrance polls... half say they made up their minds early...Back to that Hiawatha, Iowa precinct: turnout DOUBLED from last cycle, to 333...from 129." Hiawatha is a suburb of Cedar Rapids.

8:32 - Context on "half are first-timers:" The percentage of first-time caucus goers reported on the 2004 entrance poll was 55%. However, between 2000 and 2004, turnout grew from 61,000 to 122,000. Most of the pre-caucus polls the reported it put the share of first-time caucus goers between 20% and 35%. But the last Des Moines Register poll put it at 60%.

8:35 - Via The Page: "AP reports large crowds could delay entire process in nearly 1,800 spots throughout the state."

8:40 - MSNBC just reported actual delegate counts based on 8% of precincts: Edwards 35%, Clinton 32%, Obama 30%. However, these early reports come mostly from smaller precincts which, as I understand it, are more likely to be rural. For what it's worth, the Register poll reported vote preference among rural voters as 30% Edwards, 25% Clinton, 24% Obama.

8:44 - Ambinder again: " OBAMA CAMPAIGN PREDICTING 200,000 TURNOUT"

8:45 - MSNBC now reporting Obama and Huckabee ahead in the entrance poll, and they are starting to report specific numbers from the entrance poll. Presumably the analysts are now comfortable with their weighting and estimates to report specific numbers.

8:55 - CNN just projected Huckabee the winner of the Republican caucus.

9:15 - Yet more Ambinder: "OBAMA LEADING IN ENTRANCE POLLS / YOUNG VOTER TURNOUT LARGE"

9:26 - NBC News projects Barack Obama the winner of the Iowa Caucuses.

10:05 - CNN has now posted entrance poll tabulations for the Democrats and Republicans in Iowa. MSNBC has comparable tables for Democrats and Republicans (remember, the underlying data is the same).

Composition of Democratic caucus-goers:

  • 57% first time caucus goers
  • 40% age 17-44 (was 32% in 2004)
  • 25% Independent/Republican/Other (was 20% in 2004)

 

Comments
Hudson:

Interesting that the media is reporting Edwards in 3rd place, when the Iowa Dem party website is indicates he is ahead of Clinton with more htan 50% of precincts reporting.

____________________

Mark Lindeman:

Difference between the polls and the delegate allocations? It's hard to follow all these counties at once....

____________________

It is interesting isn't it. I have a screenshot of them declaring winners literally 5 minutes after I had cast my own vote, and I'm in Des Moines (one of the most densely populated cities) in Iowa.

____________________

Jezebel:

Well, looks like Ann Selzer is genius.

____________________

Mark Lindeman:

From the entrance poll it looks like Clinton is up about 3 1/2 (give or take, I'm just going from a few tabs) over Edwards, but it isn't translating into delegates.

____________________

CalD:

There will be no living with John Zogby after this.

____________________

Chris:

The NY Times has an interactive graphic that apparently purports to show actual popular vote totals for the Democrats in each county. I had always heard that the popular vote totals would not be tabulated or reported. The numbers look a bit suspicious, though, for three reasons. First, they sometimes look suspiciously rounded; could the vote in Taylor County really be 100 to 175 to 175 to 50? Second, they add up to (by my calculations -- the graphic doesn't seem to aggregate the totals to reach any statewide figure) to 249,460 -- a larger number for total caucusers than I had seen reported. Third, the popular vote percentage for each candidate (again, by my calculation) comes out to:

Obama: 37.6
Edwards: 29.8
Clinton: 29.53
Others: 3.1

Is it possible that the popular vote percentages could so closely parallel the delegate percentages, in a system where there were so many reasons to expect distortion?

Moreover, does anyone know where these figures come from, or how reliable they are?

The link is: http://politics.nytimes.com/election-guide/2008/results/states/IA.html

____________________

Michael:

Polling upshot of this result: the DMR retains it's strong reputation (even though it didn't get the likely voter model quite right but did get some things right, including turnout, first-timers, and Obama's margin)...while ARG's reputation, already bad, is probably in tatters now.

____________________

wxdavid:

what about those ARG polls that shwoed Hillary WINNING Iowa by 5 to 12 points over the past 3 weeks ? Do they look like crap or what?

____________________

Paul:

Both Des Moines Register and Zogby look real good to me compared to results available this morning. Is there a suggestion as to what polling service in New Hampshire might rival the work that the Des Moines Register does? Is Zogby planning to track in New Hampshire leading up to Tuesday?

____________________

CalD:

Michael,

Don't be a hater. ARG did fine in New Hampshire last time around. (That's where their based.) In 2004, the percentages in their final poll for the five top finishers were all well within 1 MoE of the actual election results. Their national polls also tend to track others well, albeit with a probable house effect in favor of Democrats.

I'd say they've got a thing or two to learn about Iowa though. But this was their first foray into polling for the Iowa caucuses, which everyone seems to agree is a murky morass, fraught with pitfalls for the uninitiated.

____________________

Anonymous:

And again, if the NYT's popular vote numbers (or estimates) are right, you'd expect a strong rural bias to the delegate results. Just one example: in more densely populated Johnson County, where I live, there were 39.5 caucus-goers for every one delegate they elected. But in very rural Adams County, there were a mere 6 caucus-goers for each delegate elected -- effectively giving their votes almost seven times as much weight as mine.

Yet the statewide delegate percentages don't come out any different than the popular vote percentages, according to the Times. A mystery.

____________________

Stephen Hendricks:

I can't resolve the Times popular vote discrepancies, but it does bear noting that the "entrance polls" suggest that if the Democrats had only counted "first choices" as the GOP does, Clinton would have been a solid second place. My back of the envelope calculations:

Obama 34.9%
Clinton 27.0%
Edwards 23.4%

____________________

Mark Lindeman:

Stephen, I have 34.4 / 26.8 / 23.0, so we're on the same page.

It seems clear that those NYT numbers are based on some sort of back-derivation from the state delegate equivalents. I haven't finished exploding the code yet.

____________________

Chris:

I'd sure like to see some kind of accurate report, or at least reality-based estimate, of those popular vote figures. If it's true that there's a strong rural bias, and that Edwards and Clinton ran stronger in rural areas (both of which seem clear from the delegate numbers), then Obama may have easily cleared the 40% mark last night in the popular vote.

____________________

Mark Blumenthal:

Close. Actually, those numbers are the state delegate equivalents.

I emailed Joe Lenski, the network's exit pollster, and he sent a helpful clarification:

The Iowa State Democratic Party provides the state delegate equivalents results to the AP in decimals to two decimal places. You can see the results by county and precinct at their web site (iowademocrats.org and iowacaucusresults.com). However the AP computer system and our computer system do not handle decimal places for votes so the AP multiple all of the county numbers by 100. So the numbers in the NYT margin of victory map have an implied decimal place so 7,832 really means 78.32 state delegate equivalents etc.

Amazingly it looks like the Democratic attendance is over 200,000 so I can see how your readers could be confused but that is just a coincidence. These county delegate numbers will add up to 250,000 � which is the 2,500 state delegates times 100.

____________________

from washingtonpost:

dems (76%) Ind (20)
O 32 41
C 31 17
E 23 23

so doing the math not including republicans (4% pro obama):

Obama 36
clinton 26
Edwards 22

____________________

according to the washigton post

using 76% dems + 20% ind

O 32 41
C 31 17
E 23 23

this leaves out 4% republicans

Obama 36
Clinton 26
Edwards 22

____________________

craigfarmer:

according to the washigton post

using 76% dems + 20% ind

O 32 41
C 31 17
E 23 23

this leaves out 4% republicans

Obama 36
Clinton 26
Edwards 22

____________________

Mark Lindeman:

Mark, I figured out most of that, but the browser ate my correction. Oh well.

However, I haven't figured out how the state delegate equivalents are allocated among counties. It's possible to get precinct-level figures. My impression is that the county totals (from precinct level) are reweighted somehow to get the final SDEs.

Chris, a place to start is the entrance poll size-of-community numbers: Obama 40% in urban, 30% in suburban, 31% in rural. There the numbers are weighted 42/27/31 respectively (e.g., 42% of voters "urban") -- but it's not clear to me whether that is a best estimate of who turned out, or whether it's adjusted for delegate allocation. Anyway, on the one hand, if the first set of numbers is close to right, then it's unlikely that Obama was much higher than he appears no matter how big the urban turnout was. (I'm not counting on the numbers meaning what they seem to mean.) But by the same token, it means that he did damn well even in the rural areas.

Craig, my numbers are an average based on every table. Doesn't mean they're right.

____________________

Chris:

Mark and Mark -- Thanks for tracking all that down. I guess that's as much as we'll know. Pretty crazy that we have to guess at how people actually voted!

____________________

Chris:

P.S. I have set myself a little (very unscientific) project: I will weight the county-by-county delegate totals to eliminate the voters-per-delegate bias, then reaggregate them and see what I find. I'll report back . . .

____________________

Mark Lindeman:

Chris, partly we can blame the system where people may end up 'voting' for someone different than they intended. But it would be nice to know those initial intentions. If the "head counts" (even the totals) are reported to the state party, I didn't manage to find them.

Just in case anyone wonders: the precinct-level riddle goes away if one just ignores the precinct level (grin). Small counties tend to have disproportionately large county conventions, but that shouldn't affect state results. The state delegate equivalents are actually "fair" relative to Democratic vote shares in recent elections. (For instance, Polk County contributed 14% of Kerry's votes and got 14% of state delegate equivalents.) Of course that doesn't mean that the SDEs fairly represent who turned out for whom.

____________________

Chris:

Mark L -- Maybe that's right. I just realized that I can't do my experiment without having the turnout numbers from each county. My earlier calculation (of a 6:1 bias) was based on the assumption that the Times was reporting actual votes, not just extrapolating from delegate totals. I need some sleep, obviously. If I ever find county-by-county attendance figures, I'll run the experiment, just to see. But in the meantime I have to assume that you're right.

As someone else pointed out elsewhere, though, weighting by performance in the general election isn't the same as one-person, one-vote. If your precinct tilts heavily democratic, but has a relatively low rate of turnout at the caucus, your vote will count disproportionately.

____________________



Post a comment




Please be patient while your comment posts - sometimes it takes a minute or two. To check your comment, please wait 60 seconds and click your browser's refresh button. Note that comments with three or more hyperlinks will be held for approval.

MAP - US, AL, AK, AZ, AR, CA, CO, CT, DE, FL, GA, HI, ID, IL, IN, IA, KS, KY, LA, ME, MD, MA, MI, MN, MS, MO, MT, NE, NV, NH, NJ, NM, NY, NC, ND, OH, OK, OR, PA, RI, SC, SD, TN, TX, UT, VT, VA, WA, WV, WI, WY, PR