Author Topic: One Week Out: The Problem With Polling Iowa  (Read 945 times)

0 Members and 1 Guest are viewing this topic.

Offline TBBT

  • Member
  • ***
  • Posts: 23
One Week Out: The Problem With Polling Iowa
« on: January 25, 2016, 12:37:21 pm »

http://theresurgent.com/one-week-out-the-problem-with-polling-iowa/

One Week Out: The Problem With Polling Iowa
By Erick Erickson  |  January 25, 2016, 05:00am

There are a number of polls out that have Donald Trump surging in Iowa at the last minute. I do believe Trump is probably ahead, but I am aware of several of the campaigns’ internal polls and none of them have anything but Cruz and Trump clustered together very closely.

The problem with polling Iowa is that a caucus is not a primary. With a primary, voters go to a polling booth, click the name of the candidate who they support, and leave. With a caucus, often not even in the same location as where voters go to vote in primaries, there is an hour commitment, the voter must be in the room by seven o’clock in the evening, and they must show their support rather publicly.

People who show up for primaries may not know the caucus location, they may not realize they have to stay an hour, and they may think they can show up late. Likewise, we know historic turnout models for caucuses and many of the late polls with a huge Trump lead have turnout models that far exceed even the most historic high turnouts of caucuses past.

My friend Steve Deace, a Cruz supporter, has some very insightful data on caucuses and polling models.

    So now we have FOX as well as CNN producing polls this week that show 300,000 Iowans are voting in the Iowa Caucuses, and therefore Trump with a double-digit lead. Allow me to put those projected turnout numbers in perspective:

        That’s about a 200% voter increase from the highest Iowa Caucus turnout ever back in 2008.
        The most voters we’ve had in a primary (which always has higher turnout) in Iowa this century is only 230,000. And our last U.S. Senate primary had only roughly 150,000 voters in 2014.
        There are actually 11,000 fewer registered Republicans in Iowa this January than in January 2015.

Ann Selzer has a very good track record in Iowa, catching both the Huckabee and Santorum trend lines. She’s now caught the Cruz trend line. Pay attention if she releases another poll, but otherwise be cautious. Caucuses are more dependent on ground games than standard primaries. The prevailing consensus of reporters in Iowa is that Cruz, Carson, and Rubio have some of the very best ground game operations.

We have one week to go. We will see how all the data holds up in the face of Donald Trump’s unorthodox campaign. If traditional fundamentals hold, he will not come in first. But if his unorthodox campaign that defies tradition puts him in first, we may see a paradigm shift in electoral politics like we have never seen before.

The danger for Trump is that, with so much late breaking polling showing a renewed and large lead coupled with Trump touting all this polling, expectations are suddenly being set very high for him, which might make him not coming in first very problematic for him.


Offline TBBT

  • Member
  • ***
  • Posts: 23
Re: One Week Out: The Problem With Polling Iowa
« Reply #1 on: January 25, 2016, 01:02:07 pm »
Howdy... Refugee from Free Republic here.

Anyway...

Looking at the 2012 and 2008 polling:

http://www.realclearpolitics.com/epolls/2012/president/ia/iowa_republican_presidential_primary-1588.html
http://www.realclearpolitics.com/epolls/2008/president/ia/iowa_republican_caucus-207.html

In 2012, no poll accounted for Santorums performance. Santorum's polling high was 18 immediately preceding the vote. He received 24.6% when the votes were cast.

In 2008. Romney was in the 10 to 16 range right before the vote. Romney's actual number ended up being 25.2%. McCain actually did far worse that what was predicted
by the polls. Preceding the vote, McCain ranged 24 to 31. After the vote was tallied, McCain's actual result was 13%. Huckabee, presumably the evangelical favorite, also
exceeded his polling on vote night. His RCP average before the vote was 29.7. Huckabee's actual number ended up being  34.4%.

Santorum, Huckabbe, and even Romney, presumably would have had support from evangelicals. In each case they out performed their polling. This round the presumed
evangelical favorite would be Cruz.

On the other hand...

Conventional wisdom so far has not applied to Trump's campaign, and there is no telling if it will still apply in Iowa.

All in all, if you are a Cruz fan, you would certainly prefer not to be looking at things from this perspective at the moment. If the CNN/Fox polls are accurate - and I have my doubts
about that for various reasons - then the concern-o-meter has to be in the red. 
« Last Edit: January 25, 2016, 01:02:50 pm by TBBT »

Offline EC

  • Shanghaied Editor
  • Hero Member
  • *****
  • Posts: 23,804
  • Gender: Male
  • Cats rule. Dogs drool.
Re: One Week Out: The Problem With Polling Iowa
« Reply #2 on: January 25, 2016, 01:20:44 pm »
Welcome in!

Polls are, to my mind, somewhat misleading thanks to their inherent self selection. Regardless of the methodology used by pollsters, and the attempts to use representative samples, They can only record people who actually answer their questions, which tends at this stage to mean people who have been paying attention (most don't until about a week before the actual election) and who don't mind having dinner or Game of Thrones interrupted by some idiot questioning them!
It's one of the factors that threw of the UK election polls so badly last year. The other, of course, being that people will troll pollsters, every chance they get!
The universe doesn't hate you. Unless your name is Tsutomu Yamaguchi

Avatar courtesy of Oceander

I've got a website now: Smoke and Ink

Offline Longiron

  • Hero Member
  • *****
  • Posts: 2,343
Re: One Week Out: The Problem With Polling Iowa
« Reply #3 on: January 25, 2016, 01:26:21 pm »
http://theresurgent.com/one-week-out-the-problem-with-polling-iowa/

One Week Out: The Problem With Polling Iowa
By Erick Erickson  |  January 25, 2016, 05:00am

There are a number of polls out that have Donald Trump surging in Iowa at the last minute. I do believe Trump is probably ahead, but I am aware of several of the campaigns’ internal polls and none of them have anything but Cruz and Trump clustered together very closely.

The problem with polling Iowa is that a caucus is not a primary. With a primary, voters go to a polling booth, click the name of the candidate who they support, and leave. With a caucus, often not even in the same location as where voters go to vote in primaries, there is an hour commitment, the voter must be in the room by seven o’clock in the evening, and they must show their support rather publicly.

People who show up for primaries may not know the caucus location, they may not realize they have to stay an hour, and they may think they can show up late. Likewise, we know historic turnout models for caucuses and many of the late polls with a huge Trump lead have turnout models that far exceed even the most historic high turnouts of caucuses past.

My friend Steve Deace, a Cruz supporter, has some very insightful data on caucuses and polling models.

    So now we have FOX as well as CNN producing polls this week that show 300,000 Iowans are voting in the Iowa Caucuses, and therefore Trump with a double-digit lead. Allow me to put those projected turnout numbers in perspective:

        That’s about a 200% voter increase from the highest Iowa Caucus turnout ever back in 2008.
        The most voters we’ve had in a primary (which always has higher turnout) in Iowa this century is only 230,000. And our last U.S. Senate primary had only roughly 150,000 voters in 2014.
        There are actually 11,000 fewer registered Republicans in Iowa this January than in January 2015.

Ann Selzer has a very good track record in Iowa, catching both the Huckabee and Santorum trend lines. She’s now caught the Cruz trend line. Pay attention if she releases another poll, but otherwise be cautious. Caucuses are more dependent on ground games than standard primaries. The prevailing consensus of reporters in Iowa is that Cruz, Carson, and Rubio have some of the very best ground game operations.

We have one week to go. We will see how all the data holds up in the face of Donald Trump’s unorthodox campaign. If traditional fundamentals hold, he will not come in first. But if his unorthodox campaign that defies tradition puts him in first, we may see a paradigm shift in electoral politics like we have never seen before.

The danger for Trump is that, with so much late breaking polling showing a renewed and large lead coupled with Trump touting all this polling, expectations are suddenly being set very high for him, which might make him not coming in first very problematic for him.

Eric Erickson = Cruzbots what do u expect him to say, go Donald :chairbang:

Offline aligncare

  • Hero Member
  • *****
  • Posts: 25,916
  • Gender: Male
Re: One Week Out: The Problem With Polling Iowa
« Reply #4 on: January 25, 2016, 01:30:14 pm »

Whatever the final tally, republican Iowans haven't selected the eventual nominee in a long, long time. Based on past performance, whichever candidate wins Iowa is almost certain to lose the general. If Trump were smart he should deliberately throw Iowa. But, knowing Trump, it's win, win, win, baby. He wants to win every state.

Democrat Iowans on the other hand, selected Obama in the caucus and in the general. That's all the history I need to know about Iowa.