Iowa GOP Presidential Nomination Feb 1st - Do the polls mean that much?

The time is fast approaching for the first of the GOP primaries - with Iowa actually a caucus on the night of Feb 1st 2016.

With some questions around the validity of the polls and how well they will predict the result I thought I would take a look at the 2012 run up to the caucus in Iowa (which took place earlier on Jan 3rd)

Going into the 2012 Caucus the polls read (RCP averages @ Jan 2nd 2012):

Gingrich 27.4% , but on a downward trend from 35%
Romney 25.2%, pretty static (maybe a slight upward trend)
Paul 12.2% , on a slight up trend
Santorum 4.0%
Bachman 6.2%
Perry 6.4%

In the final month leading into the caucus, Gingrich had actually surged and then fallen back again, Romney had made a 4% move upwards, as had Paul, but overall there was not a great deal of movement in terms of positions.

What actually happened:

Santorum won with 24.56% of the vote
Romney 24.53%
Paul 21.43%
Gingrich 13.3%
Perry 10.33%
Bachman 4.98%

So, a huge surprise with a Santorum win, given he was polling at 4% (note in the late polls he was polling higher but this post is really to judge the current polls we are seeing and the level of accuracy they may hold).

Gingrich underperformed massively - it looks like a large part of his vote swung to Santorum - who had been working extremely hard in Iowa leading up to the election night. Perhaps some switched to Ron Paul also.

Romney achieved about waht he would have expected from the poll numbers

Paul did better than his numbers suggested he would, by almost 10%.

The above poll vs result suggests to me, in Iowa at least, the polls are not that reliable and Gingrich's result will no doubt be a warning to Trump and Cruz who currently have strong leads. Its probably also worth pointing out that Iowa was not representative of how the rest of the states voted.

The current Iowa polling numbers from RCP are:

Cruz 30.3%
Trump 27.5%
Rubio 12.0%
Carson 9.3%
Bush 4.8%
Paul 2.8%
Christie 2.3%
Huckabee 2.3%
Fiorina 2.3%
Kasich 1.5%
Santorum 0.5%

There are a lot more candidates this time around - will Santorum get another lift from those that voted for him last time round? Perhaps taking a few percent off one of the leading candidates? Maybe. The number in the field also highlights the potential GOP Splitter strategy may play a role here, limiting the delegate count for the 'winner'. I do not expect any candidate to win the state with 50%+% and therefore score a state win agains the 8 win threshold to be eligible for nomination at the convention.

Overall this does lend some credence to those ignoring the polls and suggests Rubio and Bush are not quite sunk yet, despite falling poll numbers.

Whilst Iowa does not seem to be a huge influence, it does offer the advantage of some early momentum, I guess (at least some good PR as Santorum got last time round) There are 30 delegates at stake here, 27 of which are district or at-large (all 27 will be proportionally allocated based on the state wide vote). 3 more super delegates will be selected by the RNC.

Interestingly because delegates are unbound, they can vote for who they like at the convention, and in 2012 Ron Paul actually won 22 of the 30 Iowa delegates, so I would expect the anticipated delegate count reported on the night to change again come convention time.

How could Santorum win ANYTHING?

"The Science is BOGUS... tell a plant that carbon dioxide is dangerous!"

His own words, he's quite a nutty one.


EDIT: Hit enter too quick, and Ted Cruz isn't even an American born citizen!

He was born in Canada!

a reply to: RomeByFire

It was a surprise, but he did in fact win - albeit after a recount (on the night Romney narrowly won)

originally posted by: RomeByFire
How could Santorum win ANYTHING?

Ted Cruz isn't even an American born citizen! He was born in Canada!

Missing the point much?

Let's review: The OP is talking about the reliability of polls as compared to actual results. OK? He shows that the polls were inaccurate. Imagine that! The relevant question here is why were they inaccurate? Was it because of last minute stumping that was successful? Did those who were polled lie to the pollsters? Was the methodology of the pollsters deeply flawed? And how does last time compare to this time? All these are legitimate questions to ask and I don't know the answer myself.

I do suspect the nominee will be one of the top three here. But anything could upset the apple cart, including external events. If nothing else, it;s a great show that lasts longer than TFA.

It sems as though Trump is pulling well clear of Cruz in the polls leading up to Iowa...
We'll soon see if the polls are a good barometer of the result...

I think that the polls this year are less reliable than they have been in the past, both in general and for the Republican nomination in particular.

First of all, I believe the effectiveness of polling in general is less effective because of the near-ubiquity of caller IDs and because of laws regarding calling people with cell phones. People are less likely to answer a call from an unknown number than they are to either ignore the call or intentionally let it go to voicemail or an answering machine. In addition, pollsters can only call cell phones if the numbers are being dialed manually. More and more people - particularly younger people - have only a cell phone but do not have a landline. Polls in which the pollsters manually dial the numbers take longer to perform, therefore almost always sample fewer people, and have higher margins of error. For this reason, automatic polls over-sample older people and try to compensate using mathematical polling models, some of which will be proven to be more accurate than others.

Second, I think the polls for the Republican primaries this year, particularly for candidates on the lower end, are made less accurate by the unusually large number of candidates still in the race. At the moment, the most recent poll on Real Clear Politics is a Fox News poll conducted from January 18 to 21 and published on January 24. Trump has 34%, Cruz 23%, and Rubio 12%. Everyone else, 8 candidates, have less than 7% each, and the poll has a margin of error of +/- 5%. Jeb Bush, for example, has 4% in this poll. The margin of error means he could have as much as 9% or as little as zero. If you look at the Republican polls for this year, you'll see that the numbers on the low end tend to fluctuate wildly for candidates with numbers at or near the margin of error.

In the case of the polls regarding the Republican race, I'm not convinced that the numbers of people sampled can accurately represent the electorate, given the large number of candidates still in the race. While the polls on the Democratic side have the same issues with over-sampling people with landlines, etc., there are only two "real" candidates in the race (sorry, Gov. O'Malley), and there isn't a low end of the polls to get mathematically skewed over.

After the first few couple primaries/caucuses, it will be interesting to see which polling models were most accurate and whether or not (and how) polling organizations with less accurate numbers change their modeling to compensate ahead of Super Tuesday and the other, later primaries.

I wonder who the anti-trump candidate will end up being.

iowa should sink kaisch and christie a couple points in new hampshire and eventually setting them up to get wiped out in south carolina if they don't call it quits before then.

should be down to trump, cruz, carson, paul, bush, rubio after new hampshire.

After super tuesday a 4 man race
trump, cruz. rubio/bush, paul

I can see this thing going brokered especially if cruz makes a run and then the RNC throws out the birther issue 2.0 either during the campaign or at the convention.
There will also be some messed up things going on at state delegate selections like last time, just in case it goes brokered the pawns will be in place.

a reply to: jellyrev

If someone is going to be a legitimate anti-Trump candidate, they'd better get their game together fast. Trump is currently leading in every poll I've seen for the first 4 elections (Iowa Caucus, New Hampshire and South Carolina Primaries, Nevada Caucus). If Trump wins Iowa, he goes into New Hampshire with momentum; if he wins both, he's going to develop an air of inevitability.

Due to the rather convoluted rules both sides have for choosing a nominee, I think a clear win on the Republican side is unlikely before the Super Tuesday elections on March 15, as most of the early Republican races award delegates proportionally. However, it will rapidly become difficult for 4th/5th/6th place candidates to raise the money they need to keep a valid political ground game going. Imagine being Jeb Bush and going to donors hat in hand after coming in 5th - just behind Ben Carson - in Iowa.

a reply to: PhloydPhan

I honestly don't see trump getting a higher vote percentage from winning iowa going into new hampshire. statistician nate silver predicts cruz winning iowa. Looks like he estimates trumps votting getting percentage 5-8 points less than current polls.

The anti-trump candidate will likely be rubio imo. when he places 3rd/4th in iowa a lot of money will be thrown his way.

carson is on his way out