“But the polls … the polls have Trump well ahead!” you say.
Of course, with primaries and caucuses, the national polls are meaningless: Only state polls count. Here is where the candidates currently stand in Iowa, according to Real Clear Politics:
It is quite true that the polls currently have Trump ahead. However, this average includes two polls that have Trump up by 11%, and both of these polls were taken a week ago (January 20 and 21), which is an eternity in politics. The most reliable poll of the bunch–Quinnipiac, which is only two days old–shows Trump up by only 2%.
The polls are showing considerable movement on a week-to-week basis, and since October Trump has spent most of his time in second place. Further, while Trump has shown a slight, overall upward trajectory since early September, both Ted Cruz and Marco Rubio have shown a consistent increase in support since October, Cruz spectacularly so.
Yet, when it comes to the Iowa caucuses, the polls are often wrong. In 2008, on the eve of the Iowa caucuses, this was the RCP average of Iowa polls, versus the actual tally:
Huckabee: 29.7%/34% +4.3
Romney: 26.7%/25% -1.7
McCain: 11.8%/13% +1.2
Thompson: 11.7%/13%% +1.3
Paul: 7.3%/10% +2.7
Guiliani: 6%/4% -2
Thus, RCP was off by an average of 2.2%. At least it got the winner right. This could not be said in 2012. On the eve of the Iowa caucuses in 2012, this was the RCP average of Iowa polls, versus the actual vote:
Romney: 22.8%/25% +2.2
Ron Paul: 21.5%/21% -.5
Santorum: 16.3%/25% (winner) +8.7
Gingrich: 13.7%/13% -.7
Perry: 11.5%/10% -1.5
Bachmann: 6.8%/5% -1.8
Hunstman: 2.3%/.6% -1.7
Thus, RCP was off by an average of 2.7%, and missed a late Santorum surge.
So, what happened?
There is the issue of potentially flawed polling methodology, which some of the polls showing Trump well ahead seem to exhibit.
However, apart from this, turnout is everything. Candidates who are able to get their supporters to the caucus are favored over those who can’t. Since only a small fraction of the people who claim they will be voting in the Iowa caucuses will probably actually show up, the caucuses favor those candidates who have camped out in Iowa, who have a strong ground game in Iowa, and who have the support of Iowa evangelical leaders. This was the secret to Huckabee’s and Santorum’s success in 2008 and 2012, respectively, and it does not describe Trump’s campaign in 2016. Indeed, the campaign it most strongly describes is Cruz’s.
Further, with the way the Iowa caucuses work, the voters’ second choice for a candidate becomes almost as important as their first (while GOP caucus goers generally cast secret ballots in a straw poll, traditionally many tend to shift votes at the last moment if it looks like their first choice is not getting traction). Right now, Cruz is the overwhelming second choice (23%), followed by Rubio (16%), Trump (11%), and Carson (8%). If we add the first choice numbers with the second choice numbers, we get:
This is certainly not what the vote totals will look like, but it is a good gauge of each candidate’s real support. Based upon this, it would not be surprising at all to see final results something along the lines of:
I’m not saying that this will happen. I am saying that this would not be out of line with previous outcomes. Indeed, FiveThirtyEight, which has been very good at predicting election outcomes in the past, currently predicts:
Rubio: 14.8 %
While the numbers are lower than those I guessed at, the spread is about the same, though it predicts a closer match between Cruz and Trump.
My own suspicion is that Carson will under-perform his current polling, and that one or two of the current bottom tier will see a one or two percent bounce based upon the debate on Thursday.
With the Iowa caucuses a little less than a week away, it is impossible to say what will happen. The race is simply too volatile. At the same time, the likelihood is quite high that Trump will suffer an upset in Iowa and Cruz will win. Given past history, it is also unlikely that Rubio can make up enough ground to come in second.
We will see on Monday.