Diagnosing the Carson surge

A recent Elon University Poll shows Dr. Ben Carson surging past Republican Donald Trump among North Carolina Republican primary voters. Do the numbers reflect a real change in fortunes or differences in how surveys are conducted.

Posted 4:48 p.m. Nov 9, 2015 — Updated 6:55 p.m. Jul 13, 2018

RALEIGH, N.C. — A plurality of Republicans who are registered to vote in North Carolina preferred Dr. Ben Carson to all other possible candidates, including Donald Trump, according to a recent survey by the Elon University Poll. The Elon poll found 31 percent said they would vote for Carson, while 19 percent said they would vote for Trump.

These results are interesting for three reasons.

One is that this is the first poll in North Carolina to show Carson has greater support than Trump. Is Carson really "surging," as many press reports describe?

Second, this result is another opportunity to point out that polling averages deserve more attention than the results of any single poll. Third, this survey is another example of why polling a long time before the actual election needs to be taken with a grain of salt.

To start, just one week earlier, Public Policy Polling reported Trump held a solid lead in the Tar Heel State. In fact, PPP said Trump was ahead of Carson, 31 percent to 23 percent, which is about the same margin as the Elon Poll said Carson led Trump.

One explanation for the conflicting results is that these two polls took place at different times, and Carson is gaining support, which is the story told by Elon's latest poll. I went to Pollster to examine the national trend.

Indeed, as the graph shows, Carson's support has been increasing over time, but the movement is not dramatically different in the last week. Interestingly, my students and I are currently polling the North Carolina State University community, and while the survey hasn't finished, Carson is easily leading the other GOP potential nominees among undergraduates.

If the Republican presidential primary were held today, whom would you support?

This is a NC State Pack Poll re: presidential preference conducted in November of 2015.

^{Source: N.C. State University Pack Poll}

I also wanted to see if the Elon Poll matched a larger trend within North Carolina. The problem with this approach is that few polls have been taken. Nevertheless, the Elon Poll is the only one to show anyone else but Trump with the lead since summer. Either the Elon Poll captured shifting sentiment that will be validated by future polling, or it is an outlier and we shouldn't pay too much attention to it.

This is a snapshot of polls in North Carolina showing the leaders among GOP presidential contenders.

One concern I have raised about how news media report on election polls is that their headlines are often inconsistent with the results of other polls. The headlines grab viewers' attention, and the story can promote news narratives about the candidates that are artificial and misleading. I raised this criticism in my last post about The New York Times declaring Carson was in the lead nationally. It led to multiple media outlets echoing this claim. Yet, the NYT poll was the only poll in a month to have found that. Worse, in the 12 polls taken since then and recorded at Pollster, only two of them also supported the notion Carson was ahead.

Perhaps more importantly, most people have not decided whom they will vote for. Yet, the way pollsters usually ask the question results in most people giving an answer. Election polls usually ask respondents to say whom they would vote for, "if the election were held today." Yet, the election isn't being held today. People have yet to really engage in the campaign, and most will probably change their minds.

Take a look at what happens when respondents are given the chance to say how definitive their candidate preferences are.

N.C. State students: How definitive is your choice for president in November 2015?

The Pack Poll asks students if they've settled on a candidate.

^{Source: N.C. State University Pack Poll}

In the current PackPoll, we asked students if they had definitely made up their minds, were leaning towards a candidate or were still trying to decide. Only about 20 percent said their minds were made up.

When opinions are so fluid, its wise to not rush to judgment about what will transpire months from now, especially based on a single poll unmatched by any other evidence.

Suggest A Correction