Tuesday, March 8, 2022

Please Ignore Internal Polls

 by Christopher Cooper

As the primary season kicks into full gear, we can expect a barrage of internal polls purporting to show one candidate with a “commanding lead” in the upcoming primary. One such poll from Madison Cawthorn made the rounds yesterday. The press release reporting the results noted that Cawthorn “dominates” the NC-11 primary field and “holds an 80% GOP approval rating”(1).


My goal here is simple: to convince you to ignore internal polls. I’ll use Cawthorn’s recent internal poll as an example, but this is not a phenomenon limited to Madison Cawthorn or to the Republican Party. I’ve seen untrustworthy internal polls from Democrats, Republicans, and Unaffiliated candidates. The rare Green or Libertarian candidate who can afford internal polling is subject to the same critiques.


Primary election internal polling is unreliable and uninformative because of three factors: selection, lack of disclosure, and uncertainty.


Selection: When I teach campaigns and elections, I give the students a challenge: find a publicly released internal poll that looks bad for the candidate who released it and I’ll give you an automatic A. In 20 years of teaching, no student has successfully gotten the automatic A (lots have gotten non-automatic As, however. If you’re a student reading this, please don’t avoid my classes). Does that mean that no candidate has ever conducted a negative internal poll? Of course not. It means that only the positive ones are released, while the negative polls never see the light of day. Internal polls that are made public are released precisely because of their results. That alone should be enough reason for us to ignore internal polls.


Lack of Disclosure: Professional polling firms and university polling operations give lots of details about how they conducted a poll--who was in the population (the group of people whose opinions they are trying to learn about), who was in the sample (the people who the pollsters actually talk to), the mode (live caller, robo-poll, internet poll), the specific question wording and question ordering, details about the margin of error, and a host of other decisions that can help citizens understand how trustworthy a given poll is. These best practices are not secret—the preeminent organization in public opinion research (AAPOR) lists them here (2). 


The release that accompanied Madison Cawthorn’s recent internal poll included none of these 11 standards [3].  Instead, the reader is presented with the sample size, question wording on two questions (out of how many questions? The reader has no idea), and an assurance that the population is “NC-11 GOP voters.” We don’t know when the poll was conducted, whether it was the “new districts,” whether it was by phone or web, whether or how the data were weighted, how they define “GOP voters” or much else. We do know that the pollster misspelled two candidate’s names (Wendy Nevarez, not Nevarro) and Michele (not Michelle) Woodhouse. The poll also excludes one recent candidate (Kristie Sluder), leading me to believe that it was conducted before Sluder entered the race.


Uncertainty Inherent in Primaries: General election internal polls should be taken with a grain of salt, but primary internals are about as reliable as a bag of magic beans. All election polling is based on the assumption that the pollster can predict who will show up to vote (or at least the types of people who will show up to vote). Primary electorates are notoriously difficult to predict—particularly in North Carolina in 2022. In addition to standard uncertainty, Unaffiliated voters (soon to be the largest group in NC politics), can choose which primary they will vote in. Will left-leaning Unaffiliated voters in NC-11 dip into the Republican primary to vote against Cawthorn? Your guess is as good as mine. And your guess is also as good as the guess of Cawthorn’s pollster.


This little blog entry is not going to stop candidates from selectively releasing internal polls. It won’t make them release full information about how their poll was conducted. Their goal is advocacy, not information. My goal is more modest: whether you’re reading this from Mar-a-Lago or a West Asheville coffee shop, please ignore internal polls. Regardless of whether they favor or challenge your preferred candidate, they give you no new information. If you still trust them, well…give me a call because I have some magic beans I’m willing to sell for a song. 



Chris Cooper is the Madison Distinguished Professor of Political Science and Public Affairs at Western Carolina University. He tweets at @chriscooperwcu


[1] Sorry—no link. My goal is to convince you to ignore these things; I’m certainly not going to send you to the very thing I want you to avoid.

[2] We’re fortunate to have a bunch of good polling organizations in NC. I listed some of them here:

[3] Friendly reminder: Cawthorn’s poll is simply a convenient and recent example. Many of the same critiques can be levied at recent polls I’ve seen from other Democratic and Republican candidates in NC-11 and elsewhere.