How I read N.C. political polls
Dueling surveys in the Berger–Page race show why I care less about toplines and more about incentives
Everybody loves a good political poll, but they’re generally misunderstood. Polls matter not because they’ll tell you with scientific precision who’s winning a race, but because they are the best tool out there for shaping campaign and donor behavior.
Money is the oxygen of modern politics, and nobody wants to pour six or seven figures into a sure loser. National groups, especially, don’t do charity.
We learned that the hard way on the Dan Forest campaign for governor. We’d raised a good bit of money, but we were up against a fundraising juggernaut. We needed the Republican Governors Association to make a significant investment in advertising to help level the playing field.
The calvary never came, and it was mostly because of polls. The summer before the election, public polling consistently had us 10–12 points down against incumbent Gov. Roy Cooper. The RGA looked at the board, did the math, and invested elsewhere. They put more money into flipping Montana and kept some dry powder for Virginia in 2021 — the race Glenn Youngkin ultimately won.
Of course, we ended up losing by only about four points. Who knows what $10 million from the RGA would have been able to do1. That makes this all a particularly good illustration of what I’m trying to say. Polls are often “wrong” but still powerful enough to change the trajectory of a campaign.
The two questions I ask when looking at a poll
Whenever I pick up a poll, I ask myself two basic questions:
Who did the poll?
How did they do the poll?
The first question is about incentives. Before looking at the numbers, you have to understand what the poll is trying to accomplish and what the track record is of this particular polling outfit.
Is this from a university or a media organization that’s trying to be relevant — and get clicks? Or is this a candidate releasing an internal poll? (Western Carolina’s Chris Cooper has a good piece on why you should ignore internal polls. I mostly agree, but think you can still learn something from them a lot of the time.)
The corollary is then, why did they release the poll? Candidates release internal polls only when they’re positive, and strategically timed around fundraising opportunities. Consultants may release a poll to drum up business, or a PAC might do so to raise money.
The second question is about methodology.
How many people did they survey, and how did they build the sample? Are we talking “likely voters,” “registered voters,” or just anyone who answered the phone? How did they decide who counts as likely? How were people reached — live calls, landline vs. cell, text-to-web? Were results weighted, and if so, how?
In the best case, a pollster shows you all of this, including question wording. In the real world, though, you sometimes get only toplines — and that’s when you should be especially cautious.
That’s a lot of questions, and you’re never going to get clean answers to all of them. That’s why polls are never gospel, but just a clue.
Looking at two polls in the Berger-Page race
With all that in mind, let’s look at two recent polls in the Republican primary between Senate president pro tem Phil Berger and Rockingham County Sheriff Sam Page — the most important political contest in North Carolina in 2026.
The first is from Opinion Diagnostics, fielded in mid-December. It showed Page leading 42% to 32%.
The second is from Rick Shaftan’s shop, Neighborhood Research, fielded in early January. His topline flips the script, showing Berger leading 39% to 35%.
In one sense, both of these polls can’t be “true” at the same time. But that’s now how I look at it. Both polls are giving you real information if you break them down.
The incentives
Start with the incentives. Opinion Diagnostics is affiliated with Page, so it’s pretty much an internal poll. Again, that’s not automatically sinister. That’s politics. But you do have to look at the incentives behind it.
A lot of times, candidates will release an internal poll to try to dry up the other side’s fundraising. That’s clearly not the case here. Berger is going to have a funding advantage no matter what.
The more realistic goal is to make it socially and strategically acceptable for donors to cross the most powerful politician in North Carolina and give Page the money he needs to be a real contender.
A poll showing Page up 10 points is a pretty effective way to do that. We don’t have any public fundraising data from Page since mid-year 2025, but my guess is he’s struggling to raise the money he needs to win. The poll is a data point that does two things: 1) Shows that Page’s campaign is legitimate; and 2) Gives donors a reason to give.
Shaftan’s incentives are a lot different. A poll in a high-profile race is often about building a résumé, gaining credibility, and maybe even bringing in a few new clients. The track record here is not great, though. Shaftan did a poll last October in New Jersey showing Democrat Mikie Sherrill and Republican Jack Ciattarelli tied, 44%-44% in the governor’s race. Sherrill ended up winning 57% to 42.5% a few weeks later.
However, Shaftan’s poll results aren’t just toplines. He also shared some open-ended and attitudinal material that is pure gold, telling you how voters are actually narrating this race to themselves.
The methodology
Both surveyed 300-400 people, which doesn’t feel like a lot but is actually pretty darn good for a state legislative race. These are notoriously difficult to poll, mostly due to low name ID, but both Berger and Page are very well-known in their district, which helps.
Shaftan’s live phone method is a plus, especially in a low-turnout primary electorate. Opinion Diagnostics’ mixed-mode approach can still be legitimate, but it puts more pressure on weighting assumptions and question structure.
The Shaftan poll doesn’t give a whole lot of details on the sample or the questions, and it also doesn’t say anything about weighting. It doesn’t even have a stated margin of error.
Opinion Diagnostics has a little bit more information on methodology, but not a ton. It does say there was weighting done “using an iterative proportional fitting process,” but we don’t really know what that means.
I’m particularly interested in the “undecideds” in both polls. My guess is that the Opinion Diagnostics question format pushed people to actually choose more effectively than the Shaftan poll. It’s curious to me that in a race with as close to universal name ID as you’ll ever get in a state legislative race has so much undecided at this point.
The bottom line
OK, with all that said, what do I take from these two polls?
I’m mostly going to set aside the topline results. In a perfect world, you’d watch the same pollster’s results over time, or look at an average across multiple polls. We’re unlikely to get that in this race. Everything has to be taken with a grain of salt.
Instead, I’m primarily interested in the other nuggets of information we can glean from the polls. In the N&O, I wrote that the Opinion Diagnostics poll shows us that the multi-million-dollar blitz trying to push Page out of the race was unsuccessful. That’s still true. The new poll adds to that narrative.
If I’m Berger, I’m devastated to see that I’ve spent two decades of my life and $2 million in the last few weeks and am at only ~40% in my home district. That is not a good sign for the incumbent.
But if I’m Page, I’m not celebrating. These polls are not definitive enough to crack the power structure on their own. Fundraising is going to be incredibly tough, and so much of victory in a primary race is about turnout. That costs money.
The most important thing I take from the new poll is Shaftan’s analysis of his results, drawn from the open-ended responses. In a message back to the Berger War Room, he said:
“Berger’s #1 open-ended unfavorable are his negative ads against Page, which aren’t really drawing blood. Rockingham people think Page is a good sheriff, but some think he is out of his league and should stay in his lane. He has not made the case for why he should be elected, and is a function of anti-Berger hate. That was the impression I got from him when he and I spoke a year ago. This race isn’t about issues. It’s about personality. And that’s not a plus for Phil Berger. People just don’t particularly like him.”
I think if the election were held today, Page is ahead. But there’s a ton up in the air and no poll is going to give us a definitive take ahead of Election Day.
Quick hits
Want to hold Democrats accountable? Stop handcuffing the offices they keep winning. My new newspaper column argues that the General Assembly should give the attorney general real tools to intervene when systems fail. You can’t demand results from an office you’ve hollowed out. Read it free with gift link: A conservative case for stronger Democrat executives in North Carolina
At a premium
Top spenders on social media last week
Question of the week
I still get asked this question all the time, and honestly I don’t know whether it would have made the difference or not. 2020 was a weird time. Hold the election in 2021 instead and Forest is the governor right now.