Matt Shinners
Polling has become a hot-button issue over the past few election cycles. They’ve seemed to miss as often as they’ve hit, and controversies have arisen around polling firms that put out fake polls to make a point.
However, polls are still important tools to understand where a race stands, and where the electorate is moving. As with all tools, though, it’s important to understand how to use them.
Let’s dive into considerations that you should go through when analyzing a poll, and then look at some of the firms that have polled multiple independent Senate races this cycle
Reading a Poll
When you read a poll, start with who paid for it. A survey commissioned by a campaign or an allied group is released by choice, and campaigns rarely publish numbers that hurt them.
Then look at the basics: when it was conducted, how many people were surveyed, and the margin of error. A 500-person poll with a ±4.4-point margin cannot meaningfully separate candidates who are three points apart.
Next, check who was surveyed. Polls of likely voters generally tell you more about an election than polls of registered voters or all adults. It also helps to know how people were reached (phone, text, online, or a mix) and whether the pollster released its full questionnaire and crosstabs (the breakdowns by party, age, region and other groups).
Some polls will also drive intentionally toward a result. These “push polls” will, for example, give positive biographical information about their preferred candidate and negative information about the opponent. Results thus skew toward their preferred candidate.
A similar technique is an “informed voter” poll. These polls, ideally, present neutral information about both candidates to simulate likely election-day behavior, as individuals learn more about candidates as they prepare to vote. However, some of these polls will instead be push polls, with the information not presented in a neutral manner. Informed voter polls can be useful, but, again, check to see if the firm has released the language of their questions to ensure that the bios are presented in a neutral manner.
Watch for these red flags:
- No sponsor named. The release doesn't say who paid for the poll.
- Memo only. The release is a short memo with toplines but no question wording or crosstabs.
- Primed questions. Voters hear candidate biographies or other information before the ballot question.
- Outliers with no explanation. The results differ sharply from other polls of the same race.
- No track record. The pollster has no history or isn't rated by an independent grader such as the Silver Bulletin pollster ratings.
Finally, remember that any single poll is a snapshot. The average of several independent polls is a sturdier guide than any one survey.
Independent Senate Race Polling
Seven firms have polled more than one of 2026's independent Senate races, which include Dan Osborn (Nebraska), Seth Bodnar (Montana), Todd Achilles (Idaho), Brian Bengs (South Dakota), Bob Chew (Colorado), and Ty Pinkins (Mississippi).
Ratings of the pollsters below come from two sources. The first is the Silver Bulletin, Nate Silver's successor to the FiveThirtyEight grades, last updated in January 2026. They are based on polls released in the final 21 days before an election. The second is Media Bias/Fact Check, which bases its pollster ratings on Silver Bulletin data.
Public Policy Polling
This firm was founded in 2001 and has run six polls in these races, more than any other firm. Five were for the Bengs campaign, between July 2025 and September 2026. One was in Idaho for an unnamed client aligned with independents. Media Bias/Fact Check reports a Silver Bulletin grade of A- and rates the firm "Least Biased" with a slight left lean.
Impact Research
The firm's stated focus is Democratic candidates and progressive causes. In these races, Impact ran two polls for the Osborn campaign, one for the Bengs campaign, and one in Mississippi for the SPLC Action Fund. Silver Bulletin grades Impact B-, with a mean-reverted Democratic bias of 1.32 points.
Change Research
Mike Greenfield and Pat Reilly founded Change Research in July 2017. It is a Bay Area public benefit corporation that polls entirely online. It has run one poll each in three races, each for a different client:
- Osborn's exploratory committee, in Nebraska.
- Independent Candidate Action, a group backing several independent candidates this cycle.
- In Idaho, the campaign of John Stegner, an independent candidate for governor.
Silver Bulletin grades it B-, with a mean-reverted Democratic bias of 2.68 points.
Tavern Research
Tavern is a Chicago-based political tech startup founded in 2024. The firm pairs polling with AI-generated ad content and message testing, and its job postings say it works to help Democrats win. It has run three Montana polls, apparently for Contours Inc., and released one Nebraska poll publicly. It has no Silver Bulletin grade.
The Bullfinch Group
Brett Loyd is president and CEO of Bullfinch, which describes itself as nonpartisan. Loyd previously polled for Donald Trump's 2016 campaign. Bullfinch is the polling partner of the Independent Center, which was set up to organize independent voters. It has run two polls in Idaho and one in Montana. KATU described an Idaho poll matching Bullfinch's first survey as Achilles-sponsored. It has no Silver Bulletin grade.
Peak Insights
Peak Insights lists clients such as the NRCC, the RNC, and several Republican candidates on its select client list, suggesting it’s a Republican-aligned firm. It has run one poll in Montana, apparently for the Alme campaign, and two in Idaho. Its results showed Bodnar at 22 percent and Achilles at 18 percent. A June Idaho poll that still included Democrat David Roth had Achilles at 7 percent. No Silver Bulletin grade was found under the Peak Insights name.
Public Opinion Strategies
This firm, founded in 1991, has polled for the McCain and Romney presidential campaigns. It has run one poll in Montana and one in South Dakota. The South Dakota poll showed Rounds ahead by 23 points and was released through Townhall. FiveThirtyEight graded POS A- in 2021, and its current Silver Bulletin grade is unconfirmed.
Last Notes
In the runup to the election, expect to see more polls drop. Many will be released by campaigns trying to make it appear as if they have momentum, or as if they have the election on lock, thus suggesting to their opponents’ voters that it’s not worth showing up to the polls.
And those things may be true! But remember that polls can’t be perfect, and even neutral pollsters have biases in their instruments (the questions), samples, and methods (texting gets you different people than calling land lines).
Check the polls, but look into the pollsters and what they release, and you’ll have a better sense of what’s likely to happen on election day.