menu_open Columnists
We use cookies to provide some features and experiences in QOSHE

More information  .  Close

Will the 2026 midterms be a ‘blue wave’? Here’s what to watch as polling ramps up

4 0
06.03.2026

Will the 2026 midterms be a ‘blue wave’? Here’s what to watch as polling ramps up

One of the safe wagers about this year’s midterm elections is that polling of competitive congressional races will be frequent and abundant — and maybe even misleading. After all, the “red wave” of sweeping Republican victories anticipated four years ago turned out to be a very modest “red trickle.” 

While Democrats are favored to win control of the House of Representatives this year, it may be months before clarity emerges about how November’s congressional elections will turn out. Even so, it is none too early to be aware of some realities about election surveys this year, especially as polling already has been conspicuous in unfolding U.S. Senate races in Maine, Michigan, New Hampshire and Georgia. 

Pollsters in 2026 are coming off back-to-back disappointments. In 2024, polls overall understated Republican Donald Trump’s support for the third time in as many presidential elections, despite modifications to survey techniques that sought to avoid such an outcome. In 2025, polls collectively underestimated support for victorious Democrats in gubernatorial campaigns in New Jersey and Virginia — the year’s two most prominent statewide races. 

Of course, past results are no sure indicator of future polling performance, as the recent experience of AtlasIntel makes clear.  

A Brazil-based company, AtlasIntel was the most accurate pollster in the U.S. presidential race in 2024, correctly projecting Trump would win the national popular vote as well as carry the seven swing states where the Electoral College outcome turned. No other pollster was so successful. In 2025, however, AtlasIntel signaled a far closer race for governor in New Jersey than what transpired. Its final preelection poll reported that Democrat Mikie Sherrill led by a single percentage point; she won by 14 points.  

AtlasIntel also failed to anticipate the magnitude of Abigail Spanberger’s victory in Virginia’s gubernatorial race. 

“We literally run hundreds of polls every year,” Andrei Roman, AtlasIntel founder and chief executive, said in a post on X as the 2025 results became clear. “We’ll inevitably get a couple wrong, in particular if turnout patterns shift dramatically. And we’re always looking to learn and improve our methods.” 

Should their work turn out to be misleading about this year’s elections, pollsters will likely escape much public accountability — which has been the case in recent national elections. There has been little appetite among polling organizations and the news media to scrutinize or direct much critical attention to polls that veer off target in their final pre-election estimates.  

This tendency was reflected in a detailed report commissioned by the American Association for Public Opinion Research about polling the 2024 presidential election. The report studiously avoided specific reference to individual pollsters. 

As a result, the performance in 2024 of prominent, media-sponsored polls — including those of CNN and The New York Times, both of which estimated Kamala Harris led in enough swing states to win the Electoral College — received no attention in the report. 

The 2026 campaign coincides with turbulence in election polling, given the departure of some prominent entities in survey research. For example, the popular polling-data site FiveThirtyEight was shuttered last year by ABC News amid staff cuts ordered by its parent, Walt Disney Co. The polling institute at Monmouth University in New Jersey was closed last year, reportedly in a cost-cutting move. 

And one of the country’s most lauded pollsters, J. Ann Selzer of the Iowa Poll, retired following the disastrous survey she released late in the 2024 presidential campaign. Selzer’s poll indicated Harris had opened a 3-point lead over Trump in Iowa, a reliably Republican state in recent national elections. Selzer’s poll, published by the Des Moines Register newspaper, was dramatically in error, as Trump carried Iowa by 13 percentage points. That meant Selzer’s poll had missed by 16 points. 

“Polling is a science of estimation,” Selzer was quoted as saying after the 2024 misfire, “and science has a way of periodically humbling the scientist.”

The turmoil in election polling represents the renewal of a trend dating at least to 2015, when the Gallup Organization announced it was quitting election polling. Gallup was a pioneer in the survey research, having conducted its first presidential election poll in 1936. Its final presidential election poll was in 2012, when it reported late in the campaign that Republican Mitt Romney led President Obama by a single percentage point.

Obama won reelection by almost 4 points. 

Pollsters in recent years have notably experimented with and adapted diverse methodologies, especially as response rates have remained anemic to surveys conducted exclusively by cellphones or landlines. 

Many pollsters nowadays utilize a variety of sampling procedures. Mixed methodologies — in which, for example, pollsters seek respondents by sending text messages, placing robocalls and tapping panels of previously recruited participants, among other techniques — will undergird election surveys this year. 

Despite its flaws and shortcomings, polling remains the primary prism through which journalists, pundits and the public at large evaluate the ebb and flow of political campaigns. As they have in previous national elections, the polls and their results will help establish news media narratives about the trajectory and competitiveness of the 2026 elections.

W. Joseph Campbell is a professor emeritus of communication at American University. He is the author of seven books including, most recently, “Lost in a Gallup: Polling Failure in U.S. Presidential Elections.”

Copyright 2026 Nexstar Media Inc. All rights reserved. This material may not be published, broadcast, rewritten, or redistributed.

More Opinions - Campaign News

What we know about Noem’s new ‘Shield of the Americas’ role

DHS funding bill passes the House, with eroding Democratic support

Trump says Cuba’s next: Here’s how it could play out

Fox News poll: Majority says Trump’s handling of Iran has made US less safe

US Postal Service takes another wrong turn

Paxton offers conditions for potential exit from Texas Senate race 

Judge orders Trump administration to close out goods without charging emergency ...

How Kristi Noem finally lost Trump’s trust

Noem defends living on base in Coast Guard housing 

Watch video from Noem’s controversial $220M ad campaign

Trump has been right about oil all along

Senate Democrats block bill to fund DHS, spurning increased GOP pressure

US loses 92,000 jobs in February, falling far short of expected gain

Here are the 4 Democrats who opposed the war powers resolution

Talarico gives defiant victory party speech: ‘There is something happening in ...

Johnson: ‘Everybody in America better watch’ how lawmakers vote on DHS ...

GOP’s Turner rips Pentagon No. 3 at hearing: ‘Really kind of gross’

Democratic reps send letter to Patel over fired FBI agents and staff

The Hill Podcasts – Morning Report


© The Hill