Why You Should Believe the Polls This Time

This summer, I was a poll skeptic. Internal polling published by campaigns and allied political action committees shaped public perceptions, affected fundraising, forced primary candidates in Michigan and Maine to quit, and perhaps even influenced the data-weighing choices of independent pollsters. In some cases, most notably Michigan’s Senate primary and Wisconsin’s gubernatorial primary, pollsters struggled to reach certain demographics—what is known as “non-response bias”—leading to Election Day surprises.  

In this general election season, we have enough independent polling to avoid the problem of internal polls leading us by the nose. But non-response bias—with committed Republican voters shunning pollster queries—remains possible, and pollsters warn that data-weighting techniques can’t easily correct it. This has led optimistic Republicans and pessimistic Democrats to speculate that a midterm election Blue Wave may not materialize.  

The polling industry has its challenges, and no single poll should ever be treated as gospel. But we have lots of other data that buttress the horse-race polling and point strongly toward a Blue Wave.  

Off-Year State Legislative Elections: The canary in the coal mine was the spectacular Democratic record in state legislative regular and special elections since Donald Trump’s 2025 inauguration. By March, Democrats had flipped 30 seats and coughed up zero. In August, Democrats picked up one more. As I explained last spring, no such shutout has occurred in at least the last 15 years. Moreover, in this period we have seen a correlation between off-year state legislative flips and party performance in midterm elections. 

Real Disposable Income: As political scientist Seth Masket has shown, growth in per capita real disposable income—how much money you have after accounting for taxes and inflation—correlates with midterm outcomes for the president’s party. From the third quarter of last year to the second quarter of this year, real disposable income has declined, partly driven by higher food and fuel prices that followed Trump’s tariffs and attack on Iran.

Such an outright decline has happened in only four previous midterms in the last 75 years: 1954, 1958, 1974, and 2022. The president’s party lost 48 House seats in both the 1958 and 1974 midterms. In 1954, the loss was a more modest 18, but the opposition Democrats headed into the midterms with a sizeable minority already in hand, just five seats short of a majority—a similar situation for Democrats today. The anomalous 2022, when Joe Biden’s Democrats shed only nine seats, can be attributed to backlash over the overturning of Roe v. Wade, revulsion at persistent Republican election denialism following the attempted January 6 coup, and a relative improvement in real disposable income with a declining inflation rate. 

Presidential Job Approval: Ahead of that 2022 midterm,  John T. Woolley, the political scientist of the American Presidency Project, found that in midterm elections between 1936 and 2018 “[the six] presidents with approval between 40-45% saw their party lose between 63 and 13 seats, with an average loss of 36 [House] seats.” In 2022, Joe Biden was in that job approval range, so his nine lost House seats brought that average down to 32. 

However, for much of the last two months, Donald Trump’s current Real Clear Politics job approval rating has been below 40 percent.  

According to Gallup’s Presidential Approval Center, only two past post-World War II presidents had job approval in the 30s at the time of a midterm: Harry S. Truman, for the 1946 and 1950 midterms, and George W. Bush for the 2006 midterm. (Bush’s Real Clear Politics average was also below 40 percent before the 2006 midterm.) 

Unsurprisingly, the president’s party took it on the chin in each of those three midterms. Truman’s Democrats lost 58 House seats in 1946 and 28 in 1950. Bush’s Republicans lost 30 House seats. They also featured substantial Senate losses for the president’s party: six in 1950 and 2006, and 12 in 1946. And in 1946 and 2006, the president’s party lost control of both congressional chambers.  

Generic Congressional Ballot: The poll question that asks voters who they plan to vote for in their district’s House race, without names mentioned, has long been a reliable indicator of which way the midterm wind is blowing. The Real Clear Politics generic congressional ballot average has held steady in the Democrats’ favor all year, gradually widening from about a 4-point to 8-point margin. In 2006 and 2018, the last two Blue Wave years, we saw similarly steady trendlines and margins.  

All the horse race poll data fits comfortably with this additional poll, economic, and electoral data. Moreover, even if it is the case that pollsters are struggling to reach Republican voters, that may not undercut their findings.  

We know depressed base turnout can factor into wave elections. In 2010, dominated by the Tea Party, the electorate’s share of voters under 30, even though they powered Barack Obama’s winning coalition two years earlier. In 2016, Trump won 46 percent of the popular vote, but in the 2018 midterm, only 40 percent of the electorate were Trump voters. 

So it may be that some Republican voters are not just shunning pollsters but shunning the election, fed up with the president’s broken promise to cut prices yet too conservative to switch parties and vote Democratic. Rural voters and farming communities, GOP hotbeds, have been hit particularly hard by tariffs and diesel prices, which helps explain the surprising competitiveness in polls by Democratic Senate and gubernatorial candidates in places like Iowa, Kansas, and Texas. 

That said, believing the general thrust of horse-race polling is not the same as prematurely betting Democrats will run the table on Election Day. Every poll is a snapshot of public opinion with a stated margin of error that may miss late movement among undecided voters. Systemic poll misses are always possible when pollsters fail to model turnout demographics accurately. As pollster Stefan Hankin told the Washington Monthly’s Anne Kim last week, “The polling industry has oversold what the results of a poll mean,” suggesting a certainty that no poll can provide.  

But setting aside the innate uncertainty of poll data, the state-by-state Senate poll data is already telling us how uncertain the outcome is. 

If we review all the Senate polls sampled over the last three weeks (not including internal polls released by campaigns or political action committees), only a few races believed to be competitive at the onset of the election season now seem out of reach for Republicans, with average Democratic margins of at least seven points—the two Democratic-held seats of Georgia and New Hampshire, and Republican-held North Carolina. (Republican campaign operations are reportedly shifting campaign cash away from Georgia and North Carolina.) 

Democratic candidates have average leads of three points in Republican-held Texas and Ohio, plus Democratic-held Michigan. That Democrats are at all competitive in Texas and Ohio, firmly Republican throughout the Trump era, is a testament to the leftward political winds. Abdul El-Sayed’s underperformance relative to Michigan’s Democratic nominee for governor—who has a 6.5-point average lead over the last three weeks of polls—likely stems from some of his controversial positions, lingering intra-party divisions from his bruising primary, and Islamophobic attack ads from the GOP. Narrow three-point leads in polls should never be considered safe leads, and a narrow Republican win in these states wouldn’t mean the polls were wrong, given their margins of error. 

The next tier of even tighter Senate races, with average Democratic leads between one and two points, includes Republican-held Alaska, Iowa, Kansas, and Maine, plus Democratic-held Minnesota. Whether Minnesota is that close is questionable, as very few polls—and no first-rate pollster—have sampled the state yet. But the Democratic nominee, Lieutenant Governor Peggy Flanagan, is from an administration rocked by a Medicaid fraud scandal, so it’s plausible. And I’ve previously discussed why Troy Jackson is the latest Democrat to struggle with overcoming Maine’s five-term incumbent Susan Collins. Otherwise, the deep red states in this tier further make the case that a Blue Wave is building. Regardless of whether the tide will crest high enough to flip Alaska, Iowa, and Kansas and net Democrats a gain of as many as seven Senate seats, the mere fact that it is within the realm of reason, and that Republicans have been forced to spend to defend this turf, proves the point. 

A giant wave could even carry Democratic and independent candidates in other conservative territory. Limited polling shows South Carolina Democrat Annie Andrews and independent Nebraskan Dan Osborn trailing only slightly, well within margin of error. There’s no independent polling in South Dakota and Mississippi. Still, Republican dark money groups supporting incumbent Mike Rounds must have internal poll data worrying enough to air attack ads against independent candidate Brian Bengs. And Mississippi’s incumbent Senator Cindy Hyde-Smith is airing attack ads against the Democratic underdog Scott Colom. 

Still, the only thing that appears to be a lock is that Democrats will take the House. If Democrats can’t net the three seats required with the president’s popularity circling the drain, then they should close up shop for good. Whether they can flip more than the post-World War II opposition-party midterm average of 26 seats remains unknown. As in 1954, Democrats are starting with a near-majority already, which limits their target opportunities and requires wins in deep Republican territory to post big gains.  

The poll data is in sync with history. Trump’s abysmal job approval and economic record should lead to healthy Democratic gains, including in places where, a year ago, most would not have expected Democrats to win. That Democrats are not guaranteed a sweep of every contested Senate race, let alone an above-average net House gain, is not because the polls are untrustworthy. The polls themselves tell us to remain uncertain and tell the parties not to be complacent.  

The post Why You Should Believe the Polls This Time appeared first on Washington Monthly.