...
By Shane Croucher
Senior Editor
0
Share
Newsweek is a Trust Project member
See more of our trusted coverage when you search.Prefer Newsweek on Googleto see more of our trusted coverage when you search.
It’s looking like clear blue skies for Democrats as they fly toward the midterms, ready to land their campaign in favorable conditions. President Donald Trump, on the other hand, is piloting his Republicans through a turbulent cloud of his own making.
Just look at those numbers. Democrats lead the generic congressional ballot by 6.7 points, 49.2 percent to 42.5 percent, on the RealClearPolitics average of polls. And they are polling competitively in Senate races in Ohio, Texas, and Alaska.
The Republican pollster Patrick Ruffini calculates that Trump’s abysmal net approval is eight points worse than Joe Biden’s at the same stage of the 2022 cycle, and seven points worse than Trump’s own before the 2018 wave that cost Republicans the House.
Read More on Analysis
But, before anyone starts allocating committee gavels, please consider DAD. It’s a useful pinprick to the expectations inflation that has blown full-strength into Democrats ahead of many recent elections, and once again this year.
DAD stands for Democrats Always Disappoint. Its claim is simple and cool-headed: whatever advantage the party holds in September will have shrunk by November, because Democratic polling is a summer flower that wilts by fall.
Taken literally, DAD is also wrong. But, as a rule of thumb, it holds up pretty well, and is a healthy vaccine when a pandemic of blue fever runs rampant before elections.
DAD Math
In August, Ruffini pulled 3,312 Senate polls from the last four election cycles out of the old FiveThirtyEight archive.
He found that, in every cycle, the polls overstated the Democratic candidate. The average error ran 3.7 points in the Democrats’ favor, and 73 percent of all polls missed in that direction.
Nor is the error spread evenly across the calendar. It peaks at 5.9 points too Democratic in late July, holds through August, and settles at 2.6 points by Election Day.
Related Story
Trump the Messiah Sees Judas Everywhere
Many of the Senate polls now fueling Democratic optimism, in other words, were taken at the point in the cycle where Ruffini’s sample shows the largest pro-Democratic miss.
The national picture chimes with this perspective.
The American Association for Public Opinion Research (AAPOR) found that polls in the final two weeks of 2024 overstated Democratic margins by 2.7 points across all offices, following 4.6 points in 2020 and 3.1 points in 2016.
Three consecutive presidential cycles of the same directional miss, a run AAPOR notes is historically rare, though the scale has declined as pollsters keep their methodologies under review.
Still, there has been a clear pattern in recent years, and it points to disappointed Democrats.
Republicans Miss Too
But this has not been a one-way street.
In the midterms sandwiched between those elections, the same analysis found polls overstated Republicans by 0.6 points in 2022 and Democrats by 0.1 points in 2018.
Persistent Election Day bias has lately been a presidential-year problem.
Ruffini’s own data contains this very distinction. Midterm polls leaned heavily Democratic in summer, then came close to accurate by November.
And last year the miss ran the other way, hard. Silver Bulletin found an average absolute error of 7.1 points across 2025’s qualifying polls, alongside a 6.1-point bias toward Republicans, meaning surveys understated Democrats.
New Jersey’s gubernatorial polling was the state’s worst since the turn of the century. Abigail Spanberger won Virginia by 15.4 points. So DAD is not an iron law.
The data journalist G. Elliott Morris points out that Ruffini’s summer pattern does not survive the full historical record back to 1998.
Anyone who had discounted Democratic polling in 2022 on the strength of the previous four cycles would have called that election worse, not better.
Forecasting the direction of the next polling error from the last one is, in his phrase, a "fool’s errand."
And further still, Morris makes the case that polls may actually now be underestimating Democrats. If that proves true, we should brace for a blue tsunami.
Less Trump Is More
So where does this leave the DAD theory?
One plausible explanation for why the Democrat-leaning bias keeps appearing in presidential years and thinning out in midterms is partisan nonresponse.
Surveys have long struggled to reach low-propensity voters, so people who only vote in years when a presidential candidate is on the ballot. And these people have moved toward Trump.
In presidential years, if they turn out at higher rates than samples capture, polls can understate Republicans, simply because so many of their voters are not surveyed.
Related Story
Thomas Massie’s Last Grandstand Is a $38 Billion Trump Dividend
But in midterms, when habitual, highly motivated voters dominate the electorate, that source of error should matter less in the polling.
A paper published in the American Journal of Political Science in April found differential nonresponse among Trump supporters large enough to account for the error in the 2020 American National Election Studies.
If that explanation is right, it cuts against DAD this November for the obvious reason: Trump is not on the ballot (even though he has encouraged Republicans to campaign like he is, a risky strategy for a man of his approval rating).
Special Elections
Recent Democratic victories between the major elections are telling a real-world story reflected in the polling and adding to the sense of inevitability about a blue wave.
In one eye-grabbing example from red America, Democrats flipped a Texas state senate seat Trump carried by more than 17 points. Yet there is another big caveat for Democrats in these wins.
Race to the WH has rebuilt how it uses special elections, noting that Democrats performed far better in them than in the general elections of 2022 and 2024, because of the electorate those contests tend to attract.
And Ruffini argues that a familiar geographic signature is appearing again now: Democratic overperformance that is concentrated in red states and districts, even after pollster house effects are stripped out.
Related Story
Donald Trump Is the President of Buyer’s Remorse
None of which means Trump’s Republicans are quietly cruising to a surprise win, which remains unlikely. There is genuine annoyance and anger among voters about the state of the economy, especially the cost of living, made sharply worse by Trump’s war on Iran.
Civiqs tracking cited by Ruffini on September 4 had voter perceptions of the economy inverted from the same stage of 2018: 75 percent good and 18 percent bad then, against 33 percent good and 64 percent bad this year.
Democratic primary turnout, on John Couvillon’s compilation, is running ahead of 2018.
Ruffini’s own read is that the environment is worse for his party than it was that year, and that the open question is whether redistricting and a forgiving map limit Republican seat losses, particularly in the House.
Those efforts to redraw congressional maps ahead of the midterms—by Republican states encouraged on by Trump and checked, where possible, by Democratic gerrymandering—are yet another complication for attempts to forecast the election.
What DAD offers instead of a forecast is cautionary wisdom, the true wiseness of which will be put to the test over the coming weeks.
It’s the Slope That Kills You
For now, the Democratic summer high has not yet begun to decay. By this stage in 2022 it had already started turning back toward Republicans. In 2018 the decline began at roughly this point on the calendar.
On September 4, Ruffini wrote a sentence Democratic campaigners should pin above their desks not as a celebration of what’s to come, but a motivator to stay focused on making sure it does: "If this line doesn’t budge, that’s bullish for Democrats."
Watch for the slope, if it comes. The wisdom of DAD would mean the polling advantage for Democrats drifting back towards the Republicans before late October, similar to previous late-campaign movements.
Either way, the acronym is not a forecast that Republicans will win. It’s a reminder that Democrats have not won yet.
View 0 Comments
Submit Correction
View Editorial & AI Guidelines