If it feels like every poll you’ve seen this year came from a Democratic campaign and showed a toss-up, you’re not imagining things. Democratic-aligned pollsters account for 56 percent of House polls released so far in 2026, up from a 24 percent average over the previous four cycles, and 77 percent of those polls show a race within five points, the highest share on record.
It doesn’t seem to matter what race was surveyed either. In elections to the U.S. House of Representatives and U.S. Senate, spanning seats that Kamala Harris won easily in 2024 to those overwhelmingly Republican-leaning, these new polls almost always seem to show a close race.
Those numbers come from FiftyPlusOne’s database of House polls going back to 2006, which tells a clear comparative story about who is releasing surveys this year. My analysis finds that part of the surge is the ordinary signal of a favorable Democratic environment. But my investigation also reveals an apparent recent change in why sponsors release internal campaign polls — that traces back to a niche, nerdy polling project run by the New York Times’ data-journalism vertical, The Upshot, in 2018.
Paid subscribers to 50+1 get access to premium analysis, plus sortable tables and complete data access on our polling website. If you want to follow the 2026 cycle with the best data at your fingertips, become a paid subscriber.
There are a lot of internal polls this year
Our data includes all polls of U.S. House races held from 2006 to 2026 and published between 300 and 54 days before the election. I then counted the number of polls in each cycle that were sponsored by Democrats, Republicans, or from a nonpartisan source. The chart below shows the counts of the total number of polls from each source in each cycle.
So far, 2026 has had the third-highest number of House polls released overall, and the second-highest number of Democratic internals. The total number of released Democratic internals matches the Democratic “wave” year of 2018, rather than, say, 2010 when Republicans had the edge in the national environment (and when GOP challengers released a lot of surveys showing them ahead of incumbents for fundraising purposes).
When comparing the statistics in 2026 to the overall number of Dem/Rep/nonpartisan releases for the whole cycle historically, you can see that we should expect the total number of polls to increase considerably as the election approaches. In 2018, for example, polls released after Labor Day made up about 65 percent of all polls released over the whole year. The increase is particularly large for nonpartisan/public surveys.
Democratic internal polling so far makes up 56 percent of all polls released this cycle, higher than the 48 percent of polls released at this point in 2018. By the end of that cycle, Democratic internals made up just 27 percent of the total number of surveys.
Internal House poll releases have historically been an indicator of which party has an advantage in the national electoral environment. The next graph plots the Democratic share of partisan internals released (we omit nonpartisan polls). There is a strong relationship between the share of all internals that are released by one party, and the advantage that party has in the national House popular vote. The current composition of partisan surveys, at 80.7 percent Democratic, is more D-leaning than every cycle since 2002 — larger even than 2018.
How The New York Times changed incentives for House polling forever
The dynamics of House polls have changed significantly in the last decade. In the fall of 2018, The New York Times and its pollster Siena College teamed up for an ambitious project to poll dozens of competitive House races and update the results live on the internet as interviews were completed. In an earlier analysis of that project, I found that there was a strong relationship between the competitiveness of a NYT/Siena House poll and the change in donations given to the Democratic candidate in the two weeks after the poll was released.
The chart below plots the Democratic margin in each NYT/Siena poll against the change in donations to the corresponding race in the two weeks before and after each poll was released. A higher point means donations in that race rose in the two weeks after the Times released its poll in the race, and a lower point means donations fell. Donations were highest in races where polls indicated a Democratic candidate was coming up just short against their opponent, and lowest in races the poll showed to be uncompetitive. (The measure of donations here is the average amount given to the Democratic candidate each day.)
Campaigns quickly caught on to the effect these polls had on their fundraising and released their own surveys showing competitive races.
In House races in the 2018 election cycle, I found that the release of a NYT/Siena survey was associated with a large increase in the share of other polls showing the Democratic candidate within 5 points of their opponent.
The graph below groups polls by whether they were released just before or after a NYT poll of the same district, and the Democratic candidate’s margin of victory in the poll. I categorized each poll according to the following margins:
Double Digit Deficit: Democratic candidate behind by 10 points or more
Behind But Not Out: Behind by 5 to 10 points
Battleground: Within a 5-point margin
Lead But Not Safe: Ahead by 5 to 10 points
Double Digit Lead: Ahead by 10 or more
The chart shows a statistically significant increase in the share of “battleground” polls released after a NYT/Siena poll of the same district:
Moreover, this effect has persisted in every election cycle after 2018. The bar chart below shows the competitiveness of every House poll released since the 2006 election cycle grouped by sponsor type. In and after 2018, the share of House polls showing competitive races increased regardless of sponsorship: The share of Democratic-aligned competitive polls increased by 18 percent on average, Republican-aligned by 20 percent, and nonpartisan by 0 percent.
It is not surprising that Democratic campaigns, which are much more reliant on small-dollar donations than their Republican opponents, have seen the biggest shift toward released competitive surveys.
Democratic polls are popping up in redder and redder seats
We have learned that the volume of Democratic internal polls has not changed from 2018 to 2026. But the frequency of Democratic internals that show a close race has risen significantly.
Overall, 77 percent of Democratic internal polls released this cycle have shown a race within 5 points, a new high across all sponsorship categories and election cycles.
A final reason why it seems like there are more Democratic internals this cycle is that there has been a sharp increase in the number released in heavily Republican districts. In the chart below, I have grouped Democratic internals by cycle based on the margin of the vote Kamala Harris had over Donald Trump in the seat in 2024 and the competitiveness category of the poll result. The chart shows a large increase in the number of Democratic internals released in heavily GOP seats, especially when looking at the 2022 and 2024 cycles, where Democratic internals were released mostly in battleground or Biden-won districts. Some of these polls show Democrats outrunning Harris’s margin by 20 points or more!
Overall, this analysis does not make any claims about the accuracy of current House polls. It may very well be right that Democrats are leading in seats Republicans typically win by 5-15 percentage points.
But it is clear that motivation for releasing a poll has changed significantly over the last decade as a response to fundraising incentives. It seems the original NYT/Siena polling project has turned pre-election surveys into a tool for raising campaign funds, not merely measuring public opinion.
Article credits:
Katie Marriner created the charts for this piece.
Mary Radcliffe provided research and data.
Cooper Burton copy-edited.
G. Elliott Morris was editor.
Paid subscribers to 50+1 get access to premium analysis, plus sortable tables and complete data access on our polling website. If you want to follow the 2026 cycle with the best data at your fingertips, become a paid subscriber.
Thank you Noah and Elliott and Mary and Katie and Cooper. This was the second most instructive post of this election season for me. (First was Andy Borowitz posting that the best way to donate is a recurrent donation to a candidate in an important election so that the campaign can count on the funds as needed.) I kept wondering why I was getting emails and texts that were at odds with the Strength in Numbers figures. They kept me wondering if there had been some drastic change in public opinion. Now I know that my feeling that the candidates were using the poll numbers to generate contributions was correct.