Massive primary turnout points to biggest year for Democrats since Watergate There’s a remarkably close correlation between the primary vote and the House popular vote Daniel Donner Sep 22 ∙ Throughout this year’s just-concluded primary season, there’s been one consistent storyline: Democratic primaries have seen much higher turnout than their Republican counterparts. But does this meaningfully indicate what we might expect in November? It turns out the answer is very much yes. In midterms over the past 80 years, there’s been a surprisingly tight relationship between voting behavior in a given cycle’s primaries and the House popular vote in the general election that follows. Based on that relationship, this year’s primary turnout suggests that Democrats would win the House vote by an enormous 13-point spread. That’s much larger than what the generic congressional ballot forecasts. That metric, according to a polling aggregate calculated by FiftyPlusOne, currently gives Democrats a smaller 7-point edge. Either would be more than enough for Democrats to take back the House, and neither outcome would be surprising. Both measures have great predictive power, but like any other forecasting tool, they both come with uncertainty attached to them. Below we’ll explain what precisely the primary results are telling us, and why they matter. The bottom line is that Democrats need to be prepared to capitalize on the potential for an electoral tsunami on Nov. 3. Uncovering the national primary Assessing a state’s total primary turnout in midterm years is not as straightforward as it might seem, however. That’s because every state hosts a mix of different races, and for many contests, it might not even hold a primary at all. As we gathered data, we had some choices to make. We decided to start with elections for governor, which are the most common type of statewide election in midterm years. If a state didn’t hold a gubernatorial primary in a given year, we then looked to see if there was a primary for U.S. Senate. Failing that, we moved on to any other statewide races, then the total for all U.S. House races, then the total for all state legislative races, and finally, the total for as many county-level races as possible. Sometimes, we had to mix and match, using, say, a Democratic primary for governor and a Republican primary for Senate. On rare occasions, we were unable to find any data at all, but any gaps are largely immaterial. With robust data at hand stretching back almost all the way to when the 17th Amendment began mandating the direct election of senators nationwide, we developed the graph at the top of this post. It shows the total number of Democratic and Republican votes across all state primaries from 1918 to 2026, in midterm years only. We’ve dubbed this the “national primary.” We can see a century of dramatic political history playing out in the graph above: a tripling of Democratic primary votes from 1930 to 1938 following the end of the Progressive Era and the birth of the New Deal; the mid-century domination of the Democratic Party, when it comprised an uneasy coalition of southern segregationists and northern moderates and liberals; the long, slow realignment from the Civil Rights Era to Barack Obama’s presidency; a surge in turnout for both parties since 2014 that well outpaces population growth; and the recent see-sawing lead, ending with the domination of the 2026 cycle by Democrats. Taking it one step further, we can calculate a percentage vote share for Democrats and Republicans in this synthetic national primary, then take the difference between the two. The chart below shows that difference for every midterm since 1918. As you can see, this year’s margin favors Democrats by about 11 points. Primary margins move in concert with the House vote Now that we have a margin for our national primary, we can compare it to the margin of the House popular vote in midterm elections: Remarkably, from 2006 to 2022, the two measures were nearly identical. It’s tempting to take these five data points and run with them, but a closer look shows there’s also a good relationship between the two metrics before 2006—just with an offset. We can get around this problem, however, by plotting the change in margin for each of these measures. That is to say, if the Democrats’ margin in the national primary was 5 points in a given midterm but 11 points in the next midterm, we’d graph that data point as a 6. The same applies for the House popular vote. And once we do this, something even more remarkable emerges: Now, we have two measures that, for the most part, follow each other closely across 80 years of data. Prior to the 1940s, the relationship falls apart during the rapid changes in party composition outside of the South, but the ensuing eight decades give us a lot to work with. Making a prediction Now we get to the final step of the analysis, which reveals an impressively tight correlation between the change in national primary margin and the change in House popular vote stretching all the way back to 1942. The diagonal line shows the relationship between the two measures over the past eight decades. The point in the upper right, for instance, represents 2018. The primary margin that cycle was 14 points better than in 2014, and the House popular vote margin was also 14 points better than in 2014. This puts the point for 2018 just above the diagonal line, which is known as a regression and describes the historical correlation between the two measures. This means that in 2018, Democrats did just a little better than history—at least, going by this particular metric—would have predicted. (For visual clarity, data points from 1986 onward are shown in red while earlier data points are shown in orange, though there’s no difference in behavior between the earlier and later portions of the dataset.) A handful of years (1966, 1970, and 2006) might be considered outliers, but excluding them also results in essentially no difference in the equation that defines the regression line we’ve shown above.¹ As you can see on the right side of the graph, though, we’re quite literally in uncharted territory this cycle. The difference between the national primary margin in 2022 and 2026—18 points—is greater than any other such gap going back to 1942. To figure out what the House popular vote might look like based on the 2026 figures, then, we have to extrapolate beyond the regression line in the graph above. This results in a prediction that Democrats will do 16 points better than 2022, when they lost the House popular vote by 3 points. That pencils out to an expected margin of 13 points for Democrats in the House vote this year. To put that in perspective, that would be the best performance by Democrats in a midterm since the Watergate elections of 1974, when they won by 17 points. Because we’ve gone beyond the regression line, however, there’s greater uncertainty than usual about this estimate. If you look at the spread of data points around the regression line, in fact, you’ll see that a House popular vote anywhere from 4 to 6 points higher or lower is not unusual. That said, there’s nothing in the data that suggests 2026 is likely to be an outlier with a departure much greater than that. For example, unlike in 2006, this year the median of primary margin changes at the individual state level is nearly the same as the margin change for the national primary. What’s more, 43 states showed an increase in the Democratic primary margin compared to 2022, so the vast bulk of our data is pointing in one direction.² We can also go back in time and see what this technique would have predicted in years past. The earliest date we can reasonably reach back to is 1962, which gives us five prior data points we can use to run a regression. Things start out a bit rough without much data to work with and then run headlong into a couple of potential outlier years, but after that the predictions are pretty useful aside from an expected complete flub in 2006. The final November generic ballot numbers from pollster Charles Franklin are also shown for comparison. Both measures have done well in midterms since 2006. You might wonder why this relationship hasn’t been noticed before. The answer is that it has—at least, in terms of vote share, rather than the change in vote share. Most recently, analyst Ethan Chen, using data compiled by Republican pollster John Couvillon, showed a very close relationship between the midterm House margin and the primary margin going back to 2006 in a subset of states. Similarly, David Byler used Couvillon’s data to analyze the implications of primary turnout in 2022. Also in 2022, Lakshya Jain and Kraz Greinetz looked at turnout in midterm and presidential years as a predictor for November results. And back in 2018, Nate Cohn analyzed the data from 2004-2016. So far as we’re aware, though, no one has gone back as far as we have, nor has anyone else examined the change in vote share between elections in the way we have. What does it all mean for 2026? The estimate we get from the national primary this year, although highly favorable to Democrats, is consistent with the special elections data. In other words, two completely independent methods yield results in the same ballpark, and both are in the “as good as or better than 2018” camp along with Donald Trump’s approval ratings, right track/wrong track numbers, economic opinion, and party ID. The generic congressional ballot polling, however, is not better for Democrats than it was in 2018—at least, not yet—but it’s close. Below you can see how each of these metrics line up compared to the same point eight years ago. Taken all together, the indicators show a political environment that is very favorable to the Democrats. However, one thing that all these indicators cannot account for is the possibility of a major disruption in the final stretch of the campaign season. That includes a veritable tidal wave of Republican campaign cash that is only just now getting unleashed. But it also includes the potential for the bottom to truly fall out of Republican support should something like a generational energy crisis emerge—which we may be seeing with the eye-popping spike in gas prices. What we can say, though, is that Republicans would need an epic turnaround in their fortunes to retain control of the House. Democrats, meanwhile, can expect a sizable win—and, if the national primary proves predictive, a massive one. Thanks to Quinn Yeargain for assistance gathering data on early 20th-century elections. 1 There is, however, an increase in the R2 value from 0.68 to 0.84 when these three years are excluded. 2 The seven states that did not see an increase in the Democratic primary margin compared to 2022 were Delaware, Florida, Kentucky, Oklahoma, Oregon, South Dakota, and Tennessee. Thank you so much for being a paid subscriber to The Downballot! If you’d like to further support our work, you can donate on our ActBlue page. Donate to us at ActBlue