Skip to main content
Here’s The Best Tool We Have For Understanding How The Midterms Are Shaping Up

If you’re interested in how the 2018 midterm elections will go, you may want to keep an eye on the generic congressional ballot, one of the best tools for predicting how many U.S. House seats each party will win. The FiveThirtyEight generic ballot tracker, which we launched today, can help you do that. The generic ballot question, posed by pollsters for decades, is simple: It asks which party voters would support in a congressional election1; there’s usually no mention of specific candidate names — that’s what makes it “generic.” We hope you’ll check it out every day between now and November 2018. (Occasionally is fine too.)

So what does the generic ballot show right now? That Democrats are in a historically strong position, with a 44 percent to 37 percent lead over Republicans. That is an incredible gain from the eve of the 2016 election, when our generic ballot estimate put Democrats up by only a single point, 45 percent to 44 percent.2 Indeed, at no point during the summer or fall leading up to the 2016 election did Democrats have as large an advantage on the generic ballot as they do now, and the generic ballot was essentially tied by November. In other words, the political environment seems to have become a lot worse for Republicans since last year’s presidential election.

As I’ve written previously, the generic ballot, even this early in a midterm cycle, can be quite predictive of the outcome of the following year’s House elections. Once you control for which party is in the White House, the generic ballot about 18 months before a midterm election is strongly correlated (+.78) with the eventual House result — i.e., the share of votes cast for the president’s party versus the share of votes cast for the opposition party. Here’s all the generic ballot polling we have going back to 1942:3

Generally, in the runup to the midterms, the party that doesn’t control the White House (now, the Democrats) generally sees its position on the generic ballot improve — or remain stable. Given the Democrats’ current 7-point advantage, they’d be expected to win the 2018 national House vote by about 9 percentage points (assuming, of course, that past trends hold and the forecast is perfect, which is very unlikely). But sometimes the political environment changes and the party in the White House makes gains on the generic ballot. Ahead of the 2002 midterms, for example, when George W. Bush was the president, the Democrats held a small lead at this point in the cycle, but Republicans took back the advantage after the Sept. 11 terror attacks.

As we approach the 2018 midterms, expect the generic ballot to become even more predictive. For the 18 midterm elections that have taken place since 1946, I compared the final generic-ballot polling of the cycle by Gallup or the final polling average from RealClearPolitics4 with the results of the national House vote and found that the final polling missed by an average of only 2 percentage points. That’s about as accurate as the final national presidential polls before a presidential election.

Simply winning the national House vote isn’t enough for Democrats to win a majority of seats in, and control of, that chamber, however. Because of incumbency advantage — which helps Republicans, who hold most of the seats now — and the way that congressional district lines are drawn — which also benefits the GOP — the Democrats need to do more. How much more is not exactly clear. (That largely depends on how many Republicans run for re-election and the Democrats’ coalition — for example, can they increase their support in suburban and rural areas?) Rough estimates suggest that in 2018, the Democrats may need to beat Republicans by 5 to 8 percentage points nationally to win a majority of House seats.

So if the Democrats maintain their 7-point advantage in the generic ballot, does that mean they are on their way to taking back the House? Not exactly. The generic ballot isn’t gospel. Even the final generic-ballot polls of a midterm election cycle aren’t perfect at forecasting the final national House margin. The true margin of error for generic ballot polls is about +/- 5 percentage points, even for those done at the end of the campaign. So that 7-point advantage we see for the Democrats could indicate a national House win of only 2 points, which would fall far short of what the Democrats need to regain control of the chamber. (Or Democrats could win by as many as 12 points.) In the end, a final generic-ballot average showing the Democrats up 7 points would suggest that they’d be about a 50-50 proposition to take back the House.

Finally, a few points about how we calculate the generic ballot numbers you see in the interactive. Essentially, we use the same methodology that we do for our presidential approval estimates, but with two minor exceptions:

  • Because generic ballot polls are issued much less frequently than presidential approval polls, we use a longer time window in computing our generic ballot estimates. Polls conducted as long as 60 days ago may still have some influence on our estimate, whereas we use a 30-day cutoff for presidential approval numbers.
  • Whereas the presidential approval model defaults toward using the widest sample universe — that is, it uses polls of all adults before polls of registered voters, and polls of registered voters before polls of likely voters — our generic ballot estimates reverse this, and prefer likely-voter polls to registered-voter polls, and registered-voter polls to adult polls. This is because while presidential approval ratings are traditionally conducted among all adults, the generic ballot’s main purpose is to predict the upcoming race for Congress, and there’s often a significant “enthusiasm gap” in which party’s voters actually turn up at the midterms.

As each party recruits candidates and the dynamics of individual district take shape, how the generic ballot will translate into seats won by each party will become clearer. For now, the generic ballot shows the Democrats in a stronger position at this point in a midterm election cycle than any party without control of the House since 1942. That’s about all a minority party can ask for at this point.


  1. Some pollsters ask slightly different versions of the question, such as whether voters prefer that Congress be controlled by Democrats or Republicans.

  2. Although our final estimate had the Democrats ahead by 1 point, Republicans actually won the House popular vote by a point (a 2 percentage point error). That translated into Republicans holding onto their large majority because of other advantages they have in House elections.

  3. No data was available at this point for the 1946, 1950, 1970, 1978 or 1990 cycle.

  4. I used Gallup surveys from 1946 to 2002 and the RealClearPolitics average from 2006 to 2014. No generic ballot polls were taken in the final month of the 1942 campaign.

Harry Enten is a senior political writer and analyst for FiveThirtyEight.