We have said it over and over again – turnout matters when it comes to control of the House of Representatives! Of course, turnout is only part of the game. There are two aspects to elections: who will turnout to vote (turnout) and, conditional on turnout, who will voters vote for (preference)? So, why so focus on turnout, you ask? Especially in this cycle, predicting sentiment conditional on turnout, while certainly not easy, is at least easIER. In this highly politicized environment, voters have made up their mind. True, there is still some model error (smart talk for us mis-estimating sentiment conditional on turnout), but we have pretty good reason to believe that our top-lines are close to true. First, we leverage close to 200,000 responses and more than 30 Million behavioral data points to estimate directional vote – an unprecedented effort. And, our methods have produced very accurate results in the past (such as the 2016 general election), when preferences were not as easy to predict.

Turnout is a different story, especially in midterms, when the drop-off in turnout, i.e. those turning out in presidential but not in midterm elections, depends on circumstances. In short, turnout matters because a vast part of uncertainty of the outcome is now down to the partisan composition of the actual electorate come November. Do Democratic leaning demographic subgroups show up in higher numbers than usual? For example, usually only around 20% of 18-24 year-olds vote in midterms, a highly Democratic group. How high will that number be in 2018? More importantly, turnout is not deterministic, it is instead affected by late-cycle developments, especially in this cycle, and ultimately, whether YOU (and you girlfriend/husband/neighbor) shows up to the polls!

And even more crucially, the effects of different turnout scenarios are especially large this year. To help you gauge what this means in practice, we at PredictWise, in collaboration with data scientist Boris Yanovsky, have built an interactive tool, allowing you to play around with different turnout scenarios, and explore the effects on election results come November. Go check it out – it is live right here


Now, the first thing you see is the map come November that we think will happen under a conservative/bad turnout (baseline) scenario. Specifically a likely voter universe derived from turnout models of the biggest Democratic voter file vendor, TargetSmart. This likely voter universe is adjusted for turnout in 2017 and 2018 special elections, but is ultimately built on fundamentals such as previous turnout history and reflects very little of the attitudinal enthusiasm gap we talked about earlier. Of all voters in this likely voters space identifying with a political party, 49.9 percent of voters identify as Democrats (including leaners), and 50.1 percent of voters identify as Republican, again including leaners, according to our estimation. In some ways, as far as turnout is concerned, such a scenario is the most pessimistic realistic outcome for Democrats come November. Assuming that we estimate vote intention conditional on turnout correctly (read more above), this scenario sees Democrats taking the national popular vote by 1.8 percentage points, and sees Democrats effectively winning 9 seats, falling short of the 23 seats they would need to take the House, accounting for currently six vacant seats.


From here, you can select a state from a drop-down menu, and then select the district you are interested in by clicking on the map. Second, you can select the demographic that you are most interested in. For example, if you are most interested in exploring the effect of elevated Millennial turnout, select “Age- 18-24” from the demographic drop-down menu. In all cases, the starting value for the slider will be set to its “empirical truth”, i.e. will reflect the percentage of voters falling into the demographic category based on our conservative/base-line scenario. From here, you can move the slider to any number between 0% to 100%. In our example, moving the slider to 0% would mean that out of all voters, 0% will be 18-24 year olds, and moving the slider to 100% would mean that out of all voters, 100% will be 18-24 year olds.

You will see that the two-party vote share in this district will change based on slider movements, reflected by changing colors of the district map. Below the map, you can compare the new demographic composition of voters in this district with our baseline scenario, and see the adjusted two party vote share.

One word on methods: without going into two much depth here (for a more detailed take on our methods centered on dynamic Mr. P., the affectionate term for “multi-level regression and post-stratification, read our white paper), this tool is based on two assumptions: (a) if we increase the turnout rate of certain demographics, we assume that those new voters will vote exactly the same as the cohort of actual voters falling into the same demographic; (b) If we increase the slider for a certain demographic, it not only means that we are increasing its share in the composition of the electorate, we are also decreasing the share of non-selected demographics proportionally. For example, if we increase the share of 18-24 year olds by 10 percentage points, the share of 25-34 year olds, 35-44 year olds, 45-54 year olds, and 55+ will be reduced by 10 percentage points multiplied by their share of the overall voter pie in our baseline scenario: If 35-44 year olds make up 40% of all voters in our baseline scenario, their share will be reduced by a factor of 10 * .4 percentage points.

Disclaimer: Politics is not a spectator sport. It is not deterministic. When we say there is 60% likelihood of the Democrats winning the House we are assuming a certain level of commitment by staff, volunteers, and, most importantly, the voters themselves. If Democrats get super motivated: they will win. If Republicans get super motivated: they will win. You are part of the game!