What Voter Turnout Numbers Actually Predict

For decades, conventional wisdom held that high voter turnout favored one party. Polling, journalism, and political strategy all operated under the assumption that getting more people to the polls would tilt the result toward Democrats and away from Republicans. That assumption was correct, more or less, from the 1960s through the early 2000s. Then it stopped being correct, and most of the analysis has not caught up.

Voter turnout is one of the most-cited numbers in election coverage and one of the most-misunderstood. The headline number tells you something specific but limited. What it does not tell you is which party benefits from the people who showed up — and the answer to that question is now different from what it used to be.

What turnout is actually measuring

Voter turnout is the share of eligible voters who actually cast a ballot. The "eligible" qualifier matters. A turnout figure can be calculated against the voting-age population, the voting-eligible population (excluding non-citizens and people with felony disenfranchisement), or registered voters. The numbers come out differently depending on which denominator is used, and different organizations use different conventions.

For practical purposes, the most informative measure is turnout among the voting-eligible population. It tells you what share of the people who could legally vote in a given election did so. In modern presidential elections, that number has typically run between 55 and 67 percent. In midterms, it drops to 40 to 50 percent. In off-year and local elections, it can fall well below 30 percent.

A single headline number — "turnout was 67 percent this year" — sounds informative but conceals enormous variation across demographic groups, geographies, and types of voters. Younger voters turn out at roughly half the rate of older voters in most elections. College-educated voters turn out at substantially higher rates than non-college voters. White voters turn out at slightly higher rates than non-white voters, on average. Almost every demographic line has a turnout gap on it.

Why the partisan calculation has changed

For most of the late twentieth century, the marginal voter — the one most likely to be missing from a given election — was more likely to vote Democratic than Republican. Working-class voters, minority voters, and younger voters all turned out at lower rates than their share of the eligible population suggested they should, and all of them leaned Democratic. Higher turnout meant more of those voters showed up, and the Democratic share of the vote went up.

That calculus has changed because the coalitions have changed. In recent cycles, working-class white voters without college degrees have moved decisively toward Republicans, while college-educated suburban voters have moved toward Democrats. The marginal voter — the one most likely to be missing in low-turnout conditions — is no longer reliably one party’s voter. In some recent elections, higher turnout has helped Republicans more than Democrats, because the new low-propensity voters in the Republican coalition turned out when conditions favored them.

This is a relatively recent realignment, and most of the operational assumptions of campaigns and election forecasts are still adjusting to it. Get-out-the-vote operations that used to be reliable Democratic infrastructure have become less straightforwardly partisan. The same is true of voter-ID laws and other turnout-affecting policies; the partisan effect of any given change depends on which voters get added or removed at the margin, and that calculation has flipped in some states.

High turnout, low turnout, and what each predicts

A high-turnout election is not automatically a more representative election. It is an election in which a larger share of the eligible population participated. Whether that share is more or less demographically representative than a lower-turnout election depends on who turned out.

In the 2020 election, turnout reached its highest level in over a century — and the result was still extremely close, with the partisan coalitions roughly mirroring each other in size. High turnout did not produce a landslide because the new participants split more evenly between the parties than was historically expected. In other words, higher turnout did exactly what skeptics of the old wisdom had been predicting: it produced a more representative electorate, and that electorate split.

In midterm elections, where turnout drops by 15 to 20 percentage points, the composition of the electorate changes significantly. Older voters, college-educated voters, and politically engaged voters become a larger share of the actual electorate even though they remain the same share of the eligible population. The party whose coalition is older, more educated, and more engaged tends to do better in midterms — but which party that describes can change across cycles.

Why turnout coverage is often misleading

Election-day turnout coverage tends to focus on a few visual cues: long lines at urban polling stations, high-volume mail-ballot returns, early-voting totals. These cues feel informative but are weakly predictive of the final result.

Long lines are partly a function of demand and partly a function of polling-place allocation. A long line might mean enthusiastic turnout or it might mean an overwhelmed location with not enough machines. Mail-ballot returns by party registration tell you who returned a ballot but not how they voted — and an enthusiastic mail-ballot operation in one party may simply be cannibalizing election-day voting that would have happened anyway. Early voting numbers run into the same problem.

A more reliable read of turnout comes from comparing actual cast votes against the historical baseline for similar precincts. If a precinct that usually casts 80,000 votes in a presidential election casts 95,000, that is meaningful. If the same precinct casts 85,000, the increase is small enough to be within normal variation. Coverage that aggregates these comparisons across the state, weighted by demographic patterns, gives a useful read. Coverage that just reports raw turnout numbers without comparison gives almost nothing.

The democratic legitimacy argument

There is a separate argument about turnout that has nothing to do with which party wins. Higher turnout produces an outcome that more closely reflects the preferences of the eligible population. In a country where roughly a third of eligible voters routinely sit out presidential elections and around half sit out midterms, the elected officials represent a subset of the population whose preferences may differ significantly from the population as a whole.

Whether that gap matters depends on your view of how representative democracy is supposed to work. Defenders of low-turnout elections argue that the people who vote are the ones who care enough to be informed, and that compulsory or near-universal turnout would dilute the electorate with people who are not paying attention. Critics argue that the framing is backwards — that lower turnout entrenches the preferences of an unrepresentative subset and that policies designed to suppress turnout exploit this dynamic.

The empirical question — does higher turnout produce different policy outcomes? — has been studied repeatedly and the answer is, modestly, yes. Higher-turnout electorates tend to produce slightly different elected officials and slightly different policy mixes than lower-turnout electorates from the same underlying population. The magnitude of the effect depends on the specific election and the specific policies, but it is consistent across enough studies to count as real.

What to take from a turnout number

A turnout headline tells you participation, not outcome. To get more out of the number, look at three things underneath it: the demographic composition of the electorate compared to historical baselines, the partisan registration breakdown of mail ballots if available, and the regional variation in turnout compared to past cycles. Those three slices contain almost everything useful that the headline number conceals.

The relationship between turnout and election outcomes is more complicated now than it was a generation ago. Higher turnout does not automatically favor one party. Lower turnout does not automatically suppress one coalition. The composition matters more than the volume, and the composition can be different from what the headline number suggests. Read turnout the way you would read a polling average — as a starting point, not a verdict.