Brief us

WP-35: Terrain

The city votes a generation older than it lives

In the mayoral elections of America’s thirty largest cities, the median voter was 57 and the median eligible resident was 42. Registered voters over 65 turned out at five times the rate of those under 35.

By Brushworks research deskNovember 2026Working paper

Abstract. Local elections decide schools, housing, policing and the power bill, and they are decided by a small and much older electorate. Portland State University’s study of mayoral elections in the thirty largest U.S. cities found turnout below 15 percent in ten of them and a median voter fifteen years older than the median eligible resident. Most of the gap comes from timing: cities that vote off the national calendar vote with a fraction of their residents. We set out the record and what it means for any campaign, measure or institution that has to win a local vote.

In brief

The finding
Most local elections are decided by a small electorate that is about a generation older than the city. The calendar, more than apathy, decides who shows up.
For a local campaign or a measure
Plan for the electorate that will vote, and measure every effort to enlarge it against a control. The voters you need are often the ones who vote in November and skip May.
For an institution facing a local vote
The people who decide your siting, your levy or your franchise are mostly homeowners over fifty-five. Speak to their concerns first.

A city’s most consequential decisions (its schools, its zoning, its police budget, its taxes) are made by the people who vote in its local elections. Portland State University’s Who Votes for Mayor project measured who those people are in the thirty largest U.S. cities. In the elections it studied, the median voter was 57 years old. The median eligible resident was 42.1, 2 This paper sets out the record and what it asks of anyone who has to win a local vote.

Who votes

In ten of the thirty largest cities, fewer than 15 percent of eligible residents voted in the most recent mayoral election the study covered. In Las Vegas, Fort Worth and Dallas, turnout was in single digits.1 Across the thirty cities, 46.7 percent of registered voters aged 65 and over voted, against 9 percent of registered voters aged 18 to 34.1

Turnout of registered voters in mayoral elections in the thirty largest U.S. cities, by age.

The gap is wider in some cities than others. In Houston, 6.6 percent of registered voters aged 18 to 34 voted for mayor, against 42.9 percent of those over 65, and the median voter was 61 in a city whose median adult was 41.3 In San Francisco, 62 percent of registered voters over 65 voted, against 28 percent of those aged 18 to 34; the median voter was 52 against a median voting-age resident of 43.4

The median voter and the median resident. Each bar runs from the median age of eligible residents to the median age of mayoral voters.

The calendar decides

The strongest predictor of low local turnout is the date. Of the eighteen cities in the study with turnout below 15 percent, fourteen held their mayoral elections in years without a national election, and sixteen held them on different dates from national elections.5 Portland, which elected its mayor alongside the presidential election, had the highest turnout in the study at 59.4 percent and the youngest median voter at 46.2, 6

Of the eighteen large cities with mayoral turnout below 15 percent, how many voted off the national calendar.

The finding is not new. Studies of municipal elections have shown for two decades that moving a local election onto the national calendar raises turnout substantially, and that off-cycle elections favor organized groups with a direct stake in the outcome, such as public employees and developers.7, 8

What an older electorate decides

An electorate that is older, more often a homeowner and more often long-resident weighs local questions differently from the city as a whole. Housing supply, school bonds, transit and the siting of large projects all draw different majorities from the voters who show up than from the residents who live with the result. A campaign or institution that reads the city’s demographics instead of its electorate will write for people who are not voting.

What changes when the electorate is a generation older. A qualitative reading for a local program.

What a local program does

  • Plan for the electorate that will vote. Build the model from the local voter file and from behavior in past local elections, not from the census profile of the city.
  • Find the November-only voters. Many residents vote in national elections and skip local ones. They are the cheapest people to add to a local electorate, and their behavior shows who they are.
  • Measure mobilization against a control. An effort to enlarge a small electorate is the easiest program to overrate. Hold out a slice of every cohort and count the difference.
  • Choose the date when you can. For a measure, the election date is often a choice. Choose it knowing who will vote on it (WP-17).

What the research says about timing

The evidence on election timing is among the most consistent in the study of American local politics. Comparing cities that hold local elections with national ones against those that do not, researchers have found turnout in on-cycle municipal elections far higher than in off-cycle ones, and the difference does not disappear when the comparison accounts for the size, wealth or politics of the cities.7

Anzia’s work explains why the date matters beyond the number of voters. In an off-cycle election, the share of the electorate with a direct stake in the outcome (public employees in a school board race, for example) is much larger, because they vote while most of their neighbors stay home. Using the timing of school board elections, she shows that teacher salaries tend to be higher where boards are elected off-cycle, which is consistent with organized groups carrying more weight in small electorates.8

Why the date matters: the share of a small electorate that a stakeholder group can supply, as the electorate shrinks. Illustrative arithmetic.

The arithmetic is simple. A group of five thousand reliable voters is two percent of a hundred-thousand-vote November electorate and ten percent of a fifty-thousand-vote spring one. A program that ignores the calendar is, in effect, choosing which organized groups will decide the result.

Some states now push in that direction: California’s Voter Participation Rights Act of 2015 requires many local governments to move their elections onto the statewide calendar when turnout in their own elections falls well below the state’s. For a campaign or an institution, the policy debate matters less than the practical one: on the date the vote is held, who will be there, and who could be.

The November-only voter

Between the residents who never vote and the regulars who vote in every local election sits a large group that votes for president and skips the mayor. In most cities they are younger than the regular local electorate, more often renters, and more recently arrived. They are already registered, already in the habit of voting, and already reachable. They are the cheapest people to add to a local electorate.

From registration to the local ballot, per hundred registered voters in an invented large city. Illustrative of the pattern in the record.

A program finds them by behavior: votes in presidential years, no record in recent local elections, a stable address. It reaches them with the date, the place and one sentence about why this election touches their life, and it measures the result against a holdout, because nothing in a low-turnout election is easier to overrate than a turnout program.

A worked example: one levy, two dates

The figures in this example are illustrative. A school district can put a levy on a May special election or on the November general. In May, the electorate is older, more often without children in the district and much smaller. In November it looks more like the district.

Who votes in a spring special election and a November general, share of the electorate. Illustrative.

Neither date is right in general. A levy that depends on families with children usually does better when they vote. A levy that depends on long-time homeowners may do better in a smaller, older electorate that it can reach directly. The choice is made on the evidence of who votes, and the campaign plans for the electorate the date produces (WP-17).

Objections

Moving local elections to November buries them under the national ballot. Voters do leave some local races blank on long ballots. Even so, more residents decide the local race in November than in an off-cycle election, which is the finding the research has repeated for two decades.7, 8

Older voters are simply the engaged ones. They are the ones who show up. Engagement in local life (schools, parks, transit, rents) is spread across ages, and the younger residents who live with a decision are often the ones not voting on it.

Local campaigns cannot afford measurement. A holdout costs nothing to draw and saves the cost of repeating a program that did not work.

Questions to ask before a local vote

  • Who voted in the last three local elections, by age, tenure and home ownership?
  • How many registered residents vote in November and skip local elections?
  • If the date is a choice, who votes on each available date?
  • What control will say whether the turnout program added voters?

Further reading

  • Anzia, S. F. (2014), cited above: how off-cycle elections favor organized groups.
  • Hajnal, Z. L., and Lewis, P. G. (2003), cited above: municipal institutions and turnout.
  • Portland State University, Who Votes for Mayor (2016): the city-by-city data behind this paper, mapped to the census tract.

Implications

For a candidate. The local electorate is small enough to know by name. Its priorities are those of an older city than the one you see on the street.

For a measure or an institution. The date of the vote shapes the result as much as the text of the question. Ask who votes in that month before you ask what to say to them.

For a city. Holding local elections with national ones is the single most reliable way to make the local electorate look like the city.

Notes and references

  1. Portland State University, Who Votes for Mayor, October 2016, findings for the thirty largest cities.
  2. Keisling, P., Who Votes for Mayor, October 27, 2016.
  3. Houston Press, reporting the Portland State findings for Houston.
  4. San Francisco Chronicle, reporting the Portland State findings for San Francisco.
  5. Harvard Data-Smart City Solutions, summarizing the Who Votes for Mayor findings on election timing.
  6. Portland State University, Who Votes for Mayor, as reported by Philadelphia Magazine, October 28, 2016.
  7. Hajnal, Z. L., and Lewis, P. G. (2003). Municipal institutions and voter turnout in local elections. Urban Affairs Review, 38(5), 645–668.
  8. Anzia, S. F. (2014). Timing and Turnout: How Off-Cycle Elections Favor Organized Groups. University of Chicago Press.

Basis of figures. Turnout and age figures from Portland State University’s Who Votes for Mayor project, covering the most recent mayoral election in each city between 2011 and 2015; figures for individual cities as reported from the same study. The exhibit on what changes is a qualitative reading; the city funnel and the levy example are illustrative. No Brushworks program figures appear in this paper.

Next paperThe Review

Start a conversation

Brief us.

Tell us the race, the calendar and the margin. Within a week we will tell you whether we can move it and what it would take. Candidate work and independent-expenditure work are taken by separate teams; say which you are. Governments and institutions: tell us the mandate and the jurisdiction, and we will tell you what registration it requires before we say anything else.

Send a briefbrief@brwrks.com

  • Candidate practice
  • Independent-expenditure practice
  • International practice

Brushworks Strategies Inc., a Wyoming corporation · Washington, New York, Ottawa, London, Munich, Dubai, Tokyo