Unequal participation: who votes, who does not, and whether it changes what government does

Unequal participation is the gap between the people eligible to take part in elections and the people who actually do. In the 2024 U.S. presidential election, Census Bureau data show that voters were older, more educated, and higher income than eligible citizens as a whole: adults with a high school diploma were 28.6% of eligible citizens but 23.0% of voters, while families earning $150,000 or more were 24.8% of eligible citizens in families and 29.7% of voters. Scholars agree these gaps exist and disagree about how much they change what government does.

In plain words: the people who vote are not a mirror of the people who could vote. Older people vote more than younger people. People with college degrees vote more than people without them. People with higher incomes vote more than people with lower incomes. If officials pay more attention to voters than to nonvoters, some people’s needs may count for less. Researchers are still arguing about how big that effect is.

The problem of unequal participation, defined precisely

Unequal participation needs careful definitions, because the same election can produce very different turnout numbers depending on what is counted.

Turnout. The share of some population that voted. The denominator matters. The MIT Election Data and Science Lab describes three: the voting-age population (VAP), which “includes individuals who are ineligible to vote, such as non-citizens”; the citizen voting-age population (CVAP); and the voting-eligible population (VEP), which also removes people ineligible under state law. MIT states that “VEP is the most preferred denominator, followed by CVAP, and then VAP”.⁠¹³

Participation gap. The difference in turnout rates between groups, such as people with and without a college degree.

Over- and underrepresentation. The Census Bureau’s measure: a group is overrepresented when its share of voters is significantly larger than its share of eligible citizens, and underrepresented when it is smaller.⁠² This measure asks whether the electorate looks like the eligible population, which is the question that matters for representation.

Self-reported versus validated turnout. Surveys ask people whether they voted; validated studies match respondents to official records of who voted. Self-reports run high.

Participation beyond voting. Contacting officials, attending meetings, donating, and organizing are also participation, and their gaps can be larger than voting gaps. This page concentrates on voting, where the data are best. Money in politics is treated on its own page.

Responsiveness. The degree to which policy follows the preferences of a group. Unequal participation is a problem for representation only to the extent that it produces unequal responsiveness, and that link is the contested part. See Representation and The representation gap.

Scale and evidence of unequal participation

Participation in the November 2024 U.S. election, as measured by the Census Bureau’s Current Population Survey Voting and Registration Supplement, was 65.3% of the citizen voting-age population, about 154 million people, with 73.6% registered.⁠¹ Voters made up 59.3% of the voting-age population, 65.3% of the citizen voting-age population, and 88.8% of registered citizens.⁠² Once registered, most citizens vote; much of the gap is in who registers.

Turnout differed sharply by education: 82.5% of citizens with an advanced degree voted, 77.2% of bachelor’s degree holders, and 52.5% of high school graduates. Women voted at higher rates than men, 66.9% to 63.7%.⁠¹

The Census Bureau’s representation table for 2024 shows how these rates change the electorate’s makeup:⁠²

Income shares are shares of citizens living in families, a smaller denominator than all eligible citizens. Residence shares include a large “not reported” group, which inflates the long-tenure difference.

These figures carry known limits. The Census Bureau acknowledges that vote overreporting is not unique to its survey and that other surveys “consistently overstate voter turnout,” and an estimated 14.1% of the citizen voting-age population did not answer the voting question in 2024.⁠² Pew Research Center addresses overreporting by matching survey respondents to commercial voter files that aggregate official turnout records; it counted a respondent as a validated voter only if both the respondent and the records said they voted.⁠³

Reasons for not voting point in several directions. Among registered nonvoters in 2024, 19.7% said they were not interested or felt their vote would not make a difference, 17.8% said they were too busy or had a conflicting schedule, and 14.7% said they disliked the candidates or campaign issues.⁠²

Causes and competing explanations

Unequal participation has several explanations, and they imply different remedies.

Resources and education. The strong association between education and turnout in the Census data,⁠¹ is consistent with the view that voting requires time, information, and confidence that come more easily to some citizens. The data show association, not cause; education may stand in for income, networks, or interest.

Registration costs. Because registered citizens vote at high rates⁠² barriers to registration matter. The National Conference of State Legislatures reports that multiple studies place the turnout effect of same-day registration between 3% and 7%, with an average of about 5%, while cautioning that isolating its effect is difficult.⁠⁹

Mobility. People who have lived at their address longer are heavily overrepresented among voters.⁠² Moving often means re-registering, and young and lower-income people move more.

Election timing. Elections held off the federal calendar draw far fewer voters, and the voters they draw are less representative. Hajnal, Kogan, and Markarian found that on-cycle November elections “generally double local voter turnout,” and that moving to on-cycle elections in California produces an electorate that is “considerably more representative” by race, age, and partisanship.⁠¹² See Low turnout in local elections.

Legal change. In Shelby County v. Holder (2013), the Supreme Court held that the Voting Rights Act’s coverage formula “can no longer be used as a basis for subjecting jurisdictions to preclearance,” with Chief Justice Roberts writing that “Our country has changed”.⁠¹¹ Scholars and advocates disagree about how much later changes in state voting rules have affected turnout gaps. See Access and exclusion.

Interest and efficacy. Some nonvoters report lack of interest or a belief that their vote would not matter.⁠² This can be read as a free choice or as a rational response to being ignored by parties and officials.

Who is harmed by unequal participation

Whether unequal participation harms anyone depends on whether it changes outcomes or policy. The evidence is mixed and should be read carefully.

Evidence that turnout changes policy. The strongest causal evidence comes from abroad. Fowler studied Australia’s adoption of compulsory voting and found that “the policy increased voter turnout by 24 percentage points,” increased the vote and seat shares of the Labor Party “by 7–10 percentage points,” and significantly increased pension spending at the national level.⁠⁶ In that setting, bringing in nonvoters changed both who won and what government spent.

Evidence of unequal responsiveness. Gilens and Page analyzed 1,779 policy issues and reported that “economic elites and organized groups representing business interests have substantial independent impacts” on U.S. policy, “while average citizens and mass-based interest groups have little or no independent influence”.⁠⁴ Their study concerns preferences and influence, not turnout directly, but it is the central piece of evidence for the claim that unequal voice produces unequal results.

Evidence against a large effect. Enns responded that even where the preferences of the affluent and the middle diverge, “policy ends up about where we would expect if policymakers represented the middle class,” because the two groups usually rank policies in the same order; he concludes that “the opportunity of unequal representation of the ‘average citizen’ is much less than previously thought”.⁠⁵ In presidential elections, Pew found that nonvoters in 2024 would have split 44% for the Republican candidate and 40% for the Democratic one, and concluded that if all eligible Americans had voted, “the overall margin in the popular vote likely would not have been much different”; in 2020, by contrast, nonvoters favored the Democratic candidate 46% to 35%.⁠³

Who bears the risk. Taken together, the groups most exposed if participation gaps do matter are younger adults, people without college degrees, lower-income families, Hispanic citizens, and people who move often.⁠² The risk is greatest in low-turnout elections, where the gaps widen.⁠¹²

What has been tried to reduce unequal participation

Reforms aimed at unequal participation mostly lower the cost of voting or move elections to dates when more people vote. Each has measured effects and clear limits.

Automatic voter registration. NCSL counts 24 states plus Washington, D.C. as having enacted or implemented automatic voter registration, a process in which eligible people are registered when they deal with certain government agencies unless they opt out. Oregon enacted it in 2015 and implemented it in 2016.⁠⁸ A study for the Center for American Progress co-authored by Reed College political scientist Paul Gronke estimated that the program added about 116,000 people who would have been unlikely to register otherwise, of whom about 40,000 voted in November 2016, and that the program increased voting among ethnic minorities and among younger, lower-income, and rural residents.⁠⁷ The limit is visible in the same numbers: roughly two in three of the new registrants did not vote.

Same-day registration. Twenty-four states and D.C. offer same-day or Election Day registration.⁠⁹ The estimated turnout gain of 3% to 7% is real but modest.⁠⁹

Compulsory voting. Australia’s experience shows large turnout effects and policy consequences.⁠⁶ The Stanford Encyclopedia of Philosophy’s summary of the research is cautious: compulsory voting “gets citizens to vote, but it’s not clear it does much else”.⁠¹⁰ It has no serious prospect in the United States and raises the liberty objections discussed below.

Moving elections on cycle. Consolidating local elections with federal ones roughly doubles local turnout and makes the electorate more representative,⁠¹² with costs discussed on the local elections page.

Federal registration law and voting rights enforcement. These are covered in Access and exclusion and are not repeated here.

The record suggests that cost-lowering reforms shrink gaps at the margin, that timing reforms produce the largest changes in local electorates, and that no American reform has made the electorate look like the eligible population.

The strongest arguments about unequal participation across the political spectrum

Unequal participation is judged differently depending on what one thinks voting is for.

Equal voice as a democratic requirement. Many on the left argue that systematic gaps by income, age, and race mean government hears some citizens more than others, citing work like Gilens and Page.⁠⁴ and Fowler’s evidence that turnout changes policy⁠⁶ On this view, the state should make registration automatic and voting easy.

Participation as a responsibility, with integrity first. Many conservatives argue that voting is a civic responsibility citizens should choose to exercise, that easy-voting reforms should not weaken safeguards, and that reforms sold as neutral often have partisan motives. They point out that nonvoters’ preferences shift between elections,⁠³ so claims that higher turnout favors one side are unreliable. See Trust in elections.

No duty to vote, and a duty to vote well. A libertarian and epistemic argument, associated with philosopher Jason Brennan, holds that “there is no general duty to vote (abstention is permissible)” but that it is wrong to vote badly.⁠¹⁰ On this view, low participation by uninformed citizens is not a harm, and compulsory voting is coercion. Lisa Hill, who defends compulsory voting, acknowledges that voters may experience it “as yet another unwelcome form of state coercion”.⁠¹⁰

Representativeness through institutions. The demographic argument for compulsory voting, attributed to Arend Lijphart, holds that it “would tend to ensure that the disadvantaged vote in higher numbers”.⁠¹⁰ Institutionalists who reject compulsion often favor the same goal through timing and registration rules.⁠¹²

Coincidental representation. A skeptical empirical view, represented by Enns, holds that because groups usually rank policies in the same order, unequal participation changes less than its critics claim, and that party identification explains more of who gets represented.⁠⁵

Unresolved questions about unequal participation

• Does turnout change American policy? Australian evidence says yes for one reform;⁠⁶ U.S. evidence on responsiveness is disputed.⁠^4,5
• Which gaps matter most? Age, education, and residence gaps are larger in the 2024 table than some gaps that receive more attention.⁠² Should reform target the largest gaps or the ones tied to historical exclusion?
• How much is measurement error? Overreporting and nonresponse.⁠² can distort group comparisons if they differ by group.
• Do newly registered voters stay? The Oregon figures describe one election⁠⁷ Long-term turnout among automatically registered citizens needs more study.
• Is nonvoting a choice to respect or a barrier to remove? The answer differs by person, and survey reasons⁠² cannot fully separate them.
• Application to this framework. Can any opt-in priority tool reach the people who do not vote, and how would anyone know? See Collective priorities and Questions.