Can High Schools Increase Voting and Civic Participation Later in Life?

Mumma (2026) examines whether the high school a student attends affects civic engagement, especially voting, in adulthood. She analyzes education records for nine cohorts of Indiana public school students, linking them to birth records, parental voting histories, and nationwide voter records. After accounting for student, family, school, and community characteristics, she finds meaningful differences in schools’ effects on later voting. A one-standard-deviation increase in a school’s civic effect raised the probability of voting within the first four eligible election cycles by 2.1 percentage points, or 5.3 percent, and increased the number of times students voted by about 9.4 percent.

Why This Article Was Selected for The Policy Scientist

Civic participation is a central outcome of public education because schools may shape whether young people develop habits of political participation that persist into adulthood. This question is especially timely given the persistent turnout gap between younger and older Americans and renewed attention to civic education. The study makes an important contribution by extending research on education and voting from particular programs to differences across ordinary public high schools. Its unusually rich linked administrative data are a major strength, as are the value-added models, sibling comparisons, mover analyses, and extensive robustness tests supporting a causal interpretation. Still, evidence from 335 Indiana schools may not generalize nationally, and future randomized or stronger quasi-experimental studies would further strengthen causal inference.

Full Citation and Link to Article

Mumma, K. S. (in press). High school effects on civic engagement. American Economic Journal: Economic Policy. American Economic Association. https://www.aeaweb.org/articles?id=10.1257/pol.20250610

Central Research Question

The study asks whether the high school a student attends has a meaningful effect on civic engagement in adulthood, with particular attention to voting. The central issue is whether differences in adult political participation reflect only the characteristics of students, families, and communities or whether high schools themselves contribute independently to later voting behavior. The analysis focuses on traditional public high schools rather than specialized civic programs, making the question broader than whether a particular civics curriculum or charter-school intervention works. It also examines possible mechanisms that might explain school effects, including academic achievement, noncognitive outcomes, college-going behavior, peer civic norms, civics coursework, and extracurricular activities. Because adolescence is an important period of political socialization and many students become eligible to vote shortly after high school, the study treats secondary schooling as a potentially consequential environment for the development of lasting civic behavior.

Previous Literature

The study builds on several distinct bodies of research. One establishes a general relationship between education and political participation. Dee (2004), Milligan, Moretti, and Oreopoulos (2004), and Sondheimer and Green (2010) examine whether additional education increases civic participation and voter turnout. This literature generally finds that education and civic engagement are positively related, although identifying the causal contribution of schooling has been difficult.

A second literature emphasizes political socialization during adolescence. Niemi and Sobieszek (1977), Jennings and Niemi (2015), and Campbell (2006, 2013) describe schools, families, peers, and social networks as environments in which young people develop political beliefs and habits. Gidengil, Wass, and Valaste (2016), for example, document a strong relationship between parents’ voting and their children’s subsequent political participation. This literature provides a basis for examining parental civic engagement and peer civic norms as potential influences on student voting.

The paper also builds on research examining specific educational institutions and interventions. Gill et al. (2020) find that attending Democracy Prep charter schools increased voter registration and turnout. Cohodes and Feigenbaum (2021) similarly find positive effects of Boston charter-school attendance on later voting. By contrast, Carlson, Chingos, and Campbell (2017) find no significant effects of private-school vouchers on registration or turnout. Research on civics instruction itself has produced less consistent results. Manning and Edwards (2014) and Jung and Gopalan (2023) find limited evidence that broad civics coursework reliably increases later political participation.

Finally, the empirical strategy draws heavily from the value-added literature. Chetty, Friedman, and Rockoff (2014) demonstrate how teacher value-added measures can be connected to long-term student outcomes, while Deming (2014) examines whether value-added measures recover causal school effects. Jackson et al. (2020) and other recent studies extend this approach beyond standardized test scores to longer-term and nonacademic outcomes.

Data

The analysis combines unusually detailed administrative records from several sources. The education records cover students enrolled in Indiana public schools between the 2006–07 and 2021–22 school years. The primary sample consists of nine cohorts of students who first entered ninth grade between 2007–08 and 2015–16. These records contain enrollment histories, demographic characteristics, standardized test scores, attendance, disciplinary outcomes, Advanced Placement participation and scores, and other measures of student performance.

Education records are linked to Indiana birth records, which provide information on students and their parents. Birth records make it possible to identify parents and subsequently connect both students and parents to voting histories. Approximately 70 percent of students in the initial sample could be matched to Indiana birth records.

The study then links students and parents to nationwide commercial voting records from L2. These records identify voter registration and participation in federal, state, and local elections. Multiple annual voter files are combined to improve the ability to track individuals who move or change names.

After sample restrictions, the principal estimation sample contains approximately 456,700 students attending 335 high schools. Civic engagement is measured over students’ first four age-eligible election cycles. Outcomes include registration, voting in general and primary elections, participation in other elections, and the number of elections in which an individual voted. These measures are also combined into a standardized civic-engagement index. Similar measures of parental voting before the student entered high school provide unusually strong controls for family civic behavior.

Methods

The primary statistical strategy adapts school and teacher value-added models to estimate each high school’s contribution to adult voting. The models control for extensive student characteristics measured before high school, including eighth-grade math and English scores, attendance, suspensions, demographic characteristics, special-education status, English-learner status, and eligibility for subsidized meals. They also incorporate parental voting behavior, characteristics of each school cohort, and characteristics of the student’s county of birth.

The models include cohort and first-eligible-election fixed effects to account for statewide differences across years and elections. School effects are estimated using a leave-cohort-out procedure in which outcomes for other cohorts at the same school are used to predict the school effect for a particular cohort. This approach reduces the possibility that random fluctuations in one cohort are mistaken for persistent school effects.

The causal interpretation nevertheless depends initially on selection on observable characteristics: after conditioning on the extensive controls, students attending different schools must not differ systematically in unmeasured characteristics that independently predict voting. Several additional analyses test this assumption. The study examines forecast bias using variables excluded from the main models, tests whether estimated school effects predict outcomes among siblings who attended different schools, and examines students who moved between high schools. These analyses provide stronger evidence than a conventional cross-sectional multivariate regression, although school assignment itself is not randomized.

Findings/Size Effects

Adult civic participation differs substantially across high schools even before adjustment. About 60 percent of students from schools at the 90th percentile of the turnout distribution voted at least once, compared with about 37 percent at schools at the 10th percentile—a 23-percentage-point difference. Voter registration rates differ by approximately 15.2 percentage points across these groups.

After adjusting for student, family, school, and community characteristics, schools continue to have statistically significant and practically meaningful effects. A one-standard-deviation increase in a school’s estimated civic effect increases the probability that a student votes during the first four age-eligible election cycles by 2.1 percentage points. This represents a 5.3 percent increase relative to the sample mean. The same increase in school civic effectiveness raises the number of times a student votes by approximately 9.4 percent.

Effects are also evident in the first general election for which students are eligible. A one-standard-deviation increase in school civic effects raises turnout in that election by about 1.7 percentage points, or 7.4 percent relative to the sample mean. The effect on voter registration is approximately 1.4 percentage points, equivalent to about a 2 percent increase. Relative effects are particularly large for participation outside major general and primary elections, approaching 20 percent, although the absolute changes are comparatively small.

The validation analyses strengthen the interpretation that these differences reflect exposure to particular schools rather than only differences in the students attending them. School civic effects remain predictive in within-family comparisons among siblings attending different schools and among students who move between high schools. Forecast-bias tests also produce estimates close to zero.

Evidence concerning mechanisms is less definitive. Schools with stronger parental civic norms among the student body tend to have larger civic effects, suggesting that peers and social environments may matter. School effects on college-entrance-exam participation are also positively associated with civic effects. Evidence linking AP U.S. Government performance to civic effects is suggestive but sensitive to model specification, while the presence of civics-related extracurricular activities does not consistently predict larger voting effects.

Conclusion

The study concludes that high schools contribute measurably to civic participation after students leave school. The effects are not large enough to explain most variation in voting or eliminate major age-based turnout gaps, but they are sufficiently large to indicate that civic engagement is another dimension along which schools differ in effectiveness.

The results broaden the concept of school quality beyond test scores, graduation, and college attendance. Schools that produce stronger academic outcomes are not necessarily the same schools that generate stronger civic outcomes, suggesting that civic effectiveness represents a distinct dimension of institutional performance.

At the same time, important questions remain unresolved. The analysis identifies school-level effects more clearly than it identifies the specific practices producing them. Peer civic norms appear relevant, while evidence concerning formal civics programs and extracurricular activities is considerably weaker. The findings are also based on 335 schools in one Midwestern state, so the magnitude and mechanisms of the effects may differ in jurisdictions with different electoral institutions, school systems, or political environments. Future experimental and quasi-experimental research could more precisely identify which school practices produce lasting changes in civic engagement and whether those effects generalize across different educational and political settings.