How Mobility Among Young Voters Shapes Turnout: the role of college counties in North Carolina
Introduction
Why This Matters in Today’s Democracy
The personal aspirations and motivations of individuals are closely tied to their sense of self-agency and to how actively they participate in political life. In today’s era of globalization and heightened interconnectedness, people have become increasingly mobile—not only across countries, but also within states and even between counties. This project examines how rising mobility intersects with political and electoral involvement, and how these patterns relate to educational communities such as public and private universities, as well as the personal motivations that shape voter registration and election participation.
Young voters are critical to the future electorate, and their turnout patterns signal the direction of democratic engagement in the coming decades. Understanding how and why young people participate in elections is therefore essential for anticipating the future health of democratic participation.
Geographic movement and housing relocation are known to disrupt civic routines, often reducing turnout and weakening ties to local political environments. Moves associated with education, early careers, or personal development can break existing civic connections and lead to lower participation rates. At the same time, educational environments play an important role in supporting voter participation and engagement. Higher levels of education are typically associated with more informed and engaged voters, and college communities may provide conditions that encourage young people to stay connected to civic life.
Studying how educational settings interact with youth mobility therefore offers valuable insights into whether the proximity of higher-education institutions helps reinforce political engagement during a period of significant transition. College environments can act as civic anchors, offering structure, community, and shared experiences that support ongoing involvement even when individuals are navigating major life changes. In many ways, the academic and social aspects of college life work together to strengthen civic awareness and voter participation.
For these reasons, exploring voting trends and disruptions in voting behavior is especially important among young adults ages 18–25. This group is only now entering the labor market, gaining decision-making power, and forming long-term civic habits. As mobility increases, it becomes more challenging for young people to maintain consistent voting routines or stable patterns of civic behavior.
Modern American democracy is being reshaped by unprecedented levels of mobility. As young people relocate between counties, states, and educational institutions, the stability of their emerging civic habits becomes increasingly fragile. These early transitions carry long-term consequences, shaping whether young adults develop consistent voting routines or disengage from political life altogether. By understanding how mobility patterns influence turnout—and how the presence of educational environments may offset or amplify these effects—we can better anticipate the future of democratic participation among the next generation of voters.
Hypothesis
Young movers (18–25 y.o.) who relocate into college counties are expected to vote at higher rates than both movers entering non-college counties and non-movers, because college environments provide stronger civic and social engagement opportunities. In contrast, movers entering counties without colleges are expected to show lower turnout, highlighting differences in how local educational environments shape civic integration.
Data Used
The primary data set for this project used was North Carolina Voter Registration and History File (NCVHIS). This administrative data provides individual-level information on voter registration, county of residence, registration dates, and voting history in general and local elections.
North Carolina is one of the few states that provides broad public access to detailed voter registration and voting history records, making it an ideal data set for analysis using programming tools such as R. North Carolina was selected because its election data is both comprehensive and easily accessible through downloadable files. The NCVOTER_statewide file contains registration information—including name, address, registration date, phone number, and demographic variables; while the NCVHIS_statewide file provides election-specific records for each unique voter registration number, including which elections a person participated in, their voting method, voting date, and precinct. This level of record keeping extends across multiple years of current and previous elections, allowing to examine historical snapshots that reveal where individuals lived at different points in time, when they moved, and how their age changes across election cycles. Because of this depth and accessibility, North Carolina’s data set is especially well-suited for studying voter mobility and turnout patterns.
There are total of 100 counties in the state of North Carolina, housing over 11,046,000 people statewide (as of late 2024).The geopolitical and industrial focuses of the state range from traditional industries such as agriculture and textiles to rapidly expanding sectors like pharmaceuticals, banking, and information technology, benefiting the statewide economy that way.
Lastly, North Carolina is a state which consists of various types of higher education institutions and offering all of the types (public & private colleges and universities, community colleges, residential and non-residential housing — living on campus or not). According to the state’s public university system alone, nearly 248,000 students are enrolled across its campuses. Obtaining the locations and types of these institutions was relatively easy to access since such type of information is accessible on the state’s education department. In order to more easily manipulate and process the data, the lists and county relations were built via excel to simplify computational steps within R.
Additionally, finding county maps across the state & their borders has been beneficial and helpful in the analysis to more effectively communicate findings because it creates visual representation of the analysis and clear output of the information.
Map of North Carolina Schools Counties
Analytical Framework
There are several types of elections in United States politics, each serving a different function in the electoral process. General elections determine which candidates ultimately assume public office and include contests for positions such as President, U.S. Senate, U.S. House of Representatives, governor, state legislature, and various municipal offices. Presidential elections occur within the broader category of general elections, but general elections also take place in non-presidential years for many federal, state, and local offices. Primary elections, by contrast, are intra-party contests used to select each party’s nominee who will appear on the general-election ballot.
For the purposes of this analysis, our research focuses on the elections taking place within the years covered by the data and does not differentiate between general and primary elections in the current dataset. This approach enables consistent measurement of turnout patterns across time, particularly when examining mobility and voting behavior. However, distinguishing between general and primary participation represents an important opportunity for future research, as primary-election data may reveal additional insights into civic engagement, partisan dynamics, and participation patterns among movers and non-movers.
Elections Turnout Rate measures the percentage of eligible or registered voters who participate in a given election. It is one of the most important indicators used to analyze voting behavior across different groups. Turnout can be examined by demographic characteristics, party affiliation, geographic location, and other relevant factors, making it a central metric in both political science and election studies. By comparing turnout rates across groups and over time, researchers can identify behavioral patterns, evaluate civic engagement, and assess how participation changes under different conditions or experimental “treatments.” These analyses help illuminate broader trends in political participation and the factors that shape voter engagement.
Setup & Working With R
Throughout the Election Data Science course at the University of Florida, I strengthened my proficiency in R programming by applying a wide range of analytical tools and techniques. These included data visualization, dataset merging, geographic shapefile manipulation, and the calculation of turnout and mobility measures. All of these skills were essential in developing this project and have become an invaluable part of my technical toolkit.
This analysis relied heavily on the capabilities of RMarkdown and RPubs for organizing, documenting, and presenting the results in a reproducible format. In addition, several core R functions played a central role in the workflow, including filter(), select(), inner_join(), left_join(), distinct(), ggplot(), and various tools for working with shapefiles. Together, these functions enabled efficient data cleaning, transformation, and visualization, ultimately allowing me to explore voter mobility and turnout patterns across North Carolina.
Some of the libraries used in this project are:
library(tidyverse)
library(sf)
library(tigris)
library(ggpattern)
library(patchwork)
library(scales)Naming Conventions & Preparing the Data
To structure the analysis, the dataset first needed to be organized chronologically and filtered by the target age group. Because most individuals move and pursue higher education between ages 18–25, I aligned both snapshots of the voter file to capture the same cohort over time. The 2022 (Year 1) data was filtered to include voters aged 18–23, and the 2024 (Year 2) data was filtered to include the same individuals at ages 20–25. This approach ensured continuity across both datasets while increasing the likelihood that voters appeared in both years. Although this age filtering reduces the size of the dataset, it creates a more accurate and consistent set of observations for analyzing mobility and turnout among young voters.
Filtered2022 <- read_csv("~/Downloads/filtered2022.csv") |>
select(county_2022, voter_reg_num, age) |>
rename (age_2022 = age) |>
distinct(voter_reg_num, .keep_all = TRUE)Filtered2022 contained total of 764,528 distinct voters (from 2022 voter file snapshot).
Filtered2025 <- read_csv("~/Downloads/filtered2025.csv") |>
select(county_2022, voter_reg_num, age_at_year_end) |>
rename (county_2025 = county_2022,
age_2025 = age_at_year_end) |>
distinct(voter_reg_num, .keep_all = TRUE)Filtered2025 contained total of 761,503 distinct voters (from 2025/current voter file).
ncvhis <- readRDS("~/Downloads/ncvhis1.rds")Once the primary voter-registration datasets were prepared (filtered2022 and filtered2025), I identified individuals who appeared in both years by creating combined_base, using an inner_join() on the unique voter_reg_num. This step produced a unified file containing all voters relevant to the analysis and ensured continuity across the two snapshots. In total, 472,909 voters met the targeted age criteria and appeared in both filtered datasets.
Identifying movers and non-movers
Movers - are people within interest group of voters for analysis, who have changed counties from year 1 to year 2. There were total of 54,565 people who have relocated their counties within the time frame; 20,719 moved to schools (38%), making 33,846 of non-school movers (62%).
Below is the map of counties to which movers are relocating.
Non-movers - are voters who have not changed their location of residency between the two years used for analysis. There are total of 418,344 of this type of voters within that time frame.
Below is the map of counties in which non-movers are staying most often.
Then, the following data sets have been developed in consequence of manipulating filtered lists created earlier, combining among voted lists, belonging to specific counties and county types (school or non-school):
- 13,160 movers to public counties only
- 8,034 movers to private counties only
- 21,194 movers to college counties in general (combining both public and private)
Findings
“Differences and Indifferences” in Election Data Science
In election data science, we often look at how turnout changes over time for different groups.
A difference means that one group’s change in turnout (its delta) is noticeably larger or smaller than another group’s change. That suggests that something about that group’s context or “treatment” (e.g., moving, living in a college county) is affecting participation.
An indifference means that two groups’ deltas are very similar. Their turnout moves in parallel across elections, so the factor we’re interested in (for example, moving to a college county vs. not moving) doesn’t seem to change behavior much beyond the general trend.
To calculate a delta the formula below has been used:
Δ turnout = turnout in 2024 − turnout in 2022 (in percentage points)
Then we compare these deltas across groups to see where there are differences (divergent changes) and indifferences (similar changes).
Δ - change in turnout:
Non-movers: Δ+18.8% delta increase in turnout between 2022 and 2024
All movers: Δ+11.4%
Movers to counties with schools (combined private and public): Δ+20.1%
Movers to counties with public schools: Δ+20.6%
Movers to counties with private schools: Δ+16.3%
To evaluate differences and indifferences in turnout behavior, I compare the change in turnout between 2022 and 2024 across mobility groups. Non-movers experience a +18.8% increase, which serves as a baseline for natural turnout growth when civic ties remain intact. All movers show a significantly smaller increase of +11.4%, indicating a clear difference: residential mobility disrupts civic engagement and reduces the turnout boost typical of a presidential cycle.
However, movers who relocate to counties with colleges show a very different pattern. Movers to any college county increase turnout by +20.1%, nearly identical to, and even slightly higher than the non-mover baseline, indicating an indifference in the effect of moving when the destination contains a college. Movers to public-college counties show the strongest increase (+20.6%), suggesting that public institutions provide civic infrastructure that supports electoral engagement. Movers to private-college counties increase turnout by +16.3%, higher than all movers but still below the non-movers, implying a moderate but weaker civic reinforcement effect.
Overall, the differences and indifferences across deltas demonstrate that mobility alone suppresses turnout, but moving into counties with colleges, especially public colleges, appears to rebuild civic ties and support electoral participation among young voters.
Future Research Consideration
Although this research provides evidence that mobility and college environments shape young voters’ turnout patterns, several important dimensions remain outside the scope of this analysis. The study does not differentiate between primary and general elections, does not account for precinct-level variation, and does not examine the broader socioeconomic or psychological factors that influence a young person’s likelihood of voting. Because the focus was primarily on county-level trends and the presence of higher-education institutions, many other mechanisms that shape political behavior remain unexplored. These limitations highlight valuable directions for future inquiry.
County Characteristics and Migration Motivations: Future research could examine why young people choose certain destination counties. Colleges may attract movers for reasons unrelated to civic engagement—such as employment opportunities, affordability, or access to social and cultural life. Understanding these motivations would help reveal whether the observed turnout patterns are driven by institutional environments or by underlying characteristics of the individuals who relocate there.
Presidential vs. Non-Presidential Election Dynamics: This analysis focuses on turnout change during a presidential election year. A valuable extension would be to compare turnout deltas in midterm or local election years to determine whether the observed effects persist regardless of election salience.
Generational Differences in Political Participation: The mobility patterns and civic behaviors of today’s young adults may differ significantly from previous generations. Comparing the turnout dynamics of young movers across different decades could clarify whether the relationship between relocation and participation is stable or evolving over time.
Institutional Differences: Universities vs. Community Colleges: Counties with higher education institutions are not uniform. A deeper analysis could distinguish between university counties and community college counties to assess whether the size, mission, and civic culture of the institution create different turnout effects.
Understanding Motivation, Personal Traits, and Social Integration: Mobility is often tied to personal characteristics—such as ambition, desire for new opportunities, and belief in one’s capacity to change their situation. These characteristics also correlate with political participation. Future research could explore how personal motivation interacts with moving, and whether certain types of young people self-select into high-engagement environments like college towns.
Conclusion
Overall, movers to school counties show a noticeably larger increase in voter turnout compared to movers in general. This suggests that higher education institutions play an important civic role, offering structure, information, and social networks that help young voters stay engaged even after relocating. In contrast, movers who do not enter college counties experience a much smaller turnout boost, reinforcing the idea that mobility alone can disrupt civic habits. Taken together, these findings indicate that college communities may serve as stabilizing anchors for democratic participation among young adults. As mobility continues to shape the experiences of this generation, understanding the civic impact of educational environments becomes increasingly important.
References
Meredith, M. (2009). Persistence in political participation. Quarterly Journal of Political Science, 4(3), 187–209.
Niemi, R. G., & Junn, J. (1998). Civic education: What makes students learn. Yale University Press.
Squire, P., Wolfinger, R. E., & Glass, D. P. (1987). Residential mobility and voter turnout. American Political Science Review, 81(1), 45–66.
Wolfinger, R. E., & Rosenstone, S. J. (1980). Who votes? Yale University Press.
Hillygus, D. S. (2005). The missing link: Exploring the relationship between higher education and political engagement. Political Behavior, 27(1), 25–47.