Research
Dissertation Papers
Should I Mail or Should I Go: Voting Behavior After a One-Time All-Postal Election
With Valentin Lindlacher · CESifo Working Paper No. 12075 (2025) · Latest version (PDF)
Abstract. Can a one-time change in election administration affect voting behavior after regular rules return? During Bavaria’s 2020 mayoral runoffs, postal voting was mandatory and ballots were automatically mailed to eligible voters. Subsequent elections returned to the usual voting rules. Using municipality-level election data from 2013 to 2025 and an event-study difference-in-differences design, we find a temporary increase in total turnout of approximately 0.4 percentage points and a persistent shift of about one percentage point from in-person to postal voting, lasting at least five years. We also document temporary changes in party support, including gains for the CSU. The results distinguish short-lived changes in overall participation from more durable changes in voting-mode choice.
After the Breakdown: Persistent Changes in Voting Modes and Electoral Participation
Single-authored · Latest version (PDF)
Abstract. Can failures in election administration affect participation in later elections? I study the aftermath of Berlin’s 2021 electoral breakdown using court-documented disruptions and harmonized precinct returns from 2016 to 2026. An event-study difference-in-differences design compares in-person, postal, and total turnout in affected and clean-control areas. In-person turnout remains lower in court-linked precincts, while postal voting increases in several subsequent elections. In the 2025 federal election, estimated in-person turnout is 1.77 percentage points lower and postal turnout 1.25 points higher; the net total-turnout difference is smaller and statistically imprecise. The evidence documents persistent differences in local voting-mode composition following the breakdown and its legal aftermath, without identifying individual switching or changes in perceived voting costs.
When Votes Are Already Fixed: Electoral Reopening and Voter Turnout
With Valentin Lindlacher · Latest version (PDF)
Abstract. Does turnout differ when part of an election result is already legally fixed? Berlin’s 2024 partial repeat of the 2021 federal election allowed voters in selected districts to replace their ballots, while valid votes elsewhere remained in the tally. We reconstruct retained first-vote totals and examine turnout in 294 repeat-affected reporting areas across twelve constituencies. Turnout declined less where a larger share of the constituency electorate could vote again. A one-percentage-point higher repeat share is associated with a 0.055-percentage-point smaller turnout decline, or 0.039 points after adjusting for an extrapolated historical differential trend. The geographic allocation of the repeat and earlier turnout trajectories limit causal interpretation. The findings concern the legally revisable scope of an election, not directly observed beliefs about pivotality.
Other Research Projects
School Reassignment and Political Adaptation
Single-authored · Work in progress
Abstract. Do people become politically more similar to communities to which they are institutionally connected, even without moving? I reconstruct public-school reassignment events in North Carolina and link historical school-attendance boundaries to longitudinal voter-registration records, geocoded residences, and neighborhood characteristics. The design compares reassigned residents with residents remaining in the original school community, relating subsequent party-registration changes to the destination community’s prior political orientation. Results suggest directional adaptation in adult party registration. Endogenous boundary changes, limited district-level variation, and incomplete validation of reassignment events mean the evidence does not yet establish a causal effect. The project examines whether institutional ties can shape political geography independently of residential mobility.
A Year of College, a Lifetime of Party? Evidence from North Carolina Voter Registrations
With Sarah Listabarth · Work in progress
Abstract. How does exposure to a college environment before a first election relate to initial party affiliation? Using repeated North Carolina voter-registration records from 2008 to 2026, we identify likely college students from address histories and infer differences in the time spent at college before their first general election. We compare students with longer and shorter predicted exposure, controlling for their home areas and using inferred family connections for additional comparisons. In the main specifications, longer exposure is associated with a 1.2–1.4-percentage-point higher probability of Democratic registration, with some persistence into later elections. The study investigates political socialization during the transition into the electorate, while recognizing the limitations of inferred college attendance and nonrandom exposure.
The Politics of Parental Protection: Sex Offender Proximity and Political Preferences
With Sarah Listabarth · Preliminary research
Abstract. How do local safety threats shape parents’ political preferences? We investigate whether exposure to nearby registered sex offenders affects political preferences, examining whether responses differ between mothers and fathers and by the gender of their children. The project explores how parental concerns about family safety and local social environments translate into political behavior.
The Destination Side of Migration: Evidence from Hurricanes
Single-authored · Preliminary research
Abstract. How do disruptions to potential destinations affect migration decisions, even among people who are not directly exposed to a disaster? Using longitudinal North Carolina voter-registration data, historical migration flows, and inundation from Hurricanes Matthew and Florence, I investigate how damage to historically important destinations affects migration from unaffected origins. Preliminary evidence suggests fewer moves from origins whose typical destinations experienced greater exposure, particularly for shorter-distance moves. I also examine whether these disruptions change who moves and the socioeconomic and political characteristics of the neighborhoods migrants select into. These latter analyses are descriptive rather than causal estimates of neighborhood effects.
Policy and Applied Publications
