A decade of GOTV impact

Unequal long-term effects of inducing young people to vote
EPSA 2025

Kasper M. Hansen

Søren Damsbo-Svendsen

June 24, 2025

Motivation

• Can GOTV efforts increase turnout and also improve participatory equality?
• Unclear what the downstream (long-term) effects are of GOTV campaigns
  • requires large field experiments in the past
  • highly demanding downstream data requirements
  • some evidence of long-term effects,
    primarily through habit (Coppock & Green 2016)

Motivation

• The (immediate) effects of GOTV campaigns have been shown to be moderated by baseline voting propensity
  • (e.g., Bhatti et al. 2018; Enos et al. 2014; Hirvonen et al. 2025)
  • Ongoing debate!
  • Key disagreement:
    • mostly low-mid (Europe) or mostly high (US)
    • high exacerbates inequality
• None have studied long-term effects on participatory inequality (low, mid, high)
• We analyze conditional long-term effects of two large Danish GOTV field experiments

Data and design

• Two GOTV field experiments targeting young voters (18-29 years)
  • 2013 n = 95,570 voters
  • 2014 n = 153,661 voters
  • total n = 249,231 voters
  • sampled randomly from among vast majority of young electorate
• Combined with all available downstream turnout records
  • 8 downstream elections
    • 2014-2024
    • local, national, EU
  • 2013 total n = 710,386 turnout decisions
  • 2014 total n = 1,152,522 turnout decisions

Data and design

• Panel attrition in downstream elections
  • 4-44 %
  • causes:
    • death and migration (rare)
    • non-eligibility (more common)
    • lack of turnout data from entire municipalities
      (common in two elections: 2014-2015)
  • no influence across elections
  • virtually no self-selection
  • results quite robust to use of restricted/balanced panels

Data and design

• Treatment
  • SMS text message sent out 0-6 days before Election Day
  • Short reminder about upcoming election + frame/tone variations
• Outcome
  • binary turnout
  • individual, validated, administrative
  • basis of actual election results

Data and design

• Moderator: baseline voting propensity
  • low, mid, high (~equal-sized)
  • based on predicting turnout in control group
  • logit model with predictors: admin sociodemographics (e.g., age, gender, social status, income, education, origin country), past turnout, same for parents
• Analysis: OLS regression
  • total effects: (essentially) difference-in-means → ITTs
    • election FEs, marginalizing over election
  • conditional effects: interaction between treatment and voting propensity
    • election FEs, marginalizing over election + voting propensity

Results

Effects of 2013 text message GOTV campaign

1. Strong immediate effects, especially on low-propensity voters
2. Different picture for downstream effects:
   • low: no downstream effects
   • mid: strong downstream effects across five elections
   • high: downstream effects in next local election
   • total: positive downstream effects in 3+ elections
3. Cumulative effect = increasing participatory inequality?

(Limited) effects of 2014 text message GOTV campaign

1. No evidence of downstream effects
2. Strong immediate effects only for low-propensity voters
3. Two findings replicate:
   1. “stairway” of immediate effects
   2. no downstream effect for low
4. Treatments possibly too weak, on average
   • evidence of total downstream effect for particular timing-tone treatment combos
   • possibly similar downstream effects conditional on voting propensity

Conclusions

• Studies of GOTV downstream effects are rare
• Studies of GOTV effects conditional on voting propensity are few
• Combination can tell us a lot about long-term benefits/risks to participatory (in)equality
• We show evidence on a decade of GOTV impact on low, mid, and high propensity voters
  • immediate effects strong(est) for low-propensity voters
  • but no downstream effects!
  • multiple downstream effects on mid-high propensity voters
  • likely, not improving equality long term – and may harm it
• Questions: (a) mechanism, (b) case + propensity definition, (c) cumulative effect

Thank you