What actually gets people to vote.
Voter mobilization has the largest body of randomized experimental evidence in political science. The finding that social norms — showing people their neighbors' voting history — produce the largest turnout effects ever documented has become one of the most cited and most ethically contested results in behavioral research.
7
experiments
7
positive results
0
null or negative
6
replicated
Key Findings
01
Social pressure — showing voters their neighbors' turnout records — produces the largest GOTV effect ever documented, at 8 percentage points.
The 2006 Michigan Social Pressure Experiment mailed 344,000 voters one of four letters: civic duty appeal, self-monitoring (own past voting), neighbor comparison (neighbors' past voting), or control. The neighbor comparison condition increased turnout by 8.1 percentage points — the largest effect ever documented in a randomized GOTV experiment. The civic duty condition increased turnout 1.8 points; self-monitoring alone, 4.9 points. The finding has been extensively replicated, spawning a cottage industry of social norm mailers by political campaigns.
02
Structural reforms — vote by mail, automatic registration — produce larger and more durable turnout effects than behavioral interventions.
Oregon's vote by mail experiment — which moved the entire state to universal mail voting — increased turnout by 2–10 percentage points depending on the election type, with the largest effects in local and primary elections where turnout is typically lowest. Automatic voter registration studies find consistent increases in registration rates and modest turnout effects. These structural changes require no repeated intervention; the benefit compounds across election cycles.
03
Text message and SMS reminders reliably increase turnout but with small per-contact effects that require large scale to matter.
Pooled analysis across six US text message GOTV experiments found consistent effects of 0.3–0.9 percentage points per message. At small campaign scale the effects are negligible; at large scale (statewide programs reaching millions) the effects translate to tens of thousands of additional votes at a cost well below that of door-to-door canvassing. UK voter registration SMS experiments produced similar small but consistent effects on registration completion.
All Experiments in the Registry
Voter Registration Postcards to Unregistered Eligible Citizens
positiveUnited States · 2019
Voter Registration SMS — Local Authorities
positiveUnited Kingdom · 2022
Automatic Voter Registration — Oregon
positiveOregon, United States · 2016
Michigan Social Pressure Voting Mailer
positiveMichigan, USA · 2006
Oregon Universal Vote by Mail
positiveOregon, USA · 2000
Door-to-Door GOTV Canvassing — New Haven
positiveNew Haven, CT, USA · 2000
Text Message Get-Out-the-Vote Campaigns
positiveCalifornia and multiple US states · 2011
What the Evidence Cannot Yet Tell Us
Do social norm mailers produce backlash effects? Some recipients report feeling their privacy was violated. The long-run effects on trust in civic institutions are unknown.
What is the durability of turnout effects across elections? Most experiments measure a single election; habitual voting may compound initial effects or they may fade.
Do GOTV interventions increase turnout by mobilizing non-voters, or do they primarily increase repeat voting among already-likely voters?
How do different vote-by-mail implementations compare? Oregon's system differs significantly from states that require an excuse for absentee voting; the effect size varies substantially.