Main Results
Norms, environmental attitudes, and Green support barely move in the immediate post-survey. Bystanders give about €4.45 more to the environmental cause.
When Protest Enters Public Space
The Effects of Direct Exposure on Bystanders
Typical Perspective
Bystander Perspective
Case: Friday’s for Future
Youth-led climate movement since 2018; recurring strikes; large, peaceful, highly visible
Experimental Setup
- Three-wave field experiment (N = 262; September ’23 + March ’24 + September ’24)
- Pre-, Post- and Follow-up surveys
- Screening: unexposed, not planning to join, at least 16, fluent in German
- 50€ Compensation + 10€ for follow-up
- Cover task: count park benches, for a supposed study of street accessibility
- Placebo items on environment, immigration, and redistribution
Manipulating incidental encounters
Route and Rally at the Brandenburg Gate showing salient sensory cues during W1 (15 September 2023)
Pedestrians were randomly assigned to walk one of two routes (≈ 2.6 km):
- treatment group observed and/or heard protest activity
- control group did not see or hear any protest activity
Randomization App
People rarely walk the streets alone!
Sequential block randomization by group size (Moore and Moore 2017), assigned in real time.
Local average treatment effect (LATE)
Since not every treated participant noticed the protest, we instrument self-reported exposure with randomized assignment (2SLS).
Within-design
Y_{t_1i}=\alpha+\beta\,\widehat{\text{Exposure}}_i+\gamma Y_{t_0i}+\varepsilon_i
Between-design
Y_{t_1i}=\alpha+\beta\,\widehat{\text{Exposure}}_i+\varepsilon_i
- Clustered SE by group ID
- Results for individual waves
- Precision-weighted average effect pooled across waves
Heterogeneous effects
Attitudes stay mainly zero across groups. Donations are largest among people already predisposed toward environmental protection.
Acknowledgments
21 RAs on the ground full day for each wave. Thank you!
Noah Badenhoop, Alma Below, Jana Krischke, Felix Heimburger, Eren Aytac, Madlen Beck, Antonia Briel, Nora Chirikure, Ahmed Elebrashi, William Fernandez, Maurice Baudet von Gersdorff, Janosch Glöckner, Inga Haudan, Annina Hermes, Philipp Heyna, Lukas Kern, Ruth Meissner, Teresa Hofmann Muñoz, Lennard Naumann, Cristian-Liviu Nicolescu, Anaïs Nimubona, Abigail Peña, Sophie Randerath, Magdalena Römischer, Alena Scheinert, Luca Schenk, Claudia Schliepe, Charlotte Schlüter, Johan Schlüter, Evelyn Schneider, Cole Simon, Jonathan Stahl, Marlene Steenken, Quentin Strelli, Frederik Thieme, Lara Türmer, and Lukas Wessling.
Study Overview
Two recruitment stands in waves 2 and 3. One stand in wave 1.
Materials
Envelope with the pre-treatment survey, the route map, and a sealed post-treatment survey.
Routes, 15 September 2023
Global climate strike, about 24,000 participants.
Routes, 1 March 2024
Co-organized with a union and an NGO, about 1,000 participants.
Routes, 20 September 2024
Global climate strike, about 5,000 participants.
Did they see it?
Control routes stay clear of the protest. On treatment routes, recognizing an environmental protest is nearly universal in wave 1 and uncommon later.
Light bars: saw a protest. Dark bars: recognized an environmental focus. Control is reverse-coded, so higher bars mean a successful manipulation in both arms.
Covariate balance
Pre-treatment covariates by route assignment.
Compliance shares
Compliers, never-takers, and always-takers.
Who complies
Pre-treatment covariates across compliance groups.
Pre-registered power
252 clusters were set for 80% power at a standardized effect of 0.15.
Donation amounts
The increase is broad participation at low and moderate amounts.
Norms by wave
Wave-specific LATE for descriptive and injunctive norms.
Attitudes by wave
Wave-specific LATE for environmental attitudes.
Green support by wave
Wave-specific LATE for propensity to vote for the Greens.
Behavior by wave
Wave-specific LATE for newsletter sign-up and donations.
Norms, intent-to-treat
Effect of assignment to the treatment route, whether or not the respondent reported seeing the protest.
Attitudes, intent-to-treat
Intent-to-treat estimates for environmental attitudes.
Green support, intent-to-treat
Intent-to-treat estimates for propensity to vote for the Greens.
Behavior, intent-to-treat
Intent-to-treat estimates for newsletter sign-up and donations.
Placebo norms
Immigration and redistribution. Same LATE specification.
Placebo attitudes
Immigration and economic attitudes.
Placebo parties
Propensity to vote for parties other than the Greens.
Placebo newsletters
Newsletter sign-up for immigration and redistribution organizations.
Placebo donations
Donations to organizations working on immigration and redistribution.
Two-sided tests
Main outcomes, with two-sided tests. The preregistered tests are one-sided.
Covariate-adjusted estimates
Controls for age and left–right self-placement.
Follow-up subset
Immediate effects re-estimated on respondents who also completed the follow-up.
Random subsamples
Immediate effects in random subsamples the size of the follow-up.
Issue salience
At follow-up, treatment respondents are no more likely to name the environment as the most important problem.
Social desirability
Donation effect by a pre-treatment social-desirability index. Wave 3.
Heterogeneity
LATE by gender, age, ideology, prior Green support, and prior environmental attitudes.