Democracy & social cohesion
Polarization, misinformation, racism and social cohesion. Real interventions, each with its sources: what worked, what didn't, and why.
10 worked7 mixed3 didn't work
What worked
- WorkedDemocracy & social cohesion
Media literacy for all ages: Finland
ChallengeCitizens exposed to online disinformation need the skills to judge sources and content.
Finland built media literacy into teaching from early childhood to upper secondary school.
1strank on Europe's Media Literacy Index every year since 2017ObservationalRead case - WorkedDemocracy & social cohesion
Bad News fake-news game: online, worldwide
ChallengeDebunking after the fact is slow; people need resistance to misinformation before they meet it.
A free online game lets players act as fake-news producers to learn common manipulation tricks.
−21%perceived reliability of fake headlines after playingQuasi-experimentalRead case - WorkedDemocracy & social cohesion
Prebunking videos shown as YouTube ads
ChallengeFact-checks cannot keep up with the volume of misleading content on large platforms.
Short videos explaining manipulation tricks were tested in lab trials and as real YouTube ads.
+5%recognition of manipulation techniques in YouTube field testRCTRead case - WorkedDemocracy & social cohesion
Fact-checks tested in Argentina, Nigeria, South Africa and the UK
ChallengeWorry that corrections fail or backfire, and that evidence on fact-checking came almost only from the US.
Fact-checks written by national fact-checking groups were tested in parallel experiments in four countries.
−0.59drop in false belief on a 5-point scale after a fact-checkRCTRead case - WorkedDemocracy & social cohesion
America in One Room: national deliberative poll
ChallengeDeep partisan division and hostility make it hard for US voters to find common ground.
A random sample of over 500 US voters spent a weekend deliberating on five divisive policy areas.
45% → 25%support for sending undocumented immigrants homeQuasi-experimentalRead case - WorkedDemocracy & social cohesion
Deep canvassing to reduce transphobia: Miami, Florida
ChallengePrejudice against transgender people, and fear of a backlash against a new anti-discrimination ordinance.
Volunteers held 10-minute, two-way doorstep conversations with voters about transgender people.
+10 ptswarmth toward transgender people, lasting 3 monthsRCTRead case - WorkedDemocracy & social cohesion
Narrative-sharing canvassing on immigration: Tennessee and California
ChallengeHostile, exclusionary attitudes toward unauthorized immigrants and transgender people.
Canvassers swapped personal stories with voters to test what makes prejudice-reducing conversations work.
4+ monthsdurability of reduced exclusionary attitudesRCTRead case - WorkedDemocracy & social cohesion
Random squadron mixing and interracial friendships: US Air Force Academy
ChallengeRacial separation in social life, even on diverse college campuses.
Freshmen were randomly assigned to squadrons where they live, eat and train together.
+30%chance a white male cadet picks a Black roommate, per extra Black squadron peerQuasi-experimentalRead case - WorkedDemocracy & social cohesion
Greening vacant lots to restore neighborhood life: Philadelphia
ChallengeBlighted vacant land that breeds fear and crime and keeps residents indoors.
Trash was cleared and grass, trees and low fences added to randomly chosen vacant lots.
-29.1%gun assaults near treated vacant lots (greened or cleaned up) in neighborhoods below the poverty lineRCTRead case - WorkedDemocracy & social cohesion
Reconciliation radio soap opera: Rwanda
ChallengeDistrust, deference to authority and intergroup prejudice after the 1994 genocide.
A radio drama about two rival villages modeled dissent, cooperation and cross-group love.
<20% vs 60%+group comments that denied mistrust: soap opera listeners vs comparison groupRCTRead case
Mixed results
- MixedDemocracy & social cohesion
Strengthening Democracy Challenge: 25 interventions tested in the US
ChallengePartisan animosity and support for undemocratic tactics are rising, but no one knew which fixes work.
A crowdsourced contest picked 25 short interventions and tested them head-to-head on 32,059 Americans.
23 of 25interventions that reduced partisan animosityRCTRead case - MixedDemocracy & social cohesion
Braver Angels depolarization workshops: US campuses
ChallengeAmericans increasingly dislike and distrust people from the other party.
Workshops modelled on couples therapy brought Republican and Democratic students together to talk.
6+ monthslasting rise in donations to depolarization groupsRCTRead case - MixedDemocracy & social cohesion
Network Enforcement Act (NetzDG): Germany
ChallengeIllegal hate speech and defamation spread on social media while platforms acted slowly on complaints.
Large social networks had to remove clearly illegal posts within 24 hours of a complaint or face fines.
11%share of NetzDG complaints Twitter acted on, early 2018ObservationalRead case - MixedDemocracy & social cohesion
TV drama to encourage reporting electoral corruption: northern Nigeria
ChallengeVote buying was widespread in Nigerian elections, but very few people ever reported it.
Anti-corruption storylines written into a hit Hausa-language TV drama before the 2023 election.
+53 ptsbelief that reporting helps, among viewers the payment got watching (imprecise)RCTRead case - MixedDemocracy & social cohesion
Blind auditions behind a screen: US symphony orchestras
ChallengeBias against women in hiring for top orchestras, which were almost all male.
Orchestras auditioned musicians behind a screen so judges could not see who was playing.
+50%chance women advanced in some rounds (contested)Quasi-experimentalRead case - MixedDemocracy & social cohesion
Mixed Christian–Muslim soccer teams: northern Iraq
ChallengeDeep Christian distrust of Muslims after ISIS displaced Christians from northern Iraq in 2014.
Christian teams displaced by ISIS were randomly given Muslim or Christian extra players for a league.
61% vs 17%Christians training weekly with Muslims 6 months onRCTRead case - MixedDemocracy & social cohesion
Mixed Christian–Muslim vocational classes: Kaduna, Nigeria
ChallengeRepeated Christian–Muslim riots and deep religious division in Kaduna city.
Young men were randomly placed in religiously mixed or single-faith job-skills classes.
849young men randomized; less discrimination, same prejudiceRCTRead case
What didn't work
- Didn't workDemocracy & social cohesion
Anti-Fake News Act: Malaysia
ChallengeThe government said false online information threatened public order ahead of the 2018 election.
Malaysia made spreading 'fake news' a crime punishable by up to six years in prison, then repealed it.
6 yearsmaximum prison term for spreading 'fake news'ObservationalRead case - Didn't workDemocracy & social cohesion
Ban the Box and racial gaps in hiring: New Jersey and New York City
ChallengeCriminal records shut many Black men out of jobs, widening racial employment gaps.
Laws stopped employers asking about criminal records on job applications.
7% -> 43%how many more callbacks white applicants got than similar Black applicants, before vs after the box was removedQuasi-experimentalRead case - Didn't workDemocracy & social cohesion
Implicit bias training for police: New York City
ChallengeRacial and ethnic disparities in police stops, frisks, arrests and use of force.
The NYPD gave its officers a one-day implicit bias course and tested it in a randomized rollout.
14,471officers in trial; no detectable change in enforcement disparitiesRCTRead case