A University of Potsdam field experiment found that following a stranger’s Instagram profile for just three days measurably reduced prejudice among German users toward people of Middle Eastern descent, with effects spilling over to improved attitudes toward LGBTQIA+ individuals.
The study, published in the Journal of the Association for Information Systems, provides the first randomized field evidence that the “contact hypothesis”—a cornerstone of social psychology traditionally tested in physical spaces—operates effectively inside algorithmic feeds, according to Wissenschaft.
Three Days, Four Profiles, One Algorithm
Researchers Georg Voronin, Olga Abramova, Stefan Stieglitz, and Chee-Wee Tan constructed four fictitious Instagram accounts. Two bore German-sounding names; two bore Arabic-Turkish-sounding names. All posted identical, neutral content—photos of house facades, parks, everyday scenes. No political messaging, no identity discourse, just ambient daily life.
Two hundred eighteen German Instagram users were randomly assigned to follow one profile for 72 hours. Pre- and post-intervention surveys measured warmth and competence perceptions, behavioral intentions (active/passive harm vs. facilitation), and attitudes toward a secondary out-group not represented in the feed: LGBTQIA+ people.
“The Instagram contact led members of an ethnic group—in our case Germans—to evaluate another ethnic group—here people of Middle Eastern origin—as warmer and more competent afterward.”
The shift was small to medium in effect size, comparable to face-to-face contact studies.
Spillover Beyond the Feed
The most striking finding emerged in the spillover analysis. Participants who followed the Middle Eastern–presenting profiles showed reduced prejudice toward LGBTQIA+ individuals—a group entirely absent from the experimental manipulation. This “secondary transfer effect” has been documented in offline intergroup contact literature but, as the idw-online.de report notes, “for digital contexts there was hardly any evidence until now.”
The study frames this as social media-enabled intergroup contact (SMIC), a theoretical advance distinguishing passive algorithmic exposure from active, user-initiated following behavior.
Design Choices That Mattered
Several methodological decisions sharpen the causal claim. Random assignment to profile condition rules out self-selection bias—users who choose diverse feeds may already be less prejudiced. The neutral content controls for narrative persuasion; the effect stems from mere exposure to a normalized out-group member, not counter-stereotypical messaging. The three-day window captures immediate malleability, though durability remains untested.
“How long the positive effect lasts is still unclear,” wissenschaft.de reported, citing the authors’ own caution. The study also relies on self-reported attitudes, not observed behavior—a standard limitation in prejudice research.
Platform Architecture as Policy Lever
The findings arrive as the EU’s Digital Services Act (DSA) enforces systemic risk assessments for very large online platforms. Co-author Stefan Stieglitz argued that regulators should not focus exclusively on mitigating harms—hate speech, disinformation, polarization—but also on “the as-yet largely untapped potential to promote social cohesion, if digital spaces are designed responsibly.”
This reframes algorithmic design: recommendation systems that optimize for engagement often amplify homophily and conflict. But a “contact-aware” ranking signal could surface cross-cutting content without coercion.
Boundaries of the Effect
The authors are explicit: “A staunch supporter of right-wing extremist groups will probably not be converted by Instagram either.” The effect sizes (Cohen’s d ≈ 0.2–0.4 across primary outcomes) indicate meaningful but modest shifts. They do not replace structural interventions—education, anti-discrimination law, economic integration—but they reveal a low-friction, scalable adjunct.
Funding came via the CYLENCE project (grant 13N16636) from Germany’s Federal Ministry of Education and Research. The full paper is available at Journal of the Association for Information Systems.
The University of Potsdam, Brandenburg’s largest university, was formed in 1991 through the amalgamation of the Karl Liebknecht College of Education and the Brandenburg State College alongside several smaller institutions. Its origins in teacher education continue to shape its academic profile, which now spans more than 150 degree programs across three main campuses.
Parts of the university occupy the New Palace of Sanssouci, a UNESCO World Heritage site, while the Golm campus hosts one of the region’s largest science parks alongside three Max Planck Institutes and two Fraunhofer Institutes. In 2009, the university won the “Excellence in Teaching” initiative of the Stifterverband für die Deutsche Wissenschaft.
“Our data show clearly that social media can reduce prejudice,” Voronin said. “The exciting question is why the functions that could do this have hardly been used so far.”
Worth a look
- NEOWIZ Reveals New Trailer and Early 2027 Release Window for Chasing the Light of Doomsday
- Small Group of Isolated Awá-Guajá Briefly Emerges in Brazil, Asks for Food
- RBA's Fifth Hike Adds About A$94 a Month on A$600,000; Three Voted to Hold (daybreakwire.com)
- Letterkenny University Hospital Restricts Visiting Over Flu and COVID-19 Outbreaks in Six Wards (news-usa.today)