Laboratory

Migration and Mobility

At a Glance Projects Publications Team

Project

Analyzing Social Media Discourse and Spatiotemporal Patterns in Attitudes Toward Migrants

Asli Ebru Sanlitürk, Carolina Coimbra Vieira, Jisu Kim, Abigail Tun Mendicuti, Emilio Zagheni; in Collaboration with Toomet Ott Siim (University of Tartu, Estonia), Neal Marquez (University of Washington, Seattle, USA), Weber Ingmar (Saarland University, Saarbrücken, Germany), Garimella Kiran, Sudhamshu Hosamane (both: Rutgers, The State University of New Jersey, New Brunswick, USA), Kyle Wang (University of Delaware, USA), Clara H. Mulder (University of Groningen, Netherlands), Kyoung-In Baik (Sungkyunkwan University, Korea, South), Jiho Kwak (Seoul National University, Korea, South)

Detailed Description

Migration is a significant and widely debated issue, attracting sustained attention from both political and social perspectives. The prominence of this issue has increased in recent decades due to large-scale displacement, heightened political discourse, and the spread of media narratives and propaganda surrounding immigrants. In the contemporary world, a considerable share of public discussion occurs on online platforms. These digital spaces provide researchers with a valuable source of data, offering timely insights and enabling the observation and analysis of dominant patterns of discourse, as well as of emerging trends in attitudes toward migrants and refugees. Understanding these patterns and trends supports the use of evidence-based approaches in migration studies and contributes to the development of more informed and adaptive policy regarding public attitudes toward migration.

The studies conducted under this project use different sources of online discussions and often merge different types of digital trace data for their analyses. The microblogging platform Twitter (now X) is one of these sources. In one study, Twitter data have been used together with anonymized cellphone call records made available by the “Data For Refugees (D4R)” challenge. Focusing on interactions between Turkish natives and Syrian refugees, it combines measures of spatial and social segregation derived from mobility and sentiment expressed on social media. Findings show that during periods of strongly negative online discourse, Syrian refugees are more socially isolated from the host population. These findings highlight how offline social dynamics and integration are closely linked to fluctuations in online discourse. A second study has combined geolocated Twitter data with data on the number and average tone of online news articles from the Global Database of Events, Language and Tone (GDELT) and census and survey data to examine attitudes toward migrants in transit; in this case, toward Central American migrant caravans. Using topic modeling, sentiment analysis, and regression models, the study shows that while similar themes emerge across countries, the drivers of sentiment intensity differ, particularly between the United States and Mexico. These findings underscore the role of national context and media environments in shaping migration-related attitudes.

Adopting a similar data approach, two studies focus on the online discussions on the emerging microblogging platform Bluesky. The first study analyzes country-level heterogeneity in migration discourse on Bluesky using posts from the United States, the United Kingdom, and Germany. By applying topic modeling, sentiment analysis, and word embeddings to identify thematic and linguistic differences across contexts, the study reveals substantial cross-national variation in discourse, with consistently higher levels of negative sentiment and distinct country-specific narratives. The second study examines how political disruption shapes the migration discourse online on Bluesky in response to major US policy shifts associated with Donald Trump, including debates on deportation. Combining large-scale topic modeling with multi-LLM stance annotation, it investigates dominant policy topics, their volatility over time, and cross-linguistic variation in discourse.

Moving beyond public perceptions of migration on social media, the fifth study adopts a causal experimental approach to examine how visual cues influence attitudes toward migrants. Using a conjoint design with AI-generated images that systematically vary attributes such as race, religion, and appearance, it aims to measure perceptions of citizenship, legal status, deportation deservingness, and willingness to report individuals to authorities. This approach will provide causal evidence on how stereotypes and visual markers shape migration attitudes, offering insights beyond traditional text-based conjoint designs.

Publications

Baik, K.; Kim, J.:

Analyzing the discourse around Russo-Ukrainian war in Germany: understanding variances in public stances

SocArXiv papers. unpublished. (2025)    

Baik, K.; Kim, J.:

Analyzing the discourse around Russo-Ukrainian War in Germany: understanding variances in public stances

In: Advances in social networks analysis and mining: proceedings of the 17th International Conference on Advances in Social Networks Analysis and Mining - ASONAM 2025, 315–331. Cham: Springer. (2025)    

Tun Mendicuti, A.; Kim, J.; Mulder, C. H.:

Understanding opinions towards migrants in transit: an analysis of tweets on migrant caravans in the US and Mexico

In: WebSci '24: proceedings of the 16th ACM Web Science Conference, Stuttgart, Germany, 22-24 May 2024, 1–10. New York, NY: Association for Computing Machinery (ACM). (2024)    

Marquez, N.; Garimella, K.; Toomet, O.; Weber, I. G.; Zagheni, E.:

Segregation and sentiment: estimating refugee segregation and its effects using digital trace data

MPIDR Working Paper WP-2019-021. (2019)