Tweeting migration: A decade of shifting sentiment across local areas of Great Britain

Public opinion towards migration increasingly shapes political discourse and policy agendas. Yet, traditional surveys lack spatial and temporal granularity needed to detect local variation and rapid shifts in sentiment. Leveraging a dataset of 925,000 tweets from 2013-2022, we examine how local contexts influence sentiment towards migration across Great Britain. Using natural language processing, we extract sentiment from tweets, linking them to demographic and socioeconomic characteristics at the district level. Our analysis reveals stark geographical divides: anti-migration sentiment is more prevalent in areas that are older, less diverse and with lower rates of higher education. We find an intensification of polarisation after the 2016 Brexit referendum, with rises in both pro- and anti-migration expression and identify short-term fluctuations in sentiment corresponding to real-world events. Contrary to group threat theory, we find no association between short-term migration and negative sentiment. Instead, long-term demographic factors – particularly the presence of migrants – are positively associated with sentiment, supporting contact-based explanations. By capturing both the structural and dynamic dimensions of migration sentiment, we advance understanding of how local contexts mediate public sentiment. These findings demonstrate the value of digital trace data in complementing surveys and offer a methodological blueprint for future research.