Data Download

We have built a novel dataset linking the Twitter profiles of 100,000 academics to their respective academic profiles, encompassing data from 12,000 institutions across 174 countries and 19 disciplines. The dataset captures all posts, comments, shares, metadata, and social following networks of these accounts from 2016 to 2022. We also collected similar scale data for random set of users in the U.S.

Data Collection and Analysis

Using scalable large-language model (LLM) based classification techniques, we categorize and label the content, focusing on both the substantive content (the "what") and the language and tone of expression (the "how"). Our analysis covers topics such as climate crisis, economic policy, cultural dimensions, and the tone of language used.

Download and Code (Replication Package)

Visit the Zenodo Repository to download data and code: https://zenodo.org/records/15115397 

Using the data

This work is licensed under an Apache License 2.0 [LICENSE}. The following paper has to be cited in all publications that make use of or refer in any kind to the files provided:

Garg. P & Fetzer. T (2025). Political Expression of Academics on Social Media. Nature Human Behaviour. Available Open Access 🔓.

We also kindly ask researchers using this to send us their paper, or a link to their paper. We will continuously upload or add links to these papers to inform the research community about ongoing progress and related work.