Abstract
Data-driven research is key to producing evidence-based public policies, yet little is known about where data-driven research is lacking and how it can be expanded. We propose a method for tracking academic data use by country of subject in English-language social science and medicine articles, applying natural language processing to a large corpus of academic articles. The model's predictions produce country estimates of the number of articles using data that are highly correlated with a human-coded approach, with a correlation of 0.99. Analyzing more than 140,000 academic articles, we find that high-income countries are the subject of ∼50% of all papers using data, despite only making up around 17% of the world's population. Finally, we classify countries by whether they could most benefit from increasing their production or use of data, with the former applying to many poorer countries and the latter to many wealthier countries.