Beyond Surveys: Leveraging Real-World Events to Validate Behavioral Measures of News Exposure

Do people learn about the political world through online media? We address this question by exploiting an unexpected global news event—the Russian invasion of Ukraine—and by developing exogenous measures of media exposure based on three months of web-tracking data from five advanced democracies. Our analysis differentiates between visits to general news domains and visits to politically relevant and Ukraine-specific articles, with the latter identified using machine learning techniques. We validate these exposure measures through multiple approaches, including their capacity to predict knowledge about the Russian invasion. Our findings underscore the importance of granularity: visits and time spent on Ukraine-related articles emerge as the only significant predictor of surveillance knowledge, while broader measures, such as domain-level visits, show no significant impact when controlling for self-reported exposure and other key predictors. We conclude by discussing the substantive and methodological implications of these results.