This letter addresses whether attendance estimates from news or social media are biased.
Data & Methods: Utilized over 10 million individuals' cellphone location data during the 2017 Women's March protests.
Key Findings: Cellphone estimates strongly correlate with both mainstream media reports and geolocated tweet counts (including text-based and image-based methods). Inferences about protest size drawn from these diverse sources align closely.
The study demonstrates that measures of the Women's March derived through news coverage, social media analysis, or cellphone data all capture true attendance variation effectively. This finding is significant because accurate measurement helps explain policy responsiveness to protests.
