When campaigns call, who answers? Using observational data to enrich our understanding of phone mobilization

For decades, campaigns have used phone calls to move voters to the polls. Political scientists have made great strides using field experiments to study whether campaign calls effectively increase turnout. However, due in part to limited access to observational data, some of the most basic questions about this mobilization strategy have gone overlooked. In this paper, we take a step back to provide a rich descriptive analysis of a novel dataset of millions of campaign phone calls made in California during the 2016 election. We use this dataset to shed light on three important questions: Whom do campaigns call? When campaigns call, who answers? Are those who answer more likely to turn out to vote? Our analysis reveals patterns consistent with previous theories, but also sheds light on new patterns. For example, we find that about two-thirds of campaign calls are to landlines, but those who are called on a mobile phone are twice as likely to answer. We conclude by using a matching analysis to examine the relationship between answering the phone and turning out to vote. We find that those who answer the phone are 5.9–6.8 percentage points more likely to turn out to vote. The rich descriptive analysis included in this paper provides empirical validation of prior theories of campaign mobilization, and opens avenues for future field experiment research.