As an early Christmas treat, Mona Krewel (Victoria University of Wellington) will present her current work on targeted social media adverts deployed during the 2024 campaign in our colloquium. All interested parties are welcome!

When: 18 December, 12-2pm

Where: GFG 02-617

Abstract: The Anatomy of Influence: Unveiling Predictive Patterns in Targeted Electoral Advertising on social media in the 2024 US Presidential Election

This paper examines the influence of data-driven campaigning by analysing the use and visibility of targeted electoral advertising on social media during the 2024 U.S. Presidential Election. Despite widespread concerns about voter manipulation, privacy, and polarization, systematic empirical research on campaign microtargeting strategies remains scarce. To address this gap, the study uses the Meta Ad Targeting dataset, linked with the DIME and OpenSecrets databases, to provide an in-depth analysis of over 740,000 verified political ads from 4,271 party-linked advertisers between November 2022 and November 2024. The analysis proceeds in two stages. First, we descriptively identify demographic, geographic, and behavioural targeting strategies employed by the campaigns and examine how these vary by ideological orientation and spending level. Second, predictive models—a logistic regression and a decision tree classifier—estimate which targeting and spending features best predict ad visibility. 

Findings reveal that while both campaigns engaged in extensive microtargeting, their targeting strategies are sometimes far less sophisticated than widely assumed. Ad spend turns out to be the most powerful predictor of ad visibility, suggesting that financial capacity, rather than a nuanced targeting strategy, largely determines exposure. By integrating large-scale digital trace data with campaign finance information, this paper advances the empirical understanding of how political influence is exercised through social media advertising. It also highlights methodological challenges inherent in studying microtargeting based on the data provided by the platforms and the limitations of impression-based measures of impact.