An estimated $10bn was spent on advertising for the 2024 US election cycle. Of this, the vast majority – around 70% – was spent on television ads. But the question is: does this advertising actually drive changes in voter behaviour? We look at a study from The Quarterly Journal of Economics, which tests the hypothesis using alternative data from three election cycles.

The authors test if the amount of partisan TV advertising affected voter turnout and voter choice across three US presidential election cycles (2004-2012).

To test this hypothesis, they exploit variations in the number of ads placed in neighbouring counties. Due to US TV broadcasting laws, neighbouring counties can receive a completely different subset of local TV channels.

The authors compare ~6k distinct combinations of neighbouring counties, analysing data for ad occurrences and voter behaviour.

The authors use advertising data from Nielsen Media and Vivvex CMAG (formerly Kantar), and compare this to official voting results.

The authors find...