Spillover Presidential Ads and Campaign Contributions in a Polarized System
13 Pages Posted: 28 Dec 2019 Last revised: 19 Feb 2020
Date Written: November 1, 2019
Estimating how campaign advertising affects contributions is a challenging problem. Urban and Niebler (2014) address this question by focusing on “spillover” zipcodes — zipcodes in non-competitive states which receive Presidential campaign ads because they share a media market with a competitive state. We revisit their results and find the relationship they find between contributions and ads is driven by mistakes in propensity score matching. With corrected matching, combined with regression, their result becomes insignificant. However, if we model contributions as a two-stage process — a decision to contribute followed by a decision on amount — we find evidence that spillover ads predict higher contribution amounts, but not the decision to contribute. We also study the separate effects of Democratic and Republican ads, and find evidence for both positive and negative partisan responses. We also find, consistent with asymmetric polarization, that Republicans respond more strongly to Democratic ads than reverse.
Suggested Citation: Suggested Citation