Political Advertising and Public Opinion: A Study of the 2016 Presidential Primaries

Abstract

Using a newly released public data set of all television advertisements run in early primary states of Iowa and New Hampshire for the American 2016 presidential election, I dynamically estimate the return of both positive and negative advertising on a candidate's standing in polls. This study uses a first-of-its-kind methodology, exploiting the frequent polling of early primary states to measure the near-instantaneous effects of television advertising. Among Republicans campaigning in Iowa and New Hampshire, I find that, for any two candidates, an additional 233 positive advertisements or 440 fewer negative advertisements in the span of a week would be needed to increase relative poll numbers one percentage point. These estimates are highly statistically significant and robust to controls. In the smaller and tightly-contested Democratic race, I only find a weak, statistically insignificant result. These findings point to modest but measurable returns to political advertising, but only when large imbalances in advertising exist between candidates. These results provide further evidence against the long-debated "minimal effects hypothesis," which holds that political advertising and other campaign activities have little impact on public opinion.