https://www.sciencedaily.com/releases/2020/09/200902152157.htm

The study, published Sept. 2 in the journal Science Advances, measured the persuasive effects of 49 high-profile advertisements from the 2016 presidential campaign on a nationally representative sample of 34,000 people through a series of 59 randomized experiments. Expanding on prior research suggesting that political ads have little impact on voters' preferences, the study shows that those weak effects are consistent irrespective of a number of factors, including an ad's tone, timing, and its audience's partisanship.

"There's an idea that a really good ad, or one delivered in just the right context to a targeted audience, can influence voters, but we found that political ads have consistently small persuasive effects across a range of characteristics," said Coppock, an assistant professor of political science in the Faculty of Arts and Sciences. "Positive ads work no better than attack ads. Republicans, Democrats, and independents respond to ads similarly. Ads aired in battleground states aren't substantially more effective than those broadcast in non-swing states."