Danny Hayes is an assistant professor of government at American University and a contributor to Behind the Numbers.
Occupy Iowa Airwaves has begun.
The swift ramp-up of the ad wars in the Hawkeye state has prompted broad speculation about whether big ad buys might widen former House speaker Newt Gingrich’s lead, solidify Texas Rep. Ron Paul’s support, help former Massachusetts governor Mitt Romney stop the bleeding or revive Texas Gov. Rick Perry’s chances. But these ads, coming three weeks before the Jan. 3 caucuses, may be popping too early to make a difference: the effects of political advertising tend to be as fleeting as a Kardashian marriage.
A series of recent political science papers demonstrates just how fast political information loses its influence. A randomized field experiment analyzing advertising during Perry’s 2006 reelection bid in Texas found that the governor’s ads improved his standing among voters during the week they aired. But when follow-up surveys were conducted one week later, the effects had dissipated entirely. The ads worked -- they boosted Perry’s vote share -- but only temporarily.
A group of researchers from UCLA drew the same conclusion in a study of 2006 gubernatorial, Senate and House races. Again, ads helped the sponsoring candidate win over voters, even some who identified with the opposing party. But within a few days, those effects had essentially evaporated. Call it the Keyser Soze pattern: and like that, they were gone. (A study of the 2000 presidential race by the same group yielded similar findings.)
As a result, the only ads that appeared to shape voters’ choices at the ballot box were the ones run at the end of the campaign. “Politicians may engage in endless campaigns,” the authors write, “but advertising in the home stretch -- perhaps just the last day or two -- appears to have a greatly disproportionate impact.”
These studies find support from other work on the duration of communication effects. People often make judgments -- political or otherwise -- based on information they’ve recently encountered. Political scientists Dennis Chong and James Druckman conducted two studies in which they presented individuals over the course of several days or weeks with arguments on different sides of political issues. One study focused on the Patriot Act, and a second involved debates over urban growth.
When subjects were asked about their opinions on these issues at the end of the experiment, Chong and Druckman found that most people leaned disproportionately toward the argument they had heard most recently. That was true even when they had been exposed to earlier frames providing contrary points of view.
The ability of advertising campaigns or clever communication efforts to alter election outcomes shouldn’t be overestimated. If voters are exposed to similar numbers of competing ads, the effects will probably cancel out. And many voters are anchored by long-standing support for a candidate. No amount of advertising is likely to move them.
Of course, advertising can have significant effects, especially in an unsettled environment like the fight for the Republican nomination. But the research suggests that the ads potential Iowa caucus-goers are seeing this week won’t have much to do with the outcome. Instead, campaigns will get the biggest bang for the buck with a barrage of ads in the days immediately preceding the caucuses. Better late than early.
Read the full poll results