Verified by 
We've all experienced the barrage of political ads on television in the final days of important and competitive elections. Last year, here in California, it was literally impossible to turn on the television during the primary and general elections and not see a political advertisement --- with ads for the gubernatorial, U.S. Senate, statewide candidates, and ballot measures filling up nearly all of the available advertising space on commercial television stations.
But do voters pay attention? Does this barrage of political ads influence the outcome of an election?
Candidates and political consultants think the answer to both questions is yes. For example, candidates running for office in big states like California pump amazing amounts of money into their television advertising budgets. Take a look at Meg Whitman's campaign expenditure reports, available online. You'll see entry after entry for multimillion dollar expenditures to organizations that are producing and placing media buys for her campaign --- just like all other candidates running for office in statewide elections in California.
And it's not just candidates running for statewide offices, even local candidates are on television. Recently we had a contested city council election in my home city of Pasadena, and in that race our incumbent city councilmember produced and aired a television ad in his re-election bid, and this was an election in which about 4,000 votes were cast.
But many political scientists have questioned the extent to which television advertising --- indeed, pretty much any type of campaigning --- changes voter perceptions and election outcomes. The argument that political campaigns "don't matter" has a long and distinguished pedigree in political science. One of the earliest and most important books in the study of political behavior, an analysis of the 1940 presidential election by Lazarsfeld, Berelson and Gaudet, argued that voter's decisions were largely consistent with a small set of demographic characteristics, and that few voters seem to have been swayed by the information disseminated during that campaign.
This argument was one that I took on in my first book, "Information and Elections." There I argued: It isn't surprising that voters don't change their minds during presidential elections. I also provided substantial evidence showing that voters learn important information in a presidential election and that the information they pick up matters for their behavior.
But most of this research, ranging from the early work of Lazarsfeld, Berelson and Gaudet to my own, used surveys and methodologies that social scientists call "observational" (in other words, non-experimental research). That has a number of implications for the scientific study of how campaigns influence political behavior, especially establishing cause and effect.
But in a novel study that was recently published in the American Political Science Review, Alan Gerber, James Gimpel, Donald Green and Daron Shaw were the first to implement a sweeping experimental study of the effects of paid political advertising. In their paper, "How Large and Long-lasting Are the Persuasive Effects of Televised Campaign Ads? Results Form a Randomized Field Experiment," they were able to deploy about $2 million in television and radio advertising experimentally in a gubernatorial election in 2006. A gubernatorial campaign was willing to randomly assign television and radio advertising in most of the media markets in the state, and Gerber et al. were then able to assess the effects of advertising using traditional survey techniques.
Gerber et al. find in their data that the television ads have a strong, but short-term, effect on voter opinions. The authors focus their discussion of these results on their theoretical implications for the study of political behavior and campaign strategy, suggesting that their results argue for a perspective where advertising primes voters to think about candidates in different ways. While that will be the subject of much debate once other scholars have read and begun to test these ideas in other ways, these results have important substantive implications.
If media advertising campaigns have strong but transitory effects, this suggests that in a highly competitive campaign environment a candidate will need to engage in a costly media campaign just to keep up with their opponent. This is consistent with what we typically see in such races, and is quite likely why candidates spend such enormous sums of money on advertising.
The application of experimental techniques to study the possible effects of campaign advertisements on voter perceptions and behavior is an important development. We'll need to see more such studies, done in a broader range of campaign contexts, to get a better understanding of how ads work and why candidates spend so much money airing them.
