Background on the Effectiveness of Political Advertisements
It has become conventional wisdom among elected officials and campaign professionals that political advertisements are a key element of a successful election bid. This wisdom holds most true for elections at the national and state levels, where it is assumed that advertisements, and particularly television spots, are necessary to reach large populations of voters. Not surprisingly, increasingly large sums of money are spent on political advertisements each year.
Stephen Ansolabehere, Roy Behr, and Shanto Iyengar found in 1991 that as television became an option for campaign ads, spending grew from $140 million in 1952 to nearly $2 billion in 1988, and according to the Alliance for Better Campaigns, $1.6 billion dollars was spent on television ads by candidates, political parties and independent groups during the last election--almost double the previous record of $771 million established during the 2000 election campaign. This increase in spending was most noticeable in the race for the White House, where approximately $500 million was spent on TV ads by the campaigns, parties, and outside groups.
Not only has the amount of money spent on television commercials increased, but the share of campaign expenditures devoted to TV spots has also skyrocketed. In 1952 less than 5 percent of campaign expenditures was for television advertisements. By the 1988, election, 20 percent was being devoted to the purchase of airtime, with the share of 2004 campaign spending likely in the 40-50 percent range, according to Opensecrets.org.
While much is known about the level of effort campaigns put into political ads, there is less known about the effectiveness of these ads in altering voter preferences, largely because of the difficulty in isolating and testing the impact of ads on voters from the many different factors that shape their choices. This daunting task is made more daunting by the limitations of one of the most common tools employed in measuring voter attitudes and behaviors--public opinion surveys. In particular, self-reported measures of exposure to political advertisements are often prone to error, thus restricting the ability to make causal inferences. Put more simply, if a respondent to a survey can't accurately recall if he or she saw a particular advertisement, there is little confidence in isolating the ad's impact on the respondent's preferences in an election.
To overcome these limitations, many researchers have turned to more experimental designs where the exposure to ads is more controlled. In particular, numerous studies by Ansolabehere and Iyengar argue that campaign advertising can be effective in shaping individual preferences. Their experiments find evidence that political advertising can be persuasive, allowing "voters to develop differentiated images of the candidates, images that play an important role in shaping voting choices." Experimental approaches to the study of ad effectiveness do offer the researcher the ability to disentangle factors such as exposure to political messages and partisan attitudes through the introduction of experimental controls, but they are constrained by a number of factors.
First, the samples used for experimental designs can often be challenged on their failure to represent adequately the characteristics of the wider voting population. Faced with high logistical hurdles, researchers are often limited to running experiments on easily available groups such as college students (I plead guilty). Thus, even if one is to isolate the effect of a stimuli on participants in the experiment, broader inferences to the overall population may be problematic.
Second, the very essence of an experimental setting may limit the ability of a researcher to generalize findings. In particular, there is a concern that the laboratory setting where stimuli can be controlled is significantly different from that in which individuals usually encounter political ads. While efforts to replicate conventional "home" settings in experiments by Ansolabehere and Iyengar are quite impressive, concerns remain about the realism of experimental environments.
Additional reading |
