The paper is based on a unique data set gathered by the authors. We used video and spending data on presidential advertising in all 210 media markets, and advertising for the Senate and Congress in Minnesota, on television and on radio to allow a large-sample, systematic analysis of the content, visuals and sound used in political advertising in 2008.
Content: we checked the accuracy of all the claims made in more than 500 different ads. This included sourced claims, such as the way a candidate had voted on a particular piece of legislation, and claims such as "is a leader in Congress", where we checked how many bills the candidate initiated or simply signed on to. Visuals: we noted the presence or absence of distorted candidate images, mise-en-scene, use of different montages, any text on screen, and the use of graphics. We also coded the use of different visual distortions. Sound: we noted features of the ads such as the presence and nature of music cues, characteristics of voiceovers such as gender, and whether it was the candidate or a third party.
Our aims with these data were to examine: 1. The extent to which the claims made in political advertising are factually correct and whether this varies by tone. 2. Visual techniques in ads and the association of particular visual methods with ads that mislead. 3. Sound techniques in ads and their association with advertising tone.
We find that negative advertising is more likely to be inaccurate. We demonstrate that visual compositing is associated with ads that are factually inaccurate and that this association is independent of the ad’s tone. We find that negative ads are more likely to manipulate sound, not only than positive ads but also than contrast ads.
