Introduction
One of the most memorable aspects of the 2004 Presidential election was the
Swift Boat Veterans for Truth campaign, which questioned John Kerry’s stated record
during and immediately after the Vietnam War. The 527 organization spent $22,565,360
(OpenSecrets.org 2005) producing and airing nine television commercials, one book, and
a DVD documentary. Prior to the first Swift Boat ad running on August 4
th
, nearly every
poll had John Kerry ahead of George Bush by a couple of points. By the time the last
round of advertisements were aired on October 13
th
, John Kerry was several points
behind George Bush in every major poll (Realclearpolitics.org 2005). This negative
advertising campaign was widely cited in the media as a major cause of John Kerry’s fall
in the polls.
Conventional wisdom holds negative advertisements accomplish two feats
simultaneously: 1) negative messages drive down support for the target of the
advertisement, and 2) turnout declines among supporters of the target. Colorful
anecdotes about negative advertisements are certainly memorable, but it is far from clear
how successfully negative messages accomplish these two goals. This paper draws upon
two randomized field experiments to shed light on this issue.
The tone of campaign messages and voting behavior is one of the most researched
topics in the fields of political science and communications. Despite intense effort, the
research has proffered a spate of inconsistent findings. Different observational studies
provide evidence for all of the three possible relationships between negative
advertisements and voter turnout: negative messages decrease turnout (Ansolabehere, et
al. 1994; Ansolabehere, Iyengar, and Simon 1999; Kahn and Kenney 1999), increase
2
Convention