For several years, Ken Goldstein and his team at the Wisconsin Advertising Project have been compiling and analyzing data on televised campaign ads. The distinguishing characteristics of their work are their coding of the “tone” (positive, negative, or contrastive) of “virtually every political advertisement broadcast in the top 75 markets in 2000 and in the top 101 media markets from 2001 to 2004,” and their incorporation of data on the frequency with which each commercial was actually aired in each area. Fortunately for the rest of us, the Wisconsin team has made their data sets publicly available, so others can use them in their own research (as David Park and I are doing in a soon-to-be-completed paper about which we’ll be posting). Now, after completing many papers and articles, the Wisconsin team has brought out Campaign Advertising and American Democracy, a new book in which they present a wide array of results — some old and some new, some consistent with conventional wisdom about campaign advertising and some not. In coming posts, I’ll discuss key findings from this volume. For now, just a couple of nuggets from an early chapter:
* Ads sponsored by parties and interest groups are far more likely to be negative than are those sponsored by the candidates themselves.
* As campaigns have gotten underway longer and longer before election day, the common practice of using Labor Day as the opening bell in research on campaign advertisng has become increasingly problematic. In 2004, there were more than half a million airings of ads in the presidential race alone before Labor Day.
Stay tuned for more on this study.