This is the conclusion of the Wesleyan Media Project:
bq. In an analysis of close to 900,000 airings from January 1 to October 5, 2010, the Wesleyan Media Project finds that the distribution of positive, negative and contrast ads is comparable to 2008 in proportion, if not in volume. The Project distinguishes between ads that focus primarily on attacking a candidate, ads that promote a candidate, and ads that include a promotional and attack message (termed contrast ads). Compared to 2008, attack ad percentages remain constant, with only slight changes in positive and contrast ad percentages.
Find the full analysis “here”:http://election-ad.research.wesleyan.edu/2010/10/14/release3/. See also “yesterday’s release”:http://election-ad.research.wesleyan.edu/2010/10/13/spending-on-u-s-house-and-u-s-senate-advertising-approaches-200m-in-last-five-weeks-367m-total-since-jan-1/ on spending. Ben Smith summarizes that “here”:http://www.politico.com/blogs/bensmith/1010/Three_to_two.html?showall and also “notes”:http://www.politico.com/blogs/bensmith/1010/What_the_ad_trackers_miss.html one caveat about the ad tracking.