linear - 8:44 pm on Jul 19, 2005 (gmt 0)
It's a real hair fire to constuct a rigorous controlled experiment involving adsense. I'd think the only way to do it would be to set up two channels, then show channel A to half your visitors (maybe use the last octet of the IP address modulo 2?) at frequency F1 while you show channel B ad units that are otherwise identical at frequency F2. If you do it consecutively, changes in the ad mix over time can foul your results up.