Over dinner one evening at a Philadelphia restaurant, following a daylong Internet conference at the University of Pennsylvania, we shared with each other our individual plans for future research. Each of us had been studying and writing about new media in politics for several years. We found that we were both interested in studying the role of the Internet during the 2000 campaign. One of us had plans to content analyze candidate Web sites to examine how candidates presented themselves to voters. The other wanted to survey voters to determine how they react to candidate messages online. Hearing the other's plans, we each realized that our own work would provide just half of the picture of the Internet in elections. We decided to join forces. Our aim would be a fuller picture: how candidates use Web sites and how voters react to them.
Both of us had been frustrated at the paucity of systematic, detailed evidence about Americans' online behavior. Especially in the early years of the Internet, scientific data about voter use of Web sites were largely unavailable. The best available clues came from logs of