Regular readers of COTTonLINE know that we follow the polling work of Scott Rasmussen as reflected in the Rasmussen Report. You may also have noticed that the Rasmussen numbers tend to be somewhat less positive for the President than some other polls.
This article in The Atlantic explains why Rasmussen's poll results may be so, and does so without either supporting or criticizing them. The factors are: automated polling, giving respondents four choices rather than two, and selecting individuals to be polled based on likely voting.
Automated polling may provide a way around the "Bradley effect," a reluctance or fear some white voters have to admit to an interviewer that they do not plan to vote for a black candidate.
I'm not sure what gives with four choices. Clearly, depending on how Rasmussen selects his likely voters, he might tend to underpoll likely Obama voters. I guess the question is this: how did his numbers predict the 2008 election?
