This is interesting. Sadly, as an opt-in poll, it suffers from the most obvious and extreme selection bias, and raises more questions than it answers.
Does it suggest that atheists are most likely to use twitter? Or perhaps that, without religion, we look elsewhere for silly memes, and thus are more susceptible to polls and the like? Perhaps this higher memetic availability explains why more atheists might be using twitter in the first place. Perhaps it’s really just a measurement of religiousity in the Bay Area and a few other places where twitter is popular.