Interesting file, thanks, but I think you misread some of the data. In the self-described religious affiliation numbers, there's actually no data for atheists in 1990. Atheists were included in the agnostic totals, making differentiation between them impossible. It wasn't until 2001 that they had any data for atheists. There is a marked increase between 2001 and 2008, but also more respondents, so looking at percentages is better.
2001 - 4.3%
2008 - 7.1%
Also of some interest is this: Percentage of no religion specified (this includes atheist, agnostic, and "no religion"😉
1990 - 8.17
2001 - 14.17
2008 - 14.97
An increase, yes, though not a gigantic one.
The most interesting thing from that data has nothing to do with non-religion. For example, Protestantism got walloped. Evangelicals and Non-denominational Christians experienced huge increases. Judaism was one of the few to experience a decline in totals both times despite more respondents. And a few others of note.
I also tend to slightly inflate non-religious numbers in my mind, and deflate religious ones, because of what I call "census Christians." Those who will check the "Christian" box despite never attending church, never really praying or worshiping, and generally leading secular lives. The label is just a social identifier, either out of indifference, habit from upbringing, or specifically to be identified with a group so as not to be outcast. I can't imagine anyone checking "non-religious" or "atheist" if they aren't, but I can with religion.
...
However, my point for the last couple pages has been the social perception of atheists, not numbers. Muslims are a far greater percentage of the world's population than atheists, but the perception of them is almost universally below other religions. I love reading through that kind of information, but it doesn't change what I've posted. Any cursory increase to the number of atheists isn't the catalyst that will change public perception of them. Muslims or Tea Party'ers might have overtaken atheists in the "least trusted" category for now, but the larger point is that none of them are particularly trusted in the cultural zeitgeist.
And as interesting as that data is, it's at best 3 data points. Identifying long-term trends instead of random fluctuation or short-term responses will take much more than what the data currently offers.
Thank you though, I'm bookmarking that.