The statistic of low Firefox use is based on accessing US government websites. Could it be that there is significantly LESS government site access by the population of users that prefer Firefox? As a corollary I recently read that game companies observed significantly HIGHER bug reporting from Linux users on Steam, not because there were more Linux-related bugs, but simply because that set of users were more likely to initiate bug reports. Of course Firefox is not Linux and Steam is not the world, but a statistic from a relatively narrow segment of the internet should not be assumed representative of the whole.