Gary Marcus posted an article in the Newyorker concerning science, religion, and faith.
(...) Any agnostic is free to believe that his favorite religion has not yet been completely disproven. But anyone who wishes to bring science into the argument must acknowledge that the evidence thus far is weak, especially when it is combined statistically, in the fashion of a meta-analysis. To emphasize the qualitative conclusion (X has not been absolutely proven to be false) while ignoring the collective weight of the quantitative data (i.e., that most evidence points away from X) is a fallacy, akin to holding out a belief in flying reindeer on the grounds that there could yet be sleighs that we have not yet seen.This is one of several prongs that is useful in understanding religion. Religions have been looked at an enormous amount over the course of human history. Every time we have looked at them, they have failed. The accumulation of all those failures add a statistical weight to just how unlikely it is that any given religion is true. And all that is before we even weigh any claim from a specific religion.
The idea that any religion is true is so miniscule that it would take an enormous amount of verifiable evidence to swing the scales back the the idea that believing in religion was scientifically reasonable.