Several years ago, Gregory W. Lester wrote an article for the Skeptical Inquirer entitled Why Bad Beliefs Don't Die. John Loftus has posted a representative piece from the article, which is still pertinent.
Also of interest is this one comment from the post:
The great thing with the scientific method is that its self-correcting mechanism would clear it (referring to a previous comment) in a short term. The problem with religion is the necessity to discard logic and reason from the beginning. It does not even start to work. As papalinton posted, if you start with a false and absurd premise, reason and logic may even appear to work for it.