I don't completely agree with Haidt on all topics...but he's reaching the pont of being one of the best synthesists around...and I'm more supsicious when he disagrees that I might be wrong.
Having read another 10% of Haidt's book...he basically covers ground that is well-understood by readers of this blog...but more clearly than I do. The core, meta-idea underlying much of his book, and this blog...is that there are a lot of very good reasons for people to believe things that have *nothing* to do with the truth value of the positions. Rather evolutionary success rates of various cognitive predilections are the key indicator regarding what people believe...and *not* what is true.
The sixth tenth of Haidt's book looks at the research on religion...works with it inside his evolutionary understanding of ethics...and points out that the religious experience of a UVA football game, a church service, an extasy-fueled rave, and an aztec human sacrifice appear to line up real well. They're all effective group-bonding experiences...and indeed the social science on religion works out real well to confirm this analysis. Religious folks are effectively indistinguishable from non-religious folks on every test and metric (not involving talk) we can find...with one big exception: In-group cooperation. Religious folks cooperate with other coreligionists better than do athiests.
Key takeaway: Don't assume that beliefs and truth are...or should be...about truth. Rather, remember that we are evolved creatures with minds designed to win in the highly groupist evolutionary environment. Truth is a luxury.