I've found the article referenced below to be enlightening as to why the Bernoulli explanation is wrong. It's quite readable. I'm not an expert in this field, so I count vouch for its accuracy. Enjoy.

http://jef.raskincenter.org/published/coanda_effect.html