What is really worrying in Dawkins' approach, however, is that he himself makes a very dubious use of evidence.  He challenges various religious interviewees for evidence, but then neglects evidence for his own more eccentric claims.

The very choice of interviewees undermines his whole approach.  He disregards an elementary point of scientific method.  For a good survey, observation needs a random sample not a biased one.  Dawkins never applies this fundamental principle here.  He seems to have first decided what to prove and then chosen interviewees to confirm his hypothesis.  He never bothers to consider contrary evidence, even though he recalls his former professor whom he commends for doing exactly that.

Is it possible that Dawkins couldn't find individuals who represent a healthy view of religion? In Christianity, for instance, did he ever consider prominent people like Mother Teresa of Calcutta, who spent her life helping the poorest of the poor precisely because of her deep religious faith?  He did not.

Did he ever think of the innumerable, unnoticed, under-estimated priests, nuns, and lay people who dedicate their entire lives to bringing health, hope, and joy to countless families in the most derelict shanty towns imaginable, precisely because they are driven by faith?  He did not.

Instead, he tried to convince television viewers that the few exceptions he interviewed justify a general claim: that religion is bad for society.

One would expect better reasoning from a prominent scientist.