In the unkindest cut of all, Dawkins even argues that the persistence of belief in God is itself an outcome of natural selection – acting perhaps on our genes, as argued by Dean Hamer in The God Gene, but more certainly on our “memes”, the bundles of cultural beliefs and attitudes that in a Darwinian though non-biological way tend to be passed on from generation to generation. It is not that the meme helps the believer or the believer’s genes to survive; it is the meme itself that by its nature tends to survive.Now, goodness knows I'm sure that anyone who has spent much reading work by materialists has heard this argument before. And there are definately some elements to the "meme" theory which seem to have explanatory power. Other elements, however, strike me as rather lacking.
For instance, the persistence of belief in a particular religion is naturally aided if that religion teaches that God punishes disbelief. Such a religion tends to survive if the threatened punishment is sufficiently awful. In contrast, a religion would have trouble keeping converts in line if it taught that infidels are subject after death to only a brief spell of mild discomfort, after which they join the faithful in eternal bliss. So it is natural that in traditional Christianity and Islam, disbelief became the ultimate crime, and Hell the ultimate torture chamber.
It seems that one of the primary appeals of the "meme" concept (as exemplified in the above quote) is that it provides an handy explanation of why other people believe things contrary to what the speaker believes to be the truth. Thus, when Dawkins calls religion a "mental virus", he is explaining away its existence by stating that the reproductive powers of the meme (threatening its host mind with damnation if they don't believe and pass on the belief to others) outweight both the meme's falsehood and its alleged detrimental effects upon the host mind.
Similarly, Weinberg above theorizes that if a religion threatens sufficiently bad consequences for unbelief, mostly people will simply believe it rather than risk the consequences. I think this glosses over what is for most people the main criteria for whether or not they believe something: Whether the belief fits with reality as they see and feel it.
Convinced materialists who try to use meme theory to explain religion tend to weight that factor in belief selection pretty lightly, since from their point of view religious belief seems to bear no relation to reality. And yet, if the oft-cited church of the Flying Spaghetti Monster were to start threatening unimaginably horrible tortures to be inflicted upon all those who do not believe in the noodliness of their deity, they wouldn't find their membership increasing one bit. That's because, the vast majority of people (except for a group of atheists so tone deaf to any kind of supernatural beliefs that they all seem equally silly) the claims about the Flying Spaghetti Monster do not seem to bear any relation to reality, while the claims of the historic monotheistic religions do, to varrying extents.
Thus, it seems misplaced to me to cite the degree to which non-believers are predicted to suffer for their non-belief as a primary factor in the spread or persistence of religion.