Moving on from the elite scientists of the National Academy and the Royal Society, is there any evidence that, in the population at large, atheists are likely to be drawn from among the better educated and more intelligent? Several research studies have been published on the statistical relationship between religiosity and educational level, or religiosity and IQ. Michael Shermer, in How We Believe: The Search for God in an Age of Science, describes a large survey of randomly chosen Americans that he and his colleague Frank Sulloway carried out. Among their many interesting results was the discovery that religiosity is indeed negatively correlated with education (more highly educated people are less likely to be religious). Religiosity is also negatively correlated with interest in science and (strongly) with political liberalism. None of this is surprising, nor is the fact that there is a positive correlation between religiosity and parents' religiosity. Sociologists studying British children have found that only about one in twelve break away from their parents' religious beliefs.

As you might expect, different researchers measure things in different ways, so it is hard to compare different studies. Meta-analysis is the technique whereby an investigator looks at all the research papers that have been published on a topic, and counts up the number of papers that have concluded one thing, versus the number that have concluded something else. On the subject of religion and IQ, the only meta-analysis known to me was published by Paul Bell in Mensa Magazine in 2002 (Mensa is the society of individuals with a high IQ, and their journal not surprisingly includes articles on the one thing that draws them together).⁵⁷ Bell concluded: 'Of 43 studies carried out since 1927 on the relationship between religious belief and one's intelligence and/or educational level, all but four found an inverse connection. That is, the higher one's intelligence or education level, the less one is likely to be religious or hold "beliefs" of any kind.'

A meta-analysis is almost bound to be less specific than any one of the studies that contributed to it. It would be nice to have more studies along these lines, as well as more studies of the members of elite bodies such as other national academies, and winners of major prizes and medals such as the Nobel, the Crafoord, the Field, the Kyoto, the Cosmos and others. I hope that future editions of this book will include such data. A reasonable conclusion from existing studies is that religious apologists might be wise to keep quieter than they habitually do on the subject of admired role models, at least where scientists are concerned.

The great French mathematician Blaise Pascal reckoned that, however long the odds against God's existence might be, there is an even larger asymmetry in the penalty for guessing wrong. You'd better believe in God, because if you are right you stand to gain eternal bliss and if you are wrong it won't make any difference anyway. On the other hand, if you don't believe in God and you turn out to be wrong you get eternal damnation, whereas if you are right it makes no difference. On the face of it the decision is a no-brainer. Believe in God.

There is something distinctly odd about the argument, however.

Believing is not something you can decide to do as a matter of policy. At least, it is not something I can decide to do as an act of will. I can decide to go to church and I can decide to recite the Nicene Creed, and I can decide to swear on a stack of bibles that I believe every word inside them. But none of that can make me actually believe it if I don't. Pascal's wager could only ever be an argument for feigning belief in God. And the God that you claim to believe in had better not be of the omniscient kind or he'd see through the deception. The ludicrous idea that believing is something you can decide to do is deliciously mocked by Douglas Adams in Dirk Gently's Holistic Detective Agency, where we meet the robotic Electric Monk, a labour-saving device that you buy 'to do your believing for you'. The de luxe model is advertised as 'Capable of believing things they wouldn't believe in Salt Lake City'.

But why, in any case, do we so readily accept the idea that the one thing you must do if you want to please God is believe in him? What's so special about believing? Isn't it just as likely that God would reward kindness, or generosity, or humility? Or sincerity? What if God is a scientist who regards honest seeking after truth as the supreme virtue? Indeed, wouldn't the designer of the universe have to be a scientist? Bertrand Russell was asked what he would say if he died and found himself confronted by God, demanding to know why Russell had not believed in him. 'Not enough evidence, God, not enough evidence,' was Russell's (I almost said immortal) reply. Mightn't God respect Russell for his courageous scepticism (let alone for the courageous pacifism that landed him in prison in the First World War) far more than he would respect Pascal for his cowardly bet-hedging? And, while we cannot know which way God would jump, we don't need to know in order to refute Pascal's Wager. We are talking about a bet, remember, and Pascal wasn't claiming that his wager enjoyed anything but very long odds. Would you bet on God's valuing dishonestly faked belief (or even honest belief) over honest scepticism?

Then again, suppose the god who confronts you when you die turns out to be Baal, and suppose Baal is just as jealous as his old rival Yahweh was said to be. Mightn't Pascal have been better off wagering on no god at all rather than on the wrong god? Indeed, doesn't the sheer number of potential gods and goddesses on whom one might bet vitiate Pascal's whole logic? Pascal was probably joking when he promoted his wager, just as I am joking in my dismissal of it. But I have encountered people, for example in the question session after a lecture, who have seriously advanced Pascal's Wager as an argument in favour of believing in God, so it was right to give it a brief airing here.

Is it possible, finally, to argue for a sort of anti-Pascal wager? Suppose we grant that there is indeed some small chance that God exists. Nevertheless, it could be said that you will lead a better, fuller life if you bet on his not existing, than if you bet on his existing and therefore squander your precious time on worshipping him, sacrificing to him, fighting and dying for him, etc. I won't pursue the question here, but readers might like to bear it in mind when we come to later chapters on the evil consequences that can flow from religious belief and observance.

I think the oddest case I have seen attempted for the existence of God is the Bayesian argument recently put forward by Stephen Unwin in The Probability of God. I hesitated before including this argument, which is both weaker and less hallowed by antiquity than others. Unwin's book, however, received considerable journalistic attention when it was published in 2003, and it does give the opportunity to bring some explanatory threads together. I have some sympathy with his aims because, as argued in Chapter 2, I believe the existence of God as a scientific hypothesis is, at least in principle, investigable. Also, Unwin's quixotic attempt to put a number on the probability is quite agreeably funny.

The book's subtitle, A Simple Calculation that Proves the Ultimate Truth, has all the hallmarks of a late addition by the publisher, because such overweening confidence is not to be found in Unwin's text. The book is better seen as a 'How To' manual, a sort of Bayes' Theorem for Dummies, using the existence of God as a semi-facetious case study. Unwin could equally well have used a hypothetical murder as his test case to demonstrate Bayes' Theorem. The detective marshals the evidence. The fingerprints on the revolver point to Mrs Peacock. Quantify that suspicion by slapping a numerical likelihood on her. However, Professor Plum had a motive to frame her. Reduce the suspicion of Mrs Peacock by a corresponding numerical value. The forensic evidence suggests a 70 per cent likelihood that the revolver was fired accurately from a long distance, which argues for a culprit with military training. Quantify our raised suspicion of Colonel Mustard. The Reverend Green has the most plausible motive for murder.*) Increase our numerical assessment of his likelihood. But the long blond hair on the victim's jacket could only belong to Miss Scarlet. and so on. A mix of more or less subjectively judged likelihoods churns around in the detective's mind, pulling him in different directions. Bayes' Theorem is supposed to help him to a conclusion. It is a mathematical engine for combining many estimated likelihoods and coming up with a final verdict, which bears its own quantitative estimate of likelihood. But of course that final estimate can only be as good as the original numbers fed in. These are usually subjectively judged, with all the doubts that inevitably flow from that. The GIGO principle (Garbage In, Garbage Out) is applicable here — and, in the case of Unwin's God example, applicable is too mild a word.