Should theists accept higher risks of death?

The ever-provocative Will Wilkinson opines:

In a fit of Beckerite rational choice reasoning, I decided that theists ought to have higher rates of death by accident. If I believe that heaven is infinite bliss, then I should be quite eager to join my maker. Suicide is a disqualification for paradise, but dying in a car accident isn't. So, one should expect that theists who believe in perpetual Miami would take more risks than those who do not so believe, and that thus, death-by-accident ought to be higher among believer than non-believers.

My guess is that there is no difference in rates of death-by-accident among believers and non-believers. If my guess is correct, then there's another reason to believe that many people don't really believe in God, even though they think they do. Or, at least, there's a reason for rational choice economists to believe meta-atheism.

My take: Most of all, theists should have stronger reasons to live.  They have their own selfish reasons, plus whatever role they think they are supposed to be playing in God's plan.  So they ought to take fewer chances; indeed the data suggest that both religious belief and religious participation are correlated with longer lifespans.  And even if theists believe death is paradise, that will come sooner or later in any case.  In other words, heaven brings an "income effect," not a "substitution effect."  We need of course two auxiliary assumptions.  First, theists, given their perceived roles in God's plan, do not feel a strong impatience to arrive in heaven.  Second, the method of death under consideration should not affect the probability of heaven vs. hell.

That all being said, we don't have a good theory of how to rank-order infinities (e.g., "infinity plus three" is not mathematically larger than "infinity").  So how can anyone who sees any chance of infinite utility satisfy standard choice axioms?  Even Nick Bostrom can't answer this question.  (And should theists accept Tabarrok's Offer?)  But I won't blame this problem on theism per se.  As Nick argues, atheistic cosmologies can easily have problems with infinite expected values.  And arguably theism could be used to define limits on time, physical space, or the scope of possible worlds.  So both empirics and theory suggest that theists should be more eager to live, and less willing to die (now).