Bill Barnwell has had me blocked on Twitter ever since I write a post for Deadspin about why his piece "Mere Mortals" was flawed.

I'm pretty sure at least one of those has never been popular by consensus.

I'm sure that there are plenty of papers that should consider correcting for multiple testing whose authors aren't even aware of the problems posed by multiple testing. They've been taught that 0.05 is the magic number and that nothing else matters. I'd also guess that the awareness of this problem varies widely by…

I absolutely agree that to do this study correctly, you should be using survival analysis (as is done in the original NIOSH study that Barnwell criticizes). All I was trying to point out was one of the flaws in Barnwell's argument.

You're absolutely right. All I was trying to do here is demonstrate one of the problems of using mortality rate. You can't answer the question of which group is dying younger, if either, with this type of analysis presented here. One way to answer that question would be to, as you point out, estimate a hazard model.