> The error bars suggest there is too much in-population variance to come up with useful general guidelines.

I’m not sure I’d agree there. What’s the argument for that?