Robin Hanson has been blogging Ken Lee’s PhD dissertation and saved the best for last: Jobs Kill.
The big result: death rates depend on job details more than on race, gender, marriage status, rural vs. urban, education, and income combined!
He presents this table:
Two comments on the chart. As I understand it, the higher the factor above the more a job characteristic contributes to death. So, “Overall:Physical Demands”, at 1.699, is a big killer. Also, more stars means a higher statistical significance.
Ok, so I want to talk about “Context: Socially Challenging”. Here’s Ken Lee (this link may some day break) describing this factor a bit:
Work Context: Socially Challenging… has such attributes [such] as as impact of decisions on others, frequency of conflict situations, stress tolerance, and dealing with physically aggressive or angry people.
One thing Robin has taught me (though this perhaps isn’t his insight) is that intelligence evolved to deal with the social complexity in our society. This means that jobs that are socially challenging are jobs that tax the human mind more than any other job in the world.
The most socially complex jobs, in my opinion, are sales jobs. Remember, the best salespeople are those that are best at two things: one-on-one persuasion and accepting rejection, two incredibly socially stressful activities.
If my take away here is that salespeople have high mortality, then I completely buy it. It’s brutal work.
It would be cool to correlate these ‘death factors’ to wage and employee turnover in the occupations. I bet wage is related to status and turnover will be highest in low-status, high-danger jobs.
Not Unreasonable 
Numbers below one are helpers, not killers. So 0.508 is the biggest effect.
So the point is that, overall, the negative effect on health from socially-induced stress is dominated by positive attributes of those same jobs.
But would it not still follow that among People vs Things jobs, the most socially complex ones are still the most ‘deadly’?