The most dangerous jobs in America

Robin Hanson writes:

On the criteria of potential to help people avoid death, this would seem to be among the most important news I’ve ever heard.

[In his recent Ph.D. thesis, Ken Lee finds that] death rates depend on job details more than on race, gender, marriage status, rural vs. urban, education, and income combined! Now for the details.

The US Department of Labor has described each of 807 occupations with over 200 detailed features on how jobs are done, skills required, etc.. Lee looked at seven domains of such features, each containing 16 to 57 features, and for each domain Lee did a factor analysis of those features to find the top 2-4 factors. This gave Lee a total of 22 domain factors. Lee also found four overall factors to describe his total set of 225 job and 9 demographic features. (These four factors explain 32%, 15%, 7%, and 4% of total variance.)

Lee then tried to use these 26 job factors, along with his other standard predictors (age, race, gender, married, rural, education, income) to predict deaths in the 302,890 people for whom he had job data. Lee found that his standard predictors didn’t change much, and found these job factor risk ratios (Table 34, column 2):

KenLeeHugeJobEffects

Ten of the 26 estimates are 5% significant, and five are 1% significant – this isn’t random noise (*** p<0.01, ** p<0.05, * p<0.1). Each factor is scaled to range in value from 0 to 1 across the 806 occupations; its risk ratio is an estimated ratio of death rates when that factor has its max value of one, relative to death rates when that factor has its min value of zero. And these are huge risk ratios!

If you take all of Lee’s standard non-age predictors (race, gender, married, rural, education, income), and multiply together their risk ratios, you’ll find that a poor badly-schooled unmarried urban black male dies 17.7times as often as a rich well-educated married rural asian woman (of the same age), with a lifespan roughlythirty years shorter on average. (A risk ratio of 1.57 costs roughly five years of life.)

Yet big as this effect is, the top five job factor risk ratios give a total ratio of 19.7, bigger that all the other non-age effects put together! And the top ten job factor ratios give a total risk ratio of over 100!  (All twenty six factors together give a total risk ratio of 563.) Jobs are clearly a huge and neglected influence on who lives and who dies.

Hanson summarizes:

If you cared about preventing death, rather than just signaling your concern, these results suggest you stop wasting your efforts on tiny effects like medical insurance, auto accidents, crime, recreational drugs, radiation, or food safety, and focus on: jobs. Yes a lot of job-death variation must come from different types of people doing different types of jobs, but a great deal of this variation is also likely causal – some jobs kill folks much more than others.

At the very least we should try to tell people about the huge life and death consequences of their job choices. Then workers could demand higher wages for more deadly jobs, which should induce employers to seek ways to substitute less deadly for more deadly jobs.

I’m suspicious of that factor of 19.7, though–I think you can get big numbers when you multiply together noisy estimates. In any case, the general point is interesting. Especially in light of the recent NYT story reporting that reseachers attribute much of obesity to sedentary jobs, moving the focus away from home life to work life.

9 thoughts on “The most dangerous jobs in America

  1. “Especially in light of the recent NYT story reporting that reseachers attribute much of obesity to sedentary jobs, moving the focus away from home life to work life.”
    I bet that irritates Gary Taubes. His argument about why in nutrition/health people seek out a variety of explanations for different (correlated) symptoms rather than One Big Factor (simple carbs, according to Taubes) reminds me of Karl Smith on long-term thinking.

    • Well, I came here to mention Taubes, too. To elaborate on why the NYT story is likely wrong, Taubes argues (compellingly, in my opinion) that eating too much and not exercising enough are caused by obesity, not vice versa. He makes an interesting argument by analogy, pointing out that it is obviously absurd to say that children grow because they eat a lot. Rather, their hormones cause growth, which makes them hungry, which makes them eat a lot. With obesity, it’s apparently been well established for quite some time that insulin signals fat cells to store fuel (glucose and its ilk), which reduces the fuel available to muscles. Lack of fuel for muscles causes feelings of hunger and lack of energy. Simple carbs play their role by inducing an insulin response, and lots of simple carbs over time cause insulin resistance, which causes an even stronger insulin response to simple carbs. The long term effects are things like obesity, diabetes, and a host of other diseases.

      In case anyone is interested, here’s an interesting interview with Taubes: http://www.econtalk.org/archives/2011/11/taubes_on_fat_s.html

  2. I think this is the case where couple of examples can be a tremendous help. Can we get a list of life expectancies for
    people with different occupations? What about factoring out the pay grade?

  3. Sick people and well people choose their jobs differently (even more so in America where health insurance comes with certain jobs).

    If people are at lower risk if they have a job with fine motor ability doesn’t that say more about the people who don’t choose that job rather than the safety of that job? I mean, noone with Parkinson’s is going to choose to become a cartoonist or watchmaker or be able to stay in that job long term after diagnosis.

  4. Ten of the 26 estimates are 5% significant, and five are 1% significant – this isn’t random noise

    Well, one or two might very well be. But which ones? Such a mystery!

    In any case, just to be safe, I’m going to stop cooperating with my co-workers. Wait, crap. That might make my job more socially challenging. Oy.

  5. Hanson’s insistence that other people are motivated primarily by signalling seems like a classic case of projection to me. Or at the very least, if he really cared about the results of this paper, he could do a hell of a lot more than state some dubious conclusions and point at some at some statistical significance tests.

  6. I took a quick look at the original post by Robin Hanson and the PdD thesis by Lee. Neither of them clearly point out that risk ratios do not apply to job factors (which cannot die), they apply to people. The only meaning of the risk ratio of 19.7 that they got by multiplying the top five factors is this: someone in a job that has all five factors will have 19.7 times higher odds of dying than someone in a job that has none of the five factors. In reality, there are not likely to be enough people with those two sets of characteristics to make that comparison meaningful.

    As an aside, I think someone who can write “poor badly-schooled unmarried urban black male dies 17.7 times as often as a rich well-educated married rural asian woman (of the same age)” has not thought clearly about the analysis. That “poor badly-schooled unmarried urban black male” can only die once.

    • But you can say the hazard of death is 17.7 times as high for one individual as for the other.

      The causal reasoning here seems like the main problem. The supposition seems to be that you can find groups of people who are otherwise identical but one of which has “socially challenging” jobs while the others do not, and the ones with the “socially challenging” jobs die at higher rates. But that isn’t the case at all– the only “controls” being used in this study are the characteristics of the jobs themselves.

      This will be particularly a problem if you compare two jobs that have similar descriptive characteristics but very different levels of social status: the analysis will attribute the different death rates to the small differences in job characteristics, when really it is because one job is full of people who are richer and healthier than the others.

  7. Hmm.. large risk ratios and other measures of correlation don’t ensure that the relationship isn’t confounded (even if a few available covariates are thrown into the regressions). That should be old news to everyone here, but I don’t see much discussion here making a case that the direction of causality is correct.

Comments are closed.