Heroic Job Bias

Friday Eliezer described a Superhero bias:

The police officer who puts their life on the line with no superpowers, no X-Ray vision, no super-strength, no ability to fly, and above all no invulnerability to bullets, reveals far greater virtue than Superman.

Superman risks less, yet we celebrate him more.  Eliezer followed with:

John Perry was a New York City police officer … the fact that an atheist and a transhumanist can still be a police officer, can still run into the lobby of a burning building, says more about the human spirit than all the martyrs who ever hoped of heaven.

Actually the hero bias is far worse than Eliezer imagines, because in fact police and firefighters do not risk their lives more for you!  Yes, these workers do die more often on the job, but they die much less often because of the job.  You see, these are relatively high status jobs, and having low status is quite deadly for humans.  So in fact people risk their life more by becoming waiters, bus drivers, truck drivers, plumbers, painters and lots of other jobs!  A 1999 Demography article shows how male mortality risk varies with job category:


Larger numbers (to the right) represent higher death rates.  The darker line adjusts for age and race, while the lighter line adjusts for age, race, income and education.  As you can see, many familiar jobs have higher death rates that the police and firefighter category, highlighted with a red arrow.  Yes, correlation need not be causation – but a small fraction of this correlation is far more than on the job police/fire deaths.  And even if police did die more often you wouldn’t know they were doing this for you, rather than to keep their job.  FYI, here are female jobs: 


FYI, last year I posted about how helping professions don’t help more

GD Star Rating
Tagged as:
Trackback URL:
  • Cynical Masters Student

    So how about high status, high stress jobs (think investment banker)? They certainly do take a toll on health. Furthermore, I’m not at all sure whether police officer is a high status job (depending on your viewpoint, that may both strengthen or weaken the above argument). It may be in the US, but around here (Switzerland), few people like the cops or think particularly high of them.

  • Stuart Armstrong

    So in fact people risk their life more by becoming waiters, bus drivers, truck drivers, plumbers, painters and lots of other jobs!

    But if we are looking to reward the virtue or courage of police officers, then what is important is not their actual risk but their perceived risk – unless the people who beceome officers have access to these sort of statistics, or intuitively perceive part of it (very unlikely) then we can still say that it takes more courage to be in the police than to be a waiter.

    And even if police did die more often you wouldn’t know they were doing this for you, rather than to keep their job.
    True, and there are many other reasons an office would take risks – “deriving self-image through the risks they take” or “craving the respect of their comrades” are the first ones that spring to mind.

    However, I’d wager that that an officer would be more likely to think “I’m risking my life for the citizens of Gotham New York” more often than a waiter would – just because waiters don’t have a narrative that lets them think that way, while officers do. So in the virtue/courage front, officers are still more likely to intent to die for your than other proffessions.

    Your other points are valid, though. Police officers may be more personally admirable than waiters on average, but that doesn’t mean they’re more usefull to you.

    • haydos6

      mate shut up would you ever go into a burning hot house or building to save someone. as soon as you do i would like to hear about it ok.

  • Cynical, the health benefits of high status vastly outweigh any “high stress” cost for jobs such as investment banking or lawyering.

  • Stuart Armstrong

    Moving away from the police (traditionaly blue-collar) to, say, FBI agents (traditionally white-collar) may be instructive. I didn’t see any FBI agents listed on that chart, but my understanding is that being an agent is higher risk and lower pay than equivalent jobs with the same qualifications.

    If we conclude, based on the above arguments, that police officers do not risk their lives more for you, then the same would imply that FBI agents do risk their lives more for you.

    Would you agree with that, or am I misunderstanding your point?

  • former burger flipper

    So if we grant the investment banker or lawyer has a more stressful job than the burger flipper (which I’m not sure about) and we grant medicine doesn’t account for the difference in mortality, where does it come from?

    Is it that high status equates to better nutrition or a lower daily baseline of stress?
    I feel like a gotta be missing something here.

  • future burger flipper

    …and fascinating by the way. Really intriguing stuff.

    I found it interesting that farmers are among the lowest mortality rates.
    I read something in the popular press somewhere a year or two back about how the dangers of being a cop were over-rated. I believe they just used on-the-job death rates.
    But their claim was that being a farmer was considerably more dangerous than being a cop.

  • Matthew


    I didn’t see any controls on what kinds of people take low-status vs. high-status jobs.

    For example, low testosterone levels are correlated with higher mortality, as is low IQ.

  • michael vassar

    Robin, I agree with Matthew that we don’t have crucial controls on this study. Also, I flat-out dis-beleive the claim that investment banking doesn’t have greater costs in terms of stress. Spend some time with those people. It’s clear that a major thing that their job is selecting for is ability to survive levels of stress that most people simply couldn’t endure for more than a few weeks. Even with intense pre-selection most leave as soon as they can.

    Also, how are the status levels of these jobs decided. Sewers is high status?!

    Finally, if mortality is that high for waitresses maybe we should ban waitressing and food service work. Unlike construction, we don’t need those professions to function as a society. I wonder how giving up children for paid adoption effects mortality risk. I’m not absolutely joking here. It’s much more clear to me that consistency is in principle a good idea than that anti-paternalism is. In practice, paternalism in the US seems to cause more harm than in Sweden, possibly because of less benevolent elites. I wonder how job status and life expectancy relate in Sweden.

  • Floccina

    I am thankful for those heroic Alaskan King Crab fishermen who risk their lives to bring may favorite food to me. It would be nice if someone learned a way to farm Alaskan King Crab.

  • Gray Area

    “I didn’t see any controls on what kinds of people take low-status vs. high-status jobs.”

    It’s not just that. I often see the following two assumptions about confounders:

    (a) Finding more confounders to adjust for improves your estimate of the causal effect, and
    (b) If you DO find a confounder, the proper way to incorporate it is to adjust for it (e.g., if you want the effect of X on Y, and you have a confounder Z, you want sum_z P(y|x,z) P(z)).

    Both (a) and (b) are false. Even if you assume (b), (a) is false by simple counterexample. The reason (b) is false is because adjusting may not remove the confounding (it depends on the causal structure of the domain). Pearl discusses this at much greater length.

  • Roy Haddad

    Shouldn’t job seekers biased towards heroic jobs have the effect of lowering the wages of heroic jobs? If so, then taking a heroic job is charitable (not to claim it is a good thing).

  • Floccina

    Correlation is not causation is there strong evidence that if you took a people who could be lawyers and put them in a jobs as a waiters that they would tend to die younger on average?

  • Floccina

    BTW my father was a fireman who often worked with the police and used to say that “Barney Millar” was the only realistic cop show on TV.

  • Everyone, there is a huge literature on the health status correlation, with many strong indications that a big part is causal from status to health.

    Stuart, I’m sure FBI are included in the category described.

  • The point is that not just anybody will be able to land high status jobs, and not just anybody ends up in low status jobs. One obvious point is that the barriers to entry are vastly different. Low status jobs tend to have very few barriers to entry, and the opposite for high status jobs.

    I’m an actuary, and every so often a member of the Society of Actuaries runs a mortality study on our members, and unsurprisingly we have lower mortality than that of the populace at large. We’re a highly educated bunch, having to pass about 8-9 math-based exams, and we have pretty low risk lifestyles and low stress jobs (relatively). Our barriers to entry are pretty high, and you’ve got to be conscientious to get through years of self-study (it usually takes about 7-10 years to get through all the exams.) If we weren’t actuaries, we’d probably be accountants or math teachers or the like. It’s unlikely that one of us would end up as a construction worker; though possible, it’s highly improbable.

  • Doug S.

    I’ve heard that lawyers have unusually high rates of depression…

  • Psychohistorian

    How the hell do death rates increase across the board when you add controlling factors? Shouldn’t the totals have to be the same?

  • I think this website could use more specific examples.

    If, say, New York male firemen are less likely to die of, say, AIDS than New York male waiters do, does that really have much to do with the status of the fireman’s job vs. the waiter’s job? Or does it have more to do with the type of man that wants to become a fireman versus the type of man who wants to become a waiter?


  • Gray Area

    “How the hell do death rates increase across the board when you add controlling factors? Shouldn’t the totals have to be the same?”

    You are thinking that ‘controlling factors’ are just conditioned on, but it’s not that simple.

    Say you want to know if smoking causes cancer. You can look at the conditional probability P(cancer|smoking), and if it’s higher than P(cancer), you can argue that smoking causes cancer. The problem is, there a lot of variables which can influence both how likely you are to smoke, and how likely you are to develop cancer, for instance age. Older folks are more likely to smoke, and (perhaps) more likely to smoke. So intuitively what we want to do is look at cancer/smoking interaction in every age group. This interaction would just be P(cancer|smoking,age), the probability of cancer given smoking (at a particular age). We combine these estimates by weighting them by the prior distribution on age, to obtain sum_age P(cancer|smoking,age)P(age).

    So now we ‘added an additional factor’ to our guess at the effect of smoking on cancer. It turns out that this new guess doesn’t have to bear any relationship to our original guess. (It’s a useful exercise in probability theory to figure out exactly why; the intuition is that there could be an arbitrarily strong influence from smoking to cancer via the age variable, and this influence is precisely what gets cancelled by adjusting for age).

  • Gray Area

    I meant ‘Older folks are more likely to have cancer.’ Apologies.

  • douglas

    I’m wondering where I can find a list of jobs and if they are low status or high status.
    I work with people who have all kinds of jobs, doctors, lawyers, farmers, some don’t work at all. I wonder if it would be helpful if I knew something about job status.

  • Floccina

    Sorry for so many posts but this subject is very interesting to me. I have been reading some paper on social status and health. In my crowd lawyers have very low status in a way by this I mean to imply that there are multiple ways to view social status? Doctors seem by any measure that I can think of to have higher status than lawyers. But lawyers seem to do significantly better than doctors in the above graph.

    What happens to social status when one retires, when housewives quit working?

    Are their ways to raise the status of people around us? How about our own status?

    Does a bad boss lower the life span on his long term employees significantly?

    Does taking welfare lower or raise ones social status?

  • Carl Shulman


    Doctors spend time with sick patients and in hospitals, so we should expect high mortality.

  • Jeff Borack

    I think a Superman type character is a greater hero than his law-enforcement counterpart. “With great power comes great responsibility” – Uncle Ben If I had super-powers, villainy would look like a pretty good option. Why would Superman or Spider-man waste their time, sacrifice their relationships, and potentially risk their lives (against other super-powered foes) when they can join the side of evil and do it for a profit?

  • Cynical Masters Student

    By what channel exactly should higher status lead to lower mortality?

  • Cynical Masters Student

    Something went wrong with my last post, so just the gist of the rest: how exactly should income and perhaps IQ be good enough to control for factors like genetic fitness which may very well both increase chances to get a high status job AND for living a long life…

  • douglas

    Floccina- here are a couple web sites that deal with the bad boss. I don’t know that there are quantified studies to the health effects, but there seems to be plenty of anecdotal evidence to the effect that a bad boss can lead to bad health.


    You can raise the status of others around you by showing them respect.

  • Stuart Armstrong

    Stuart, I’m sure FBI are included in the category described.

    Probably, but their relative numbers would be so small that the result would not be representative for them.

    But I asked that question about FBI agents for a different reason – I hope you don’t mind. You’ve previously blogged about how helping profession don’t help – a position I disagree with, but which is coherence (the market rewarding people precisely according to their utility).

    Then you present this post, which shows that the police is less altruistic/useful/admirable than we think. Fine; but the measure used (overall death rates adjusted by certain criteria) allows us to say that some professions are less admirable than others; but by the same token, we must be able to say that some are more admirable than others. Since public admiration of jobs is rather biased and random, it is nearly certain that there exists admired jobs that are in fact admirable.

    That’s why I brought up the FBI. I don’t have the data on death-rates, but FBI agent is not that high status a job, is relatively low paid, and requires relatively high education attainment. Patriotism would also tend to make it more attractive to some people than it is inherently, further depressing pay, conditions ans status. It seemed a prime candidate for an admired job that is in fact admirable on your scale.

    I just wanted to see whether your contrarian urge was driving this post, more than your desire to overcome bias. Sometimes I feel you don’t keep that contrarianism in check, even when it should be. Sometimes conventional truths are actually true 🙂

    Anyway, apologies for that, and keep up the blogging.

  • savagehenry

    I think a lot of the reason people tend to perceive police as being admirable for their risk even if it is in actuality lower than many other jobs is because of the nature of the risk itself. When a police officer dies on the job it’s not uncommon for it to be because he was killed by a criminal he was trying to subdue. The same goes for fire fighters, when one dies it’s often because he was trying to rescue someone. But, when a waiter dies on the job it’s probably because they did something stupid and got in an accident (I’m trying to think of ways a waiter can die on the job and they’re all hilarious for some reason), very seldom is it because they were rushing into a burning building to save 8 children, 7 puppies, and a grandma all at once. In general I think people see dying while helping others as more admirable than simply dying of natural causes or because of a job related accident.

    However, I have to say, my girlfriend is a notable exception haha. Whenever a police officer dies and it gets covered on the news for a week she gets furious because she says that just because they’re a cop does not necessarily make them a better person than the 10 other people that died that week but only got 10 seconds on the news.

  • Doug S.

    Another confounding factor: some jobs have physical fitness requirements. Firefighters, especially, have demanding fitness requirements, and I think there must also be a fitness requirement for police officers.

    Questions of the form “Why X?” often can be answered with large numbers of plausible sounding guesses; I don’t know how much benefit there is to idle speculation, though.

  • KS

    This is stupid…. just because they die more on the job than police or firefighters doesn’t make them HEROIC…. firefighters and police do their job and risk their lives everyday because they want to….. waiters and plumbers DO NOT!!

  • Jannette Dickinson

    What I want to know is if those numbers take into account the number of people who actually work in that particular job. If there are more people working as waiters or waitresses, then of course more people are going to die on the job. Besides that, I don’t think that people who are brave enough to run into a burning building are thinking, “man, I wish I didn’t have to go in there and save some one, but I don’t want to lose my job.” If that were the case, I think they would just find another job.