Violence, the Fundamental Attribution Error, and Contempt for the Poor

(Addendum 7/26/12: This post is actually getting read, so I’ve made a quick editing pass to fix some of the late-night stylistic infelicities. The prose is now slightly better, the content is unchanged.)

GMU economist Bryan Caplan has a series of blog posts about the putative irrationality of the poor. Most of them are linked in this post, which gives what I take to be his basic argument. That basic argument goes as follows:

1. Other people often say that poverty causes all sorts of social ills, because many of those social ills are traceable to or constituted by self-harming behavior that correlates with poverty, leading those other people to think that poverty (rationally) drives that behavior.
2. But rather than poverty providing an rational incentive to engage in that behavior, poverty actually provides a rational disincentive to do so, because that behavior has worse consequences for the poor than for the non-poor.
3. Therefore, it can’t be that poverty rationally leads to those behaviors.
4. Instead, we should look for common causal factors for both poverty and self-harming behavior. Obvious hypotheses include irrationality, stupidity, irresponsibility, etc., and those hypotheses are worth considering. (Caplan: “Low IQ, low conscientiousness, low patience, and plain irrationality.”)

This form of argument gets applied to a variety of different behaviors. For example, Caplan points out that being single is financially costly, s.t. (thanks to the diminishing marginal utility of wealth) poor people should be single less often rather than more often. He concludes that poor people are single more often because they’re irrational. Similarly, Caplan points out that the consequences of drug and alcohol abuse and non-renumerative crime are worse for the poor (who, e.g., can’t just shrug off a DUI, pay for the lawyers or go into fancy rehab). So if the poor abuse drugs and alcohol more or commit more crimes it can’t be a rational response to poverty, but must reflect irrationality, stupidity, laziness, etc.

What Caplan ignores in all these cases is that the poor face a different and markedly worse set of choices than the non-poor. In terms of the version of the argument I gave above, Caplan misses that 2 does not lead to 3.

In the most abstract terms, suppose that a non-poor person has three choices: A, B, and C. And let’s assume that A is the best choice, B is the second-best choice, and C is the worst choice. Suppose that she in fact rationally takes choice A. Now suppose that she becomes poor, and it so happens that being poor makes choice B even worse than it was before (but still not as bad as choice C). Yet, all of a sudden, when she becomes poor, she starts choosing B. “IRRATIONALITY!”, Caplan cries. But what if becoming poor eliminated choice A, or made it even worse? Then choice B becomes rational, because she’s become poor, even though her being poor made choice B worse than it would have been if she were rich.

This is a general problem with Caplan’s style of argument, and there are many concrete ways in which it shows up. As a whole, Caplan very badly suffers from the fundamental attribution error: he attributes the choices of the poor to their personal failings rather than attending to how poverty radically constrains their choice sets. But for the purposes of brevity, I want to focus on just one feature of poverty and how it screws up Caplan’s argument all over the place. That feature is this:

The poor are radically more vulnerable to violence than the non-poor.

Many of the poor grow up in less safe neighborhoods and continue to live in them as adults. Many of the poor are homeless, and thus lack the basic physical protections of walls and locks. Many of the poor have associates who are primarily also poor (thanks to a lack of social capital), and in a vicious feedback loop from the above-noted correlations with drugs/alcohol/crime/etc., thus have associates who abuse substances and commit crimes, and are more likely to be violent. etc.

Now let’s look at how some of these facts constrain poor people’s choices and shatter Caplan’s argument. I’ll consider several of his examples.

First, singleness. Poor people are already less likely to be desirable marriage partners, so the choice to marry is already more costly for the poor than for the rich. But let’s think particularly about the situation of poor women. Poor women are particularly vulnerable to domestic violence. Not only are many of their marriage prospects potentially violent (for reasons given above), but they are also more likely to become economically dependent on their spouses and hence have much more trouble escaping violent relationships. (The empirical association between economic dependence and domestic violence has been well-known for decades.)

No surprise, then, that poor women report dramatically more domestic violence than non-poor women. In light of these facts, Caplan seems extremely, irresponsibly hasty to attribute poor women’s singleness to their irrationality. It may be that because poor women are so vulnerable to violence, marriage is such a bad choice for them that it’s even worse than taking the big financial hit from singleness. And since poor people often lack social connections to non-poor people, if poor women aren’t marrying, poor men have less chance to marry too.

Ok, that takes care of singleness. What about crime? This is a really easy one, considering the well-known phenomenon of joining gangs for protection. In the words of my least-favorite police department: “many members join because they live in the gang area and are, therefore, subject to violence by rival gangs. Joining guarantees support in case of attack and retaliation for transgressions.” And, of course, being poor makes you more likely to living in a gang area. And, of course, joining a gang often obliges one to commit crimes (initiations, pressure from other gang members, etc.).

More generally, consider that in a community where everyone else is carrying a gun, it becomes more rational to do so oneself. Carrying a gun allows you to credibly threaten retaliation if someone else tries to victimize you. (This is a point I believe I first heard from, or attributed to, Glenn Loury, though I can’t track it down anywhere in his papers online.) In a community where others are committing crimes, it becomes more rational to do so oneself in order to signal qualities like toughness and, again, avoid becoming victimized. In a violent community in which people challenge you for dominance, you’d better not back down from the bar fight or, again, your likely outcome is routine victimization. All of these are particularly true when the police can’t be relied upon to protect you, as is the case in many poor communities.

“But,” someone might object, “your argument depends on the assumption that lots of poor people are violent in the first place, in order to give all the rest of the poor people an incentive to be violent.” To which I answer: “it only requires a few.” Intuitively: introducing a few violent people into a community, where police protection is inadequate, can give many others an incentive to resort to violence or to the tools of violence (e.g. carrying guns) in order to protect themselves. And, of course, this leads to a tipping point phenomenon in which those newly violent people give still others an incentive to resort to violence, etc. The game theory is left as an exercise to the reader. (Or just fucking read Leviathan.) Moreover, the poor are often crammed into closer proximity with one another than the non-poor are (the poor don’t get to hide out in the suburbs or in big houses, etc.), and are thus more vulnerable to this kind of spillover effect.

Again, we see the same problem with Caplan’s argument: he ignores how poverty can eliminate or greatly worsen the supposedly “better” life choices, like refraining from crime.

Finally, consider substance abuse. In his most recent post, Caplan argues that the poor don’t rationally abuse drugs and alcohol to dull the pain of poverty, because substance abuse trades off a little short-term pain dulling for much greater long-term pain.

But again, let’s introduce vulnerability to violence into the equation and things suddenly look different. Someone who is unusually vulnerable to violence has good reason to discount the future much more heavily than someone who isn’t, because the future is much more likely to contain a quick death or life-ruining disability or imprisonment. If you’re a teenager who has seen multiple of your acquaintances killed by gang violence or locked up before you even get out of high school, it’s perfectly reasonable (for a Bayesian updater) to think that you’re not likely to make it past 30, and hence to focus on short-term pleasures with long-term negative consequences. (This also can contribute to an explanation of things like irresponsible sex, failure to invest in human capital, etc.)

In all these cases, we see that a single negative consequence of poverty — a higher vulnerability to violence — radically constrains the choices of the poor, and makes supposedly irrational choices look much more rational. And this is so even though those choices are more harmful to the poor than to the non-poor.

Now think about all the other negative consequences of poverty that might constrain choices in a similar way. Lack of education (poor neighborhoods = shitty schools). Lack of social capital. Lack of access to affordable, nutritious food. Lack of access to affordable credit. And so on, and so on, and so on.

Instead, Caplan leaps to a single explanation for all of the self-harming behavior of poor people: it’s because they’re irrational and stupid. This reflects less on the supposed inability of the poor to run their lives than it does on Caplan’s inability to imagine the circumstances of anyone other than himself.

(n.b. comments are open but set to 100% moderation because I’m sick of spam; I’ll approve all non-spam comments as quickly as possible.)


20 Responses to “Violence, the Fundamental Attribution Error, and Contempt for the Poor”

  1. abgornish Says:

    Why are there such differential responses to poverty across ethnic and cultural dimensions? Consider that there are plenty of studies showing that poor Koreans with stores in black neighborhoods often produce children whose savings and school behaviors are far more middle class than middle class blacks who earn substantially more? I believe that economists have also looked at consumption patterns and — correcting for income — show that black americans show more impatience, less saving behavior, and more preference for items like clothing over books and schooling for kids, especially when compared to poor Asian Americans. And of course, we know that test scores of Asian American kids from the bottom quartile of the income distribution beat out the kids of black Americans from the highest quartile of the income distribution. So poverty may increase stress, but differential rates across groups suggests that cultural patterns, genetics, training, and social conditioning can either overcome or exacerbate their problems.

  2. Caplan v. Gowder on Blaming the Poor | Bleeding Heart Libertarians Says:

    [...] Paul Gowder responds here. [...]

  3. ben Says:

    I think you have a point in showing that poor people’s choices are constrained/modified compared to rich people’s choices, and that therefore Caplan’s arguments does not fully hold.

    However, it feels like you go from one extreme to the other in persistently trying to absolve the poor from any responsibility whatsoever.

    In particular, one interesting thing from Caplan’s articles that you *didn’t* respond to was his question:

    “If you were a social worker, would you advise the poor to turn to drugs and alcohol?”

    One could also add:
    If you really think it’s such a rational choice, would you advise a poor friend of yours to engage in gang violence?

    If you really think it’s such a rational choice, would you advise a poor friend of yours to engage in unprotected sex with uncommitted partners?

    And so on.

    If not, why not?

  4. Eli Says:

    Caplan has no contempt for the poor, and does not see them as irrational. In fact, he’s trying to resolve what appears to be an irrational response to poverty, which is turning to drugs or alcohol.

    Do you think that poor lottery winners usually quit their drug and alcohol abuse after they win the lottery? I wonder if any research has been done on the subject.

  5. Floccina Says:

    The poor are radically more vulnerable to violence than the non-poor.

    But the poor are only more vulnerable to violence because the poor are more violent.

    Also some of the lowest income people in the USA are Amish and Hasidic Jews and the exhibit none of the pathologies that the destroy the poor. Their are also some hippies (permies) in Montana with very low income who live fine lives.

  6. Paul Gowder Says:

    abgornish: some of this has to do with the worse overall life expectations of black people. If you suffer continuing employment and police discrimination, again, you’re not likely to see much point in middle-class behavior. Some of it may also be due to something similar to the crime dynamic I suggest in the post. That is, poor people of a given race who mostly associate with others of that race can be subject to bad strategic equilibria in which they anticipate others in their community making bad choices, and hence rationally make those bad choices themselves. And a community can be stuck in one of these bad equilibria for a long time without it being attributable to the personal qualities of any individual members or the group as a whole. This is basically Hobbes’s story in Leviathan.

    Ben: that’s a fair point. One answer is that expectations look different from the inside than from the outside. I don’t know what it’s like to watch my friends get killed by gang violence, so I don’t have a real understanding of how little hope for the future it might lead to. So no, I wouldn’t advise a friend growing up in a poor neighborhood to start hitting the pipe, but I also try to maintain enough epistemic humility to recognize that they can (rationally) see the world in a very different way than I do.

    Also, there might be some circumstances in which I would advise someone to join a gang. In prison, for example. Or in a totally collapsed neighborhood where the police have basically fled and joining a gang is the only way to get any social support.

    Eli: That’s an interesting question. I suspect that just by the nature of the lottery the n would be too low to get really useful statistical inferences, but would love to hear differently.

  7. Floccina Says:

    So from your post the solution should much more and better policing of the poor. Is that right?

  8. Paul Gowder Says:

    Floccina: Note that if the Hasids and others are a counterexample to my argument they’re also a counterexample to Caplan’s, since presumably their poverty is not caused by irrationality, stupidity, etc. (Actually, I’d suggest that Hasids, Amish, hippies, etc. are severe outliers because their very high social capital ameliorates most of the bad effects of poverty.)

    And yes, as much as I dislike how policing is done in this community, more and better policing of the poor would help. But emphasis here is on the “better.” That is, fewer pointless arrests for nonviolent crime, much more aggressive policing of violent crime, and rebuilding the relationships between police and community so that people could actually trust that the police were genuinely interested in putting a stop to violence rather than, e.g., carrying out their own racial agenda. Even then, it would only be a small part of the solution (because some of the crime is remunerative, and poor people often lack other economic opportunities). But it might help.

  9. Richard Says:


    Bryan’s question doesn’t really get us anywhere. Because the next question is, if the you advise the poor person against joining a gang, or against unprotected sex, was it good advice or bad advice? And that will depend exactly on whether these behaviors are rational.

  10. Floccina Says:

    I think that we know why those other groups are poor and it is not the same reason that most poor Americans are poor.

    I totally agree with you about policing. I think that the poor policing shows how little government serves the needs of the poor.
    I think we should end the war on drugs, cut back on other areas of spending and concentrate more effort/spending on better first and more policing second. I think that might help the poor more than the rest of the welfare programs combined, I would at least like to give it a try.

  11. LemmusLemmus Says:

    Hey, I was going to write my own critique of Caplan’s argument, despite the fact that I tend to agree with his conclusion that more of an emphasis needs to be put on the explanations of the type that he proposes.

    As I’m probably not going to get round to it, here’s the short version: Much of Caplan’s argument suggests that he doesn’t understand the concept of opportunity costs. Which is kind of creepy, given that he’s an economist.

  12. TGGP Says:

    I agree that there are semi-realistic situations in which I would advise a person to join a gang and/or carry a gun. But for lots of other common behaviors the most sensible explanation is hyperbolic (not merely severe) discounting: they will live to experience regret. Such discounting has been demonstrated to exist even among the non-poor, I don’t think it’s a stretch to use it as an explanation here.
    Regarding your Bayesian point: in most neighborhoods even if there is a lot of dysfunction, there will be people who behave differently and have better outcomes and an observer could see that conditional on such behavior they should expect a better life. Those people have taken choice B in the absence of choice A, choice C is the hyperbolic discounting one.

  13. Ak Mike Says:

    I’m sorry, but few of your conclusions follow from your premises, and moreover your conclusions are falsified by well-known facts.
    As an example of the first problem, you say that because it is dangerous in poor neighborhoods, more people will carry guns for protection. But carrying a gun does not make a person violent, does not make it more likely that the person will use drugs, nor that the person will be a poor employee, nor that the person will be less likely to marry. You probably know that persons who legally carry concealed weapons are less likely to commit crimes than the average person.

    Your conclusions moreover are falsified by the experience of immigrant populations in the early twentieth century. In poverty far more dire, in violence just as bad or worse than today’s violence, with discrimination directed against them that was not at that time illegal, with no government assistance, nearly everyone growing up in the crowded, unhealthy, violent and impoverished immigrant communities wound up in the middle class.

    In short, you have not made your case that poor people have no alternatives that allow them to escape poverty.

  14. Daniel S. Goldberg Says:


    This is a great post, and it’s a major, major area of scholarship and research in my domains, and one I actually write on (because we have such a tendency to conceptualize health and its distribution in human populations along methodologically individualist terms rather than taking stock of the overwhelming evidence that social and economic conditions powerfully determine risky behaviors).

    So indeed it is no surprise that so called ‘risky behaviors,’ what have historically been called ‘vicious habits’ are strongly and disproportionately concentrated among the least well-off — this is exactly what we would predict if indeed deleterious social and economic conditions constrain choices and generate disadvantage.

    Related points: the social epidemiology on these points is powerful and really well-supported, specifically the concept of clusters of disadvantage. This point shows that groups subject to one social disadvantage are highly likely to experience other such disadvantages (at least in the aggregate). Powers and Faden write about this to great effect in their book on social justice and public health, and obviously Jo Wolff is magic on this point, too.

    Also, stigma. While individual agency does exist — people in bad situations can sometimes still make better or worse choices — the individualism resident in Caplan’s posts (and highly prevalent in American society) facilitates devastating stigma of already marginalized and vulnerable groups, which is Not A Good Thing, to put it crudely. (Actually, not only is stigma corrosive but turns out to be really bad for your health, too).

    Anyway, sorry for the pedantry, but this topic is one very close to my scholarly heart.

  15. Aaron Says:

    Let’s go to the science:

    Why G Matters: The Complexity of Everyday Life –

    Intelligence and Social Inequality: Why the biological link? –

    Heritability of IQ –

  16. Paul Gowder Says:

    Even assuming those (hotly contested) results are correct, to say that lower intelligence (or, I should say, lower IQ test results) is correlated with poverty and to say that poverty is correlated with certain behaviors is not to say that lower intelligence causes those behaviors. Nor is there any necessary relationship between IQ test results and the kinds of cognitive-bias type irrationality, like hyperbolic discounting, that might (on a Caplan-esque theory) lead to things like drug abuse. Moreover, even if the poor had lower intelligence, if you follow the “all these behaviors are the result of stupidity” theory and deny that there are environmental constraints controlling the decisions of the poor, you seem to be committed to the theory that the poor have much lower intelligence than these data suggest. There are lots of people who can’t get through college but can figure out that condoms prevent babies.

  17. Aaron Says:

    Oh I would definitely say it is multivariate. I think everything you say in this post is right. But I’m just making the point the IQ thing shouldn’t be discounted either.

  18. Dan in Euroland Says:

    I am very late to the party, but for a reference to strategic complementarity in gun violence look at Sethi & O’Flaherty’s paper “Homicide in Black and White”

  19. kr Says:

    yo, i think you are right on point. i agree.

    i’m taking a class with caplan right now… he is clearly an intelligent guy – but he reeks of entitlement – he’s got this cocky self righteousness that makes it hard for me to listen charitably to his arguments.

    keep up the awesome, peace.

  20. Britta Says:


    Good post. Sociologists and anthropologists doing studies of urban ghettos have made similar points, i.e., that behaviors which allow for survival in extreme poverty are not conducive to middle class success. For example, survival generally requires some sort of collective pooling of resources, which then makes individual saving, the backbone of middle class morality, impossible, because those who come into an amount of money or a higher paying job are embedded in a network of obligation. Likewise, studies on teen pregnancy show that teen pregnancy and single motherhood is beneficial for the very poor, while deleterious for the middle class or working poor. So, the decision to get pregnant at 15 is generally option A for a girl in an urban ghetto, but option C for the people commenting on this blog.

    Secondly, your point on prison gangs should be highlighted. Joining a gang in prison is option A, the best option, for most people in prison. However, gang affiliation does not end at the prison walls and is very hard to exit. Given the incarceration rates of black men in the US, gang affiliation, this is not a trivial point. Likewise, incarceration also affects the other issues, such as singleness. When most men in your cohort are in prison, scarcity of mates makes singleness a rational option in addition to the lack of desirability. Secondly, it means that initially stable and long-term relationships are far more likely to end with the incarceration of the man in the relationship.

    Finally, racism. Comments comparing black people unfavorably to immigrants miss that urban ‘white’ ethnics engaged in widespread violent gang activity and organized crime and lived lives that were ‘nasty, brutish, and short’ in the tenement slums of NYC and other urban ghettos. However, the ability to assimilate into WASP society coinciding right about the time of mass migration of blacks from the South to the North (for precisely the reason that they were whitER than black people) allowed for Italians/Irish/etc. to escape cycles of poverty after few generations. As black migrants never had that option, the standard upward immigrant trajectory was unavailable to them.

    Finally, the creation of Northern urban ghettos is also predicated on the widespread collapse of manufacturing and the layoff of black workers, who then for reasons including discrimination had fewer options than white factory workers. Lack of gainful employment options over generations completely changes the landscape of what is a ‘rational’ goal in life. My guess is most commenters here, if the options were 1) lifelong unemployment & borderline homelessness, 2) stressful minimum wage jobs, or 3) a comparative interesting and remunerative career in the black market would choose (3), particularly if they were intelligent. If left with (1) or (2), an early death as a side effect of using drugs, alcohol, and drug foods as analgesics also seems more rational than a long physically and psychologically miserable life.

Leave a Comment