Thou shalt not sit with statisticians nor commit a social science

Robin Hanson writes that “social scientists know lots” and then asks “Why then do so many people think otherwise?” Like Robin, I have worked in both the physical and social sciences and I have a few thoughts on the comparison, first a big thought, then several little thoughts.

Social scientists haven’t done much

Robin mentions various cognitive biases for why people disparage the social sciences. But he misses (I think) one key reason to disparage social science which is that social scientists may know a lot, but they haven’t done that much.

Compare to physical and biological sciences and engineering. Research in these areas has given us H-bombs, chemical fertilizers, laptop computers, vaccinations, ziplock bags, etc. etc. And social science has given us . . . what? An unbiased estimate of the incumbency advantage? The discovery of “nonattitudes”? A clever way of auctioning radio frequencies? The discovery that sumo wrestlers cheat? Not much “news you can use,” I’d say. I guess there’s been some work in epidemiology that’s been useful. Certainly some interesting things—I’d agree with Robin that “social scientists know lots”—consider Milgram’s experiment, or just the existence of polls which give us a sense of the distribution of opinions on lots of issues—but I don’t think this comes close to comparing to the achievements of the so-called hard sciences.

Social science is important, though. It gives us ways of looking at the world, for example in economics the ideas of Smith, Ricardo, Keynes, etc., frame how we think about policy questions. Perhaps because of their experience studying small effects and individual variation, social scientists are often good at understanding statistical ideas. For example, sociologists tended to see right away what I was getting at in my critique of a sociologist’s sloppy biological ideas on sex ratios, even while biologists themselves were fooled. But, to be sure, “ways of looking at the world” is pretty weak. The dollar auction is an impressive demo and the median voter theorem is cool, but it’s not like the hard sciences where, for example, you can point to a cloned sheep and say “hey, we did that!”.

Comparison to the study of history

Rather than comparing social science to physics, chemistry, biology, and engineering, a more useful comparison might be to history. Historians know lots, both about specific things like what products were made by people in city X in century Y, or who signed treaty Z, and also about bigger trends in national and world events. But historians haven’t given us any useful products. History has value in itself—interesting stories—and helps us understand our world, although not always in a direct way. Once people start trying to organize their historical knowledge, this leads into political science. Again, it’s not giving you anything useful such as you’d get from the study of chemistry etc., but it’s the logical next step to organizing your knowledge. Assuming you accept that history is important and interesting in its own right.

Similarly, social psychology organizes what would otherwise be episodes of personal stories of social interaction, economics organizes what would otherwise be anecdotes of business, etc. I think this stuff is interesting (otherwise I wouldn’t do it), and ultimately I justify it by the way I justify the study of history.

We’re not in a war economy and people do all sorts of things that are less useful than the development of effective pesticides, high-grade plastics, etc. etc. To compare social science with physical/biological sciences and engineering is like saying to somebody, “Why do you repair lawnmowers? You should be a paramedic, that would save more lives!”

Various other thoughts

– I think physical science is more difficult than social science. Now, I can’t say that _all_ (or even most) physical science research is harder than most social science research. In fact, I would imagine that routine physical science research (for example, taking measurements in a lab, which I’ve done, way back when) requires care, such as the importance of sterilizing equipment and knowing the best time to use serological pipette, but less thought than routine social science research (for example, conducting field interviews). But to do the serious work, my impression is that the ideas in social science are much closer to the surface than in physical and biological sciences, which are just a lot more technical. I think it was a lot easier for me to slide from physics to political science than it would’ve been to go the other way.

– That said, I don’t think that, when physicists etc. decide to work in a social science, they necessarily make helpful contributions. I’m thinking here of social networks, where some physicists and applied mathematicians (for example, Duncan Watts) have done important work, but others have notoriously just assumed that their simplified models are relevant to reality.

– The boundaries of social science are not so clearly defined. Social science certainly includes political science, sociology, economics, and anthropology (for example). But only some of psychology is a social science, similarly with history.

– As a researcher who does a lot of work in social science, I do get annoyed at “hard science snobbery.” For example, I had a colleague where I used to work, a very nice guy, but he had this attitude that statistics applied to biology was the real thing, and applications in almost any other field were inferior. My feeling was: yeah, sure, but are you out there curing cancer?? Just because you work in a field where some people are doing important stuff, it doesn’t mean that you are.

22 thoughts on “Thou shalt not sit with statisticians nor commit a social science

  1. I totally agree with you that physical science is much more difficult than social science. The works of social scientist are so simple (if we compare with high-experienced physical scientists). Social scientists go around, accumulate data, making statistical estimation, and give away some useful conclusion.
    Physical scientists actually contribute many powerful theory as well as very nice new products.
    In otherwise, social scientists contribute to the background the world. If you consider Darwin as a social scientist, you can see that his theory contribute to the building of society building, such as US society.

  2. Predicting human behavior requires accounting for many variables simultaneously, and generally results in explaining only a modest amount of the variance of any given outcome. I have a PhD in psychology, not in a "hard" science, but I think that the behavior of a child is vastly more complicated than the behavior of a chemical or a molecule. Chemicals react the same every time. For a social scientist, variability is a way of life.

  3. Though I haven't read Hanson's post here are my thoughts on this one:
    Is it possible that the lack of big impact contributions of the social sciences is because it is harder? It is much harder to form laws about human behavior that hold with strong predictive power. In this sense the "hard sciences" are easier.
    Also, I think you are under-selling the contributions of the social sciences. Studies by psychologists, economists and others assessing the impact of policies or programs have an impact on future policies and programs. This happens A LOT and has a huge impact I think on the quality of life for many people.
    To some extent I think you are fooled by the "hard sciences" because they have more tangible contributions (like ziplock bags).

  4. Robin,

    Well, I did say "physical and biological sciences and engineering." In any case, physicists developed the H-bomb, chemists developed cellophane, and social scientists developed . . . the SAT's???


    I agree that the social sciences are important. Heck, I'm 50% in the political science department here. And check out the title of this blog, along with the topics of my forthcoming two books! I was just answering Robin's question, "Why then do so many people think [that social scientists don't know lots]?" To my mind, a natural reason for thinking that is that the products of social science are not as impressive as the product of physical and biological sciences. That's one reason why I suggested the comparison to the study of history.

    As the saying goes, the alternative to "good statistics" is not "no statistics," it's bad statistics.

    Unfortunately, the discussion at Robin's site (Overcoming Bias) devolved into political arguments–there's clearly another thing going on, which is that social science is associated in many people's minds with the political left, and many of Robin's commenters are coming from the right. That's interesting in its own right but a little off the topic of what social scientists know.

  5. Kobi,

    Yeah, but there's also the social science that's used, for example, to make those personality tests that they give to prospective employees to see if they're going to steal.

  6. I think if more Americans had a fuller understanding the impact the social sciences has had on the nuts and bolts of political campaigns and just politics in general then they might be a little disturbed about just what it is social scientists know.


  7. While I wouldn't agree that the physical sciences (+ engineering) are "harder" in an abstract sense — obviously they have different requirements, like a lot more of math when you do math (pardon the pun) or physics, but in order to do psychology, you need to be more concerned with understanding people —, I do very much believe that social scientists and psychologists haven't yet done very much that is really useful.

    But I also think there is a reason: In some ways, research is easier for the physical sciences, because they usually have a very good idea of what they are analyzing; e.g. it is rather clear that DNA "is there" in a realistic way and that it is really important, even if you don't really know of all the things it does. In the social sciences, however, you have to first define what you want to analyze; you can not really say there is "conservatism" as a natural fact like there is DNA.

  8. Social Science is actually harder than physical science, and that is why it hasn't produced "nice products" that are useful to society.

    In physics or biology, you can conduct experiments, whether it is introducing a certain bacteria to an acid environment, or observing the behavious of sub-atomic particles when exposed to very low temperature.

    In Social Sciences, you can't conduct experiment, albeit on a very small scale, and even then the results are likely to be biased, the agents will know what's expected of them and adopt a different behavious. In a free country where information is available to all, this sort of experiments can almost never be conducted. On a national scale, the fields of politics, history and social conditions will also affect the outcome a the result.

    In a totalitarian country however, these things are possible. Stalin's Russia for instance was a direct experiment on the possibility of rapid industrialisation and the radical reorganisation of society around a marxist framework. The statistics were biased for political reasons, but methods have been devised to overcome this problem and the results of the experiment can be quantified and assessed.

  9. Andrew, I could have said "social scientists and business" to try to lump in all the contributions of all businessmen, but that wouldn't have been very fair. Similarly, lumping in all engineers with physical scientists doesn't seem fair to me. Same for lumping in all doctors with biological scientists.

  10. Wolf,

    Yeah, I agree that social science is fundamentally more difficult in the sense that intangibles like personality and political attitude are important but can never be precisely pinned down. But, in my experience, "doing physics research" is much harder than "doing social science research." The physics research just needs lots lots more preparation.


    Point taken. But, still, I think examples like the H-bomb and some plastics are clearly the product of physics research, in a way that impressive business achievements are not so clearly the product of social science research.

    As I said, I think social science is important, or I wouldn't do it, and it can be difficult too. But I can understand why outsiders might not be so impressed with social science. Especially when compared with all the cool technology that's coming out nowadays.

  11. I personally find the difference to lie more in what is possible to generate. In the social sciences, anyone can argue anything. Some arguments and theories are well thought-out and researched, some aren't. In the hard sciences, the proof is a result of an experiment which either worked or didn't work. The set-up can be messy and can be criticized as such, but the results seem much more firm to me. This is why I personally enjoy mathematical models of social science phenomena. Though one can argue that statistics can be twisted just as much, there still doesn't seem to be quite as much "leeway" as in purely social science theories.

  12. I think it is important to define the word "harder". It seems that its meaning is obvious but I'm not so sure. Math is hard. But is math harder than, say, painting? When you think about it, it's no so obvious. Quality work in both fields require a high level of aptitude that rules out 99.99% of the human race. A top painter is unlikely to become a top mathematician. But it's not going to be any easier for a top mathematician to become a top painter.

    So I guess I would like to see people be more explicit about what they mean by "harder".

  13. I don't quite know if it qualifies, but two "social science" inventions that have had quite a large impact have been governments and money.

    You might argue that these predated social science as a science, but I would counter that they merely predated the split between philosophy and science, but they clearly should go with the social science side after the divorce (i.e. they aren't physics).

  14. Ted,

    Governments and money are social inventions but I wouldn't call them social _science_ inventions. Similarly, I don't think of the wheel as a product of the study of physics or cooking as a product of the study of chemistry.

  15. It is interesting that you give as an example of "product" of hard science a thing like the H-bomb, or, later, that you point at cloned animals. Now, the difference for me is that there hasn't been any evolution in the sense of "Big Science" (i.e. for the social sciences. I can imagine that Physics during the '20-30s of last century probably classified as social sciences do now, in terms of "achievements": all those physicists talking about the structure of matter and so what? Then the question for me should be if social science will be ever capable to do the same feat.

    p.s. a very interesting blog indeed

  16. It is completely ridiculous to say that physical science is harder to do than social science. The reason we associate 'difficulty' with the hard sciences is because it has been easier to develop technical/refined/elegant approaches to understanding physical systems than it is to social systems. Physicists can work in vacuums. They can ignore frictions of any kind. They don't have to worry about history (usually). Systems with more interacting parts that can assume a wider range of states – systems that are open – historically contingent – scale-dependent – unstable and/or chaotic – have inertia – are much more difficult to study than the typical physical system. This characterizes ecological systems as well and then social systems are just even more difficult.
    The more interesting issue is why techniques of physical science are not more frequently applied to social science in order to find points of entry into otherwise impenetrably complex systems. There are major biases in the social science that make your other depictions in this blog post accurate. These have to do with historical baggage – like the widely held belief that humans are special, outside of nature, and have irreducible or non-patterned behaviors. These beliefs keep social science on the short bus – so to speak.
    Its also the case that really smart and mathematically inclined undergraduates are more likely to gravitate toward physical science because the history of achievement creates a certain inertia, gives physical science momentum – and its been around longer so has more accomplishments to point to and more to build on. We can hope that the tide is turning in social science but its slow going.
    But to say that social science is easier is absurd if you are talking about the nature of the systems being studied. I wonder if the people arguing that have ever thought beyond the statistical techniques they were applying and moved to the nature of the phenomena at work.

  17. When my dad was in grad school in physics, at some point he was talking with his advisor and the subject of philosophy came up. My dad's advisor said "it's not that there aren't some interesting problems in philosophy, but…I mean, how much do they get done, per philosopher-man-hour?"

  18. As a sociologist, I completely agree with your first point and with the reason given by the 2nd and 3rd commenters for why this is so. But I think there are two (related) other reasons.

    1. Only a minority of sociologists agree that explaining – and thus predicting – should be the ultimate aim of the social sciences. Someone like Luhmann who merely thinks up a hugely complicated framework for thinking about social life would never have got tenure in the hard sciences, but in wide parts of the social sciences he is seen as a demi-god.

    2. Evaluations of social programmes is not highly regarded in sociology. If you have a paper that shows that intervention X reduces, say, adolescents' use of violence by 30%, surely this is an important article? But try landing that in the AJS or ASR. Chances are they'd rather publish an article on the evolution of beard fashions, 1880-1950. Habermas – widely respected – uses "social engineering" as an insult! What we need is a change in attitudes!

  19. I think it's worth noting that the physical sciences have have a head start of several hundred years, and during that time they've been able to pick a lot of low-hanging fruit. And still there are those who deny the germ theory of disease, think they can beat the roulette wheel, and think that a single pulley can make lifting easier. And physical sciences deal with small things like our bodies, our luck, and our work. However social sciences deal with the most thing important of all: our selves, souls, or however you want to call it.

    It's relatively easy to give up our beliefs about how things external to the self work. Social sciences demand we give up our beliefs about how we, the very core of ourselves, operate, and I think that's tough for most people. I think that creates a bias against social sciences.

Comments are closed.