sorry, but social science is actually a science

Every so often, you get the journalist, or academic, who loves trashing social science. The complaints are ritualistic – you can’t do experiments, people use jargon and math, and so forth. Well, Forbes has a nice article called “Enough Already with the Sweeping Claims that Economics is Unscientific.” It makes some obvious, but important points. Yes, some academics become divorced from reality with their models, but do you actually want people to study the economy without quantitative data or theory? These complaints also seem to ignore that economics actually does use experiments and much strives toward policy relevance:

Let me just start by pointing out that it is not the case that “almost nothing in economics is actually derived from controlled experiments”. Look at the CV’s of economists like John List and Esther Duflo and you can see there are plenty of experiments being done. In 2013, the study selected as the best paper from American Economic Journal: Applied Economics was for a randomized trial on how teenagers respond to HIV risk information.  If you want a concrete example of where this has made a difference, randomized treatment has been a central part of the research on the effects of charters schools. Unlike the field of astronomy, which Gobry must also think is not a science, economists do sometimes have more than observational data to go on.

And while it is true that a lot of research doesn’t use actual randomized trials, it’s also true that other kinds of research are very useful and informative. If his point was simply to argue that experiments and replication are important, and whether or not a body of research includes this should be one input among others in weighing the evidence, I’d have to agree. But of course you’d have to include external validity in there, which often counts against randomized trials. Instead of a relatively common claim about how it would be nice to have more experiments in economics, as is his style, PEG boldly overstates his case and makes incorrect absolutist claims about the importance of randomized trials.

Yes. Here’s the implication of this argument. Nearly every other social science, except history (which is a weird social science and humanities border case), has the same properties. We have ideas, we have data. Sometimes we do experiments. We collect other data. Sometimes we can replicate results. Sometimes we make progress and accumulate evidence, but other times not. This is, essentially, how science is done. The next time you hear someone trash sociology, economics, or another social science as unscientific, you have my permission to write angry tweets about them.

50+ chapters of grad skool advice goodness: Grad Skool Rulz/From Black Power 

Written by fabiorojas

October 10, 2014 at 12:01 am

11 Responses

Subscribe to comments with RSS.

  1. this is an interesting blog, but with no formal definition of what science is, I guess you could argue anything and be correct.



    October 10, 2014 at 12:43 am

  2. Nope. Read the last paragraph. OR: give me your definition and let’s figure it out.



    October 10, 2014 at 1:16 am

  3. I agree with the argument but I really dislike the tone of your words. Your tone reminded me the science wars case. In my opininion, there is no need for such a defense.


  4. Juan, I hope it wouldn’t be needed but people keep writing these articles and social scientists should pusu back.



    October 10, 2014 at 2:19 am

  5. I got your point…but to be precise, ¿You mean people like Piketty? see here: “There is no such thing as economic science. There are social sciences, economic processes involved social control…
    We should teach ‪#‎economics‬ much more in conjunction with economic ‪#‎history‬, social history, political history, political science. It’s just impossible to study issues such as dynamics of ‪#‎income‬ and ‪#‎wealth‬ ‪#‎distribution‬ in a purely economic manner. It’s very important that students in economics don’t lose all the energy in abstract mathematical models… [But] too often economists have been doing the opposite: which is [using] very sophisticated mathematical model to explain very little empirical material or sometimes no empirical material at all…
    [In order to promote economic ‪#‎justice‬] the first important thing to do is ‪#‎democratization‬ of economic knowledge. Too often bad economic ‪#‎policy‬ and economic policies in the interests of the wealthy come from the fact that we, sort of, abandon economic knowledge to group of ‘specialists’ and ‘exerts'”


  6. As for randomized controlled trials, I think the section on “no causation without manipulation” from Bollen and Pearl’s “Eight Myths about Causality and Structural Models” is relevant:

    Click to access r393.pdf

    Liked by 1 person

    Chris M

    October 11, 2014 at 8:47 pm

  7. Pretty laughable–the blogger is taking his/her philosophy of social science from, er… Forbes. Why not actually consult a scientist who writes on the topic? Eg oxford handbook on philosophy of economics. Also funny that the author uses examples of experiments that are not really about economics–they involve economists studying other fields with method atypical of their own discipline. If experiments are gold, marketing is more science any social science. Totally agree–this blog seems reactionary and mean. A lazy summary of a superficial argument.

    Liked by 1 person


    October 12, 2014 at 10:03 pm

  8. Oh yeah, and Forbes guy just lumps for a naive popperian position. Do we think popper really had the last word on this question?

    Liked by 1 person


    October 12, 2014 at 10:07 pm

  9. Reblogged this on Notas desde la Orilla del Camino and commented:
    This is an interesting discussion…


  10. PEG isn’t getting his ideas (directly) from Popper so much as from Jim Manzi, who does private sector A/B testing as his day job and has also written about A/B testing in academia and policy. I leave it to you whether Manzi qualifies as a “scientist,” but I think it is meaningfully to describe him as a private sector social scientist.

    I read Manzi’s book Uncontrolled and found it really interesting, though I found his vision of A/B testing various specific contingencies interesting and think we should do more of it, but I worry that taking it to the extent he wants would lead us to an extremely atheoretical endpoint. It’s actually not unlike the forecasting/modeling issue in the sense that forecasting, like A/B testing, can give more accurate predictions but is less elegant and parsimonious in its description of reality than modeling and so which one you want depends on whether you want to know what happens or why it happened.

    (also, FWIW, I’m friends w PEG and a friendly acquaintance w Jim).



    October 20, 2014 at 4:54 pm

  11. Reblogged this on SocioTech'nowledge.


    Pedro Calado

    October 28, 2014 at 2:26 pm

Comments are closed.

%d bloggers like this: