what economists can teach sociologists

A while back, I wrote a post called “what economists should learn from sociology.” Consider this a follow up post – what sociologists can learn from economists. Let’s start with substantive topics:

  • Micro 101: The basic tools of microeconomics are very useful. Supply and demand, comparative advantage, marginal analysis, opportunity costs, expected utility, etc. Sure, in the real world, people aren’t perfect calculators, but they aren’t morons either. If a situation is fairly well defined, people will compare options, assess costs and benefits, and so forth.
  • Game theory: Interactionism is very popular in sociology, yet interactionist theories are often ad hoc when you get down to it. Game theory is a nice way to model interactions, even if it has limitations. The other nice thing about game theory is that the basics are fairly easy to learn compared to other topics.
  • A focus on outcomes: Sometimes, I feel like sociologists are a little too focused on process and not enough on outcomes. Economists have developed ideas, like welfare analysis, that could help sociologists guide their thinking on how the things we study might have policy implications.

Let’s switch to professional practices:

  • I really like how most economists can easily recite the core models and theories of the discipline. Because we teach social theory through original texts, we focus too much on “Weberian theory” than the theory that Weber actually believed. Experienced researchers can, of course, extract the theories and models, but we make it too hard for students. We need our core to have a succinct presentation.
  • I think it would be good to have a very modest amount of formal models.
  • Economics programs have a reasonable time to degree – 4-6 years. Except for those who require extensive travel, there is little reason to believe that sociology can’t be the same way.
  • I also like how economists maintain links between the academy and the worlds of business and policy.

Now, let me switch to things that should not be learned from economists:

  • The belief that math makes you scientific and that un-mathematical ideas are inherently vague or useless. Any belief that implies that Charles Darwin was a bad scientist should be immediately rejected.
  • We should not praise people just because they are good at math, as admirable as that may be. Social science is about understanding how people behave. Math can be a tool, but too much theorem proving will distract you from developing intuitions about the social world. Most theorems will be quickly forgotten, while a powerful empirical finding can resonate for years. We should praise people for helping us understand the social world.
  • The “econ rules” for talk and interaction. We may humor ourselves by believing that aggressiveness implies authority. But, honestly, we’re propping up our self-image. Let the guy get past the first slide, please. There will be more than enough time for slash and burn during the Q&A.
  • A religious belief in the neo-classical model of decision making. As any computer scientist or psychologist can tell you, there are many models of decision making that can be tested or used for theory building.
  • An obsession with clean identification. You can have great scientific work with observational data. Otherwise, we’d have to fire all astronomers, meteorologists, and any other scientists who study large complex systems.

I really hope that sociologists can tap into the good side of economics. I think there’s a lot to be learned.

Adverts: From Black Power/Grad Skool Rulz

Written by fabiorojas

October 18, 2011 at 2:06 am

6 Responses

Subscribe to comments with RSS.

  1. nice set of lists! from the perspective of one coming through an interdisciplinary program phd (econ, org theory, political science), these thoughts very much resonate. it has been interesting watching economists learn how to teach student (such as myself) who want to learn the methods & theories but have no interest in being socialized as such.



    October 18, 2011 at 5:14 am

  2. With some of the substantive topics you mention, you could make the case that they’re actually much more useful if you employ them like a sociologist or anthropologist rather than like a typical economist. That is, rather than taking rational utility-maximization as a premise about human behavior, treat it as a specific pattern of interaction that can arise given a particular system of social relations.



    October 18, 2011 at 8:29 am

  3. About this “obsession with identification”:

    1. a big part of economics deals with highly value-laden questions, as does sociology. I do not think there is such thing as being too careful with making claims in this environment.

    2. social interactions usually do have selection problems, unlike meteorological and astronomical phenomenon.

    I would thus argue that close attention to identification is almost as important here as was the introduction of double blind testing in medicine.


    Indrek Seppo

    October 18, 2011 at 10:51 pm

  4. Bayesian games



    October 19, 2011 at 11:58 am

  5. @indrek:

    First, I accept identification as an important issue, for the reasons that you described. If you read soc journals, you’ll see a fair number of experiments, IVs, and so forth. Sociologists believe in identification.

    What I reject is the belief that data analysis is useless unless the process has been completely identified. There are a few reasons for my belief.

    First, there are some processes where an experimental design would be really, really hard to create. This is true when you deal with large complex systems. For example, a lot of macroeconomic questions are hard to approach through experiments. If we wanted to know if the Obama stimulus package helped the economy, we can’t rerun our economy under different conditions. Observational data is all we got.

    Second, identification assumes that you have a well defined research question. However, there are a lot of research areas where people are in the process of just figuring out the state of the world. For example, my own research area, protest movements, has a paucity of data. Protests are hard to track for a variety of reasons. Thus, we are happy when we get decent observational data.



    October 19, 2011 at 2:46 pm

  6. in a bayesian hierarchical model, couldn’t one model one’s belief in a given identification as a higher layer of the hierarchy? a notion of “complete” ldentification seems chimerical, but then it begs a question of what sort and degree of identification one wants to claim.



    October 22, 2011 at 5:39 am

Comments are closed.

%d bloggers like this: