sociology 2043 – guest post by graham peterson

This is a guest post by Graham Peterson. He has just finished up his master’s degree in economics at UIC and will begin the PhD program in sociology at the University of Chicago. He is interested in economic sociology and extended blog comments.

It’s a shame sociology took its final-swoop quantitative turn just about the same time economics reissued its permissions to do comparative history and started publishing discourse studies in its mainstream journals.  It’s as if estranged siblings would be forever doomed to blow past one another and reissue each other’s mistakes.  Politics.

The turn for sociology was of course both theoretical and empirical, but that damned dissertation in 2021 really sealed it: Foundations of Sociological Analysis (Paul Samuelson glowed proud from his grave).  It was as if sociologists had been waiting for someone to come along and resolve all the definitional issues in theory and compose an axiomatic graph-theoretic derivation of sociological principles.  Now all the theorists do is applied combinatorics on graphs.  Useless.

Of course existence theorems aren’t actually quantitative, even with the Greek.  No, the empiricists test the theorems.  They measure and decide what the Big Variables are — ostensibly.  However it doesn’t take a genius to see that entire careers are being made on increasingly useless precision in metrics – one more marginal regression on the NSA’s Limited Access Comprehensive Social Network to seal sociology’s feckless irrelevance.  The newspapers used to say the same things about economics and national income accounts.  Shame.

We were at one point some of the greatest champions of a healthy philosophy of science.  As late as 2013, in fact, methodologists and practitioners in both sociology and economics started talking again because of the rebirth of institutionalism and history in economics, and the excitement around quantitative methods in sociology.  But the mainstreams of both disciplines were too narrowly obsessed with their own insular histories.  Too many barking dogs.

We now face criticism from every side for the failures of our policy recommendations, which governments eagerly gobbled up for their scientific appearance.  Who can blame them, they were fed up with economists.  But even after the quantum computing turn, even after the reallocation of educational resources with Intergenerational Path-Dependence Structural Interruption hasn’t moved the Gini Coefficient much.

It seemed to me in 2013 that sociology was taking the theoretical and methodological turn that economics had in the 1930s and 40s, when much was gained for economists, but so too lost.  If only sociologists and economists had talked to each other more, we wouldn’t be, in 2054, the same mess economists had found themselves eventually, in 2013.

Adverts: From Black Power/Grad Skool Rulz

Written by fabiorojas

July 12, 2013 at 12:01 am

11 Responses

Subscribe to comments with RSS.

  1. Ha, thanks Professor Rojas. I hold a BA in economics from UIC though.



    July 12, 2013 at 4:57 am

  2. So at least two people who read that said they had nearly no idea what I was talking about. Cutesy creative writing is not my strength.

    Regardless the usually sophisticated methodological and epistomolgical discussions, people’s choices of methods often distill into basic pllitical a d institutional heuristics. One of these is of course “math is scientific” Another one is “math is hard therefore rigorous.” To be sure, sociologists are more aware than likely any group of social scientists how ridiculous those ideas are. Yet I see them taking root as sociology experiences a sort of cascade of quant methods in terms of both theory (simulations, networks) and empirics (big data, advanced metrics).

    Economics is now at a point where its mainstream has begun to heed the advice of its critical methodologists. Mathematical theory offers nothing in the way of “precision” that literary theory cannot. Changing the assumptions of a model changes its conculsions even if one is “bound” by the “rigor” of deductive implications to arrive at its conclusions. Assumption-making is just as ad hoc and arm chair as any other sort of theorizing. This is not a problem, and can complement literary theory, but the idea that it enjoys a priori priviledge is ridiculous. Clearly sociology isn’t going full blown applied graph theory any time soon. But it’s worth noting the enormous mess economics got itself into with the excitement over game theory: behavioral economic theory is now being subsumed into mainstream micro theory in econ, and these guys are arguing about whether or not the models are fully generalizeable mathematically, which seems mind blowingly stupid to me, and defeats the entire purpose of the behavioral turn (i.e. let’s pay attention to what people actually do). Macroeconomic theory also gets called story telling now — of course story telling is all social scientists really do period, but mathematical theorists born in the land of Derp will not admit that. Sociological theorists interested in the usefulness of information, graph, game, and other math dept manifestations ought to pay attention to what can happen institutionally if sociology loses its edge in front-and-center critical methodology. Mathematical theorizing is at some threshold, easily met, just jargon-making like any other runaway band of armchair nose pickers.

    As for empirics, social scientists forget that they’re unable to “measure” anything – we estimate things. And estimations come with error bounds. And it is the judgment of those error bounds relative to the social relevance and urgency of a question that determines the loss function of a sample. The natural experiment revolution in economics has led to researchers ignoring the economic significance of questions and shooting for identifiable models for their own sake. But precise measurement of nothing means precisely nothing. Precision, representitiveness, randomness — all of these mean absolutely nothing if pursued with the “blindfolded researcher’s” faith that they exempt her from making judgment calls of taste and fashion in picking and interpreting data. Then there’s P-values. Virtually all that matters is the Beta of a variable. I saw someone complain at a big data workshop that no p-values were presented for an estimation – estimation with millions of observations. This is the kind of knee-jerk methodological nonsense that sociology is just as liable to become infected with as economics.

    Sociologists whine a lot that their work doesn’t have more policy influence (I think this position denies the just staggering degree to which pop-sociological translations get bandied around, while for instance your garden variety market-defending conservative can’t make a basic supply and demand argument, but I digress). Sociology can and will, with greater quant work, have more influence on policy, because the man in the street finds statistics magical and impressive, but sociology will also find itself the subject of the same criticisms economics now does for “not predicting problems,” and making awful recommendations.

    Interestingly, groups all over use themes about purity and cleanliness to define the in-group from the out-group. Quants you see talking this way, championing “clean” parsimony, and “cleaning up the mess” qual theorists and empircists have made. Be wary. Heterodox economists and crtical methodologists have been secod class citizens in economics for decades (making messes), because a large portion of their corpus is merely a translation of center stage epistemological and methodological discussions from sociology and the humanities into economic terms.

    To turn the economic phrase, qual and quant methods are complements, not substitutes.



    July 12, 2013 at 1:17 pm

  3. So, six paragraphs, and then six more long paragraphs, and the take home is:

    “qual and quant methods are complements, not substitutes”

    This is a perfectly reasonable position to hold. I hold it. Many actual sociologists hold it. Perhaps most. What I don’t get is why you present the position as if it’s an attack on some established order that thinks otherwise. It is not. Your position is the established order.

    An even bigger problem is your assumption that sociology’s undergoing some quantitative turn. The quantitative turn in US sociology happened decades ago. What US sociology is dealing with is a turn toward the internet, i.e. Big Data. Big Data is important because it rewards new skills, makes new data available, requires new theories, and potentially moves the center of sociological practice from the university campus to the private sector. (I said, potentially.)



    July 12, 2013 at 7:11 pm

  4. eh. the “quantitative turn” in sociology that you suggest is severely hindered by the fact that most undergraduates choose sociology as a way of running away from math and other “hard courses”, rather than to try and understand the staggering complexity of social life.



    July 12, 2013 at 10:24 pm

  5. Graham,

    In contrast to Austen and Anon, my reaction was hardly “eh.” But perhaps that’s because I’m a heterodox reader (well, lurker, really) on this blog. The issue, I think, is that you’ve waded beyond the sloganeering lip service paid to mixed methods and into the deep waters of how this clashes with the pseudo-hard-headed “this is how the world is” assumptions about the current academic state-of-play. I take you to be saying that this latter point is what reveals “political heuristics.” (Although why “heuristics”? “Commitments”? Some less jargonistic word?)

    Anyway, you can take these two dismissive comments as actually demonstrating your case neatly. Austen writes:

    “What US sociology is dealing with is a turn toward the internet, i.e. Big Data. Big Data is important because it rewards new skills, makes new data available, requires new theories, and potentially moves the center of sociological practice from the university campus to the private sector. (I said, potentially.)”

    Essentially all of the skills, theories, and movements-of-the-center-of-practice to which Austen is referring (at least that I’ve seen) are quantitative and technical-engineering. If this is *not* true, please disabuse me. But literally, institutionally, if this shift (potentially) takes place, where is the qualitative counterpoint to be going to be housed? Part of Graham’s point is that while we in sociology love us some heterodox economics (since its basically sociological), its influence in economics itself has been nugatory, so that hardly seems like a desirable example.

    Moreover, Anon writes a comment with nearly the identical meaning, but phrased more bluntly:

    “eh. the “quantitative turn” in sociology that you suggest is severely hindered by the fact that most undergraduates choose sociology as a way of running away from math and other “hard courses”, rather than to try and understand the staggering complexity of social life.”

    This comment (echoing Fabio’s complaints about lack of programming skills in sociology) strongly implies that “to try and understand the staggering complexity of social life” one ought to be steeped in math. This, to me, anyway, suggests in turn that this is the privileged means of getting to understand social life in all its “staggering complexity.” Although, once again, correct me if I’m wrong.

    (And of course: “But we already have a mandatory theory course and usually some kind of qualitative logic-of-inquiry! Why don’t we also require everyone to learn python and advanced algebra?” Well, fine, but what’s almost always missing from these discussions is how we can improve BOTH qual and quant training. Instead, the claim usually goes something like: “Sociology is bad at both qual and quant. Let’s improve quant.”)

    Anyway, I hardly ever comment here because while Org Theory is many things (many useful!), it is hardly a pluralistic methodological or academic-political discussion. But you seem anxious, and I wanted to underscore that, these two comments aside (which validate one of your central points), at least one reader here understood your point, and thought it was a good one.



    July 13, 2013 at 2:05 pm

  6. I come from a discipline, political science, where the vast majority of established scholars pay lip service to “mixed methods” as Anonnity describes it. In public, they wax poetical about how quantitative and qualitative methods are complements; and, in reality, most understand little about the advantages and disadvantages of either and communicate only with their methodological allies. There is little realization and concern when quantitative methods are grossly misapplied to study problems that they cannot provide meaningful answers to; and equally little recognition that quantitative methods provide powerful solutions to some problems formerly considered unsolvable. This narrowing of methodological perspective – and the reflexive urge to avoid hard and uncomfortable debates about methodology – is what I believe Graham to be criticizing.



    July 13, 2013 at 2:31 pm

  7. “Essentially all of the skills, theories, and movements-of-the-center-of-practice to which Austen is referring (at least that I’ve seen) are quantitative and technical-engineering. If this is *not* true, please disabuse me.”

    The move toward Big Data is not a move into quantitative methods from something other than quantitative methods. It is a move to quantitative methods from other quantitative methods. Any lamenting over a quantitative turn is decades late.

    “where is the qualitative counterpoint to be going to be housed?”

    There is qualitative work being done on campuses and in the private sector. No need to despair, or devolve into melodrama. If you want to do qualitative work, that possibility does exist and will continue to exist.



    July 14, 2013 at 1:42 am

  8. “my reaction was hardly ‘eh.’ But perhaps that’s because I’m a heterodox reader (well, lurker, really) on this blog.”

    I think you and the author of this post overrate how radical — sorry, heterodox — your positions are. Methods are complementary. Yes. True.



    July 14, 2013 at 1:46 am

  9. In my quite limited experience I’ve found that undergrads and some grad students make a big row about some qual/ quant divide but for most working academic sociologists (or people in other disciplines that use multiple methods) these debates are a bit silly. I think the perception of some inexorable methodological divide is largely a result of how methods are taught in distinct sequences.
    Since at least Durkheim, sociology has always been both qual and quant. I don’t really see what the big deal is.
    As an aside, the best and most useful critiques of quantitative methods tend to come from people who are good quantitative methodologists. I’ve read various critiques of statistical methods from various standpoints (which ranges from strands of feminism, postmodern, “Austrian” economics, etc.) and the critiques tend to ring a bit hollow for me. I never know what I am supposed to do afterward…..


    Silly Wabbit

    July 14, 2013 at 4:31 pm

  10. While I agree that sociologist tend “to avoid hard and uncomfortable debates about methodology,” I think its a mistake to see this as indicating latent conflict. I think most people have a gut feeling that their method is better but it rarely goes beyond this. Like @Austen said, quantitative methodology has long been dominant in the field but qualitative methodology is established and legitimate enough that there is no conflict really.

    And I think its an even bigger mistake to hold up the “Big Data turn” as an example of how this fundamental divide keeps widening. In the last couple of posts we had about Big Data, it seems that most of the discussion was about why it was not catching on in Sociology fast enough, even though (theoretically) it opens us to a new found wealth of information about social life (i.e. its something that sociologists should be super-excited about). From there you get into a discussion about how the people with the necessary skills to handle this data (i.e. engineers) often never become comfortable within academic sociology. I think its because our quantitative sociologists (those using regression and a few other methods) are probably more likely to cite the findings of ethnographers (without actually questioning their methods)–and certainly of other regression-based quantitative sociologists–than computational sociologists. We can have a big discussion about why that is true some other time. For now I would say that this to me certainly doesnt seem a like a “quantitative turn” leaving qualitative researchers in the dust.

    As a side note I would add that political science does seem to have debates between quantitative and qualitative sociologists about methods; I’m thinking of books like “Redesigning Social Inquiry,” and the more recent “A Tale of Two Cultures” by James Mahoney. It seems like this literature is really trying to establish a shared standard between quantitative and qualitative methodologies.


    soft scientist

    July 14, 2013 at 6:04 pm

  11. Thanks for all the comments. I haven’t been able to post any furious and obsessive responses because I’ve been in an electricity-less cabin in Canada (where we bathe in the bay, and where happy hour starts at five sharp).

    Sef really capitulated my point well I thought. Everyone else got it (whew!), but I of course don’t agree with Austen and soft scientist. Silly Wabbit is absolutely right – most critical methodology is trite, old hat, and empty of recommendations. It’s like listening to hippies whine about The Man and offer absolutely no alternative for how we ought to organize society other than “everything’s a circle, man. Are you going to Red Rocks? The acid out there is wicked right now.” Wait, that reminds me of my intro soc courses at community college years ago. Anyway.

    Back on track: wabbit is right — methodology is obnoxious, usually done by people who aren’t themselves practicing the method, and tends to start incredible fights among people who are married to the professional tools and epistomological approaches they’ve chosen. It’s very personal and political. But as the mathematical economist Ariel Rubenstein says (who’s himself severely critical of mathematical economics): “if the only people who were allowed to criticize a belief were people who subscribed to it, there would never be any criticism.”

    This is the second time that soft scientist has essentially said that my imagined universe of deep conflict over methods, and especially those chiefly used by economists vs sociologists, is old family business that got worked out in counseling 20 years ago, and I’m late to the party and just kicking up dust because I like attention. Now, clearly I like attention. But the position above espoused that this issue is in any way resolved in human sciences (or “human studies” if your a humanities leaner), is just wrong. And I think it’s potentially Austen’s and soft scientist’s lack of work at these boundaries which conveys the impression to them that it’s just no biggie anymore.

    A non-zero number of sociologists will flip out if you categorize them in the humanities – this while a majority of European sociologists continue to do work that is nearly indistinguishable from some of the work in modern American departments of English and Anthropology. A bunch of people here took offense to my lack of recognition of the long history of statistics in sociological demography, and now increasingly othet subfields. And that pomo thread where Rojas, Rossman, and others took on the softies (with notable philosophical erudition you wouldn’t hear from most economists), was one of the longest in this blog’s history – and totally epic regardless. So much for a collected resolution over methods. I wonder if the fact that the modal first year sociology graduate course in metrics is taught at the level of a T30 economics BA econometrics course has anything to do with this defensive insistence that “WE USE NUMBERS AND HAVE BEEN FORVER SHUT UP!”

    But my point was that the point for sociology shouldn’t be, as it had been for loads of economists, that if we just make it more mathy everythig will be fine, because you know the high priests said so. There is I think a cascade that’s taking place now in sociology where for instance its junior demographers have a keen interest in Bayesian statistics and other advanced methods, and it should be met with caution. Cascades, after all, are predicated on rational imitation – and I’ll tell you what sounds a lot dumber than my two poorly constructed posts above – the imitative drool that comes out of a lot of people’s mouth’s when they’re challenged to defend their epistemological first pinciples.

    This is one area I actually have some business commenting on: I’ve been working for a very well regarded methodologist for two years. Do you know anyone who holds professorships in both economics and English? Also, I’m not sure my more fervent critics on this blog have ever read any heterodox economics — heterodox economics (no, not idiot Austrians and their gold bug conspiracy theories and pure-philosophy-no-empirics dogmas) is essentially a sociological response, in its theoretical base and empirical methods, to economics. Add to that, institutional economics and economic history, and you’ve got not just the fields I’ve trained in mostly, but essentially a little sociology melanoma on economics’ right hand. (Yet people here are crying because I haven’t yet read Durkheim cover to cover. You sound like the mouth breathers on EJMR telling undergraduates to “go read MWG [Mas Collel, Winston, and Green – the standard graduate micro text] and GTFO” – you can do better guys.)

    If anyone here thinks there isn’t still a lot of tension between qual and quant, they haven’t for instance put a positive political theorist and classic political theorist in a room together lately and listened to the silence. Nor have they heard the warnings I have, either personally or publicly conveyed by senior and middle aged sociologists and economists (several of them senior posters here and at Scatterplot) that reviewers and conference attendees are liable to freak out when they smell (for economists) something other than revealed preference empirical methods (e.g. experiments or price estimations) or (for sociologists) rational choice models which attempt to usurp sociological territory. Don’t forget that economic and sociological demography split into fully two different departments because of arguments over the theoretical base, and these people have had trouble getting along at conferences for the last two decades. I’m not making this stuff up because I want to be a maverick, pace John McCain. The bit about demography comes from one of my UG referees, who’s worked on religion and family (not McCloskey).

    I was trying to get past some of these institutional details, and get people talking about some actual nuance and substance. For instance, formal and verbal theorists do EXACTLY the same thing (though note that if you use mathematics, you get to call your work formal, which confers status). Theorists use themselves and accumulated experience as an inductively- derived pool of bits of data, and come up with stories to explain what they feel and witness. The game theorist is no different than the verbal theorist, even if his final product is presented in deductive proofs, because everything he does depends on how he defines his payoff functions before he goes hunting for an equilibrium. Just like a verbal theorist defines his assumptions about motives and behavior before he derives, to his best efforts with logical consistency, conclusions.

    That’s just one of a range of issues in the qual quant divide. And of course there’s been progress. But true mixed methods and interdisciplinary people get, well, mixed responses. The lucky ones you actually hear about are people who have done a very difficult job reasonably well enough to get noticed – and these people, Becker for instance in economics (there are of course equivalents in sociology and poli sci I just don’t know about yet), report facing incredible hostility concomitant to their successes. These issues are extremely important, and not just among a third-tier group of cranky and failed economists or something.

    Junior scholars are the last people who should be dismissing methdological criticisms and concerns – because they are at precisely the stage where methdolgical allegiances and identities get formed. Once formed, those professional identities and networks and boundaries are the reason that conversations about methods become so diffuclt to have. I was just talking to Howie Becker about some of these issues last week — the urgency here conveyed is not the product of too much volume and too little brains and experience on my part.

    Jim Moody showed that sociology is located, of course, smack in the middle of the topology of social scientific citation networks. There is incredible opportunity for this conversation in sociology. I’m not saying the conversation didn’t start long before I showed up picking my nose — I’m saying it needs to keep going, and vigorously.



    July 18, 2013 at 7:07 am

Comments are closed.

%d bloggers like this: