is sociology a poor source of policy stories?

A few years ago, I bought a copy of Charles Tilly’s Why?, just for fun sociology reading. All the Important sociology reading got in the way, and I never read Why?

But while I was unpacking this week I came across it and thought I’d bring it along on a car ride to Providence over the weekend. Not only is it a fun read, as well as touchingly personal at times, it turned out to be surprisingly relevant to stuff I’ve been thinking about lately.

The book is organized around four types of reasons people give for things…any things: their incarceration in mental hospitals, why a plane just flew into the World Trade Center, whether the last-minute change of an elderly heiress’s will should be honored. In grand social science tradition, the reasons are organized into a 2 x 2 table:

Popular Specialized
Formulas Conventions Codes
Cause-Effect Accounts Stories Technical Accounts


Why? illustrates these types with a wide range of engaging examples, from eyewitness accounts of September 11th to the dialog between attending physicians and interns during hospital rounds.

Conventions are demonstrated by etiquette books: they are reasons that don’t mean much of anything and aren’t necessarily true, but that follow a convenient social formula: “I lost track of the time.” Stories are reasons that provide an explanation, but one focused on a protagonist—human or otherwise—who acts, and which often contain a moral edge: evangelist Jerry Falwell’s account of how he came to oppose segregation after God spoke to him through the African-American man who shined his shoes every week. Both conventions and stories are homely, everyday kinds of reasons.

Codes and technical accounts, on the other hand, are the reasons experts give. Reasons that conform to codes explain how an action was in accordance with some set of specialized rules. The Department of Public Works did not repair the air conditioning because they lacked a form 27B/6. While law is the quintessential code, Tilly shows that medicine follows codes to a surprising extent as well.

Finally, technical accounts attempt to provide cause-effect explanations of why some outcome occurs. Jared Diamond argues that Europe developed first because it had domesticable plants and animals and sufficient arable land, and lacked Africa’s north-south axis. Technical accounts draw on specialized bodies of knowledge, and attempt to produce truth, not just conformity with rules.


I’ve spent a lot of time in recent months thinking about what experts do in policy, and thinking about the different paths through which they can have effects. Lots of these effects are technical, of course. Expert opinion may not determine the outcome in debates over the macroeconomic effects of tax policy changes or what standards nutrition guidelines should be set at, but there’s no question that they’re informed by technical accounts.

But at least as important in influencing a wider audience are the stories experts can tell. Deborah Stone wrote about these “policy stories” decades ago, though she wasn’t especially focused on experts’ role in creating them. Political scientists like Ann Keller, however, have shown that scientists, too, translate their expertise into policy stories—for example, that human activity was creating the sulfur and nitrogen oxides that produce acid rain, destroying fisheries and making water undrinkable. These stories are grounded in technical accounts, but are simplified versions with moral undertones that point toward a particular range of policy solutions—in this case, doing something about the SOx and NOx emissions that the story identifies as creating the problem.

Some kinds of expertise, or rather some kinds of technical accounts, are more amenable than others to translation into policy stories. Economic models, in particular, are often friendly to such translation. For example, although this isn’t the language I use there, my book in part argues that U.S. science policy changed because of a model-turned-story. Robert Solow’s growth model, which includes technology as a factor that affects economic growth (by increasing the productivity of labor), became by the late 1970s the basis of a powerful policy story in which the U.S. needed to improve its capacity for technological innovation so that it could restore its economic position in the world.

Similarly, a basic human capital model in which investment in training results in higher wages easily becomes a story in which we need to improve or extend education so that people’s income increases.

Sociological models, even the formal ones, seem less amenable on average to these kinds of translations. Though Blau and Duncan’s well-known status attainment model could be read as suggesting education as a point of intervention to improve occupational status, it seems fairer to read it as saying that occupational status is largely determined by your father’s occupation and education. While this certainly has policy implications, they are not as natural an extension from the model itself. It hearkens back to that old saw—economics is about how people make choices; sociology is about how they don’t have any choices to make.

Blau & Duncan

I guess part of the appeal of Why? for me was that it mapped surprisingly well onto these questions that were already on my mind. Mostly I’ve thought about this in the context of economic models becoming policy stories. I wonder, though, whether my quick generalization about the technical accounts of sociology lending themselves less readily to compelling policy stories actually holds up. What are the obvious examples I’m missing?

About these ads

Written by epopp

July 9, 2014 at 7:00 pm

5 Responses

Subscribe to comments with RSS.

  1. what sociology needs, borrowing from economics, is the stylized fact! an empirical example that’s not quite true or overly simplified, but illustrative of an underlying theoretical point.



    July 11, 2014 at 3:06 pm

  2. I think we (sociologists) have a lot of stylized facts, even if we don’t call them that. People with more education get, and stay, married. African-Americans live in poorer neighborhoods than whites with equivalent incomes. Etc. They don’t necessarily tell you what to do, though, or even suggest one theory vs. another (though some of them may imply a causal relation).

    Dan Hirschman has an interesting working paper on stylized facts that touches on their role in policy, or at least in circulating ideas beyond academia. I can’t find a version online, but perhaps he’ll jump in here.



    July 11, 2014 at 5:48 pm

  3. Re: Stylized facts – I do have a working paper (based on a presentation at last year’s Junior Theorists Symposium). I’m not quite ready to post it, but if anyone is interested, send me an email and I’d be happy to share it.

    In terms of stylized facts as policy stories, I actually think we should make a sharp distinction. The term stylized fact, both as originally used by Kaldor and other economists and in more recent usage, do not refer to “an empirical example that’s not quite true… but illustrative of an underlying theoretical point.” Rather, stylized facts are simple empirical regularities in need of explanation. They are a kind of Theory3 (cf. Abend), interpretive descriptions, based on smoothing out the data (metaphorically). Three basic forms of stylized facts are rates of incidence (x% of people have y trait), trends (the share of finance in the US economy is growing over the past 40 years) and correlations (countries with high debt experience relatively low growth; children of same-sex parents do (or do not) have worse outcomes). But stylized facts are explicitly not causal theories or models.

    That said, I think stylized facts frequently play an outsized role in policy debates, although perhaps in ways that should make us nervous. Specifically, stylized facts (relatively simple empirical relationships in need explanation) often meet in the public and political sphere with folk causal theories and count as evidence for those theories. So, Regnerus claims that children of gay or lesbian parents* have worse outcomes. But he explicitly, at least in the published paper, did not claim to have a causal theory. For example, his identified stylized fact is quite consistent with the (plausible and politically quite different in tone) claim that because gay parents experience discrimination (lower income, fewer state-granted protections precisely because they are denied marriage, etc.), their children will have worse outcomes. But of course, that’s not the folk causal theory that the stylized fact met up with in public. Instead, the assumption was that gay parents were worse in some (often vague) way, and thus gay parents (rather than, say, societal discrimination against those parents) caused worse outcomes for children. This was the story that was presented in the CA same-sex marriage case as an attempt to rationalize CA voters’ behavior: (1) stylized fact of a correlation between parents’ sexuality and children’s outcomes + (2) folk causal theory about same-sex parents being bad parents –> (3) rational for CA voters to want to discourage same-sex parents.**

    A parallel case, which I also deal with in the paper, is Reinhart and Rogoff’s so-called debt threshold (that countries with debt/gdp ratios over 90% grow significantly slower). Again, in the published paper, no causal claim. In the political scene, quick jump to causal claim used to motivate austerity policies.

    So, to summarize what I’m trying to say: sociology does have stylized facts (as Beth mentions), but stylized facts are best thought of as simple empirical regularities not simplified examples to demonstrate a causal argument; these stylized facts do play a role in public debate and policymaking but not always for the better (though we can surely generate examples of more positive influences of sociology and economics stylized facts in political debates). But none of this quite directly at what I think the point of Beth’s post was: how can Sociology get its causal arguments, not just its identified empirical regularities, into better policy use? We should think about the relationship of data to stylized fact to theoretical model at this interface, but the relationship is, well, messy.

    *Note, as readers of this blog and Scatterplot surely know, that’s not quite what his data show (because of how he operationalized the categories), but that is what was claimed (roughly).
    ** This argument is bonkers for other reasons well-documented on the blogs, but again it’s the logic I want to emphasize here.


    Dan Hirschman

    July 11, 2014 at 7:26 pm

  4. Thanks, Dan, that’s very interesting. I agree that stylized facts tend to lend themselves to interpretation in light of folk theories. In effect, they are read as evidence for whatever theories we already believe, even when they don’t actually test those theories.

    I -am- interested in how sociology can get its arguments into policy use more effectively, but beyond that, I think I’m grasping toward some kind of epistemic privilege idea, a la Somers and Block (though a different kind of epistemic privilege) — that some types of knowledge claims may simply translate into policy conversation more easily than others. Mostly I’m just playing around with ideas here, though.



    July 12, 2014 at 3:59 pm

  5. I have recently read two things where sociology has been effectively “translated” into policy positions: The Case for Reparations by Ta-Nehisi Coates and All Joy and No Fun by Jennifer Senior. Interestingly, neither is a sociologist. But what they do is convey stories to which we can empathize or relate. Coates provided a face and narrative arc around well-established ideas in sociology. In fact, if you read the piece, it follows Kenneth Jackson’s Crabgrass Frontier, Robert Sampson’s Great American City and Pat Sharkey’s Stuck in Place. He gives credit where credit is due, but provides a narrative that moves sociologists’ more academic prose very convincingly. Similarly, parts of Senior’s book hit so close to home (a compliment to her writing and observation) that I felt physically uncomfortable or wanted to talk to the parent to say: “I understand, trust me.”

    I wonder if there is a way that we can convey our own work in a better way?



    July 14, 2014 at 8:29 pm

Comments are closed.


Get every new post delivered to your Inbox.

Join 1,157 other followers

%d bloggers like this: