At the request of Ann Arbor’s OrgTheory-ist in residence, I’m going to write up last Friday’s excellent ICOS lecture by Wendy Espeland. ICOS has a hilarious lecture series riddled with traditions, and they bring excellent speakers. For example, before Wendy spoke, Jerry Davis (business school prof and ICOS coordinator) introduced Greta Krippner (sociology prof and guest curator) who introduced Natalie Cotton (business and soc PhD student) who finally introduced Espeland. ICOS also has its own espresso machine, and delicious cookies. I highly recommend it. Onto the talk itself.

Warning: This report is a bit long, so if you are interested in the short version just read the next paragraph and the last two.

Espeland’s talk was based on a paper forthcoming in the European Journal of Sociology (co-authored with Mitchell Stevens) entitled “How to Study Numbers: A Different Kind of Quantitative Sociology.” The paper is intended as a programmatic statement for the growing field of the sociology of numbers/statistics/quantification. Espeland has been writing about quantification and specifically commensuration for more than a decade now, and her examples ranged over her substantive interests from water rights in the American Southwest to Law school rankings. The examples were used to justify an agenda for the sociology of quantification focusing on 5 particular aspects: Work, Reactivity, Discipline, Authority, and Aesthetics. In this report, I will briefly explain each of these five aspects, especially emphasizing the most novel aspect for me (“aesthetics”). First, though, I want to talk a bit about the theoretical link between Espeland’s proposed sociology of quantification and the performativity of economics literature.

Espeland’s sociology of quantifications starts with the insight that “numbers are deeds”. Drawing on JL Austin’s famous How to Do Things With Words, Espeland argues that numbers are more like performative utterances than descriptions of reality. She borrows Austin’s idea of “infelicities” to argue that numbers can be wrong “in ways other than their accuracy as representations of the world.” I was a little disappointed that this line of inquiry did not go further – the logical question following Austin would be, under what conditions do numbers fail to do their work? Instead, Espeland took things rather in the direction of arguing that quantification is incompatible with certain understandings of the world, as in the case of a proposed dollar value for the land of a Native American tribe in the Southwest who felt that quantifying the value of land betrayed their principles. The land was priceless.

The recent literature on the performativity of economics (Callon 1998*, MacKenzie and Millo 2003, MacKenzie et al. 2007) also draws their inspiration from Austin’s performativity. The performativity folks, however, are much more interested in epistemological question. The “killer app” of performativity is the Black-Scholes-Merton options pricing model. MacKenzie and Millo (2003) showed how this model provided only a rough guide to options pricing when the CBoE options exchange first opened, but as use of the model increased (and regulatory rules were changed to allow more flexible position-taking) the model’s predictions became more accurate. The key questions here in the sociology of finance are under what circumstances do models succeed or fail to be performed (e.g. a story about felicities and infelicities). Will any model do? What role do political actors play? Etc.

Espeland, as I already mentioned, is less interested in felicities and infelicities, or accuracy and epistemology, and much more interested in the unanticipated consequences of purposive social quantification**. Espeland’s proposed sociology of quantification looks much more like Ian Hacking’s analysis of biopolitics rooted in Foucault’s work than it does MacKenzie and Millo’s work (grounded in Latour and Callon’s Actor-Network Theory)***.

Now, onto the meat of the argument. First, quantification takes work. This point will not be novel to any sociologist who’s spent hours pouring over codebooks, weeding out missing data, tracking down coding errors, running .do files, etc. Espeland’s excellent example showed a page of graphs from a recent article on climate change that tracked changes in the average temperature of Lake Bikal. Three generations of a family of scientists had painstakingly collected data over 5 decades to generate those graphs. Years of work were condensed (seemingly condensed) into a few inches of dots and lines.

Second, Espeland names the process by which quantification can remake what it measures reactivity. Here she drew on Kinsey’s studies of sexual behavior. While Kinsey did not believe in such essentialized categories as homosexual or heterosexual, political actors on all sides drew on his work to make claims about identities. If 10% of adult men behave in almost exclusively homosexual ways, then activists can begin to claim that 10% of all humans are homosexual.

Here is a good example of another distinction Espeland emphasizes: the differences between numbers which mark and numbers which commensurate. Numbers which mark (like addresses or library call numbers) are a form of categorization, but there is little numerical about them. Any ordinal system would do. Numbers which commensurate – prices, rankings, metrics, etc. – do a lot more work and are the real object of Espeland’s sociology of quantification. In this, she sees her work as apart from existing literatures on categorization (e.g. Bowker and Star 1999). That being said, I fear there is a tighter connection between these two kinds of numbers than Espeland admitted and that, following Hacking, the categorization systems required for quantification may have a bigger impact than the eventual numbers themselves. Kinsey is no longer remembered for the exact numbers in his study as much as the categories he helped create, sustain and alter (homosexual vs. heterosexual, Kinsey “1” or Kinsey “6”, etc.). Still, Espeland is right that categorization only gets at part of the story of quantification, and a part more easily dealt with by existing sociological work.

Third is discipline (here comes the Foucault!). Most readers will probably know where this line of argument goes – rankings and other metrics are internalized and begin to alter our behavior without any direct action on the part of the rankings-makers. Espeland draws here on her recent work on US News and World Report law school rankings. For example, because the rankings counted median LSAT scores of only regular full-time students, law schools began to push weaker applicants into doing part-time and night programs. In response to a proposal by US News to start including those programs in the median LSAT calculations, law schools formed committees to ponder reforming or closing their night and part-time programs. Organizational discipline in action.

Fourth is authority. Here Espeland shows her deep gratitude to the work of Theodore Porter (especially Trust in Numbers). Numbers are used as authoritative, mechanically objective, representations of the world which can be used to do work. In Porter’s terms, numbers are rhetorical “technologies of trust”. The example given was complicated sentencing guidelines that endeavored to remove inequalities in the criminal justice system by standardizing punishments for crimes (and of course, true to Hacking or Scott, ended up creating a new, massive racial disparity). Here Espeland might have been better served by calling the aspect “objectivity” instead of authority as I think that hits closer to how she is using the term, and matches nicely with Porter’s work.

Last, aesthetics. Here is one place I think Espeland makes an especially novel contribution. She argues that numbers are powerful in part for their aesthetic qualities. She draws nicely on the work of people like Tufte to argue that making a compelling statistical graph, for example, is an art (and an art of fairly recent vintage, dating back perhaps to 18th century cartography). She analyzes as an artistic object Blau and Duncan’s famous path diagram model of the relationship between father’s education and occupation to son’s. The graph is compelling for several reasons. First, it draws on mountains of data that seem to have everything essential about them captured within the graph. Second, it seems to tell a story, or really a meta-story, about how parents and children interact, and how a father’s circumstances structure a son’s. Education and occupation have independent effects and effects through education. We read the coefficients and lines on the page as a story (in part because Blau and Duncan spend much of the book teaching us exactly how to do such a reading). A less compelling graph would make a less compelling story. The aesthetics of quantification matter.

So, to quote a famous musical, where do we go from here? Espeland argues that this framework will give sociologists traction on quantification as an ethical process, as well as stimulate research into the dependence of quantitative research on qualitative**** (like the narratives that underlie Blau and Duncan’s diagram and make it so compelling). I think Espeland’s framework will be very productive for scholars focused on quantification per se, especially those interested in aspects beyond categorization on the one hand and accuracy on the other. I’m definitely looking forward to seeing the final paper and the book to follow.

* As an aside, there are few books more in need of being reprinted than The Laws of the Markets. Amazon has only one copy available for $120, and that’s the cheapest I’ve seen it all year.
** Hmm. That’d make a pretty neat paper title. I should write that down. Oh.
*** It’s possible that I knew the meanings of almost none of the words in that sentence a year and a half ago.
**** During the Q&A, Greta Krippner questioned the opposition of quantitative and qualitative research running through Espeland’s talk. “Can they ever be decoupled?”, she asked. Espeland’s response began, “You just blew up my dualism Greta!” She went on to argue that focusing on quantification can help illuminate the important cultural work of classification that precedes it. If there’s one major criticism I have (which I will bury here in this footnote) of the paper is that it did not spend enough time addressing that cultural work, and the complex connections between categories (and numbers that mark) and measures.