A Modest Memo
TO: unreflective, unreflexive neopositivists of the social-scientific world
RE: your insulting, dismissive, and coercive use of the word "empirical"
It has come to my attention that for some time now y'all have been using the word "empirical" in very unusual and baffling ways. You say things like "can you prove that empirically?" when critiquing anthropologists and historians, and things like "you don't do empirical work" when referring to scholars who do discourse analysis, textual interpretation, and the like. You even say things like "now there's
an example of what empirical work can get you!" when someone presents a large-n quantitative study, but make no such comment when someone presents their detailed interpretation of a series of historical documents or their richly detailed discussion of the (causal
) impact of public uses of language.
News flash: "empirical" does not mean either "statistical" or "quantitative."
Empirical, according to my trusty OED, means "pertaining to, or derived from, experience," although it also carries connotations of medical quackery (e.g.: proceeding to do medical treatments without a firm scientific and theoretical foundation) and an emphasis on observation as a source of knowledge. Indeed, according to the OED the word first arose in the context of a debate about whether experiments were a valid source of knowledge -- whether one could start with observation and proceed to develop knowledge on that basis. From there the word mutates slightly so as to encompass a general orientation to the world, retaining an opposition to "theory" but also undergoing several modifications in the course of fractalized disciplinary debates about the status of knowledge.
The point is that "empirical" never
meant "statistical" or "quantitative" unless you were a statistician or quantitative analyst, and even then I think it meant something else. Statisticians have empirical data, but this is far from sufficient to define their endeavor; what makes a project "statistical" is how one organizes that data and what one does with it. Instead of a descriptive term, "empirical" used by a statistician as a critique of non-statistical work is just an insult
, and only makes sense in the context of an overarching consensus that statistical modes of reasoning are the best guarantor of Truth.
This might -- and I stress might
-- have been a sustainable position in the social sciences in the 1950s, and even then only if one ignored the work in the philosophy of science that raised doubts about precisely what the mathematical manipulation of quantitative data could achieve. But that people continue to reproduce this myth today baffles me; did y'all simply miss
the last twenty or thirty years of very public debate about this?
And come on. Seriously. Are you really
prepared to say that someone who goes out into the field and lives with a group of people for an extended period of time, learns about their cultural set of meanings, and then reports back in the form of an ethnographic account has no empirics
? What about a diplomatic historian who uses, say, the complete record of negotiations carried out between the great Powers during the eighteenth and nineteenth centuries as the basis on from which to develop their reconstruction of events and trends? Or the discourse analyst who sifts through years and years of Congressional testimony and parliamentary debates in order to precisely track the deployment of particular rhetorical commonplaces and themes?
Sheesh. "You don't really do empirical work" my ass
. Let me tell you something: "I ran regressions on a data set" is probably less
empirical than scholarship that is based on fieldwork and detailed documentary analysis. Coding data is empirical work, but in that sense it is no different than participant-observation or the disclosure of central rhetorical themes or network ties. everything that uses data is "empirical."
Social science is "empirical" by definition
, inasmuch as it is constitutively about making claims about the world that are sustainable with evidence of some kind. That means statistical analysis, sure, but it also means interpretive and relational modes of inquiry.
[Note that there can be better and worse empirical work in all of these camps. But that's a secondary issue; just because there's some really bad interpretive work out there it does not
follow that interpretive work is somehow not "empirical."]
I suggest that all of you go back and read Weber's essay on "objectivity," and then get back to me about what is and is not "empirical."[Posted with ecto]