American (Data) Exceptionalism

by Joshua Tucker on July 5, 2012 · 2 comments

in Academia,Data

University of Toronto political scientist Ed Schatz sends along the following comment regarding a recently published article. Readers should note that he managed to convince the editors of Politiy to ungate the article in conjunction with this post in The Monkey Cage – so perhaps another model worth trying to make more research accessible to a larger audience. Here’s the comment:

America may be “exceptional,” but American political scientists tend to assume otherwise. In a recent paper in the journal Polity (ungated version) Elena Maltseva and I show that political scientists writing in their flagship journal, the American Political Science Review, have generally assumed that data emerging from the U.S. context are applicable to other country-contexts. Fair assumption? Useful assumption? We think not.

Here’s our abstract:

Edward Schatz and Elena Maltseva, “Assumed to be Universal: The Leap from Data to Knowledge in the American Political Science Review,” Polity, June 2012.

The language scholars use to describe research findings has potentially enormous implications for how a science of politics develops. Consider the history of marked and unmarked terms in the American Political Science Review. Modifiers that mark reported data as spatially or temporally “different” (versus linguistically leaving the data unmarked and thus implying that the information is universal and “normal”) reflect predominant power relations. Marking, furthermore, can contribute to future power relations. Finally, knowledge claims that are made without acute attention to the marking of data are likely to be faulty. Because the implications for a science of politics are neither politically nor analytically innocent, political scientists should reveal (rather than conceal) and foreground (rather than background) the geographic and temporal origins of their data.

From this, a modest proposal: let’s “foreground” where our data come from! Whatever generalizations emerge will be far better grounded.


Tobin Grant July 5, 2012 at 12:52 pm

I worked with Ed while he was a prof here at Southern Illinois, and he convinced me of this bias long ago. I think it’s fair — and more importantly scientifically accurate — to

a) include temporal and spatial descriptors in the titles of all studies.
b) exclude them altogether. No need to say, “Legislative Committees in France…”

Or, leave it to the authors to decide if the piece should include a marker saying when and where the data is from.

Ed Schatz July 6, 2012 at 1:28 pm

Thanks. In the article, we make the case for something closer to your choice “A.” “B” strikes us as removing information of possible relevance. Naturally, authors make these decisions for a variety of reasons (your final choice listed), but we implore folks to consider the value of “a.”

It’s not hard, and perhaps the point was as plain as day even before we did this extensive coding exercise, but the broad patterns suggest something worth addressing…

Comments on this entry are closed.

Previous post:

Next post: