Social Constructions of Measurement: Three Vignettes from Recent Events and Labor Economics
Jennings, Ann, Journal of Economic Issues
Modern life is permeated by a fascination with numbers, measurement, and quantification. As Theodore Porter (1995) put it, we have come to "trust in numbers." Though we know it is possible to "lie with numbers," we nevertheless rely on them, appeal to them, and are disturbed when they seem questionable. Moreover, behind the notion that numbers can lie lurks another, undeclared notion that good numbers "tell the truth." This essay explores such beliefs by means of three vignettes, each intended to reveal some aspect of the social foundations of measurement processes that we commonly take for granted. Measurement, I argue, is inseparable from the meaning and content of the things we want to measure; both our doubts and our complacency about measurement may therefore have important social implications.
First Vignette: Voting in Florida
Consider the recent US presidential election. After 36 days of counting, recounting, and legal maneuvering in Florida, Al Gore conceded and George W. Bush became the president-elect. The final margin of victory in the state was less than .01 percent of the counted votes--far less than the statistical margin of error for the voting equipment used and than either the number of uncounted ballots or those in dispute; other polling irregularities were also reported. Though Florida's "sunshine laws" allow for further examinations of the uncounted or disputed ballots, any new measures of the vote will remain inconclusive and/or moot. As US Supreme Court Justice John Paul Stevens said, "we may never know with complete certainty the identity of the winner of this year's presidential election" (quoted in Newsweek Dec. 25, 2000, 49).
Prior to this spectacle, how many would have imagined that a "simple" matter of counting the votes cast for each candidate could be so complicated or involve so many potential snares--aside, even, from any questions of improper conduct? Part of this measurement problem, of course, is the sheer scale of the undertaking: over six million votes were cast in Florida alone, and large numbers are often hard to count. But far more is involved. Counting and measuring are not as simple as we often imagine, even when the things we want to count (this vote for Gore, that one for Bush) seem unambiguous, as is decidedly not the case in many measurement situations. All measures are social constructions, requiring a more or less elaborate social and technological apparatus to define what should be counted, for tabulation, and to identify and correct mistakes and errors. The apparatus is also necessary, however, to give the resulting measures social legitimacy. We have a winner in Florida, and in the nation, but Justice Ste vens' remark implies that a flawed apparatus may not produce a "legitimate" winner.
As the Florida case reveals, an election is not a direct, transparent expression of "the will of the people" or even the somewhat simpler "intent of the [eligible] voter." We might develop a better apparatus, more precise in its calibration, to help us with very close races like that in Florida. But Florida was so shocking, in part, because we thought we could take these measurement processes for granted. The case also raises questions, however, about what concepts like "the will of the people" or the "intent of the voter" might mean, independent of the voting apparatus itself. These concepts are themselves social constructions, and their meaning depends partly on the companion constructions by which we attempt to measure them. Without the apparatus, they would inevitably mean something quite different, something much more vague and abstract. The measurement process itself renders these abstractions "real" and usable. To raise questions about their measurement is to raise questions about the content of our u nderlying beliefs.
We prefer measurements that seem tidy and unambiguous in part because of the mutuality between the social constructions we want to measure and the constructions by which we attempt to measure them. …