Content area
Full Text
A Solution to the Ecological Inference Problem: Reconstructing Individual Behavior from Aggregate Data. By Gary King. Princeton, NJ: Princeton University Press, 1997. 342p. $55.00 cloth, $16.95 paper.
Douglas Rivers, Stanford University
A Solution to the Ecological Inference Problem was published with much fanfare. The NSF, which funded the research, issued a press release, and the New York Times and Boston Globe printed stories. Few works by political scientists, much less equation-filled tomes like this one, attract that kind of attention. The reason for the notoriety is Gary King's bold claim to be able to perform a seemingly impossible feat. Since the publication of Robinson's seminal paper in 1950, generations of social scientists have been warned about the "ecological fallacy." Has King really found "a method of inferring individual behavior from aggregate data that works in practice" (p. 1)?
The most popular method for ecological inference has been ecological regression, which was described by Leo Goodman in the American Journal of Sociology in 1959. King's proposed method is a random coefficients ecological regression model. The running example in his book involves data on precinct-level turnout rates (T^sub i^) and the proportion of voting-age people in the precinct who are black (X^sub i^). In King's notation, D^sub i^^sup b^ and B^sub i^^sup w^ are the proportions of blacks and whites, respectively, who turn out in precinct i, so It is important to understand that the assumptions required for King's estimator include all those required for Goodman's estimator and...