The difference between turnout for president and turnout for state house -- or the difference between the vote for the Democratic candidate for president and state house -- is an aggregate description. This description is politically important at the aggregate level. But suppose instead we were interested in the fraction of individual people splitting their tickets. In this case, we would need to use the ROAD data to make an ecological inference.
Similarly, models that explain or predict geographic variations in electoral vote totals for candidates as a function of economic and demographic variables could be studied as contextual effects. For example, we may wish to know whether the vote for Democratic candidates is higher in areas with universities (college towns). But if we were instead interested in the individual-level question about whether college students are more likely to vote for the Democrats, then we need to make an ecological inference.
For information on how to make ecological inferences, see Gary King's, A Solution to the Ecological Inference Program: Reconstructing Individual Behavior from Aggregate Data (Princeton University Press, 1997), and the related software programs. See also some of Bradley Palmquist's papers on ecological inference.
| Copyright © 1997-2004 | [ROAD Home] | Questions? Contact the ROAD webmaster. |