Beet
Atlas Star
Posts: 28,914
|
|
« on: August 27, 2011, 12:04:03 AM » |
|
One note, I would hesitate to draw too much conclusion from "ecological inference". That is a fallacy whereby the attributes of individual behavior are extrapolated from geographical aggregates.
Let me give an example. Suppose that in City A, the average income is $50,000, and City A votes 60% Democratic, whereas in City B, the average income is $40,000, and City B votes 60% Republican. You would conclude that the rich vote Democratic, whereas the poor vote Republican. You would also be wrong. As it turns out, 60% of the population of City A makes $30,000 and votes Democratic, and 40% makes $80,000 and votes Republican, averaging out to $50,000. In City B, 40% makes $25,000 and vote Democratic, and 60% make $50,000 and vote Republican, averaging out to $40,000. In both cities, the lower incomes vote Democratic and the upper incomes vote Republican, precisely the opposite result you would get from looking at geographical aggregates.
Instead of geographical aggregates, to correlate variables such as education and the gender gap you need to go to the individual behavioral level and get micro-level survey data directly from the people involved. Gallup has done that and found some relationship, at least under Obama:
|