there becomes a linear positive relationship between education and voting Democratic.
That's about as suprising as finding out that there is a direct link between the % of blue collar workers in a given area how likely that area is to vote for the Labour Party
Here are some actual numbers from a
quick and dirty regression based off of the University of Michigan's National Election Studies. The depent variable I used was a number ranging from -2 (strong Kerry supporter) to +2 (strong Bush supporter). N=985 observations. The actual coefficients are meaningless in comparison (in the model without taking party into account, the effect of going from a grade 8 education to grad degree is about 62% of the effect of being black; in the model with party, it is about twice the effect) so just looking at t-values:
Without party id:
education (-3.35)**
household income (4.01)**
black dummy variable (-9.13)**
R^2=.103
With party id:
partyid (36.84)**
education (-3.06)**
household income (1.68)
black dummy variable (-2.60)**
R^2=.624
The interesting thing is how more educated Republicans and Democrats alike were more likely to be for Kerry, regardless of party identification.