An Education Cleavage?
March 31, 2010 10:53 AM   Subscribe

I'm looking for studies about educational attainment as a predictor for party choice in US Presidential Elections.

I'm looking for party choice and not just turnout. I'm trying to find out whether it's possible to talk about an "Education Cleavage" in US Politics.
There has to be something out there a little more academic than CNN Exit Polls?
posted by robinhoudt to Society & Culture (6 answers total) 1 user marked this as a favorite
 
Look at eric.ed.gov, especially at the Thesaurus.

Political science journals might be helpful. If you are in school, talk to a ed/social sciences librarian.
posted by jgirl at 11:41 AM on March 31, 2010


http://www.jstor.org/pss/2090665
http://www.jstor.org/pss/2110695

&c.
posted by jgirl at 11:45 AM on March 31, 2010


Keep in mind, educational attainment also includes the ability to afford a particular level of education, and that also influences one's vote. In other words, "educational attainment as a predictor for party choice in US Presidential Elections" often includes wealth as a predictor for party choice in US Presidential Elections, and that could very much skew the results you're looking for.

If you find what you're looking for, please post it in the thread because this is fascinating stuff!

Cheers!
posted by 2oh1 at 11:55 AM on March 31, 2010 [1 favorite]


Do you have access to academic journals? I found some that seem even more relevant than those that jgirl found:

http://www.jstor.org/pss/2778029

http://www.ingentaconnect.com/content/csshe/cjhe/1998/00000028/00000001/art00002

Those came up through Google Scholar searching "political party educational attainment".
posted by parkerjackson at 12:56 PM on March 31, 2010


What 2oh1 says is not 100% true every time.

If you are just looking at a simple analysis that has the party affiliations percentages based on educational achievement then you will also be picking up all the other things that are correlated with education (one major one is wealth).

Fortunately there exist techniques that can help (partially) separate these effects. For example if you run a linear regression with (political leaning) = B0 + B1*(Years of education) + B2*(Wealth) + B3*(Wealth)*(Education) you can start to separate out how the factors influence you political views separately and how they affect once they are combined. If we lived in a nice pretty world, where those were the only two factors that influenced you views, and they influenced you views linearly, you can prove that this regression will give you the "best unbiased linear estimator" of the true effect of education on political views.

We don't live in that perfect world, so things get a lot messier. Neither education nor wealth are likely to have a linear impact, but its hard to prove/justify using any other model (a quadratic model get used relatively often). (Also I am abstracting away from how you would create a continuous variable for political leaning). Additionally many people leave out the interaction term B3*(Wealth)*(Education) when running a regression like this (implicitly assuming that the two factors can have different effects once combined.)


What I am trying to say is be careful about trusting anything that says "Y has an effect even after controlling for A,B and C" (e.g. having a clean house makes children do better in school even after controlling for race and income), but be even more careful about trusting anything that doesn't even bother to try to control for these factors.

Trying to control for other effects is imperfect, but a whole lot better than nothing.


Sorry I can't point you to any articles relevant tot he particular example you are interested in.
posted by vegetableagony at 2:01 PM on March 31, 2010 [1 favorite]


Thanks! Yes, journal articles are mostly what I'm looking for.

2oh1, I suspect that's what's going on when the exit polls I found showed that the less educated tended to vote for Obama over McCain in greater proportion than the college graduates.

I'm hoping I can find a recent study which would isolate the education variable and measure its effect, but if it doesn't exist I suppose I'll attempt to do it myself with SPSS, which I'm learning to use, if I can find usable data. Thanks vegetableagony, that's very helpful!

Does anybody know if and where any of the primary data for something like this is freely available to a student? The US census data has education, but of course doesn't list party choice...
posted by robinhoudt at 2:54 PM on March 31, 2010


« Older My grandpa passed on. This is some of his stuff!   |   Hiking around Lake Mead Newer »
This thread is closed to new comments.