Solving the climate change attitude mystery, from Statistical Modeling, Causal Inference, and Social Science, originally from Wired.
The facts appear to be as follows: 19 percent of college-educated Republicans believe that human activites cause global warming; 75 percent of Democrats believe the same. Among the non-college-educated, 31 percent of Republicans have that belief and 52 percent of Democrats.
One conclusion is that college-educated people are more likely to toe the party line.
But here's another idea. The assumption underlying this seems to be that "Republican" and "Democrat" are fixed labels -- so clearly college makes you think that global warming is more likely if you're a Democrat, but less likely if you're a Republican. But of course those labels are not fixed; people can switch parties! So maybe what happens is that education doesn't change your beliefs -- but if you are the sort of person who would believe that humanity causes global warming, but otherwise tend to agree with Republicans, a college education would flip you to the Democratic side. (And vice-versa for Democratic-leaning global-warming-skeptics.) I'm not saying that there are people doing indoctrination at our colleges, but that such an education changes the way one looks at things.
It basically seems to be Simpson's paradox in different guise. You've got to be careful when the groups you're analyzing change size!
15 May 2008
Simpson's paradox and climate change
Posted by Michael Lugo at 4:59 AM
Labels: global warming, paradox, statistics
Subscribe to: Post Comments (Atom)
I think that we usually hope that such an education changes the way one looks at things.
Interesting! Statistics are so much fun...
Is this a problem of what econometricians call "endogeneity"?
Another possibility here is that college grads tend to identify themselves with parties less, unless they're particularly passionate about some issue. So the college grads who report themselves as democrats or republicans are the ones most likely to have an opinion on climate change, one way or the other. Less educated people tend to identify with their party for other reasons.
it might be. I'm not an econometrician and so I don't know their technical vocabulary.
Mayby it's coused by choise of collage with similiar views/peaople with similiar views.
Post a Comment