PCA on political opinions

[gusl]

I'm reposting this for the sake of my new LJ friends.

PCA is a machine learning technique for creating significant axes on a data set, i.e. principal components.

If you are serious about defining political axes (left-right, etc.), you should know about this. See my previous post, and do the survey to see where you belong in this 2-dimensional political cloud.

Comments

[smandal.livejournal.com]

left/right -0.2022 (-0.0122)
pragmatism +5.6188 (+0.3382)

To your knowledge, has anyone actually done a computational PCA on political beliefs on a basis of sociological and personality characteristics?

[selfishgene.livejournal.com]

That would be interesting.

[(Anonymous)]

So you mean, the set of questions in the PCA would be a mixture of questions about political opinion and questions about demographics, including personality data (either in the form of questions the person answered or "diagnoses" like "Introvert").

[smandal.livejournal.com]

Yes. For the personality characteristics, the five factor model (http://en.wikipedia.org/wiki/Big_five_personality_traits) could be a good starting point.

(This (http://www.livescience.com/humanbiology/brain_politics_041029.html) seems relevant.)