Gabriel Rossman writes to me:
A few days ago there was a discussion on this blog about the book Conservatize Me and more broadly, about taste and politics. Many of the questions can be answered systematically since in 1993 the General Social Survey included a list of questions about musical taste. The simplest question to ask is how different types of music correlate with ideology (polviews). Generally speaking, the stereotypes hold up. Country is correlated with the right whereas classical, rap, rock music, and heavy metal are all correlated with the left. Opinions about folk music aren’t correlated with politics. Note though that even the strongest correlations are relatively weak (r<0.20) so there are plenty of liberals out there listening to country and no shortage of conservative rap fans.
Another way to look at it is to break politics into two dimensions. Let’s treat whether the government should reduce income differences (eqwith) as a measure of economic attitudes. Folk, classical, and big band music are very unpopular with redistributionists. (I guess nobody dreamt about Joe Hill the night before the survey). Rap, metal, and blues are popular with redistributionists. Country, rock, and bluegrass aren’t correlated with fiscal attitudes. For social attitudes, let’s use opinion of sex before marriage (premarsx). Folk, country, classical, bluegrass, and big band fans tend to disapprove of fornication, whereas rap, rock, metal, and blues fans think it’s fine. (If you substitute gay sex for premarital sex the pattern is the same, except for rap fans who tend to oppose it). I experimented with looking for distinctively "libertarian" taste patterns but couldn’t find any.
This is all back of the envelope stuff. A more sophisticated analysis would use factor analysis on dozens of attitudinal questions and find corresponding patterns in them.
You can find the 1993 GSS at Princeton’s Cultural Policy and Arts National Data Archive. http://www.cpanda.org/codebookDB/sdalite.jsp?id=a00006. There’s a self-explanatory web engine that allows you to compare any two variables. (Want to know how many opera fans have been in fist fights? Or how people who have paid for sex feel about nuclear power? Now is your chance.) More advanced users can download the full dataset in SPSS, ASCII, or CSV and do whatever they want with it.
Gabriel Rossman is very smart. Here is his home page. Here is a summary of his dissertation. Here is an abstract of his paper on the Dixie Chicks and where they received less play time. Here is his paper on "Who Picks the Hits on Radio"?