Reading around the web I stumbled upon National Journal‘s ubiquitous ratings of how liberal or conservative various politicians are in the domains of social, economic and foreign policy. Using the 2008 data for the House of Representatives, here are the correlations:
Social Liberal & Economic Liberal = 0.78
Social Liberal & Foreign Liberal = 0.70
Economic Liberal & Foreign Liberal = 0.66
The correlations are rather good, but not perfect. For example, assuming a linear model only 60% of the variation in economic liberalism can be predicted by social liberalism (just square the correlations). Some of this is surely the coarseness of measures Natural Journal is using; but I’m not an idealist in any case about politics insofar as there is a most liberal liberal and most conservative conservative archetype. These ideologies are embedded in the real world.
But I was curious about the residuals, that is, deviation from the trends (which are substantial). Specifically, there are the orthogonal tendencies of libertarianism and statism, which though secondary to the standard Left-Right dynamic, do exist.
I will focus on social & economic liberalism in this analysis and discard foreign policy because I don’t want to deal with other combinations right now. Additionally I am not focusing conservatism simply because it looks like it’s just the perfect inverse, so all you need to do is “change the signs.” No need to repeat. First, a scatterplot of Democrats and Republicans, colored blue and red as usual, in 2008.
I’m sure you’ve seen this diagonal hourglass before. Here is the same with a loess regression fitted to the distribution:
Some of you probably want to know who these points on the dot are. Click the image to the left. What I’m really curious about are the deviations from that line of best fit. The line itself is a standard Left-Right axis which traces the composite social & economic liberal values. But, as you can see, there are many people well above and below where “they should be.” Those above the arc exhibit more statist tendencies insofar as they are more economically liberal than their social views would predict. These would be conservative Republicans who are against free trade, or perhaps Republicans in industrial regions who vote more pro-labor than one would expect. They can also be Democrats who are socially somewhat more conservative due to the downscale demographics of their district. Below the arc are those who are more socially liberal than their economic views would predict. For Democrats these might be “New Democrat” types from professional & entrepreneur loaded districts, as well as Republicans from more libertarian leaning districts.
As it happens it isn’t difficult to extract out the residual values, the deviations, from the trend. In other words, who is the most statist and libertarian. I used a cutoff of 15 units above and below the arc. Negative values indicate libertarianism, positive values statism.