Where the Libertarians Are, Part 2

A few years ago, I did a statistical analysis of which states had the most libertarians, using data from 2004 and 2008 Libertarian Party vote shares and 2008 Ron Paul vote shares and contributions. David Boaz has prodded me to update these numbers in light of the 2012 election. This post does just that.

To come up with a single, valid indicator of how many libertarians are in each state, I use a technique called principal component analysis (PCA), which extracts the vector of data that best explains the correlations among multiple variables. Say I have a number of different measures of the number of libertarians by state. Using PCA, I can convert those different measures into a single measure. A crude way of doing this would be to simply standardize and average all of the different variables, but that method assumes that each variable is an equally reliable measure of the underlying concept. PCA actually tells us which variables are most reliable measures and weights them more heavily.

To see which states have the most libertarians, I use six measures: Libertarian Party presidential vote share in 2008 and 2012, Ron Paul contributions as a share of personal income in 2007-8, Ron Paul and Gary Johnson contributions as a share of income in 2011-12, and “adjusted” Ron Paul primary vote share in 2008 and 2012. Ron Paul vote shares are adjusted for primary vs. caucus, calendar, number of other candidates, and the like (for details see this post). Hawaii and Wyoming are excluded because they did not collect vote shares in the 2008 presidential primary. D.C. is included.

Here are the results of the PCA on these six variables:

. pca resid12 resid08 lp12 lp08 rpcpi08 libcpi12

Principal components/correlation Number of obs = 49
Number of comp. = 6
Trace = 6
Rotation: (unrotated = principal) Rho = 1.0000

--------------------------------------------------------------------------
Component | Eigenvalue Difference Proportion Cumulative
-------------+------------------------------------------------------------
Comp1 | 2.81582 1.49201 0.4693 0.4693
Comp2 | 1.32382 .517957 0.2206 0.6899
Comp3 | .805859 .266932 0.1343 0.8242
Comp4 | .538928 .0754767 0.0898 0.9141
Comp5 | .463451 .411326 0.0772 0.9913
Comp6 | .0521252 . 0.0087 1.0000
--------------------------------------------------------------------------

Principal components (eigenvectors)

----------------------------------------------------------------------------------------
Variable | Comp1 Comp2 Comp3 Comp4 Comp5 Comp6 | Unexplained
-------------+------------------------------------------------------------+-------------
resid12 | 0.1159 0.7527 0.1699 0.3288 0.5308 -0.0354 | 0
resid08 | 0.3400 0.5441 0.1240 -0.3297 -0.6750 0.0934 | 0
lp12 | 0.4360 -0.1868 0.3962 -0.6239 0.4133 -0.2408 | 0
lp08 | 0.3628 -0.3001 0.6360 0.5552 -0.1895 0.1724 | 0
rpcpi08 | 0.5218 -0.0665 -0.4366 0.2925 -0.1052 -0.6604 | 0
libcpi12 | 0.5263 -0.0897 -0.4513 -0.0152 0.2117 0.6828 | 0
----------------------------------------------------------------------------------------

“Resid*” is adjusted Ron Paul vote share, “lp*” is LP vote share, and the last two variables are contributions as a share of personal income. What this output tells us is that one single component has lots of explanatory power for the correlations among these six variables: we can interpret this component as the number of libertarians in a state. The method doesn’t give us a number interpretable as an absolute count of libertarians, but a number that we can interpret as representing how many libertarians each state has compared to all the others.

The second table of output shows how each variable contributes to each component. To the first extracted component, the one of interest to us here, the contributions variables actually contribute the most, while adjusted Ron Paul vote shares, especially in 2012, contribute the least. I have found elsewhere that in 2012 Paul did really well in states with lots of liberal voters, as he expanded his base beyond libertarians to antiestablishment liberals and moderates. As a result, his cross-state performance in 2012 isn’t actually a good measure of how libertarian each state is. Still, it contributes a little something to our measure.

Here is the extracted component, with all the states ranked from most to least libertarian:

state libertarians
Montana 5.504036
New Hampshire 4.163368
Alaska 3.586032
New Mexico 3.319092
Idaho 2.842685
Nevada 2.477748
Texas 1.632528
Washington 1.568113
Oregon 1.180586
Arizona 1.0411
North Dakota 0.7316829
Indiana 0.6056806
California 0.5187439
Vermont 0.4731389
Utah 0.2056809
Colorado 0.1532149
Kansas 0.107657
South Dakota 0.0328709
Maine -0.0850015
Pennsylvania -0.2063729
Iowa -0.3226413
Georgia -0.3296589
Virginia -0.3893113
Maryland -0.4288172
Rhode Island -0.470931
Tennessee -0.4882021
Missouri -0.4912609
Arkansas -0.5384682
Louisiana -0.5897537
Nebraska -0.6350928
Minnesota -0.7662109
Michigan -0.7671053
North Carolina -0.811959
South Carolina -0.8196676
Illinois -0.9103957
Ohio -0.9599612
Delaware -1.057948
Florida -1.072601
District of Columbia -1.091851
New York -1.225912
Kentucky -1.330388
Massachusetts -1.342607
Wisconsin -1.410286
New Jersey -1.431843
Connecticut -1.606663
Alabama -1.863769
Oklahoma -1.93511
West Virginia -2.244921
Mississippi -2.519249

Mississippi and West Virginia have the fewest libertarians, while Montana and New Hampshire have the most. Note that Montana and New Mexico will be overstated on this measure, because I have added half of the Montana Constitution Party’s vote share to the Libertarian Party vote share in 2008, because they listed Ron Paul on their general election ballot. No other state had the opportunity to run Ron Paul in the general election, however, so this choice overstates how many libertarian voters are in Montana. But excluding Ron Paul from Montana’s vote share would hurt them because he presumably drew lots of votes away from Bob Barr, the LP candidate, in that state. If I do exclude Ron Paul’s votes entirely from Montana 2008, then New Hampshire ends up just pipping them for most libertarian state. New Mexico is overstated because it is Gary Johnson’s home state, who did very well there both on contributions and on vote share.

These results are quite similar to those I found back in 2010, perhaps unsurprisingly since I included 2008 data on both occasions. Still, there are some small differences. New Hampshire has now easily passed Alaska for the #2 spot. Vermont, Maine, Kentucky, and Texas have gained, while Michigan, Idaho, Indiana, and Georgia have fallen.

21 thoughts on “Where the Libertarians Are, Part 2

  1. Anarchist walks into a bar with a Parrot on his head.

    Bartender says “Where the hell did you get that mangy thing?”

    Parrot says “New Haaaaampshire … there’s thoussssannnds of ’em!”

    1. No data on Ron Paul 2008, so it had to be dropped. On other indicators, it looked above average but not near the top.

    1. Correct. I did the same (1/2 of the vote) for Louisiana. New Mexico is overstated as Johnson’s home state.

  2. This has been observed for about 40 yrs., but applies only to those who are self-consciously, and/or radical or extremely, libertarian.

    What about “libertarian” as adjective rather than noun? Admittedly it’d take a lot more data gathering (via polling), but what about people who are significantly, although not necessarily extremely, more libertarian than the avg. person in the subject popul’n?

  3. This is great — I was hoping, despite the overwhelming control of the state by.. statists.. that there were more libertarians here in Rhode Island.. We seem to have more than the average amount of free thinkers here.

    Also, Where’s the freakin map? 🙂 With colors, of course.

    1. I know, I know! My copy of ArcGIS has expired, but one of these days I’ll get a map made, perhaps with free GIS software like GeoDA.

  4. In Kentucky we have a libertarianish US Senator in Rand Paul, a libertarianish Congresman in Massie and a libertarianish newly elected gover in Matt Bevin all of which were elected with enthusiastic Tea Party support. Anyone here who can top that?

    1. I don’t think so! Still, number of candidates actually elected is a noisy indicator, since getting elected depends on many things other than the size of one’s support base.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s