This Is What a Legislature with a Bunch of Libertarians Looks Like

NH House in 2 Dimensions
Ideal Points of NH State Legislators in 2 Dimensions (Click to Expand)

The New Hampshire Liberty Alliance does a Liberty Rating each year in which they analyze liberty-related roll-call votes of state representatives and senators and grade them. (The NHLA is a great government accountability organization, by the way, and well worth supporting; a lifetime membership is only $100.)

I used their roll-call votes for the 2014 N.H. House of Representatives but analyzed them differently (I also corrected three errors in their spreadsheet). The vast majority of roll-call votes are on economic issues, where conservative Republicans and libertarians line up. So the Liberty Rating might overstate how libertarian conservative House members really are, if those social issues that are voted on are disproportionately important. The Liberty Rating tries to assess how important each vote is, but the way they do it is arbitrary and subject to dispute (for instance, they rate a bill restricting the sharing of public school student information as highest-priority, on a level with legalizing marijuana and three times as important as a bill enacting occupational licensure of medical technicians). Furthermore, some of their bills are disputably freedom-related: they rated as negative a bill creating a new crime of “domestic violence.” Now, that bill might or might not have been a good idea, but it doesn’t seem like a liberty-related issue, unless you’re an anarchist who wants to legalize everything, including violence.

I used Item Response Theory in a Bayesian framework to estimate the ideal points of legislators in two dimensions. Let me unpack that statement for the layman. I let the data speak for itself. If legislators who generally voted libertarian voted in favor of a bill, the data are telling me that that bill is liberty-enhancing. If legislators who generally vote libertarian split on a bill, then maybe it’s not a liberty issue. Some votes might be “harder” or “easier” than others, like questions on a test. Even a pretty libertarian legislator might vote the wrong way on a hard vote, like a bill legalizing physician-assisted suicide (“Death with Dignity”, HB 1325 in 2014), which failed 219-66.

Using the R package “pscl,” I first hypothesized that all 93 roll-call votes in 2014 reflected a single ideological dimension: each legislator’s degree of libertarian-ness. We could line up all legislators’ ideal points in ideological space along a single line, and that line would be the best way to predict how all the legislators vote on any given issue.

That hypothesis ended up being wrong. On 84 roll-call votes, I couldn’t reject that hypothesis, but on 9, I could. Those were votes on which those voters who tended to vote in a libertarian direction on the other 84 votes tended to vote in an anti-libertarian direction instead – and conversely, those voters who usually voted anti-libertarian actually tended to vote libertarian on those 9 votes.

What were those 9 votes?

  1. HB1237, prohibiting local sex-offender residency restrictions (passed 231-97)
  2. HB1325, Death with Dignity (failed 66-219)
  3. HB1501, mandating licensing of outpatient abortion facilities (killed 211-86) (the NHLA generally stays out of abortion bills, but they believed, and I agree, that business licensing is the wrong way for pro-lifers to restrict abortion)
  4. HB1577, allowing alkaline hydrosis for the disposal of human remains (passed 209-116)
  5. HB1624, modernizing the juvenile justice system (passed 256-40)
  6. HB1625, decriminalizing small amounts of marijuana (passed 215-92)
  7. HB249, mandating employer use of E-Verify (killed 266-68)
  8. HB492, legalizing marijuana (passed 170-162)
  9. SB296, discriminating in favor of veterans in public employment (killed 210-128)

What do all these issues have in common? They’re social issues on which libertarians make common cause with the left! And note that apart from Death with Dignity, libertarians won on every one of these bills. Part of that has to do with the fact that socially liberal Democrats were in the majority in 2014, and part of it has to do with the fact that libertarians are numerous enough in the House to swing some close votes, like the legalization of marijuana.

There were also a few votes without a clear libertarian position; in statistical jargon, they didn’t “load” onto the first ideological dimension at all:

  1. SB318, establishing the crime of domestic violence (passed 325-3)
  2. SB336, banning deer baiting on public land (killed 200-85)
  3. SB366, establishing two casinos in New Hampshire (killed 173-172)

Arguably these roll-calls shouldn’t have been included in the Liberty Rating.

So I divided the roll-call votes into two groups: the 84 votes on which libertarians either voted with the right or lacked consensus and the 9 votes on which libertarians voted with the left. I ran separate Bayesian IRT analyses on these groups of roll-call votes, with the expected results. That procedure gave me two dimensions of ideology, allowing me to generate the scatter plot above.

When you click on the image to expand it, you’ll see legislators’ ideal points plotted in two dimensions: left-libertarian issues on the Y axis and right-libertarian issues on the X axis. The vast majority of legislators are found in two “clouds”: a conservative-Republican cloud in the bottom right and a progressive-Democrat cloud in the upper left. However, there is also a significant smattering of libertarians in the upper right quadrant. There is no corresponding set of communitarians in the bottom left, though there are a few legislators who are strikingly socially conservative given their modest economic conservatism (Democrat Roger Berube and Republican Regina Birdsell, for example).

The five legislators who are strongly libertarian on both dimensions are all Free State Project early movers. But if you draw an arc among points equidistant from the upper-right corner (an “iso-libertarian curve,” if you will) that just draws in John Burt, Richard Hinch, Todd Smith, Norman Major, Joel Winters, and Priscilla Lockwood, you enclose an envelope of legislators who really are strikingly socially libertarian given their degree of economic conservatism. Within that envelope, you’ll find 35 legislators, as good an estimate as we’re likely to have of the number of broadly libertarianish legislators in the 2014 House.

Now, left-libertarian issues all put together are probably not as important as right-libertarian issues all put together. Right-wingers are better, from a libertarian perspective, on the vast majority of economic issues and on about half of personal freedom issues (guns, tobacco, alcohol, and surprisingly, issues like privacy and police militarization). Indeed, the scatter plot above stretches the X axis to twice the length of the Y axis to accentuate this perspective. If all Republicans were like Priscilla Lockwood of Canterbury, the famed “New Hampshire Advantage” would not be long for this world.

Still, Priscilla Lockwood is the only Republican who can win in the highly Democratic town of Canterbury. Given that she is economically moderate and socially libertarian, she deserves support from groups like the NHLA. Heaven forfend she be replaced by an extreme economic leftist like Mary Gile of Concord.

Another question to consider: is there any other state legislative body in the country that would feature a pattern of roll-call votes similar to that found in New Hampshire, with a distinct, sizeable libertarian contingent? I’d venture not.

11 thoughts on “This Is What a Legislature with a Bunch of Libertarians Looks Like

  1. Except Priscilla is retired, and I’m sorry, but she wasn’t very economically moderate. Perhaps in this year you looked at, but not generally. I sat near her (and we represented the same district) and we often disagreed in 2011/2012. So your chart is suspect (perhaps more years would correct that)

    1. More years would be better, but the standard errors on the economic/right-lib dimension are small (much larger on the social/left-lib dimension). From what I’ve seen over the years, she’s on the left end of the Republican caucus and the right end of the Democratic caucus.

  2. “Furthermore, some of their bills are disputably freedom-related: they rated as negative a bill creating a new crime of “domestic violence.” Now, that bill might or might not have been a good idea, but it doesn’t seem like a liberty-related issue, unless you’re an anarchist who wants to legalize everything, including violence.”

    The point of government is to govern liberty. Any bill either enhances or reduces liberty, otherwise, why a bill, and not free choice?

    The anarchist strawman is weak. The question, at root, is not what the rules will be, but who will administer them. If the answer is an organization that claims monopoly and will kill you if you dare compete with it, that is the wrong answer.

    Also, this is interesting: Value-free correlation of voting patterns.

    1. heheh, my ears are ringing (since that linked post talks about me in particular)
      I have been discussing the difference with Jason (mostly on Facebook) between his take above and the Nominate software approach discussed there. (All credit to Will May for his work)
      I agree, Jason’s attempts to ‘clean up the data’ (my words) seem to result in the nicer graph, and an attempt to distinguish the ‘left lib’ from the ‘right lib’, and perhaps isolate some of the more cusp-y members of the dataset. But I’m not convinced that the pure Nominate graphs aren’t more useful and they are certainly less subject to any form of external bias, and are truly content neutral in that sense.

      1. These data aren’t subject to “external bias” either, except in the way I chose to present them. The results of the analysis suggested those 9 bills were ones on which conservatives tended to vote anti-liberty; I didn’t pick them out beforehand.

        The biggest flaw in my analysis is that with only 9 bills, legislators’ scores on the left-libertarianism dimension suffer from large standard errors. Adding 2013 roll-calls might help with that.

      2. Jason, by external bias, that includes the NHLA decision makers who compile that particular list of bills. (Not you in particular) One of the beauties of the original charts Will did with Nominate is that you can use every single roll call and it still works.

      3. Understood. Yeah, I approached this exercise as something the NHLA could use for its own purposes. With all roll-call votes, my guess is the standard left-right dimension would be much more dominant (which is what Will found).

    2. But the bill wasn’t about who would administer the laws. None of that would change if the bill were passed. I can’t imagine David Friedman himself would favor passing a bill to legalize murder.

  3. is there any other state legislative body in the country that would feature a pattern of roll-call votes similar to that found in New Hampshire, with a distinct, sizeable libertarian contingent?


    1. Is there a libertarian-conservative divide within the AK GOP house caucus? I haven’t heard of one, but maybe…

Leave a Reply

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

You are commenting using your 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