Free Staters' Impact on the New Hampshire Primary: A Statistical Analysis
by Jason Sorens
The Free State Project was founded on the idea that a critical mass of dedicated activists could wield political influence out of all proportion to their numbers, especially if they had good ideas that appealed to the interests and sympathies of the majority of the population. I made some attempts to quantify that influence through voter-activist ratios drawn from different but arguably comparable contexts. In the more recent study, I argued that attaining 1% of the local population constituted a critical mass.
Until now, however, we had no direct evidence about how many Free Staters would be needed to achieve political victories in New Hampshire. The recent New Hampshire presidential primary, coming now with well over 500 Free Staters in the state, affords us the first opportunity to examine our prospects with hard evidence. In this election, there was one reasonably libertarian presidential candidate running in a statewide race: Ron Paul. While not all Free Staters supported Ron Paul's candidacy, I think it is fair to say that he was far more popular among Free Staters than any other candidate, particularly before the newsletters revelations that broke on the day of the New Hampshire primary.
My method is to examine how the ratio of Free Staters to Republican primary voters in each town affected Ron Paul's percentage of the vote in those towns, by using regression analysis. This method vastly understates the true influence of Free Staters on the election, because it is unrealistic to assume that Free Staters' primary influence on the election was solely within their own towns. However, it is not possible to assess Free Staters' influences on the state as a whole directly. We can note that according to ronpaulgraphs.com, a higher percentage of the New Hampshire population has donated to Ron Paul than of any other state, and we can also note that Ron Paul's finish in New Hampshire, 8% of the vote, is double what he has consistently polled in the rest of the country. Notwithstanding these encouraging figures, we cannot know to what degree they are attributable to the efforts of Free Staters rather than to the state's pre-existing libertarian base.
Before reporting the results of the regression analysis, I should mention a few interesting town results. Among towns with a population above 500, Ron Paul's best results were in Richmond (35.5%), which he won, Wentworth (24.2%), and Grafton (23.2%). All three towns have some Free Staters; in fact, FSP participants represent 0.9% of the population of Grafton, close to the critical 1.0% mark. I think it's fair to say that if Ron Paul were perceived as having a shot at winning the nomination or even the state's primary, his numbers would have been higher in every town, because many New Hampshire voters who liked Paul appear to have voted tactically for another candidate (more on this below).
The main result from the regression analysis is that every additional Free Stater per 100 Republican primary voters resulted in approximately 2.5 percentage points improvement in Ron Paul's share of the vote in that town. Thus, in Grafton, Free Staters represented 4% of Republican primary voters (we don't know that they all voted, or voted Republican, or even supported Ron Paul, of course), and the model predicts that if no Free Staters lived in Grafton, Ron Paul would have gotten 13% of the vote, instead of 23% (23-0.4*2.5). The margin of error of the estimated "FSP effect" is about 1.0. What that means that we definitely know that Free Staters influenced the election beyond their own votes. If we assume that, say, 75% of Free Staters voted for Ron Paul, then the 2.5 figure means that the average Free Stater Ron Paul supporter brought along two and a half neighbors to support Ron Paul as well (2.5/0.75-0.75), and they definitely brought along at least one neighbor (1.5/0.75-0.75), and could have brought along as many as four (3.5/0.75-0.75). As I mentioned above, these are only the effects that Free Staters had in their own towns. (For technical details, see the end of this essay.)
I also looked at the effects of New Hampshire "Pioneers" on the Ron Paul vote. Pioneers are New Hampshire residents who sign up on our website stating that they support the goals of the Project. They are not counted in our Participants count because they didn't have to move. I find that the percentage of Pioneers in each town has a weaker effect on Ron Paul's vote share, about 1.6 instead of 2.5. One could draw the conclusion that Pioneers, freedom supporters who did not move into the state under the auspices of the Free State Project, were less effective activists in this election than Participants who either moved into the state or signed up for the FSP before we chose New Hampshire, but another possibility is that Pioneers were as a group less likely to support Ron Paul than were Participants.
The regression results yield some other interesting insights into why Paul lost the election. I find that average home value from the Census, percentage of the town's population that lives in an urban area, per capita income in the town, and seasonal vacancy rates (proxying importance of the tourism industry in the town) are all strongly negatively correlated with Ron Paul support. In other words, those parts of the state that are doing well economically did not support Ron Paul. Paul did better in poor, rural areas without a large tourism industry.
These results match up well with exit poll results that show that Paul did best among those with incomes under $30,000 and worst among those with incomes over $100,000, better among those without a college degree than college grads, and far better among those "very worried about the economy" than those with other opinions.
In other words, Paul appealed to voters who felt very economically insecure. It seems highly plausible that the reason for this was that New Hampshire voters perceived Paul as an economic isolationist, ready to shut down international trade and immigration. The idea that Paul is basically a Buchananite with libertarian rhetoric is a common perception (misperception?) among supporters of other candidates with whom I have had contact. Since New Hampshire is a dynamic, open state with an international seaport and a border with Canada, protectionism doesn't sell in New Hampshire - at least, not any longer. The wealthy and well-educated presumably viewed Paul's proposals for a border wall and withdrawal from the WTO and NAFTA with alarm, even though Paul insists that he supports unilateral free trade.
So one reason Ron Paul lost New Hampshire is that his noninterventionist rhetoric has at times painted him as an isolationist. He has failed to stress his support for a dynamic international economy. (For what it's worth, I personally disagree with his stances on immigration and international trade agreements.)
Another reason Paul lost was tactical voting. New Hampshire voters were far more likely to oppose the war and be angry at the Bush Administration than had been Iowa voters. But Paul did worse, because the exit polls show these voters going to McCain. 49% of voters with a "somewhat favorable" opinion of Paul voted for McCain, and only 3% for Paul himself! McCain, who says he is willing to keep U.S. troops in Iraq for 100 years, did far better among voters who oppose the Iraq war than among those who support it! Part of the reason for the results has to be simple voter ignorance, but part of it also has to do with the fact that the race was perceived as a battle among McCain, Romney, and Huckabee, and N.H. voters chose the candidate they thought most likely to change the Bush Administration's policies.
Finally, Ron Paul did worse than he might have otherwise because of the huge turnout. He actually received about 6,500 more votes in New Hampshire than in Iowa, and New Hampshire has fewer than half the eligible voters of Iowa! Had turnout in Iowa been at New Hampshire levels, I suspect Paul would have gotten less than 8% in Iowa, given that Ron Paul supporters are more passionate about their candidate than are supporters of other candidates (hence all the straw poll victories). One really can't directly compare results from caucuses, which have low turnout, to results from primaries, which have higher turnout.
In conclusion, the data show that Free Staters, most of whom have only moved to their location within the last year or two, have already begun to persuade their neighbors to vote for libertarian candidates. It will be interesting to see how their numbers affect the results when there is a statewide election involving a freedom-friendly candidate who actually has a chance of winning.
(Technical Appendix. Regressions are Ordinary Least Squares (OLS) regressions with robust standard errors of Ron Paul's percentage of the vote on Free State Project Participants as a percentage of the Republican primary vote, rural population percentage from the Census, and the natural logarithm of median home value from the Census. Each town is an observation, and each observation was weighted by the total number of Republican primary votes from the town (i.e., bigger towns "count more" in the regression). A few towns had no votes for Ron Paul, creating a corner solution in the data. Tobit regression did not change the results appreciably. Adding all of the following control variables also did not change the results appreciably: farm population percentage, per capita income*, percentage of population on public assistance, seasonal vacancy rate*, other vacancy rate*, median rent, median real estate taxes, 2004 percentage of the vote for Craig Benson, and a dummy variable for college towns. [* Indicates statistically significant.] Adding Free State Pioneers as a percentage of the vote caused both "Free Stater variables" to lose statistical significance, marginally, because of high collinearity between the two, but the coefficient estimate on the FSP Participants variable remained stable.)