How does the Libertarian vote split between the major parties when there is no Libertarian on the ballot? As it turned out, the election last week gave us a test case. Eric Cantor (R-VA7) resigned immediately after losing his primary a few months ago, necessitating a special election for the remainder of the current term which was on the same ballot as the general election for the upcoming session. There was a Libertarian on the ballot for the general election but not for the special.
Here was the result:
General election for VA7
Dave Brat (R) 147,897 60.9%
Jack Trammell (D) 89,793 37.0%
James Carr (Lib) 5,347 2.2%
Special election
Dave Brat (R) 148,715 62.0%
Jack Trammell (D) 91,111 38.0%
It appears from this that Brat picked up 818 votes from Carr, Trammell picked up 1,338, and the other 3,191 Carr voters didn’t vote in the special.
This is misleading, though. There is typically a dropoff from a general election to a special election on the same ballot even when the exact same candidates appear on both. There was one such race this cycle, in NC12:
General election for NC12
Vince Coakley (R) 42,128 24.6%
Alma Adams (D) 129,178 75.4%
Special election
Vince Coakley (R) 41,737 24.6%
Alma Adams (D) 128,081 75.4%
In NC12, the GOP dropoff from the general to the special was 401 votes or 0.95% and the Dem dropoff was 1,097 votes or 0.85%. The combined undervote was 0.87%.
While the GOP dropoff was slightly above the Dem dropoff in NC12, there is no reason to think that GOP voters would be generally more likely to undervote special elections than Dem voters would. Assume that if Carr had been on the special election ballot, 0.87% of the voters for all three candidates would have undervoted, with this result:
Hypothetical special election for VA7
Dave Brat (R) 146,610 60.9%
Jack Trammell (D) 89,012 37.0%
James Carr (Lib) 5,300 2.2%
Comparing this to the actual result, it now appears that Brat picked up 2,105 votes (39.7%) from Carr, Trammell picked up 2,099 (39.6%), and the remaining 1,096 (20.7%) would have left it blank.
The conventional wisdom on this board is that Libertarians primarily take votes from Republicans, but the VA7 vote this year suggests that in the absence of a Libertarian or other protest candidate they are more likely to split roughly 40-40-20 between the GOP, Dems, and undervoting.
The obvious caveat is that this “conclusion” is based on sample size of one election that may or may not be representative. (So is is the assumption that an 0.87% undervote from a general election to a special election on the same ballot is normal.) A Libertarian candidate with a relatively high profile (such as former New Mexico governor Gary Johnson, who ran for president in 2012) or who campaigns actively on specific issues might take votes disproportionately from one party or the other. Still, the VA7 vote this year provides some evidence that most Libertarian voters will in fact pull the lever for a major party candidate if faced with no other options, and that absent special factors they will roughly split between the two parties.