The issue of the effect of third party candidacies is interesting, particularly this year. Former New Mexico governor Gary Johnson is seeking the nomination of the Libertarian Party while former Louisiana governor Buddy Roemer is seeking to capture the nomination of the Reform Party.
But the most significant factor could be the campaign by the shadowy group known as Americans Elect. While claiming to be a group that is sick of partisan gridlock and favors finding an alternative candidate by means of an open national primary using the internet for voting, the organization is secretive about its funding sources and there are suspicions that its motives are not as public-spirited as advertised. The key people behind it seem to be the usual suspects who are, under the banner of ‘centrism’ and ‘bipartisanship’, pushing for the same old pro-war, pro-oligarchy type of candidate. The fact that Americans Elect has formed itself as a 501(c)(4) group, coupled with its assertion that it is not a political party, enables it to keep its funding sources secret and makes a mockery of its claim to transparency. Incidentally, Buddy Roemer is also seeking to become the nominee of Americans Elect.
One interesting feature of the ‘Bread and Peace’ model of Douglas Hibbs is that while this model calculates the share of the incumbent party vote of just the two major parties, there have been situations where a third party has had a significant share of the vote, such as with George Wallace’s candidacy in 1968, where he got 14% of the national vote and Ross Perot’s in 1992 where he got 19% and it is interesting to see what impact such candidacies might have had on the outcomes.
Going purely by the graph in Hibbs’s paper and taking the blue line as the expected vote, it looks like in 1968 the incumbent party’s Hubert Humphrey’s share of the two-party vote was not affected much by Wallace’s presence (in fact he did marginally better than the model’s prediction). What seemed to have sunk him were heavy Vietnam war casualties which reduced his predicted vote share by a whopping 7.4%. In the 1992 election, Perot’s presence may have lowered the vote share for George H. W. Bush somewhat, though the model predicts that he would have lost anyway.
The 1996 election also had a third party candidate in Ross Perot again who won about 8.4% of the vote. Although predicted to win, Bill Clinton’s winning percentage (55%) beat the model’s prediction by a large amount (4%), and this may have been due to Perot siphoning more votes away from Bob Dole than from Clinton.
So it looks like there is no convincing case to be made that third parties have swayed the election outcomes, though they may have influenced the margin of victory.