It's only natural that the political press turn to spoilers and upsets now. Election Day is a month away, and most major-party candidates have been written about enough to cover the entire state of Alaska with pixellated text. So: what if?
"If" rarely happens. We looked at every federal general election race from 2006 to 2012 and found that only 61 of the 1,800-plus races were possible spoilers -- and slightly less than half of that number actually were spoiled. In every 1,000 races, in other words, 14 of them saw a third-party candidate (or, simply, a third candidate) that likely cost someone a race.
About half of federal races in the United States are either between a Democrat and a Republican or (occasionally) uncontested. We looked at the 900 other races, ones in which there was a third or fourth (or fifth) candidate, to figure out how often the person who came in third beat the margin of victory between the first two. In other words, given candidates Adams, Baker, and Cobb, who finished in that order, how often the vote total earned by Cobb was larger than the amount by which Adams beat Baker. Without Cobb in the race, then, Baker could possibly have gotten those votes and won.
Only 6.7 percent of those three-or-more candidate races had the third candidate beat the margin of victory, landing them in what we call the "spoiler zone."
Not all of the candidates in the spoiler zone actually spoiled races, though. In some cases, they were members of the same party as the winner; often they were ideologically similar to the winner, such as a Green Party member who came in third in a race that a Democrat won. Certainly not all of the Green Party candidate's voters would have gone to the Democrat had he or she not run, but it seems unlikely they would all have backed the Republican and swung the race.
Among the races in which the spoiler effect is plausible, there's some subjectivity to this assessment -- we counted Charlie Crist's independent bid for the Senate in 2010, for example -- but in general, actual spoilers were pretty easy to spot.
Interestingly, the Senate is over-represented in the final list, compared to the number of total Senate races. In part this is thanks to the larger number of candidates that vie for those seats. The most common scenario, though, was a libertarian candidate siphoning enough votes from a Republican to give a Democrat the win. Rep. Bruce Braley (D-Iowa) in 2010, Rep. Kyrsten Sinema (D-Ariz.) in 2012, Sen. Claire McCaskill (D-Mo.) in 2006. (Lending some credence to our third example at the top of this post.)
Were all of these actually "spoiled"? Meaning, would the second-place candidate have won if the third-party person hadn't run? Maybe not. But these are the rare cases that are close enough to the "what if" that they're worth mentioning -- with an emphasis on "rare." But they're certainly fun to talk about.