There were no statewide election recounts in 2012. This is particularly noteworthy, considering the fact 419 statewide general elections (including ballot measures) took place this year. In this post-Florida 2000 political landscape, the specter of recounts continues to loom large and you will even read thoughtful people suggesting that the possibility of a recount is enough to oppose a having one-person, one-vote elections for president by national popular vote. However, when we step back and take a close look at the numbers, it becomes clear that chance of actually having to perform a recount is relatively remote. Indeed, from 2000 to 2012, 99.457% of statewide elections have been successfully held without recounts – and recounts take place consistently show minuscule changes in victory margin.

Of the 419 statewide races that took place in 2012, there were two that were almost sent to a recount. In Montana, the recount in the Superintendent of Public Instruction race was called off when the challenger failed to raise the funds required to pay for her requested recount. Initially, a County Judge ruled in favor of holding a statewide recount, after it was confirmed that a counting error was made with the ballot counting machines in some polling places, which may have led to reporting incorrect numbers. 

However, even if the challenger had come up with the requisite funds, the incumbent’s margin of victory was 0.48% after the initial count. According to FairVote’s recount data (from report covering 2000-2009 that includes a downloadable spreadsheet). a margin this large would only have been overcome in the event of a major effort of voter fraud. In the 22 statewide recounts (see chart) from 2000 to 2012, the average recount shifted the vote less than 0.03% percent, or by just a few hundred votes. Only one recount changed the margin as much as 0.11%. A 0.48% shift is 17 times the average shift found in our report and more than four times bigger than the single biggest margin change.

In the North Carolina Lieutenant Governor’s race, the challenger opted not to do a recount despite the fact that some felt the race was close enough to warrant one. The trailing Democratic Party candidate realized that a new tally was unlikely to make up the nearly 6,900 votes (a 0.02% victory margin) she needed, saying,"We just were faced with the reality of the numbers and that it's hard to flip 6,000 votes in an extended recount battle.”

While there were recounts in some non-statewide races in 2012, those recounts underscore the fact that recounts rarely change election outcomes or many votes. For instance, in North Carolina’s 7th Congressional District, Mike McIntyre lost only a single vote, with his margin of victory shifting from a 655-vote lead to a 654-vote lead, out of more than 336,000 votes cast. In Florida’s 18th Congressional District, a recount of early-voting ballots also failed to change the margin significantly.

Indeed, in the 19 statewide recounts from 2000 and 2012, only three have resulted in changed election outcomes. In these races, the initial margins were 0.009%, 0.061% and 0.010%; in other words, these were extremely close races. It is exceptionally rare to have such close elections where a recount really might change the outcome. 

Our research suggests that, absent major fraud or error, full recounts are rarely necessary. However, integrity and accuracy in elections is vital. Therefore, rather than attention to recounts, better attention should be paid to risk limiting audits that can catch potential fraud or technological error in any race. The use of these systematic post-election audits is important as a means of catching fraud or error that potentially exist in all races, not just those that are exceptionally close. 

For more recount analysis, see FairVote’s comprehensive assessment of statewide recounts in 2000-2009 and check back with us in January, when we release an update on statewide recounts from 2000 through 2012.