Posted by Emily Hellman on April 28, 2011
December 2012: You can find a brief update based on the 2012 election here. We will update a full 2000-2012 analysis in 2014.
The ability to handle a recount of votes to ensure fair, accurate and genuinely democratic outcomes is widely recognized as a critical component of election administration. Trust in elections requires trust in the recount process and ongoing vigilance in lessons to be learned about how best to do recounts and how to determine what victory margins and post-elections audit results should trigger a recount.
FairVote's April 2011 report by Rob Richie and Emily Hellman examines statewide election recount outcomes and practices in the United States, using data from the decade of elections taking place in the years 2000 to 2009 to determine how often they occur, how often they change outcomes, how much vote totals change and how these figures vary with the size of the electorate. We conclude that:
- Statewide recounts are rare: Out of the 2,884 statewide general elections in the 2000 to 2009 decade, there were 18 statewide recounts, 11 of which were deemed "consequential"(with an original victory margin no more than 0.15 percent). In other words, there was one recount for every 160 statewide elections and one consequential recount for every 262 statewide elections. This pattern was true of most subcategories of statewide elections as well, including only two consequential recounts out of the 422 elections in this period for the offices of governor, lieutenant governor, secretary of state, attorney general and treasurer.
- Outcome reversals are even rarer: Over the 2000-2009 decade, recounts resulted in three reversals out of 11 consequential recounts, or one out of every 961 statewide elections. These reversals took place in the races for U.S. Senate in 2008 in Minnesota, auditor in 2006 in Vermont and governor in 2004 in Washington.
- Margin shifts in recounts are small: Statewide recounts resulted in an average margin swing of 296 votes between the frontrunners, representing 0.027% of the statewide vote in those elections. The median average shift was 229 votes, with 15 of the 18 recounts changing the victory margin by fewer than 500 votes.
- Margin shifts are smaller and recounts rarer in larger electorates: Recounts in elections with more voters altered the vote margin by lower percentages than recounts in elections with fewer voters. In the five cases in which the total votes cast were above two million, the margin shift was on average 0.016% of the vote (less than one for every 6,400 votes cast). In the eight cases in which the total votes cast were fewer than one million, the margin shift was on average 0.039% (less than one for every 2,500 votes cast). No recount took place in our three largest states.
- Most states should revise their laws governing statewide recounts: Model state laws should establish clear post-election audit procedures and define a reasonable victory margin percentage for automatic, taxpayer-financed recounts.
Following are links to:
- Copy (PDF) of the the report, A Survey and Analysis of Statewide Election Recounts, 2000-2009
- Spreadsheet (Excel) with all of the report's data, including a comprehensive review of all statewide elections and data from the 18 elections with recounts