Election fraud comes in many flavors, but there’s a new taste test for one sort of trickery. Scientists analyzing data from several recent international contests, including the questionable 2011 parliamentary elections in Russia, have proposed a new mathematical measure to discern fraudulent elections from fair ones.
The researchers examined voter turnout and votes received by the winning party for recent parliamentary elections in Russia, Austria, Finland, Switzerland and the United Kingdom and for presidential elections in Uganda and the United States. Graphing the relationship between turnout and votes for the winner revealed unusual peaks in the data for the elections in Russia and Uganda — a signature of funny business, the scientists contend.
Ballot stuffing best explains the data, says study coauthor Peter Klimek, a complex systems scientist at the Medical University of Vienna.
“Of course, this is a statistical detection technique, not conclusive proof,” says Klimek, who, along with Stefan Thurner and other University of Vienna colleagues, reported the analysis online January 15 at arXiv.org. But the numbers need explaining, “and nothing explains them as cleanly as the fraud hypothesis,” Klimek says.
Thousands of precincts in Russia and districts in Uganda reported 100 percent voter turnout with 100 percent of those votes for the winning party, the researchers found. Graph these data various ways and the fraud signature pops out, notes Klimek. Plotting votes for the winner against voter turnout, for example, reveals a line that slopes off into a plateau for most countries, but for Russia and Uganda those lines keep climbing right off the graph.
Similar analyses of the recent elections for the Russian legislative body, the State Duma, have also found statistical aberrations suggesting the elections weren’t fair. In fact, the Russian online newspaper Gazeta.ru ran photographs of thousands protesting the elections, including a mathematician whose sign depicted one of the telltale graphs of the results. “If United Russia were an NFL team, they would win 5 percent of their games with a score of 1,000 to zero,” says Klimek.
There are possible explanations for the Vienna researchers’ results besides fraud. Some districts are “special places,” notes Walter Mebane of the University of Michigan in Ann Arbor, a statistician, political scientist and expert on Russian elections. When nearly 100 percent of the voters who turn out vote for the same party or candidate, that doesn’t necessarily indicate ballot stuffing, Mebane says. The demographics of each area have to be taken into account; for example, a very militarized area might have very high support for one party.
And there are many ways to commit fraud besides ballot stuffing. Making it difficult for particular groups of people to vote, manipulating campaign finance laws and good old coercion are just some of the other tactics that can turn an election.
Political scientist Judith Kelley of Duke University, who has been investigating election monitoring around the world, says the new analysis is “very clever.”
“I’m all for these methods — we can never have eyes everywhere,” says Kelley. “Of course the guys that are really good at stealing elections don’t really do it when counting votes. They have a much broader apparatus.”