Statistical model suggests widespread fraud in Russian election

September 24, 2012|By Jon Bardin | Los Angeles Times

A new statistical analysis argues that widespread voter fraud took place… (Vyacheslav Oseledko / AFP/Getty…)

Widespread ballot-box stuffing and fraud likely occurred in the 2012 Russian presidential election that returned Vladimir Putin to the office, according to a new statistical model. The analysis, published Monday in the Proceedings of the National Academy of Sciences, also identifies Uganda as a site of widespread election tampering.

There have been rumors of election fraud in Russia for the past several elections, and they reached a fever pitch this year. But election fraud is difficult to prove. Past approaches looked for examples of something called "Benford's law," which looks for regularities in the numbers reported in elections-- like the presence of too many zeros because someone rigging the election prizes multiples of ten. But that approach has been difficult to apply, because it requires that analysts know just how many of each digit are likely to occur in the results of a fair election.

The new model, created by a team of Austrian scientists, takes a much more rigorous statistical approach, but it relies on a relatively simple idea: If an election has areas that have extremely high voter turnout -- close to 100% -- where that turnout is mostly for one candidate, the fix is likely in.

The researchers created plots from past elections in Russia, Uganda, Austria, France and seven other countries, selected because election data was readily available. For each of the countries, voter turnout in each geographic area was plotted against the percentage of people in those areas who voted for the winning candidate.

In most of the countries -- like Romania, Finland and France -- the graph showed that most areas had roughly the same voter turnout, with about the same range of votes for the winning candidates.

But in Russia and Uganda, there was a group of points on the graph that were far from the norm areas where virtually everyone voted, and virtually everyone voted for the winner. Such results, the authors say, stink of voter fraud.

In particular, results showing that some areas had 100% turnout and 100% voting for the winning candidate suggested that not only were votes added for the winner -- Putin, in the Russian case -- but also that votes were actually removed from other candidates.

The authors take pains to write that evidence of this "bimodal" voting -- where some pockets of a country are way off the norm in terms of voter turnout and voter preference -- is not in itself enough to prove fraud. But, they say, if all the signs they discovered in Russia, or Uganda, are there, it is hard to argue that the election represents the will of the people.

Or, as the researchers write, "Should such signals be detected, it is tempting to invoke G. B. Shaw, who held that "[d]emocracy is a form of government that substitutes election by the incompetent many for appointment by the corrupt few."

You can read the paper here.

Return to the Science Now blog.

FEATURED