Vote rigging is on a lot of people's minds lately. Being able to tell whether an election has been tampered with could be the difference between the survival or destruction of democracy, but it's not always easy. Three scientists have come to the rescue, designing mathematical methods to look for certain sorts of electoral fraud and testing them on elections considered suspect.
Dodgy elections come in many forms and Dr Peter Klimek of the Medical University of Vienna isn't trying to catch all of them, at least with the one method. However, he is keen to produce a detection method for what is known as voter rigging. Unlike vote rigging, where ballot boxes go missing or have extra votes mysteriously appear in them, voter rigging involves real votes being cast under duress.
Suspicions of such events may arise when people come forward, claiming to have been the victims of harassment and intimidation, but Klimek and his colleagues wondered if they could test these statistically.
The 2011 Russian parliamentary elections and the 2013 Venezuelan presidential ballot both saw widespread allegations of voter rigging. The latter case saw electoral observers ROE-AE allege widespread use of violence and intimidation.
In Science Advances, Klimek argues smaller polling centers are more likely locations for such rigging than large ones, since they get less attention from independent election observers and there is a greater chance that individual voters' preferences will be known. The latter factor means anyone intent on coercing people into not voting, or voting against their intent, will know who to target.
The paper presents a statistical test to see if small polling stations are producing suspiciously different results from larger ones. The authors tried this out on 21 elections in 10 countries. Some of these were in places widely perceived to be models of fair democracy, such as Austria, while others have been subject to considerable suspicion.
In Austria, Finland, and South Africa, among other examples, small and large polling stations produced very similar results. The same was true in early Venezuelan elections Klimek examined. However, in some elections from Russia, Venezuela, and Uganda, turnout was much higher at small polling stations and heavily skewed towards the winning party. The smaller the station, the greater the skew. There were enough suspect votes to change the outcome of the 2013 Venezuelan election.
Results skewed by polling station size don't necessarily indicate voter rigging. For example, stations are often smaller in rural areas, which usualy vote differently from cities. Governments may also open more small polling stations in sympathetic areas, making it easier for their supporters to vote without engaging in actual rigging. Nevertheless, Klimek's work provides a test that can be used as a starting point for further investigation.