New statistical tool kit fingerprints electoral fraud (not only) in Turkey

The cheap and easy to use election observation

[Press release via AlphaGalileo]

An international research team led by Peter Klimek from the Complexity Science Hub Vienna proposes a new statistical method to diagnose voting irregularities and calculate their impact on the final results. The highly sensitive procedure shows the Turkish votes in 2017 and 2018 equally problematic.

(Vienna, Oct 5, 2018) Fair and free elections are at the core of democracy. Independent institutions like the OSCE try to monitor elections to ensure a fair and democratic process. However, a few hundred election observers cannot cover thousands of polling stations. Especially remote areas and smaller towns and villages are rarely monitored and thus prone to different sorts of electoral fraud.

A new combination of statistical methods can help uncover such irregularities and relate them to specific forms of malpractice. This includes even very subtle intimidations, like voting booths without curtains or extensive police presence near polling stations, as a new paper in PLOS ONE shows using the example of Turkey. The “tool kit,” proposed by a team of complexity scientists headed by Peter Klimek from the Complexity Science Hub (CSH) Vienna and the Medical University of Vienna, further allows them to estimate if the electoral fraud was strong enough to change the final results.

“Our method is fast, cheap, and easy to use,” explains Peter Klimek. “The only input we need is the election results.” Usually these lists are provided online within hours after an election. “Our tests display a very specific pattern: the fingerprint of the poll,” he points out. These fingerprints show places where manipulation can be excluded or hotspots where it occurred with high probability.

Unlawful polls in Turkey

In Turkey, for instance, the researchers found several districts with problematic fingerprints right after the constitutional referendum in 2017. Recep Tayyip Erdoğan used the referendum to replace the Turkish parliamentary system with a presidential system. The highly controversial poll was followed by an equally disputed snap election in June 2018 that installed Erdoğan as president with wide-ranging powers and his AKP party as the ruling force in parliament.

“We found clear statistical signs of intimidation at small polling stations in the referendum data,” says co-author Stefan Thurner from CSH Vienna and the Medical University of Vienna. “We also saw systematic and highly significant traces of ballot stuffing in about eleven percent of the stations.” Ballot stuffing means that multiple votes are submitted per person instead of one. “The numbers were just big enough to influence the outcome of the referendum towards a ‘yes’ in favor of the constitutional change,” says the complexity scientist.

The presidential and parliamentary elections in June 2018 show a similar pattern: The statistical tools produced the same fingerprints of fraudulent practices in the very same districts as in 2017. In this case, the digitally traceable aberrations were not large enough to change the final outcome, because the majority of the Turkish people voted in favor of Erdoğan and the AKP anyway. “Nevertheless, our findings confirm reports of undemocratic election procedures in Turkey,” says Stefan Thurner. “The patterns are similar to other controversial elections that we have analyzed in earlier works; for instance, the voting in Russia from 2003 onwards, in Venezuela from 2006 onwards, or in Uganda 2011.”

Election observation in the digital age

“Unlawful voting throughout the world could have been detected, or even prevented if these methods were systematically utilized for election monitoring,” says Peter Klimek. “The results from the 2017 referendum in Turkey, for example, could have been used to send observers to the most critical places. This would have helped to prevent the same kind of fraud in the 2018 elections.”

“We offer a simple and cheap method as an ideal extension to the common election observation,” Stefan Thurner adds. “Organizations like the OSCE are invited to catch up with the digital age and use the new tools of the 21^st century.”

Publication:

R. Jiménez, M. Hidalgo, A. Hinteregger, S. Thurner, P. Klimek, “Forensic analysis of Turkish elections in 2017–2018.” PLOS ONE, 5 October 2018. https://doi.org/ 10.1371/journal.pone.0204975

Additional information:

P. Klimek, Y. Yegorov, R. Hanel, S. Thurner, “Statistical detection of systematic election irregularities,” PNAS, 9 October 2012, 109 (41), 16469-16473

R. Jiménez, M. Hidalgo, P. Klimek, “Testing for voter rigging in small polling stations,” Science Advances Vol. 3, no. 6 (2017), e1602363

R. Jiménez, S. Thurner, L. R. Pericchi, P. Klimek, “Fraud Detection, Electoral,” Wiley StatsRef, 14 August 2018