Pages Menu
TwitterRssFacebook
Categories Menu

Posted by on Sep 26, 2012 in Science |

Crean method to detect electoral fraud

Crean method to detect electoral fraud

(CC) myJon

To detect a fruade electoral statisticians seeking patterns normally accede to certain constants indicating that large blocks of votes were for a certain candidate. The problem with this method is that the best is an indicator of irregularities, but nothing very definite as to declare election fraud.

To solve this problem, a team of Austrian mathematicians conducted a study which focused on the voting units.

The idea is that if a small region has a high percentage of voter turnout, and virtually all the votes from that area are for only one candidate, it’s probably because they stuffed ballot boxes with votes for, or destroyed the votes of opponent (or poorly counted, etc.).

If enough districts have this problem, it is possible that may affect the results of an election, and this is what the researchers found when analyzing in Uganda last year and early this year in Russia, where Vladimir Putin returned to power after becoming prime minister.

Crean method to detect electoral fraud image 2

(C) PNAS

As seen in the image, in elections in Uganda and Russia were numerous places where the winning candidate got figures close to 100% of the votes taken. The vertical axis is the percentage of votes the winning candidate, and the horizontal axis, the percentage of voter turnout.

Unfortunately the method is not sufficiently robust to countries with less than 100 constituencies, as we must also be clear that there are countries where there are areas with deep political differences as Canada, where the Conservatives won with over 40% of the Vote in all English-speaking provinces of the country, but only 16.5% obtained in the French-speaking province of Quebec.

Links:
- Researchers use new statistical method to show fraudulent voting in Russian election (Phys.org)
Statistical detection of systematic election Irregularities. Supporting Information (PNAS)

Tags: , , , , ,