The Electoral Integrity Project is an academic project based at Harvard University and the University of Sydney which seeks to quantify the integrity of elections worldwide. The project freely publishes its Perceptions of Electoral Integrity dataset for scholarly use. The most recent data release covers 285 elections in 164 countries from mid 2012 to the end of 2017. The founding Director is Pippa Norris. The project received media attention in 2016 when it ranked the United States last among western nations. One of the project's International Advisory Board, Andrew Reynolds, a political science professor at the University of North Carolina at Chapel Hill, noted in the Raleigh-based News and Observer that his home state's election integrity score was similar to Cuba, Indonesia and Sierra Leone. The study ranked integrity of the state's congressional districts lowest in the nation just below similar outlier Wisconsin. An editorial in the Wall Street Journal ridiculed the study, noting that "Democracy in New York and Virginia is supposedly more imperiled than in Rwanda, though Rwanda is controlled by an autocrat. The worst-performing state, Arizona, is outranked by Kuwait, Ivory Coast and Kyrgyzstan." Dylan Matthews writing in Vox agreed that "it seems foolish to infer from that that the US is less of a democracy than Rwanda" but felt that the EIP had highlighted important issues such as gerrymandering and voter registration laws. Statistician Andrew Gelman had a negative view of the PEI Index, commenting that " all seems like an unstable combination of political ideology, academic self-promotion, credulous journalism, and plain old incompetence", noting among other things that the EIP's 2014 data release has previously given the North Korean parliamentary election an 'electoral integrity' score of 65.3 and Cuba 65.6, higher than elections in EU members Romania and Bulgaria. Norris addressed the controversy in two long replies to Gelman, noting that her team had subsequently dropped the North Korean election from the dataset. Gelman, however, questioned her justification for this removal and continued to question the EIP's methodology more generally.
