The linked abstract of last year's voting study is worth looking at just for the idiocy it highlights:
The initial results of Venezuela’s April 14 presidential election returned 7,575,506 votes for Nicolás Maduro, and 7,302,641 votes for opposition candidate Henrique Capriles. This is a difference of 272,865 votes, or 1.8 percent of the two-way total between the candidates. Following the announcement of the official results, Capriles asked for a full audit beyond just the “hot audit” of 53 percent of voting machines that was done on site, in the presence of observers from both campaigns, as well as witnesses from the community, on the night of the election. Venezuela’s National Electoral Council (CNE) agreed to audit the remaining machines, although Capriles later rejected their proposed audit. The U.S. government, unlike almost all other governments in the world, has held off on officially recognizing the Maduro government until such an audit is conducted.
A statistical analysis of the audit of 53 percent of voting machines from Venezuela’s April 14 election, done on the day of the election, shows that this audit was decisive. The odds of getting the April 14 audit result if in fact the unaudited machines contained enough errors to reverse the election outcome are far less than one in 25 thousand trillion. This new paper details and explains the statistical analysis behind these results.