I spent some time today digging into the voting-center level results recently posted by the Venezuelan opposition (resultadosconvzla.com/) 1) Their dataset has ~81% coverage - doing some modeling to extrapolate the data they collected to the rest of the country, I'm estimating that the Opposition got 66% of the vote compared to 32% for Maduro. 2) Unsurprisingly, the ~20% of missing precincts are modeled to be substantially more supportive of Maduro (Opposition +19) than the precincts they did collect (Opposition + 37) and also are quite a bit smaller, lending some credence to the data generation process. Even if Maduro won all of the votes in these precincts the opposition still would win easily. 3) The opposition saw it's largest gains in lower-income precincts that went strongly for Maduro in 2013, winning ~2/3rds of precincts that Maduro won eleven years ago 4) It's hard to "prove" that a set of election results are real - but having spent a decent amount of time doing regressions on the results, they show the same kind of complex relationships with previous election results and demographics that I have seen in American elections, in a way that I think would be pretty difficult to replicate out of thin air.