For better or worse, one of the biggest stories in US politics today is the detection of election fraud, or in many cases the lack of election fraud. But determining whether fraud happened in an election can be difficult, even while proving the validity of elections for some has become increasingly important. Wouldn’t it be incredible if we could just plug a set of data from an election into a toolkit that could give us an answer if fraud occurred? Well, one political scientist from the University of Michigan, Walter Mebane believes he may have developed just such a toolkit. It’s called “election forensics”. Much like machine learning algorithms, when tested in the field it does seem to perform fantastically well, but figuring out exactly how it works can be a complicated web to untangle. We give it a shot on this episode.