Digit-based election forensics (DBEF) tries to determine whether an election was fraudulent by looking at the numerical results at the finest level at which vote counts were reported. This is appealing particularly because of its low cost, especially when local vote counts are available online. The methods are designed to classify a set of election returns as fraudulent or fraud free by inspecting their numeral distributions, and rest on (1) assumptions about the probability distribution of results in the absence of fraud or malfeasance, and (2) null hypothesis significance testing. DBEF has been advanced mainly by Professor Walter Mebane of University of Michigan, and his DBEF toolbox is considered to be the standard.