In every binary classification problem, two kinds of errors can occur, namely, misclassifying an item that belongs in the first class as belonging to the second class, and vice versa. In statistics, binary classification is often formalised as a hypothesis test. A natural framing of the classification problem in election forensics would state the null hypothesis to be that the results are clean, and the alternative hypothesis to be that the results are tainted. A false positive, also known as a Type I error, is to label a clean election “tainted”; a false negative, also known as a Type II error, is to label a tainted election “clean.”