ABSTRACT
This chapter begins the second half of the book by looking at methods for detecting various types of electoral fraud and unfairness. Here, we look at what we can test if the government only provides vote counts for each of the candidates at the electoral division level. It turns out that this is sufficient information to test for fraudulent counting. That is, the Benford test of this chapter allows us to gather evidence that the claimed vote counts were modified improperly.
This test, however, has a rather limiting assumption underlying it. To account for this, the generalized Benford distribution is developed and three methods are explored. The chapter uses the results from the 2009 Afghan presidential election to illustrate its techniques.
