| Benford's Law and its generalizations are a set of rules which describe the distribution of digits in numbers for many types of data sets. It is often used to help detect data that are fraudulently created. We investigate whether adding a subset of randomized values to a data set will alter its distribution of digits as analyzed through various Benford tests. We consider two sets of data - Medicare charges and Medicare payments for thousands of hospitals - and analyze them using various Benford tests. In addition, we remove 20% of the observations for each set and replace them with uniform random values, and examine the differences. |
Updated 03/19/2014