Forensic Analytics, 2nd Edition

CHAPTER 10Comparing Current Period and Prior Period Data : Part 2

THE PREVIOUS CHAPTER TOOK a new turn and looked at ways to compare two sets of transaction data to determine whether there had been a significant change between the current data and the prior data. The distribution and the internal diagnostics of the prior period's data formed a benchmark for the current period. The premise was that large period-to-period changes could be symptoms of a higher risk of fraud or errors. The tests included a table of numeric statistics related to a data overview, measures of central tendency, measures of variability, and the shape of the distribution. The tests also included comparative histograms and comparative first-two digit tests. The examples included the District of Columbia's purchasing card data, payroll data, debit card transactions, and grocery sales amounts.

The tests in the previous chapter could be called comparing two populations. We were essentially testing for differences between two or more independent groups of transactions. For example, in the purchasing card data it did not matter if Person A made m number of purchases for x dollars in the prior period and nothing in the current period. The comparison did not compare Person A to themselves in the two periods. There was no “before and after” comparison for groupings such as by person, or by agency, or by vendor. In this chapter we will compare data for two periods in a setting that could be called a longitudinal ...