Analytic review
Is it reasonable?
Analytic review procedures include a variety of tests which can be applied to data in order to determine of the amounts are reasonable. Often, graphic output can help the auditor better visualize the data.
Linear regression can help identify possible errors or unusual transactions which may need further review. This can be done using bounds checking (see example chart below).
In the chart above, one transaction stands out. Watch a four minute video clip which demonstrates the process.
Analytic review may also identify potential fraud. Errors often tend to be found in situations where something is different.
One technique for the identification of such potential errors is through the examination of patterns, based upon various criteria such as occurrence of round numbers, Benford's Law, distribution of days of week, transactions falling on a Federal holiday, etc.
In each of these cases it is possible to apply a metric to the entire population. Then divide the population into sub-groups and apply the same metric against the subgroup. Then compare the metrics of each subgroup against the population as a whole, using a statistically valid comparison measurement such as Chi-Squared or Kolmogorov-Smirnov (K-S).
View Power Point presentation on the whole process and the background theory.