When projecting errors for a non-statistical sampling method, the auditor can choose between extrapolating over the number of items in the population and extrapolating over the book value in the population. Typically, these methods will generate different results. Under which circumstances will the results be the same?