The Sampling Shortfall
You are testing the occurrence of 4,800 sales invoices for the year ended 31 March. You plan a statistical sample of 60 invoices using a 5% risk of over-reliance and a tolerable deviation rate of 5%, expecting zero deviations. You find three deviations: one invoice had no delivery note, one was posted to the wrong customer account but later corrected, and one related to goods shipped in April but invoiced in March. The client's controller says three out of 60 is only 5% and therefore within tolerance. Materiality is $180,000 and the total invoice value is $41 million. You have already tested controls and found them effective.
1. Evaluate the controller's reasoning and determine the audit conclusion from the sample results.
Read the worked answer
The controller's reasoning is wrong on two counts. First, the sample deviation rate is 3/60 = 5%, which equals the tolerable rate, not falls below it. In attribute sampling, the upper deviation limit at a 5% risk of over-reliance with three deviations in a sample of 60 exceeds 5% substantially. The correct conclusion is that the control cannot be relied upon at the planned level. Second, the controller is treating a statistical result as a simple percentage, which ignores sampling risk.
The three deviations are also not equivalent:
- missing delivery note = occurrence/cut-off issue;
- wrong customer account = posting accuracy, corrected, lower concern;
- April shipment invoiced in March = cut-off error, the most serious because it overstates revenue and receivables.
The planned reliance on controls is no longer supportable. The auditor should either increase the sample or abandon the control reliance strategy and perform substantive testing. The cut-off error must be projected to the population and compared to materiality.
2. Project the misstatement and explain what additional procedures you would perform before concluding on revenue.
Read the worked answer
Only the cut-off error is a monetary misstatement; the other two are control deviations. Project it as follows: one invoice in 60, so the projected misstatement is the invoice value multiplied by 4,800/60 = 80. If the invoice is, say, $4,000, the projection is $320,000, which exceeds materiality of $180,000.
Even a single projected error above materiality warrants expansion. Procedures:
- Increase the sample or test the full population of March and April invoices around year-end.
- Test cut-off directly by comparing shipping dates to invoice dates for the last week of March and first week of April.
- Review credit notes and returns after year-end.
- Reassess control risk and document why reliance was reduced.
- Consider whether the deviations are anomalies or systemic; three different causes suggest weak monitoring rather than isolated slips.
The conclusion should state the projected misstatement, whether it is material, and whether the control reliance conclusion must be revised. The controller's acceptance of 5% should be rebutted in writing with reference to sampling risk.