The Sampling Plan That Looked Fine

Deficiencies in the plan.

1. No stratification. With 4,200 accounts and $9.6 million, the population is almost certainly skewed - a small number of large accounts will carry most of the value. Simple random selection over the whole population gives every account an equal chance, which means large balances are under-represented relative to their value. Monetary unit sampling or stratification by value would have targeted the high-value accounts and given better coverage of the balance.
2. Sample size not justified. 60 out of 4,200 is 1.4% of items. There is no documented basis for the sample size - no assessment of expected error, tolerable misstatement or the level of assurance required. The assistant cannot demonstrate the sample was sufficient.
3. Non-response not followed up. 19 of 60 (32%) did not reply. Under ISA 505, non-responses must be followed up with alternative procedures - second requests, subsequent cash receipts testing, or examination of shipping documents. A 32% non-response rate is high, and the non-responders may differ systematically from responders (e.g. disputed balances are less likely to confirm). Treating them as clean is not supportable.
4. No projection of misstatement. The assistant compared the observed differences to materiality without projecting them to the population. Under ISA 530, the auditor must project misstatement results to the population and compare the projected misstatement - plus an allowance for sampling risk - to tolerable misstatement.

Why $13,500 understates the risk.

The observed differences are not the estimate of population misstatement; they are a sample result. If 3 of 41 responding accounts (7.3%) are misstated, the projected misstatement across the population is far larger than $13,500. The nil confirmation against a $9,000 recorded balance is qualitatively serious - it suggests either a fictitious receivable or a misdirected payment - and warrants specific investigation, not aggregation with a $500 rounding difference. The $4,000 and $9,000 items should be investigated individually for cause (cut-off, cash in transit, disputed invoice), while the $500 item is likely a timing difference.

The correct evaluation. Project the misstatement using the sampling method actually used, add an allowance for sampling risk, and compare to tolerable misstatement of $180,000. If the projected misstatement plus sampling risk exceeds tolerable misstatement, the conclusion must be that the balance is materially misstated unless further procedures reduce the estimate.

Additional procedures to require.

1. Re-perform the selection with stratification. Split the population into key items (accounts above, say, $50,000 - test 100%) and the remainder (sample). Document the sampling method, the parameters and the basis for the sample size. If the original sample is retained, document why it is adequate.
2. Follow up all 19 non-responses. Send second requests; where no reply, perform alternative procedures - agree the balance to subsequent cash receipts, shipping documentation and sales invoices. Document each one. If alternative procedures cannot be performed for some accounts, treat them as unresolved and consider the effect.
3. Investigate the three differences for cause.
   • $4,000: determine whether it is a pricing dispute, a cut-off issue, or a genuine error. Check the invoice and any credit note.
   • $500: likely timing - confirm and document.
   • $9,000 confirmed nil: this is the priority. Trace the sale, the invoice, the delivery docket and any subsequent receipt. If no goods were delivered or the customer has no record, consider whether the receivable is fictitious and whether fraud procedures are needed.
4. Project the misstatement. Using the sampling method applied, project the known misstatement to the population. For a stratified or MUS design, project within each stratum. Add an allowance for sampling risk.
5. Consider the effect on the audit approach. If the projected misstatement is high, expand testing - increase the sample, test the allowance for doubtful debts, and reassess the risk of material misstatement for revenue and receivables.

Evaluation against materiality.

Compare the projected misstatement plus sampling risk to tolerable misstatement ($180,000), not to overall materiality directly. If the total is below tolerable misstatement, the balance can be accepted, but the unadjusted differences still go on the summary of uncorrected misstatements and are considered in aggregate with other findings against overall materiality ($240,000). If the total exceeds tolerable misstatement, request adjustment or extend procedures. If the client refuses to adjust and the effect is material, the opinion is affected.

Documentation. The file must record the sampling design, the evaluation, the follow-up of non-responses and the conclusion. The current file does none of this, and that is a quality issue in its own right, independent of whether the balance turns out to be correct.