Audit Sampling: Methods and Monetary Unit Sampling

Audit sampling means testing less than 100% of a population and drawing a conclusion about the whole of it. The standards are AU-C 530 (private companies), ISA 530 and CAS 530, and AS 2315 for public companies.

Testing every item above a threshold is not sampling. Those are key items, and you conclude only on what you tested. Sampling is what you do with the rest.

Statistical or non-statistical

Both are allowed. The difference is how you measure sampling risk.

• Statistical sampling uses random selection and probability to measure sampling risk. It gives a number you can defend, such as "95% confidence the misstatement is under $100,000".
• Non-statistical sampling relies on judgment for size and evaluation. The sample still has to be representative, and sizes are usually set close to what a statistical plan would give.

Selecting the items

Method	How it works	Notes
Random	Every item has an equal chance, using a random number generator	Works for statistical sampling
Systematic	Pick every nth item from a random start	Watch for patterns in the population
Monetary unit (MUS)	Every dollar has an equal chance, so larger items are more likely	Suits overstatement testing
Haphazard	Pick without a structured method, avoiding bias	Non-statistical only
Block	A run of consecutive items, such as all of March	Rarely acceptable as a sample on its own

Monetary unit sampling, worked through

MUS is the most common method for tests of details on balances that could be overstated, such as receivables, revenue and expenses.

With zero expected misstatement, the sample size is:

sample size = book value × confidence factor ÷ tolerable misstatement

Common confidence factors with zero expected misstatement are 3.0 for 95% confidence and 2.3 for 90%.

Example. Expenses of $2,000,000, tolerable misstatement of $100,000, 95% confidence:

• Sample size = 2,000,000 × 3.0 ÷ 100,000 = 60 items
• Sampling interval = 2,000,000 ÷ 60 = $33,333
• Pick a random start between 1 and 33,333, then select the item containing every 33,333th dollar. Any item larger than the interval is always selected.

If you expect some misstatement, the sample gets larger. The AICPA Audit Sampling guide has the tables.

Evaluating the results

1. Find every misstatement in the sample and its cause.
2. Project it to the population. In MUS, an item smaller than the interval is projected by its "tainting": misstatement ÷ book value × interval. A $500 error on a $2,000 item, with a $33,333 interval, projects to about $8,333.
3. Add an allowance for sampling risk to get the upper misstatement limit.
4. Compare with tolerable misstatement. If the limit is above it, you cannot conclude the balance is fairly stated from this sample. Extend testing, ask the client to investigate, or propose an adjustment.

Do not correct only the item you found. A projected misstatement belongs on the summary of misstatements even if the client fixes the sampled item.

Sampling for controls

Controls are tested with attribute sampling: each item either meets every attribute or is a deviation. Sample sizes depend on how often the control runs and how much you rely on it. See our guide to the test of controls.

Where Warp Accounting fits

Warp Accounting picks samples from the GL to the firm's rules and runs the tests on each item: vouching to invoices, checking amounts, dates and accounts, and recording exceptions with the source page attached. See test of details and our guide to substantive testing.