What it means
Auditors, internal control teams and quality functions cannot examine every transaction in a large business, so they test a subset and project the result. Sampling risk is the mathematical price of that shortcut: even a perfectly executed test on a properly chosen sample can produce a misleading answer purely by chance.
The risk splits into two directions, and they carry very different consequences. Concluding that controls work or balances are correct when they are not is the dangerous direction, because it means a real problem goes unreported, while concluding there is a problem when there is not merely causes extra work.
Sample size is the main lever. Larger samples reduce sampling risk but cost more time, so testing is planned around a stated confidence level, and the sampling risk accepted is simply the remainder: a 95% confidence level implies a 5% sampling risk.
Sampling risk is distinct from non-sampling risk, which covers everything that can go wrong for reasons unrelated to size, such as a tester misunderstanding a control, applying the wrong procedure or failing to spot an error that was in front of them. Non-sampling risk is reduced by training, supervision and review, not by testing more items.
Sensible testing strategies also reduce reliance on sampling altogether. High value or unusual items are often examined in full and only the remaining population is sampled, which concentrates sampling risk on the part of the balance where an undetected error would matter least.
In practice
Real-world examples.
Example
An internal audit team tests 60 of 3,000 purchase orders for evidence of approval and finds every one properly signed. It reports reasonable assurance rather than certainty, noting explicitly that a small number of unapproved orders could exist without appearing in a sample that size.
Example
A retailer's stock count team counts 5% of stock keeping units and projects a shrinkage rate of 1.2%. The finance director increases the sample to 15% for high value electronics, because a projection error in that category would move the accounts far more than one in stationery.
Example
An external auditor testing revenue selects a statistical sample weighted towards larger invoices. Two errors turn up in the small item stratum, and because those items make up only a fraction of the total value, the projected misstatement stays well below materiality.
Think of it
“Sampling risk is the chance your sample doesn't represent the whole-not testing everything.
Formula
Calculation
Sampling risk = 100% - confidence level. A sample result is projected as: projected misstatement = (error found in sample / value of sample) x value of population.
An auditor is testing a trade receivables balance of $8,000,000 spread across 4,000 invoices, with materiality set at $100,000. She selects 200 invoices with a combined value of $500,000 and finds errors totalling $2,500.
The error rate in the sample is $2,500 / $500,000 = 0.5%, so the projected misstatement across the whole balance is 0.5% x $8,000,000 = $40,000. That sits comfortably below the $100,000 materiality threshold.
Because the test was designed at a 95% confidence level, the sampling risk is 100% - 95% = 5%. In plain terms, there is a one in twenty chance that the true misstatement is materially different from the $40,000 projected, which is why the auditor also allows headroom between the projection and materiality before concluding.Case study
Seen in the real world.
This is an illustrative and entirely fictional scenario. Ravensmoor Distribution, an invented wholesale business, had its supplier payments tested each year by an internal audit team that selected 40 invoices at random from a population of about 25,000. Those tests came back clean for four years running.
In the fifth year, a whistleblower disclosed that a purchasing manager had been approving payments to a supplier he controlled, worth roughly $460,000 spread across 180 small invoices. Because the fraudulent invoices represented under 1% of the population and were deliberately kept below the value threshold that attracted attention, a 40 item random sample had almost no chance of catching one.
The fictional company's response was to change how it sampled rather than simply to sample more. It layered targeted testing over the random sample, focusing on new suppliers, round-sum amounts and invoices just below approval limits, and it accepted that random sampling alone can never be a reliable defence against a determined fraud.
Watch out
Common mistakes.
- Treating a clean sample as proof that a whole population is clean, when a sample can only ever provide a level of assurance short of certainty.
- Confusing sampling risk with non-sampling risk, and responding to a tester's mistake by increasing sample size rather than improving training and review.
- Picking sample items by convenience, such as the first items in a file, which destroys the statistical basis on which any projection depends.
Questions
People also ask.
How do you reduce sampling risk?
Increase the sample size, stratify the population so higher value items are tested more heavily, and test every item above a chosen threshold in full.
Does a bigger sample eliminate the risk entirely?
Only testing 100% of a population removes sampling risk, and even then non-sampling risk remains because people still make mistakes.
Why not just test everything?
Cost and time, since full testing of a large population is rarely proportionate to the benefit, and modern data analysis is increasingly used to screen whole populations for exceptions instead.
From the founder's library
Take it further with the book.
Build your financial confidence beyond this definition. Shihan's full-length guide, Accounting Fundamentals, takes the same plain-English approach and turns it into a complete, practical playbook for non-finance managers, business owners and students - with chapter-end quiz answers and presentation slides included.
25% off with code MMHQ25, applied at checkout. Priced in USD - checkout may show the equivalent in your local currency.
View the book and save 25%