What it means
Sampling is a method rather than a result. Where a sample is the group of items you end up with, sampling is the technique of picking them, and the technique determines whether the conclusions can be trusted.
There are two broad families. Probability sampling gives every item a known chance of selection, and includes simple random sampling, systematic sampling (every nth item) and stratified sampling (splitting the population into groups and sampling each).
Non-probability sampling, such as judgemental or convenience selection, relies on someone's choice and cannot support the same statistical conclusions. Auditors also use monetary unit sampling, where each dollar in the population is a possible selection, so large balances are more likely to be tested.
This is efficient when the main worry is overstated amounts, because it concentrates effort where the money is. Attribute sampling, by contrast, counts how often something goes wrong, such as the proportion of invoices with no approval signature.
Good practice is to define the population clearly, choose the method, work out the sample size and record how items were picked. That record lets a reviewer reproduce the selection and shows the results are not cherry-picked.
The method also has to match the question being asked. Testing whether all transactions exist calls for a different approach from testing whether the controls operate, and a poor match produces tidy-looking numbers that answer the wrong question.
Sampling is also used well beyond audit. Credit teams sample loan files, operations teams sample products for quality and marketers sample customers for surveys, so understanding the basic methods helps a non-specialist question the evidence behind a headline statistic.
In practice
Real-world examples.
Example
An external auditor uses monetary unit sampling on a $20,000,000 receivables balance so that larger customer balances are more likely to be chosen for confirmation, which concentrates the audit effort where the money is.
Example
A retailer's finance team uses stratified sampling for stock counts, dividing items into high, medium and low value groups. It counts a high proportion of the high-value group and a small proportion of the low-value group, which saves days of counting without ignoring the items that matter most.
Example
A bank's compliance officer applies systematic sampling to 10,000 account-opening files, reviewing every 100th file to check that identity documents were collected. She begins from a randomly chosen starting file so that no one can predict which files will be examined.
Formula
Calculation
Two calculations are common. The first is the sample size for estimating an average, and the second is the sampling interval for systematic sampling.
Sample Size n = (Z x s / E) squared
Sampling Interval = Population Size / Sample Size
Worked example: an auditor wants to estimate the average invoice value within $100. The standard deviation (the typical spread of invoice values) is estimated at $500. Using a Z value of 2 for roughly 95% confidence (the exact figure is 1.96):
n = (2 x $500 / $100) squared = (1,000 / 100) squared = 10 squared = 100 invoices
For a population of 5,000 invoices, the sampling interval is 5,000 / 100 = 50, so the auditor picks a random starting point between 1 and 50 and then every 50th invoice.Case study
Seen in the real world.
Ironwood Telecom is an illustrative, fictional company whose controller wanted to test approval of 24,000 purchase orders. The team first tried reviewing the orders that were easiest to retrieve, which turned out to be mostly small ones.
An internal reviewer pointed out that this convenience approach would say nothing reliable about the big orders. The team switched to stratified sampling, testing all orders above $100,000 and a random 2% of smaller ones.
In the illustrative result, two large orders lacked proper approval, a finding that the convenience sample would almost certainly have missed. The story shows why the selection method is part of the evidence. The reviewer's report also recorded the selection rules, so that an external auditor could reproduce the exact same sample later.
Watch out
Common mistakes.
- Using convenience selection, such as the easiest files to find, and then treating the results as if they covered the whole population.
- Choosing a sampling method before defining exactly what population is being tested.
- Failing to record how items were selected, which makes the work impossible to reproduce or defend.
Questions
People also ask.
What is the difference between random and systematic sampling?
Random sampling picks items by chance, whereas systematic sampling takes every nth item from a random starting point, which is simpler to run but can be thrown off by hidden patterns.
What is stratified sampling?
It splits the population into groups, such as by value, and samples each group separately so that important segments are not missed.
Does sampling replace checking everything?
No, it is a cost-effective alternative when a full check is impractical, and some high-risk items are still tested individually.
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%Related