Definition
AU-C 530 describes stratification as splitting a population into groups of items with similar traits. The most common basis is monetary value, with the largest items placed in a stratum tested at higher coverage. Other bases include type of customer, location or age. Results from each stratum are projected separately and then added together.
Why it matters in an audit or review
A population with wide variance in item size needs a larger sample if treated as a single group. Stratifying focuses effort on the items where a misstatement would matter, and tends to reduce the sample needed for the same level of assurance. It also lets the auditor test the top stratum at 100 percent and sample the rest.
Example
A $9 million inventory listing has 2,000 lines (illustrative). The 60 largest lines make up $4.5 million and are tested in full. The remaining 1,940 lines, totaling $4.5 million, are sampled using random selection. If the sample covers $450,000 of the lower stratum and finds a $2,400 overstatement, ratio projection gives $24,000, applied to that stratum only.
Common mistakes
- Projecting errors found in the sampled stratum to the 100 percent-tested stratum.
- Building strata after seeing where the errors are.
- Forgetting to reconcile the strata back to the total.
FAQ
Does stratification change the sampling method?
It changes how the population is cut, not whether sampling is statistical.
Sources
- AICPA, AU-C Section 530, Audit Sampling (definition of stratification and projection of misstatements)
Definitions paraphrase AICPA and FASB guidance; read the standard itself before relying on a requirement. Example figures are illustrative. This page is guidance, not professional advice. Report an error