An invoice has the correct amount, but nobody approved it. Did the audit test pass or fail? This lesson covers CISA Domain 1, Section 1.6: how an audit test turns an objective into evidence, and how sampling lets you conclude about a whole population.
1.6.1 Compliance versus substantive testing
- A compliance test (test of controls) asks whether a required control operated as intended. Did an authorized person approve the invoice before payment?
- A substantive test examines the transaction, balance or information itself. Is the amount correct?
A correct amount and a missing approval can occur together, so record the substantive result and the control deviation separately.
Substantive work is either a test of details on individual records, or a substantive analytical procedure that compares a reliable expectation with the reported result.
Direction of testing
Choose a starting population that contains the exception you are looking for.
| Audit question | Start with | Compare with |
|---|---|---|
| Were all receipts recorded? | Goods receipt records | Inventory or accounting records |
| Are recorded purchases genuine? | Recorded purchases | Orders, invoices and receipts |
| Were production changes authorized? | Actual deployments and direct changes | Approval records |
1.6.2 Sampling
The population is the entire defined group, a sampling unit is one item, the sampling frame is the list you select from, and the sample is the subset you examine. Validate coverage before selecting anything: a register of completed reviews cannot show reviews that were never performed.
- Statistical sampling uses probability-based selection and supports a quantified sampling risk.
- Nonstatistical sampling uses professional judgment and does not quantify that risk statistically.
- Attribute sampling measures presence or absence, often a deviation rate.
- Variable sampling estimates a quantity, such as misstatement.
Remember ACVS: Attribute with Compliance, Variable with Substantive. It is a common association, not a rule.
Sampling methods
| Method | Use it when |
|---|---|
| Fixed-size attribute | Estimating a rate with a sample size set before testing |
| Stop-or-go | Few deviations are expected; evaluate in planned stages |
| Discovery | Even one occurrence matters |
| Unstratified mean-per-unit | Estimating one population's total from the sample mean |
| Stratified mean-per-unit | Estimating totals within groups, then adding them |
| Difference estimation | Projecting misstatement from recorded-to-audited differences |
| Ratio projection | Applying the sample's error ratio to the recorded population |
Selecting items
Simple random gives every unit the same chance. Systematic uses a random start and an interval, so check for repeating patterns. Stratified selects within relevant groups. Probability-proportional-to-size gives larger units more weight, so address zero, negative, understated or unrecorded items separately. Targeted selection picks specified items, and supports conclusions only about those items.
Sampling risk
Sampling risk is the chance a sample leads to a different conclusion from testing the full population with the same procedure. Nonsampling risk comes from other causes, such as a wrong procedure or an incomplete extract.
Worked decision
A monthly access review has a 5% tolerable deviation rate. The auditor finds 3 deviations in 100 instances, a 3% observed rate, and the statistical method gives an upper deviation limit of 7.6%. Compare the limit with tolerance: the result does not support planned reliance.
Keep going
- Read the free companion book, the CISA Complete Study Guide by Muhammad Aamir.
- Watch the full CISA Complete Course by Mirak X. Next up is Section 1.7, Audit Evidence Collection Techniques.
- Subscribe to Mirak X on YouTube. New lesson every day at 6 PM (PKT).
- Try free CISA practice questions and mock exams at https://certprepx.com/.
Free study material. Not affiliated with or endorsed by ISACA. CISA is a registered trademark of ISACA.


