Audit Sampling and Analytical Procedures
Fieldwork
Audit Sampling and Analytical Procedures
Syllabus tag: KASNEB CPA | Intermediate Level | CA24 Auditing and Assurance | Topic 7 Audit Sampling and Analytical Procedures
Lesson objectives
By the end of this topic, you will be able to:
- Explain why sampling is necessary and what risk it introduces
- Distinguish statistical from non-statistical sampling
- Select an appropriate sampling method
- Project a misstatement from a sample to the population
- Design and interpret analytical procedures
Why this matters
No auditor examines every transaction, so every conclusion rests on inference from a part to the whole. Understanding what that inference can and cannot support is the substance of this topic.
Sampling risk and non-sampling risk
Sampling risk is the risk that the conclusion from the sample differs from the conclusion that testing the whole population would have given. It is inherent in sampling and is reduced only by taking a larger sample.
Non-sampling risk is everything else that can go wrong: the wrong procedure, a misinterpreted result, an error the auditor failed to recognise. A larger sample does nothing for it. It is reduced by planning, supervision and review.
The distinction matters because the response differs. A candidate who answers "increase the sample size" to a non-sampling problem has misdiagnosed it.
Statistical and non-statistical sampling
Statistical sampling uses random selection and probability theory to measure sampling risk. The advantage is that the risk can be quantified.
Non-statistical sampling relies on the auditor's judgement. It is quicker and cheaper, and the sampling risk cannot be measured.
Both are permitted. Neither excuses the auditor from making the sample representative of the population.
Methods of selection
| Method | How it works | Suits |
|---|---|---|
| Random | Every item has an equal chance | General use with statistical sampling |
| Systematic | Every nth item after a random start | Large uniform populations |
| Monetary unit | Chance of selection proportional to value | Testing for overstatement |
| Haphazard | No conscious bias, but not random | Non-statistical sampling |
| Block | Consecutive items | Rarely appropriate |
Monetary unit sampling deserves attention. Because selection probability follows value, large items are almost certain to be picked and small ones rarely. That makes it efficient for detecting overstatement and poor at detecting understatement or omission — a balance wrongly recorded as zero has no monetary units to select.
Block selection is rarely appropriate because a run of consecutive items tells you about one period or one clerk, not about the population.
Projecting a misstatement
Errors found in a sample must be projected across the population.
A receivables population of KES 48,000,000. A sample with a book value of KES 6,400,000 contains errors of KES 192,000.
Error rate in the sample = 192,000 / 6,400,000 = 3%
Projected misstatement = 3% × 48,000,000 = KES 1,440,000
Compare that against tolerable misstatement. If tolerable misstatement is 2,000,000, the projected figure is within it — but the margin is small, and the auditor should consider whether additional work is warranted before concluding.
Projecting is not optional. Reporting only the 192,000 actually found understates the problem, because the same rate of error is expected throughout the population.
Anomalies are the exception. Where an error is demonstrably not representative — a one-off caused by a system failure on a single day — it may be excluded from the projection, but the auditor must be satisfied it is genuinely isolated.
:::checkpoint A sample of 60 items from a population of 48 million reveals two errors, one a routine pricing mistake and the other caused by a power failure that corrupted a single day's postings. Explain how each should be treated in the projection. :::
Analytical procedures
Evaluations of financial information through analysis of plausible relationships, including investigation of variations inconsistent with other information.
They are used at three stages:
- Planning — mandatory, to identify risk areas
- Substantive testing — optional, as a substantive procedure in its own right
- Completion — mandatory, as an overall review
Techniques: comparison with prior periods, budgets and industry data; ratio analysis; trend analysis; and proof in total, where an independent expectation is built and compared with the recorded figure.
Interpreting a variance
A client reports:
| Year 1 | Year 2 | |
|---|---|---|
| Revenue | 56,000,000 | 68,000,000 |
| Gross margin | 35.0% | 30.0% |
| Receivable days | 60.0 | 82.1 |
Revenue rose 21% while the gross margin fell five percentage points — worth about KES 3,400,000 of profit at the new revenue level. Receivable days rose by twenty-two.
The pattern is coherent and worth pursuing: sales appear to have been won by discounting and by extending credit. That raises questions about cut-off (were year-end sales recorded early?), about the recoverability of the larger receivables balance, and about whether the new customers were creditworthy.
The auditor's task is not to compute the ratios but to form an expectation, notice where reality departs from it, and pursue the departure. Ratios computed and then left uninterpreted earn no marks.
Effectiveness
Analytical procedures are most reliable where relationships are stable and predictable — payroll against headcount, depreciation against asset values, sales commission against sales. They are weakest where the business has changed, where the relationship is volatile, or where management can manipulate both sides of the comparison.
:::checkpoint An auditor calculates that payroll cost per employee rose 4% while the agreed pay award was 4%. Explain what this confirms, and name a circumstance in which the same result would nonetheless conceal a material misstatement. :::