Audit Sampling

Sampling risks

• Controls: overreliance (effectiveness) vs underreliance (efficiency).
• Substantive: incorrect acceptance (effectiveness) vs incorrect rejection (efficiency).
• Non-sampling risk: wrong procedure or misread evidence — fix with supervision, not larger samples.

Sample size increases when

• Acceptable risk ↓, tolerable rate or misstatement ↓, expected deviation or misstatement ↑, variability ↑.
• Population size: little effect (large populations). Stratification: decreases size.

Attribute sampling

Upper deviation rate = sample rate + allowance. Upper rate > tolerable → reduce reliance, expand substantive tests.

Probability-proportional-to-size (PPS) sampling

• Interval = book value ÷ sample size; items ≥ interval always selected.
• Small item: tainting % × interval. Large item: actual misstatement.
• Good for overstatement; poor for understatement and zero balances.

Classical variables

Mean-per-unit · ratio (proportional errors) · difference (constant errors).

Exam traps

• Attribute sampling counts deviations, not dollars: the size of the transaction doesn't matter, only whether the control failed.
• Haphazard selection is non-statistical — it isn't random selection and can't be used for statistical sampling.
• If a selected item can't be tested (for example, a missing document), treat it as a deviation or misstatement — don't just pick another item.