Audit Methodology, Sampling, and Data Analysis
What this domain tests
You must know how to design and defend an audit. Concepts: audit scope and objectives; population definition; sampling methods - random, stratified, systematic, and judgmental - and when each is used; sample size and confidence (larger samples for higher-risk populations); error rate calculation and extrapolation; inter-rater reliability and reviewer calibration; workpaper standards (what must be documented to support a finding); root-cause analysis vs. symptom counting; and the difference between a process finding and an outcome finding.
Traps that show up in questions
- Using a judgmental (targeted) sample and then extrapolating the error rate to the whole population - extrapolation requires a statistically valid random/stratified sample.
- Confusing error rate (errors ÷ claims reviewed) with dollar impact (sum of improper payments).
- Forgetting that a sample must be representative of the population it is generalized to.
- Missing that workpapers must let an independent reviewer reproduce the finding.
- Treating a single high-dollar error as evidence of a systemic problem without root-cause analysis.
Rule of thumb
Randomize before you extrapolate. If the sample was chosen because it looked suspicious, report findings as illustrative - not as a population estimate.
Worked example
You review 100 claims from a 5,000-claim population and find 8 with coding errors. Error rate = 8%. If the sample was random and representative, you may estimate the population impact (e.g., 8% × 5,000 claims × average error value). If the sample was targeted at high-risk claims, report the 8% as a finding about those claims only.