Which of the following is NOT typically used in risk assessment to select returns for audit?

• A. Data analytics and anomaly detection to flag potential noncompliance.
• B. Random, manual checks with no data basis.
• C. Historical patterns and taxpayer behavior.
• D. Scoring models to rank risk.

Answer: (B)

Risk assessment for selecting returns to audit relies on data-driven and systematic methods that identify which cases are most likely to involve noncompliance. Data analytics and anomaly detection scan large datasets to highlight unusual or suspicious patterns that deserve closer scrutiny. Historical patterns and taxpayer behavior help estimate the likelihood of noncompliance based on how similar filings behaved in the past, guiding where risk is higher. Scoring models convert these risk indicators into a ranked list so auditors can prioritize the most potentially risky cases.

The option describing random, manual checks with no data basis doesn’t fit this approach. Without data signals, the process becomes subjective and inefficient, risking misses on high-risk cases and wasting resources on low-risk ones. In practice, random checks can be used sparingly for validation or to test controls, but they are not the primary tool for risk-based selection of returns for audit.