Sampling error is the inherent discrepancy between the characteristics observed in a sample and the actual characteristics of the entire population from which that sample was drawn. In an audit, this means that the conclusions an auditor draws from a sample of records (e.g., AI system logs) might differ from the conclusion they would have reached if they had examined every single record. This is a statistical risk that exists even with perfectly random sampling and is managed by using appropriate sampling methods and sizes.

A. This describes selection bias, which is a cause of non-sampling error, not the definition of sampling error itself.

C. This describes a specific biased sampling method, which can lead to error, but it is not the definition of the error.

1. ISO 19011:2018, "Guidelines for auditing management systems," Annex A, Clause A.6, "Audit sampling," implicitly addresses this by discussing the risk that "samples may not be representative of the population from which they are selected." This potential for non-representativeness is the source of sampling error.

2. Groves, R. M., et al. (2009). Survey methodology (2nd ed.). Wiley. Chapter 2, "Inference and Error in Surveys," provides a formal definition of sampling error as a component of total survey error. (This is a standard university-level text on the subject).