Factors to Consider When Determining Precision of a Generalizable Sample
When a sample is taken and analyzed, a value of interest is obtained. However, because the value is derived from a sample of a larger population, it is not guaranteed that the value obtained exactly represents the population as a whole. Rather, the value obtained is a best estimate of the population value. The precision of this estimate varies according to the sample size.
Depending on the objective pursued by auditors, the required level of precision will also vary. Some objectives require only a minimally precise sample. For example, if the objective is to demonstrate whether a moderate to high level of error exists in a process or program, and the exact level of error is immaterial, then a small sample will be enough. In contrast, if the objective is to estimate the extent of deviation from an expected result, then the exact level of error will be relevant.
Two essential measures are used to provide information on how effective a sample is likely to be as an estimator of a population. These measures are the confidence interval (CI) and the confidence level (CL). Both are required to describe the precision of any generalizable sample estimate. (These notions do not apply to purposeful samples.)
LinkedIn
Twitter