Abstract This paper reports the results of an experiment investigating auditors' evaluation of sample results for substantive tests. The auditors 95 percent credible intervals for the amount of monetary error and for the account balance were chosen as the dependent variables because of their direct relationship with the risks of incorrect acceptance and incorrect rejection, respectively. Three sample characteristics were manipulated in the experiment: variance of audited values, mean of error amounts, and variance of error amounts. Also, relevant auditor individual differences were controlled in the data analysis -- audit experience and statistical knowledge. Overall, the results show the auditors' credible intervals were too narrow, and they were influenced by error mean and error variance.
Blocher et al. (Sun,) studied this question.