Sample Error Characteristics and Projection of Error to Audit Populations.
研究审计实践中“隔离”某些样本错误而不将其推断到总体的做法，通过规范模型和实证分析发现，审计师倾向于隔离被视为“独特”的错误，但这种做法很少合理，且错误隔离倾向与事务所和审计师经验有关。
Abstract ABSTRACT: This paper examines the practice of "isolating" certain errors observed in an audit sample and subsequent failure to project the errors in estimating total error in an audit population. An abstract (normative) model of inference is developed to analyze this practice and related descriptive theories of human information processing are reviewed. Previous descriptive research suggests that auditors may be inclined to isolate errors that are considered to be "unique." However, the normative analysis suggests that isolation is seldom appropriate. The responses of auditors to a number of hypothetical cases are analyzed to provide descriptive empirical evidence on the practice of error isolation. The responses indicate a strong tendency to isolate errors whose cause is perceived as "unique." An analysis of isolation/projection decisions as a function of firm membership, experience, and training in statistics indicates that the tendency to isolate errors is, for individual cases, related to firm membership and experience. However, the observed firm membership and experience effects are not consistent across all cases. Thus, the results are not well explained by a simple model which posits that inappropriate error isolation is limited to a few firms or to inexperienced auditors. While incorrect error isolation appears to be prevalent in practice, we suggest several factors which may mitigate its adverse effects.