SIMULATION ANALYSIS OF THE STATISTICAL VALIDITY OF THE INTERNAL CONTROL HYPOTHESIS OF AUDITING WITH IMPLICATIONS FOR SUBSTANTIVE TESTING METHODS AND LINKAGE RULES.

摘要

This dissertation reports on a study which provides evidence on the statistical validity of the hypothesis underlying the second audit standard of fieldwork. This hypothesis is articulated as the internal control hypothesis of auditing.;By contrasting the performance of these strategies in terms of substantive test sample size and actual audit risks experienced in the accounting environments, the following major findings were obtained: (1) The statistical validity of the internal control hypothesis was supported for all the substantive testing methods and linkage rules used in the study. However, this validity is dependent on the quality of internal control information. (2) Sample size for substantive tests can be reduced as much as 75% with a significant reduction in audit risks when reliable internal control information is available. (3) The "conservative" linkage rules are more likely to preserve the validity of the internal control hypothesis as more sampling and non-sampling errors are introduced to the audit process. (4) Dollar-unit sampling greatly outperformed stratified-mean-per-unit estimation for the audit environments considered in the simulation. (5) Use of a Bayesian methodology does not necessarily guarantee an improved audit sampling strategy in terms of substantive test sample size and actual audit sampling risks.;In addition, the dissertation examines the issues involved in internal control modeling, explores the implications for using a series system assumption in modeling internal control systems, and provides evidence on the performance and audit potential of methods available for statistically evaluating internal control system reliability.;The study used the simulation methodology to analyze the validity of the hypothesis. First, audit environments were constructed based on the book values of an actual accounts receivable file of a medium-sized manufacturing firm. Second, the simulation used the concept of audit sampling strategy to simulate the audit process applied to each of these accounting environments. An audit sampling strategy is an abstraction of the usual audit process as conceptualized in SAS No. 1 Sec. 320. It parallels this process by use of the sequential stages: (1) obtaining an amount of internal control information, (2) use of a "linkage rule" to determine the extent of substantive testing as a result of obtaining the information in stage (1), and (3) the substantive testing stage. The simulated auditors decision is assumed to be the same as the statistical decision of the substantive test--the last stage of the audit process and audit strategy. By considering various combinations of alternatives at each stage of an audit strategy, a total of 16 strategies were developed and simulated. Five different types and amounts of internal control information, four different linkage rules, and three different substantive test methods were used in the study.