This article discusses the nature of the assessment activity for audit decision making with citing an overview of some relevant research and some supporting evidence on the topic. The purpose of this article is to give additional exposure to a method of assessing prior probability distributions that appears to be particularly congruent with the auditor's environment. The process of quantifying qualitative evidence appears in the literature under a variety of terms, including "eliciting Bayesian priors" and "probability encoding." In this article, the process is referred to as "assessing prior probabilities or subjective probability distributions (SPDs)." According to the author, a satisfactory assessment of the prior probability function would result when the auditor, after training, believes that the resulting distribution is a good summary of his or her qualitative evidence. In a recently conducted study, all of the auditors had some difficulty in deciding where the first and third quartiles of the SPD ought to be in using the direct assessment method. There is some evidence that training in assessing SPDs may be an effective means of overcoming a common problem, that is, a possible tendency to underestimate dispersion. Results of that study indicate that even limited training may have an effect and that at least some auditors are receptive to the study and use of subjective probability distributions and Bayes method.

The results of this study indicate that a significant part of the variation in changes in performance not explained by the other variables in the model is explained by changes in productivity for the three firms studied. To the extent that the particular specification of variables is successful, this relationship indicates that the expected relationship between technical change and performance can be discerned empirically. Also, it appears that on the average the expected positive effects on earnings due to management decisions to make technical changes are being realized. While this favorable result is encouraging, it does not mean that this result can be extended to other firms even if they are drawn from the same industry. Several additional favorable replications of the experiment in the electric power industry chosen to include firms that used hydro-generated power, that are from other geographical regions, and that are of different sizes than the firms included in this study would be required to make generalized statements about the industry as a whole. An extension of this study that would make its results potentially useful for evaluating technical change decisions would be to develop a means of estimating the relationship between these decisions and measured technical change. This extension may be relatively difficult because of the likelihood of considerable time lags between the decisions and their physical outcome. If successful, a manager might then be able to evaluate on average both the effectiveness with which the technical change decisions are carried-out and the effect of that outcome on reported earnings.

Auditing, Sampling, Composite accounts, Bayesian model

ABSTRACT: Several new methods of evaluatIng dollar-unit samples have been proposed recently for use in auditing. This paper presents results of a simulation study that compared four of these new methods. Additional simulations on alternative forms of two of these are also presented. Evidence on the reliability and relative size of the upper error bounds for each method is provided for 2,160 different combinations of sample sizes, confidence levels, and types of accounting populations. The results provide a comprehensive analysis of the capabilities of these methods. All four methods have strengths and weaknesses. The Multinomial Dirichlet bound is quite reliable but very conservative and (as formulated) not able to utilize understatement errors. The two versions of the Cox and Snell bound that were tested (with different priors) varied considerably in their reliability over a range of confidence levels. While not without an occasional reliability problem, test results for the Moment Method bound and one of the two Bayesian-Normal bounds were largely favorable.

[Several new methods of evaluating dollar-unit samples have been proposed recently for use in auditing. This paper presents results of a simulation study that compared four of these new methods. Additional simulations on alternative forms of two of these are also presented. Evidence on the reliability and relative size of the upper error bounds for each method is provided for 2,160 different combinations of sample sizes, confidence levels, and types of accounting populations. The results provide a comprehensive analysis of the capabilities of these methods. All four methods have strengths and weaknesses. The Multinomial Dirichlet bound is quite reliable but very conservative and (as formulated) not able to utilize understatement errors. The two versions of the Cox and Snell bound that were tested (with different priors) varied considerably in their reliability over a range of confidence levels. While not without an occasional reliability problem, test results for the Moment Method bound and one of the two Bayesian-Normal bounds were largely favorable.]

[When making audit judgments and decisions, an auditor often relies on knowledge or information about how task variables covary. Typically, an auditor generates this information without the aid of a formal covariation model. Thus, the quality of audit judgments and decisions which rely on covariation information may depend on an auditor's ability to judge covariation in a manner that is paramorphic to a formal model. This paper reports two experiments examining the rules by which auditors integrate joint frequency data when making covariation judgments and whether their judgments are affected by context, prior expectations, and amount of auditing experience. A main result was that the subjects generally used data-integration rules that were sensitive to the objective covariation level, but often overstated or understated this level. The effects of context, prior expectations, and amount of auditing experience were generally small.]

Reviews the book "Symposium on Auditing Research: 1982," edited by David Flint.

[Auditors' self-perceptions of their judgment ability may affect how well they learn from experience and how much they rely on decision aids. This study tests the hypothesis that incomplete outcome feedback causes auditors' self-perceived judgment ability to be affected by factors that are not necessarily related to actual judgment ability. The results of an experiment that used professional auditors as subjects making internal control judgments were, for the most part, consistent with this hypothesis.]

ABSTRACT: This paper reviews auditing research on the process by which auditors form their opinions on financial statements. A structure for this research is provided to organize the review and to provide insights into areas where research is needed. This structure is obtained by cross-classifying steps in theory development and steps in the opinion formulating process. While several important lines of research have begun, the primary conclusion is that a great deal of potentially important research has not been performed.

Reviews the book "Systems Based Independent Audits," 2nd ed., by Howard F. Stettler.