Audit judgment research

Accounting Organizations and Society, Vol. 14, Nos. 1/2, pp. 83-99, 1989. Printed in Great Britain

0361-3682/89 $3.00+.00 Pergamon Press plc

AUDIT JUDGMENT RESEARCH* PAUL E. JOHNSON, KARIM JAMAL a n d IL GLEN BERRYMAN

University o f M i n n e s o t a

Abstract

This paper identifies themes in the audit judgment literature and suggests implications of the research that supports them. Research on outcome behaviors and investigations of the process of audit decision making are stressed. Issues in both methodology and theory are considered. Abrief assessment of the current state of knowledge in the field of audit judgment research is provided, and an argument is offered for a direction of future work based on the idea of context and the need to understand the meaning that tasks have for the subjects who perform them. The paper concludes with a brief discussion of data from studies currently in progress at the University of Minnesota.

AUDIT JUDGMENT: BEHAVIOR

gression a n d analysis of v a r i a n c e to draw inferences about human judgment.

T h e generally a c c e p t e d goal of a u d i t j u d g m e n t research has b e e n to u n d e r s t a n d a n d i m p r o v e a u d i t o r d e c i s i o n m a k i n g (Libby, 1981 ). In s o m e fields, it is possible to evaluate d e c i s i o n s b e c a u s e o u t c o m e s are fairly clear. In m o s t audit situations, h o w e v e r , t h e r e is little or n o k n o w l e d g e of c r i t e r i o n variables. T h e lack of a n o r m a t i v e c r i t e r i o n for audit tasks a r o u s e d early i n t e r e s t in the use of c o n s e n s u s a m o n g a u d i t o r s as a m e a n s of e v a l u a t i n g a u d i t o r j u d g m e n t . T h e basis for u s i n g c o n s e n s u s as a n e v a l u a t i o n c r i t e r i o n s e e m s to have b e e n p r o v i d e d b y E i n h o r n ( 1 9 7 4 ) w h o suggested that c o n s e n s u s is a necessary, b u t n o t sufficient, c o m p o n e n t of e x p e r t j u d g m e n t . T h e c o n c e r n a m o n g r e s e a r c h e r s as well as audit firms r e g a r d i n g different a u d i t o r s m a k i n g w i d e l y differing d e c i s i o n s in the s a m e circ u m s t a n c e s led to early studies in w h i c h c o n s e n sus was u s e d as a m a j o r d e p e n d e n t variable. T h e s e studies generally f o l l o w e d the B r u n s w i k ( 1 9 5 2 ) lens m o d e l paradigm, w h i c h relies u p o n t h e application of linear m o d e l s s u c h as linear re-

Lens m o d e l studies A s h t o n ( 1 9 7 4 ) b e g a n the lens m o d e l w o r k in a u d i t i n g b y s t u d y i n g c u e utilization, c o n s i s t e n c y a n d c o n s e n s u s of auditors' i n t e r n a l c o n t r o l judgments. Ashton's s t u d y was c o n d u c t e d using 63 p r a c t i c i n g auditors from four p u b l i c a c c o u n t i n g firms. A m a j o r i t y of the auditors had 2 or 3 years e x p e r i e n c e , a n d w e r e asked to make p r e l i m i n a r y j u d g m e n t s w h i c h w o u l d b e s u b j e c t to r e v i e w a n d possible revision at higher levels in the organization. Each s u b j e c t was p r e s e n t e d with 32 cases a n d asked to make a rating j u d g m e n t of in° ternal c o n t r o l s t r e n g t h for each case. T h e task was p r e s e n t e d a s e c o n d time, s o m e 6--13 w e e k s later, i n o r d e r to assess the c o n s i s t e n c y of judgm e n t o v e r time. A s h t o n r e p o r t e d a high c o n s e n s u s b e t w e e n auditors ( r = 0 . 7 0 ) and c o n s i s t e n c y over the t w o a d m i n i s t r a t i o n s ( r = 0.81). Main effects acc o u n t e d for o v e r 80% of j u d g m e n t variance a n d n o n e o f the i n t e r a c t i o n s w e r e significant,

*Thework reported here was supported in part by grants from the Universityof MinnesotaMicroelectronics and Information Sciences Center and Control Data Corporation. 83

84

PAULE.JOHNSONet aL

suggesting that the auditors did not look for patterns of answers to the questions. Ashton & Kramer ( 1 9 8 0 ) replicated Ashton's study with 30 auditing students as subjects, using the same six cues. They obtained the same pattern o f responses as those in Ashton's original ( 1 9 7 4 ) study. The finding of similar results for both students and auditors may have been due to the fact that the auditors used w e r e not experts, or it may have b e e n due to the restriction of range for the dependent variable (a six point rating scale). J o y c e ( 1 9 7 6 ) followed up Ashton's work by presenting subjects with a series of combinations of strengths and weaknesses regarding three internal controls in an accounts receivable system. The subjects w e r e asked to estimate the amount of time to allocate to each of five accounts receivable audit procedures. The use of time ( n u m b e r of hours) avoided the problems of restriction of range in Ashton's rating scales. In Joyce's study subjects w e r e requried to do two things: ( 1 ) evaluate internal controls, and ( 2 ) estimate amount of time to be allocated to the procedures. Subjects could agree on the evaluation of internal control (as in Ashton's study) but disagree on h o w m u c h time to allocate to specific audit procedures. Joyce reported a relatively low consensus of 0.37, based on total time judgments. He found that as subjects' experience increased, consensus decreased. Joyce also found high stability (0.86), virtually no interaction effects and main effects that accounted for approximately 75% of the total variance. In general, lens model studies have shown that cues (main effects) account for most of the variance, that interactions are not significant, and that large differences exist across individuals. Not surprisingly, the inconsistencies across individuals has been of concern to audit firrns. A major implication of the lens model research has been that decision aids developed to assist auditors should be based upon the judgments of experts, and that these judgments should be combined by mechanical (usually mathemati-

cal) models. (See Libby & Lewis, 1977, for a comprehensive review of the literature.) T w o major unresolved issues in the lens model literature led researchers away from this approach in the late 1970s. The first was the lack of an accepted normative criterion for evaluating auditor judgments. A second, and m o r e fundamental problem, however, was that the lens model work did not seem to provide insight into the processes being used by auditors in making judgments. Thus, researchers in accounting w e r e receptive to a new paradigm developed by Tversky & Kahneman ( 1 9 7 4 ) which offered a normative criterion as well as process descriptions. Heun'stic$ and biases

In their highly influential 1974 article, Tversky & Kahneman p r o p o s e d that people rely on a limited n u m b e r of heuristics that enable them to c o p e with c o m p l e x judgment situations. Tversky & Kahneman proposed that, while heuristics may be useful in many circumstances, they can lead to serious and systematic errors because they are not influenced by several factors that should affect judgments according to the normative Bayesian model. For example, in the case of a person using the representativeness heuristic, the sole criterion used for making judgments is perceived similarity. Normatively important considerations, such as sample size, base rates, data reliability and diagnosticity, are ignored because they do not affect perceived similarity. Early research on neglect of base rates was conducted by psychologists using non-expert subjects in rather abstract tasks. This w o r k was extended by researchers w h o were interested in determining whether auditors exhibit the same biases in audit settings. Results of studies such as that performed b y J o y c e & Biddle (1981), have indicated that, while auditors exhibit the same overall information processing behavior as nonauditors, they seem to be less affected by the heuristics and biases identified in the earlier work. See Libby & Lewis ( 1 9 8 2 ) for a review of these and other studies. In an important article, Einhorn & Hogarth

AUDIT JUDGMENT RESEARCH

( 1 9 8 1 ) defined optimal judgments as ones that maximize or minimize s o m e explicit and measurable criterion such as profits, errors or time, conditional on certain environmental assumptions and a specific time horizon. This definition stressed the conditional nature of optimality and suggested that the normative rules studied in the heuristics and biases w o r k m a y apply only in simple settings. Einhorn & Hogarth emphasized the importance of understanding the role of attention, m e m o r y and cognitive representation in making judgments. They recognized that the statistical models used in previous w o r k were, in fact, not optimal models in real settings and urged researchers to study behavior in c o m p l e x task environments w h e r e subjects could play an active role in the process of judgment and choice.

AUDIT JUDGMENT: PROCESS Felix & Kinney ( 1 9 8 2 ) reviewed the auditing literature and found that very little was known about h o w auditors process and combine information from compliance and substantive tests. They suggested that only limited progress has been made in the field of audit research. No strong model or theory of auditing has b e e n developed, and there have b e e n no tests of auditor behavior in realistic task settings. In order to learn m o r e about current audit practice they suggested that researchers could mail surveys, conduct field studies (interviews) and examine audit working papers. The emphasis in the Einhorn & Hogarth ( 1 9 8 1 ) and Felix & Kinney ( 1 9 8 2 ) review articles on the importance of understanding the psychological processes underlying judgment and choice, has led to an interest in studying decision processes.

Decision processes An early study of decision processes in auditing was conducted by Biggs & Mock ( 1 9 8 5 ) w h o collected verbal protocols from four senior auditors e m p l o y e d by an international accounting firm, using a case previously developed by Mock

85

& Turner (1981). Results focused on completeness of search and type of activities subjects engaged in. Biggs & Mock found that most of subject's reasoning activity took place in what they called the information search and information evaluation stages of the task. They also found two patterns of task behaviour. T w o of the subjects, A and C, performed the task using a systematic strategy, consisting of a thorough and sequential search of available information prior to making any decisions. A third subject (subject D) used a directed strategy which consisted of selecting a particular audit step and then searching for information relevant to that step. Once the decision was made for the first step, the process was repeated until all steps w e r e completed. Subject B followed a mixed strategy consisting of a systematic strategy part of the time, and a directed strategy part of the time. A comparison of subject D's behavior (directed strategy) with that of subject A and C (systematic strategy) showed that subject D took less time to do the task, made fewer changes to the planned sample sizes and obtained less information than subjects A and C. Although there was no normative guideline to conclude which one of the two strategies was m o r e effective, the auditors with m o r e experience used the systematic strategy. A review of the sample size decisions in the Biggs & Mock ( 1 9 8 3 ) study revealed an apparent lack of consensus. However, no explanation was provided for a subject's use of a systematic or directed strategy. It is not known, for example, what a good decision was for the task, what the critical cues w e r e that influenced subjects' judgments, nor h o w these cues w e r e interpreted. Despite such shortcomings, the study by Biggs & Mock was an important response to Einhorn & Hogarth's ( 1 9 8 1 ) critique of the heuristics and biases literature. The study focused upon building a descriptive model of h o w auditors make decisions and e m p l o y e d a task based upon an actual c o m p a n y that was rich in complexity. Rather than inspiring research directed towards the d e v e l o p m e n t of audit theory, however, the pioneering w o r k by Biggs & Mock

86

PAULE.JOHNSONet aL

( 1 9 8 3 ) has led to research directed towards improving decision making using c o m p u t e r systems as devices for creating auditor decision aids. Expert systems Recent studies conducted by researchers interested in developing computer-based decision aids (usually called expert systems) have adopted the Biggs & Mock approach to investigate decision processes, which is based upon the use of verbal protocol data. In a study by Hansen & Messier ( 1 9 8 6 ) the authors describe the results of a preliminary investigation of EDP-XPERT, an expert system which is intended to assist c o m p u t e r audit specialists (CASs) in making judgments as to the reliability of controls in advanced c o m p u t e r environments. These investigators conducted a protocol study w h e r e the objective was to identify IF-THEN rules that might be appropriate for the system knowledge base. The protocol study is discussed in detail in Biggs etaL (1986a), and the focus here is on what was done, what was learned, and the implications for conducting process studies to understand auditor expertise. Three managers fi'om one Big Eight accounting firm served as subjects in the protocol part of the study. Each of the managers was a c o m p u t e r audit specialist with 2--4 years of experience, including the auditing of c o m p l e x EDP systems. The case used in the research was adapted from client working papers, contained o v e r 40 pages of information and was used as part of the firm's training course materials. Each subject was told that he/she was replacing the previous CAS w h o was reassigned because of other client commitments. Subjects w e r e asked to review the client's EDP system, evaluate internal controls over the sales/receivables cycle and prepare an audit program. The public accounting firm that provided the case also had a suggested solution which served as a criterion for analyzing subject's responses to the case. The data w e r e analyzed as suggested by Biggs & Mock (1983). A microlevel analysis was done in which protocols w e r e analyzed for knowledge states and operators ( N e w e r & Simon,

1972). In addition, a macrolevel analysis was done consisting of two parts: ( 1 ) decision flowcharts and ( 2 ) episode abstracts. The flowcharts w e r e a graphic representation of the overall decision process of each subject. The episode abstract was a description of the sequence of goals set by the subject and the major information processing activities related to these goals. The results of the macro analysis w e r e not presented in the paper; the authors argued that they provided little additional information concerning the CAS's decision processes. The results reported by Biggs et aL (1986a) are the n u m b e r of lines of protocol generated and a classification of operators into three categories: ( 1 ) information acquisition, ( 2 ) evaluation, and ( 3 ) audit decision. All three protocol subjects identified a majority of the eight controls contained in the solution. The types of operators found w e r e similar to those found by Biggs & Mock (1983). No model was presented indicating h o w internal control judgments are made in advanced c o m p u t e r environments, and the study did not lead to finding many decision rules that could be used to construct an expert system. The w o r k combining a process approach to understanding decision making with the goal of building an expert system has led to the developm e n t of a n u m b e r of working systems. However, these systems have not been developed with a good understanding of what processing the auditor, whose expertise is at the heart of the system, is doing. There has been virtually no development of audit judgment theory or hypotheses to guide future work, especially in realistic task situations. (See additional studies by Biggs et aL, 1986b; Dungan & Chandler, 1985; Meservy et aL, 1986; Shpilberg & Graham, 1986.) AUDIT JUDGMENT RESEARCH: AN ASSESSMENT In this section the research described above is reviewed with the objective of identifying reasons w h y there has been only limited progress to date. Sources ofdi/~culty in past audit judgment

AUDITJUDGMENTRESEARCH w o r k are identified and an alternative approach is illustrated below with an example from w o r k in progress at the University of Minnesota. Methodology The judgment tasks used in lens model research as well as in studies of heuristics and biases w e r e highly structured, and in s o m e instances w e r e unfamiliar to subjects. In such tasks subjects play a relatively passive role, processing only the information given to them. However, as Einhorn ( 1 9 7 6 ) has suggested, real decision tasks are poorly structured. In reality, information must be searched for, data are not perfectly reliable and hypothesis formation and confirmation/disconfirmation occurs within a broad range of possibilities. In most situations, the auditor must take an active role in searching for as well as evaluating information. The processes of reasoning that w o r k on the well-structured tasks of the psychological laboratory often do not seem to generalize to settings of practice (Ebbeson & Konecni, 1980). Biggs & Mock ( 1 9 8 3 ) employed m o r e realistic tasks and focused on decision processes, using the methodology of protocol analysis. However, the research in auditing that has adopted this approach has been conducted in such a way that little was learned about h o w subjects did the task. Measures employed such as time taken to c o m p l e t e the task, percentage of information searched for, and type of activities engaged in, are all descriptions of behavior rather than statements of the reasoning process underlying this behavior. It is not known, for example, what the critical cues are, h o w subjects interpret them, h o w the interpretation of specific cues affects information search, nor h o w various cues are c o m b i n e d to make sample size decisions. Although investigations of the audit judgment process have tended to employ m o r e realistic tasks than ones used in the study of heuristics and biases, the use of only one case also makes it difficult to separate those aspects of behavior which are unique to the case, and those that are c o m m o n across cases. Despite an effort to create experimental tasks

87

based on real audit cases, the tasks given to subjects have not always elicited the expertise that investigators w e r e hoping to understand. For example, Biggs & Mock ( 1 9 8 3 ) and Biggs et aL ( 1 9 8 6 a ) constructed audit cases from the working p a p e r files and training materials used by audit firms. While these w e r e c o m p l e x cases, they appeared to be normal audit tasks in which subjects could use relatively automated reasoning. The result of such research has b e e n that investigators discover (not surprisingly) that experts do better than novices. However, no insight is provided into why experts are better since the underlying expertise is not made explicit in subjects' behavior. An additional dil~culty has b e e n the lack of use of recognized experts as subjects. Although most "process studies" in auditing have used practicing auditors there has b e e n no assurance that subjects d e e m e d to be experts have a high level of expertise in the task being studied. An additional difficulty, inherent in this type of research, is that only a small n u m b e r of subjects have b e e n investigated in a given study. Studies conducted by Biggs & Mock ( 1 9 8 3 ) and Biggs et aL (1986a) collected protocols from four subjects or less. Audit research has drawn heavily upon findings and m e t h o d o l o g y from the field of experimental psychology. Investigations such as those reviewed here, have often b e e n designed to demonstrate the conditions u n d e r which findings discovered in the psychological laboratory hold in the context of audit judgment. Such knowledge is important because it advances an understanding of the human mind, and also because it tests the generalization of findings from the laboratory to the world of everyday experience. Unfortunately, as others within the field have noted (e.g. Felix & Kinney, 1982), such w o r k has not advanced an understanding of the nature of the behavior of interest. After several years of research, there is no theory of audit judgment, and very little understanding of what auditors actu. ally do. Context Part of the reason w h y audit research has

88

PAULE.JOHNSONet aL

taken its current direction may be due to its preoccupation with providing decision aids for the practicing, auditor. Even w h e n studies m o v e d toward the investigation of process, researchers have concentrated on extracting "rules" from data to be used in the implementation of a c o m p u t e r program, rather than building a theory from which such rules might be derived. However, the rules that comprise an expert system knowledge base are seldom present explicitly in an expert's behavior. ( T h e y are, in fact, representations of the underlying knowledge.) Such rules must be created from an understanding of h o w the task is done. Research on expertise is intended to lead to the discovery of the nature of the underlying knowledge and reasoning, not as is sometimes supposed, to find "rules" in the data obtained during the performance of a task. Audit research is concerned not just with observed behavior, but also with the thoughts and knowledge of highly trained professionals w h o are carrying out c o m p l e x tasks in a tightly constrained problem-solving context. Unlike the experimental psychologist w h o is interested in describing fundamental aspects of human cognition, the audit researcher is c o n c e r n e d with discovering the knowledge and reasoning process of a skilled p r o b l e m solver. The goals of the two efforts are related, but also fundamentally different. As an example of what happens w h e n the methods of One field are adopted in another, consider again the w o r k on decision processes. The w o r k by Biggs and others (e.g. Biggs & Mock, 1983) applied methods of analysis to problem-solving data that did not describe most of the contextual knowledge developed by subjects in the course of learning to carry out the kinds of tasks investigated in the research. The k n o w l e d g e states and operators adopted in this w o r k for describing reasoning processes w e r e developed by Newell & Simon ( 1 9 7 2 ) to describe the process of solving well-structured problems. For such problems the process of problem-solving can be viewed as a state-space search (Nilsson, 1980). Despite Simon's ( 1 9 7 3 ) suggestion that ill-structured problems might be

solved by partitioning them into previously solved well-structured problems, it is unlikely that the expertise of the professional problem solver will be captured solely in these terms. The result of such work is that the process of audit judgment is described in terms of discrete units of behavior rather than concepts from a ,theory or model of the process of audit judgm e n t itself. An analogous p r o b l e m in physics would be for the motion of bodies to be described solely in terms of "pointer readings" on spring balances and m e t e r sticks instead of concepts such as force, mass and acceleration. Part of the difficulty faced in attempting to understand the world of everyday experience is that the paradigm that guides thinking and research has been behaviorism, rather than contextualism (Pepper, 1972). According to the contextual metaphor, it isn't only the behaviors that can be inferred from the data on a task, but also the meaning these behaviors have for the subjects w h o express them. The issue of "meaning" is one of the fundamental problems that has occupied psychology since the so-called psycho-linguistic revolution of the 1960s (Chomsky, 1959; Jenkins, 1968). Although a description of this issue in the field of psychology is beyond the scope of this paper, it is worth noting that meaning, and its companion p r o b l e m of knowledge, have been responsible, in part, for the development of the relatively n e w field of cognitive science. The remaining portion of this paper briefly sketches one alternative approach to conducting research on audit judgment which is based on the attempt to deal with these problems. At the outset it is possible to distinguish three types of theory: a theory of the auditor, a theory of auditing and a theory of h o w to audit. The first is a theory of persons or individuals w h o carry out a specific kind of taslc The second is a theory of that task, and the third is a theory of a specific kind of activity. The first two theories are descriptive, the last is prescriptive. The first two are developed by investigators of the p h e n o m e n a of auditing. The last can be developed by audit practitioners. The research reviewed here has been largely

AUDITJUIX~MENTRESEARCH c o n c e r n e d with auditor behavior, and though there has b e e n relatively little theory developed, what exists falls under a theory of the auditor. Books on auditing written by practitioners and scholars in the field comprise theory of the third kind. These are useful as a source of insight into what practitioners k n o w and do, but they do not, by themselves, provide theory to guide research. The second kind of theory has b e e n most neglected, and it is the second kind of theory that provides a basis for understanding what subjects k n o w that enables t h e m to perform a task. In information processing terms, a theory of auditing is a theory of what must be c o m p u t e d by any processor that attempts to perform an audit task. Different types of processors have different processing constraints and will implem e n t a given computation in accordance with these constraints (e.g. a major processing constraint for the h u m a n processor is short-term memory). In this sense a digital c o m p u t e r performs the task of auditing differently than a h u m a n being. The important point, however, is that for a computation to b e performed by either type of processor, it is necessary to have a statem e n t of the requirements for the task of auditing, a statement of what must be done as opposed to h o w this is done (Johnson et aL, 1987). The distinction b e t w e e n w h a t and h o w with regard to the p e r f o r m a n c e of a task is similar to Newell's ( 1981 ) distinction b e t w e e n the knowledge level and the symbol level in any information processing system. From the perspective of the argument presented here, it is important to k n o w first about the nature of the task before describing b o w it is performed. A theory of audit. ing is a theory of the knowledge that is required for doing auditing. At the level of an expert auditor, this operative knowledge is the expertise required to perform the audit task. If expertise is the knowledge that is required to perform a given task, then it is also expertise that enables subjects to understand the meaning of characteristics of that task. Interestingly, it is also expertise that lies at the heart of the audit researchers attempt to build decision aids for the practicing auditor. The difficulty with the approach taken by audit researchers thus far is that

89

the formalism of a representation (e.& rules and has been confused with the content of the representation. Rather than look for "rules" in the problem-solving data of a task, analysis must identify evidence of the expertise that guides what the task p e r f o r m e r does.

frames)

AUDIT RESEARCH: AN EXAMPLE The following example is based on an attempt to c o m e to grips with some of the issues raised in this paper. In one sense the w o r k fails victim to criticisms made of other w o r k in auditing (e.g. a small n u m b e r of subjects and only a few tasks). The research does, however, have three important characteristics for purposes of the present discussion: ( 1 ) the tasks developed for presentation to subjects are based upon a genuine audit issue - - namely, fraud detection, ( 2 ) the analysis of data focuses upon the knowledge that is necessary to carry out an audit task, and ( 3 ) the tasks are challenging and provide an opportunity for expert auditors to make errors. The attempt here is to develop a theory that describes the knowledge required to fred irregularities and unintentional errors in financial statements. This kind of theory, which is at the "knowledge level", is termed a theory of expertise; it describes the necessary conditions for successful task performance (Mohr, 1982). One means of generating data relevant to understanding what subjects k n o w is through the use of tasks that are challenging enough to elicit errors, even from proficient and highly expert individuals. Such tasks are useful because they provide insight into reasoning processes, enable the identification of critical cues and aid in determining h o w these cues are interpreted by subjects in the performance of a task. Constructing tasks in which subjects make errors also reveals the limits of a subject's adaptation, and permits inferences about the process that underlies task performance (Simon, 1983). The use of these kinds of tasks permits an examination of behavior uncontaminated by the relatively automated responses elicited by m o r e familiar tasks. By investigating the limits of an individual's cog-

90

PAULE.JOHNSONet aL

nitive ability in realistic tasks, it is also possible to distinguish expert and novice auditors, not only on the basis of o u t c o m e measures but in terms of the processes of problem solving as well. In the work presented here the focus is on the task of fraud detection. In such a task the detection of irregularities requires a high level of expertise. Irregularities are defined as intentional distortions of financial statements such as frauds perpetrated by managers w h o are high enough in an organization to have the p o w e r to override the accounting controls, or w h o are not bound by such controls. When management fraud exists, compliance and substantive tests are not necessarily effective. A key deterrent to the issuance of an unqualified audit report when misleading financial statements are prepared is the expertise of the auditors w h o review the financial statements before the firm issues an audit opinion. The task of concurring partner review was selected for the investigation of expertise in this context since such reviews are c o m m o n l y used by large audit firms to provide an independent check on the fairness of financial statements for publicly held companies and for sensitive audit clients before issuance of the audit report. Frauds can be considered to be deliberately constructed garden paths (Johnson et aL, in press). According to this argument, perpetrators of fxauds design them so that their auditors will be misled as to the financial position and/or operating results, and an unqualified report will be issued. Subjects The research described here is part of a longterm research project. The results presented were obtained from two subjects in a pilot study for a larger investigation currently in progress. Subject S1 is a partner in a Big Eight audit firm and has 40 years of audit experience. He is designated as an SEC Partner and has conducted numerous concurring partner reviews over an extended period o f time. He is classified as an expert for purposes of this study. Subject $2 has been a partner for about two years in another Big

Eight audit firm and has 11 years of audit experience. He is classified as a novice. Tasks The subjects were given three audit cases and asked to review them with the objective of determining whether an unqualified opinion as r e c o m m e n d e d by the engagement partner is appropriate for issuance. The cases were constructed using the annual reports and l OK reports of actual companies. In this paper, results are presented for only one case, a medical products firm, where a fraud was perpetrated by senior management. The auditors of the company, a Big Eight audit firm, issued an unqualified audit opinion. The other two cases, consisting of a training case and a case containing an unintentional error, will not be discussed further. In the medical products case, senior management of a medical products firm overstated net income by approximately 90%. The case is complex (the summary is 20 pages in length), involving a large number of issues. Subjects were provided with a narrative description of the business and a set of financial statements and notes thereto in standard financial statement format. No audit working papers were provided. Subjects were asked to review the case and indicate whether they would be willing to sign an unqualified audit opinion, and what questions, if any, they would investigate further. A garden path (Johnson et aL, in press) was present in the case due to its description as a high tech, high growth firm in the medical products industry. This description can lead the unwary auditor to expect large increases in sales and net income, which is exactly what the perpetrator of the fraud has done ~ inflated sales and net income. There are four irregularities involved in the case, all of which are income enhancing: ( 1 ) Unordered goods were shipped to cooperating distributors and recorded as sales. The goods were returned after the end of the period. ( 2 ) Cost/unit of items in Finished Goods Inventory was arbitrarily increased. ( 3 ) S o m e expenses (e.g. research and de-

AUDITJUDGMENT RESEARCH

v e l o p m e n t ) w e r e improperly capitalized during the year. ( 4 ) An accounting policy change was made to capitalize tooling costs. The change is questionable. A n u m b e r of cues in the case are consistent with what one might expect from a high growth, high tech company: ( 1 ) the c o m p a n y has a research and d e v e l o p m e n t division which is introducing new products; ( 2 ) the c o m p a n y does not pay dividends; ( 3 ) there is a large increase in fixed assets, and ( 4 ) large increases in sales and net income have b e e n reported. A n u m b e r of cues are also present which divert the subject's attention away from the actual problems of the firm: ( 1 ) the existence of foreign subsidiaries with results needing analysis, ( 2 ) tax payable shown as zero on the balance sheet; ( 3 ) tax expense declining substantially as a percentage of net income; and ( 4 ) gross profit as a percentage of sales declining substantially. These "blind al. leys" lead the thinking of the auditor away from the possibility that the c o m p a n y is trying to man. ipulate income.

Data analysis The protocol data from the two subjects w e r e analyzed to identify what cues w e r e interpreted, what evaluation of alternative solutions was undertaken and what final audit opinion was reached. This analysis required a series of transformations of the raw data of the problem-solving record into a m o r e abstract representation. The analysis employed here consisted of three levels of protocol translation. The goal of level 1 was to identify the p r o b l e m representation generated by each subject. The goal of level 2 was to determine a c o m p e t i n g set of possible solutions. The goal of level 3 was to identify the fine(s) of reasoning used by subjects in deciding upon the p r o p e r audit opinion. In level 1 analysis, each protocol was c o d e d in terms of script-ep/$ode pairs as responses to a cue or set of cues. The protocol was coded in terms of scripts to permit the identification of the information the subject processed. This is analogous to "knowledge states" analysis proposed by N e w e r & Simon (1972), though at a

91

higher level of abstraction which allows for the inclusion of domain knowledge. Each protocol was partitioned into a series of scripts separated by natural boundaries of the task such as planning documents, balance sheet, income statements, etc. Each individual financial statement was considered to be a single script. Episodes represent cognitive actions taken by the subject and are analogous to "operators", also at a level of abstraction which involves concepts from the domain of auditing. Episodes represent segments of behavior taken to achieve a goal. The episodes used by the two subjects in responding to the fraud case w e r e categorized to identify underlying audit concepts that could serve as abstractions about the data of the case. This process led to the identification of liquidity, litigation, growth, disclosure, tax and income manipulation as possible categories for evaluating cues. These categories comprised the problem representation generated by the subjects. In the level 2 analysis, the categories identiffed in level 1 w e r e again categorized. The objective here was to understand relationships a m o n g the possible categories. It was important to determine (for e x a m p l e ) which categories w e n t together, which clusters of categories c o m p e t e d with each other as possible formulations of the problem, and within each cluster, the specific set of categories from which the auditor had to choose a solution. This analysis resulted in the identification of a set of competing solutions ( c o m p e t i t o r set). This set of alternatives consisted of specific types of audit opinions which the auditor could issue, such as an unqualified opinion, various kinds of qualified opinions and an adverse opinion. In level 3 analysis, relationships among cues w e r e identified. The objective of this analysis was to identify cues that w e r e c o m b i n e d configurally by subjects as a means of detecting patterns in the data. After a p r o b l e m representation, the c o m p e t i t o r set ofpossib!e solutions and relationships a m o n g cues w e r e identified, a qualitative model of the problem-solving process used by each subject was constructed. In this model, cues w e r e assigned to solutions, and each subject's protocol was analyzed to determine h o w

92

PAULILJOHNSON et aL

each c u e was i n t e r p r e t e d . Finally, cues w e r e c o d e d ( + / - ) as b e i n g c o n s i s t e n t / i n c o n s i s t e n t w i t h the s o l u t i o n b e i n g evaluated. O n c e a qualitative m o d e l was c o n s t r u c t e d for each subject, the lines o f r e a s o n i n g used to interp r e t c u e s w e r e identified. It was a s s u m e d that s u b j e c t s might e x p e r i e n c e difficulty in solving the p r o b l e m , e i t h e r b e c a u s e they did n o t a t t e n d to critical cues o r b e c a u s e they a t t e n d e d to critical c u e s b u t failed to i n t e r p r e t t h e m correctly. K n o w l e d g e o f a u d i t i n g e n t e r e d into d e c i s i o n s m a d e at each level o f data analysis. T h e goal o f the analysis was to o b t a i n a level of a b s t r a c t i o n s u c h that the d e s c r i p t i o n o f w h a t the s u b j e c t s were doing (qualitative model) would incorporate the k n o w l e d g e of the p r a c t i c i n g auditors a n d r e p r e s e n t the s o u r c e s o f m e a n i n g used b y subjects in i n t e r p r e t i n g the data o f the case. Results As a starting p o i n t for the data analysis, a des c r i p t i o n o f the r e q u i r e m e n t s for p e r f o r m i n g the task o f c o n c u r r i n g p a r t n e r r e v i e w was constructed. 2 A c c o r d i n g to this d e s c r i p t i o n ( s h o w n in Fig. 1 ), the p r o b l e m solver first builds a pict u r e o f the firm. He/she d e v e l o p s an u n d e r s t a n d . ing of the n a t u r e of the c l i e n t ' s business, specifically, the types of p r o d u c t s and markets the comp a n y is e n g a g e d in, the financial stability of the c o m p a n y a n d any special r e p o r t i n g requirements. This r e p r e s e n t a t i o n gives the a u d i t o r a set of e x p e c t a t i o n s for use in analyzing t h e data o f the case. Overall risk analysis is d e t e r m i n e d using general business knowledge, industry k n o w l e d g e a n d k n o w l e d g e of the specific business b e i n g audited. Materiality j u d g m e n t s are

m a d e u s i n g these types of k n o w l e d g e to n a r r o w d o w n the n u m b e r of items o n w h i c h to focus att e n t i o n . In the overall risk analysis, a c c o u n t s with large balances a n d high risk a m o u n t s (large a n d u n u s u a l a m o u n t s ) are identified using k n o w l e d g e of the business. O n c e the key a c c o u n t s have b e e n identified, the data are rev i e w e d for c o n s i s t e n t a n d i n c o n s i s t e n t information. T h e data are c o m p a r e d with the representat i o n a n d c h a n g e s in o n e a c c o u n t are cross refere n c e d w i t h o t h e r related accounts. Notes to the financial s t a t e m e n t s are reviewed, and cross ref e r e n c e d w i t h the r e p r e s e n t a t i o n of the comp a n y a n d related a c c o u n t balances. T h e set of possible solutions that c a n b e realized b y p e r f o r m i n g the task in the p r e s e n t case was: ( 1 ) unqualified o p i n i o n ( t h e g a r d e n path response) ( 2 ) qualified o p i n i o n related to o n e or m o r e of the following: litigation, liquidity, disclosure a n d c o n s i s t e n c y ( f o u r blind-alleys) ( 3 ) adverse o p i n i o n ( c o r r e c t answer). T h e d e s c r i p t i o n s h o w n in Fig. 1 was used to s e g m e n t the p r o t o c o l data from each of the two subjects. Based u p o n this initial s e g m e n t a t i o n a n d a s u b s e q u e n t analysis as d e s c r i b e d above, an i n t e r p r e t a t i o n ( q u a l i t a t i v e m o d e l ) of s u b j e c t behavior was developed. This m o d e l is i n two parts: a r e p r e s e n t a t i o n used by the s u b j e c t to frame the p r o b l e m ; and a line of r e a s o n i n g used to evaluate c u e s in the case, based o n this representation. A c c o r d i n g to the analysis, b o t h subjects first g e n e r a t e d a c o n t e x t for the case in w h i c h the c o m p a n y was i n t e r p r e t e d as a high tech, high g r o w t h firm in the m e d i c a l p r o d u c t s

tit is important to establish the reliability of the data on which inferences are based in any investigationof psychologicalprocesses. It is especially important in the case of protocol data to determine that any proposed representation of a subject's problem-solving behavior can be agreed upon by multiple scorers. A second coder was employed to establish the reliability of scoring for both the identification of cues and the assignment of cues to lines of reasoning in the present study, The proportion of agreement between the two coders was 0.90 for the identification of cues and 0.82 for the assignment of cues to lines of reasoning. Both values are well within the range found in other, similar investigations (e.g. Johnson et al., 1982). A coefficient of agreement generally referred to as Cohen's K (see Cohen, 1960) of 0.72 was also computed for the assignment of cues to lines of reasoning. Cohen's K Is used to measure the proportion of agreement between two scorers in placing items into a set of K unordered categories. K is directly interpretable as the proportion of judgments in which there is agreement, after chance agreement is excluded. Although no sampling distribution exists for the K statistic, Cohen has argued that K values can be transformed into Z scores. When this is done the K value of 0.72 is significantatp < 0.001. zWe take the requirements for a task to be a description of what must be done in order to perform in rather than how these activities are carried out for a particular instance of the task at a given moment in time (see Johnson et ai., 1987,for a further discussion of this issue).

AUDITJUDGMENTRESEARCH

MOOELOF AN *~':l I

I "

IUIINE$$

I ~w~toG:

/ I-~1

r - ~ ~,T,~^s I

1

93

evaluated separately, they enable the auditor to detect the fraud. The cues which suggest that the auditor's response should be one of several qualified audit reports are also shown in Fig. 2. The cues relating to changes in accounting methods can be interpreted in one of three ways. If interpreted in terms of profitability, they suggest an income manipulation (fraud); if interpreted in terms of a qualification, they suggest a consistency qualification; and if interpreted in terms of disclosure, they suggest an unqualified (clean) opinion. The litigation and going concern cues are evaluated with respect to issuance of a qualified opinion. o

Fig. 1. R e q u i r e m e n t s for t h e task o f c o n c u r r i n g p a r t n e r review.

industry. This context led to expectations of a large investment in research and development, the possibility of inventory obsolescence, product liability litigation, and increases in sales and net income. The high tech, high growth context provided a "story" which the subjects used to interpret cues, and provided a plausible explanation for the increase in sales, sales returns, inventory, accounts receivable and fixed assets. It also made it difficult to identify the frauds perpetrated by the management of the company. Based u p o n their initial context for the case, each subject then constructed a p r o b l e m representation in which potential solutions and cues w e r e related to one another. The general form of this representation for both subjects is shown in Fig. 2. According to Fig. 2, income taxes in the case w e r e neutral ( 0 ) with respect to the firm because there is a good explanation for the change noted. The cues in the task are part of the ~growth" subcategory, provide strong evidence in support of the interpretation of the firm [confirming evidence ( + ) ] . They also show the sales returns and gross margin cues which are inconsistent with a growth company. If these cues are c o m b i n e d with the "growth" cues, they b e c o m e diluted and do not affect the outcome. If they are

GROWTH

~ I PR°~'A"'LITv ~)l

I

INCOME MANIPULATION .

/ (~ONSISTENCY

2/

I DISCLOSURE ]

IOUAUF~ATION (o) I

r

"

I UTlamK)NI

\-

Fig. 2. P r o b l e m r e p r e s e n t a t i o n for t h e task o f c o n c u r r i n g p a r t n e r review. + = consistent, - = i n c o n s i s t e n t and 0 = n e u t r a l w i t h r e s p e c t to t h e LOR b e i n g evaluated.

The expert subject (SI) noticed a decline in gross profit, which is inconsistent with a growth c o m p a n y and a large increase in sales returns. Based on his experience, he knew that companies in this industry tend to manipulate inc o m e by shipping goods to "co-operative" distributors and recording sales revenue in the current period, and then accepting the goods back and recording sales returns in the subsequent period. Based on his specific industry knowledge, the increase in sales returns was a strong indication as to the existence of the fraud. This is

94

PAULE. JOHNSON et aL

illustrated b y t h e f o l l o w i n g e x c e r p t taken f r o m his p r o t o c o l : Return policy focuses primarily on the key question of the real risk that receivables from distributors, with the high level of return, may be cycled throughco-operative distributors on the basis that after the end of the year the merchandise could be returned, and the result of that would be to inflate both the company's sales and its profits, since merchandise which is on consignment cannot be recorded as sales and certainly profit can't be recorded. However, there has been history in this industry of that kind of thing occuring on the basis of either written or unwritten agreements to take back merchandise from distributors. The high tech industry has been plagued with that! The novice subject went down the garden p a t h a n d was w i l l i n g to issue an unqualified ( c l e a n ) o p i n i o n . H e explained a w a y t h e early d i s c r e p a n t c u e s w i t h r e f e r e n c e to t h e g r o w t h story. For e x a m p l e , t h e i n c r e a s e in sales r e t u r n s was e x p l a i n e d b y a s s u m i n g that t h e c o m p a n y m u s t h a v e i m p r o v e d its w a r r a n t y policy. A dramatic increase in returns and allowances relative to the not tlmt great increase in sales. I guess the warranty quality issue remains a concern as I go through these statements. In a similar w a y t h e n o v i c e e x p l a i n e d a w a y t h e i n c r e a s e in i n v e n t o r y b y a s s u m i n g that t h e c o m pany was stockpiling inventory because they m u s t b e e x p e c t i n g a m a j o r sale. i would pose the question of whether or not we're planning for some markedly increased deliveries and if we are, do we have commitments for these deliveries,do the orders exist to absorb all of this inventory. Since h e w a s u n a b l e to find a n y p r o b l e m s in t h e financial s t a t e m e n t s , t h e n o v i c e w a s willing to issue an unqualified o p i n i o n . Domain-specific p r o b l e m - s o l v i n g strategies, t e r m e d lines o f r e a s o n i n g (LOR), w e r e n e x t i d e n t i f i e d ( J o h n s o n et aL, 1982). F o r t h e m e d i cal p r o d u c t s case, a LOR r e p r e s e n t s r e l a t e d pathw a y s o f t h o u g h t i n c o r p o r a t i n g g e n e r a l l y acc e p t e d a u d i t i n g standards, a u d i t i n g c o n c e p t s , acc o u n t i n g p r a c t i c e s , a n d i n d u s t r y k n o w l e d g e , org a n i z e d t h r o u g h a n e t w o r k o f r e l a t i o n s h i p s that specify p o t e n t i a l s o l u t i o n s to t h e task. It is u s e d

to i n t e r p r e t ( g i v e m e a n i n g t o ) c u e s in t h e task. In t h e m e d i c a l p r o d u c t s case s u b j e c t s e m p l o y e d t h r e e s e p a r a t e lines o f reasoning:

Profitability LOR. H e r e t h e c o m p a n y is v i e w e d as a firm in w h i c h a g r o w t h in n e t i n c o m e is c a u s e d b y an i n c r e a s e in sales. T h e a u d i t risk r e l a t e s to an o v e r s t a t e m e n t o f n e t i n c o m e . T h e r e is t h e p o s s i b i l i t y o f d e t e c t i n g a m a n i p u l a t i o n o f i n c o m e , in w h i c h c a s e t h e a u d i t o r will issue an a d v e r s e o p i n i o n , u n l e s s t h e s t a t e m e n t s a r e revised a p p r o p r i a t e l y . This line o f r e a s o n i n g contains an a s s e r t i o n that c a n b e tested, n a m e l y that the c o m p a n y is g r o w i n g a n d that this g r o w t h is r e s p o n s i b l e for t h e i n c o m e b e i n g r e p o r t e d . It also g e n e r a t e s e x p e c t a t i o n s r e g a r d i n g e x p e n d i ture on research and development, investment in p l a n t a n d e q u i p m e n t and i n c r e a s e s in sales, net income, receivables, and inventory. Qualification LOR. In this line o f r e a s o n i n g t h e a u d i t o r c o n s i d e r s i t e m s s u c h as litigation, l i q u i d i t y and c h a n g e s in a c c o u n t i n g p o l i c i e s as l e a d i n g to specific t y p e s o f qualifications in t h e s t a n d a r d audit r e p o r t . W h e n t h e e x p e r t a u d i t o r uses t h e qualification LOR for a h i g h - t e c h c o m pany, h e is able to a n t i c i p a t e i t e m s s u c h as litigation, w a r r a n t i e s a n d i n v e n t o r y o b s o l e s c e n c e and d e v e l o p s e x p e c t a t i o n s for t h e a u d i t risk ass o c i a t e d w i t h t h e s e items. DisclosureLOR. This line of reasoning is b a s e d on the auditor's assessment of the adequacy of d i s c l o s u r e o f t h e v a r i o u s a s p e c t s o f t h e organization. F o r a h i g h - t e c h c o m p a n y this leads t o exp e c t a t i o n s r e g a r d i n g d i s c l o s u r e o f items s u c h as o u t s t a n d i n g litigation, c o m m i t m e n t s for e x p a n sion, a n d financing a r r a n g e m e n t s . Figure 3 presents a trace of the line(s) of r e a s o n i n g a d o p t e d b y t h e e x p e r t s u b j e c t ( S I ) at p o i n t s . i n t h e case w h e r e e a c h m a j o r c u e w a s pres e n t e d . As s h o w n in Fig. 3, t h e e x p e r t u s e d t h e p r o f i t a b i l i t y LOR to i n t e r p r e t m o s t o f t h e early cues. H e r e v i e w e d all t h e m a t e r i a l to identify unusual a n d high risk items. T h e g r o w t h cues w e r e p r e d o m i n a n t l y i n t e r p r e t e d u s i n g t h e profitability LOP,, a n d c u e s s u c h as lawsuit a n d i n c r e a s e in l o n g - t e r m d e b t w e r e i n t e r p r e t e d using t h e qual-

95

AUDIT J U D G M E N T RESEARCH CUES

1. Nature of products 2.

Obsolescence

3.

Lawsuit

4,

Chanle in current assets

PROFITABILITY

QUALIFICATION

DISCLOSURE

L

S. Increase in accounts receivable 6.

|ncrease b alinwonce for XCounts

7.

Increase in inventory (f'mished loads)

8.

Increase in PPE

9.

Increase in accounts payable

receivable

10. Tax/pay = zero 11. Increase in long-term debt

12. Increase in deferred taxes 13. Increase in retained earnings

L

14. A | i n | of payables

"0--...

15. Additional pald-in capital 16. Increase in sales 17. Increase in sales returns 18. Increase in COGS 19. Small tax provision 20. Method of recordinll sales 21. Inventory in hands of distributors 22. Obsolescence (inventory) 23. Inventory tumover 24. Increase in accounts receivable

!

25. Restricted stock 26. Sale of common stock 27, Capitalization policy for PPE 28. Deferral of patent costs 29. Sale of product line

30. Increase in workinl[ capital 31. Financinll of woykinl capital 32. Statement of chanRes in S/E

!

33. Foreian currency transactions 34. ChanRe in accountinl Imolds and dies) 35. Disclosure of R & D costs

36. Capitalization of R & O costs 37. Disclosl~re of ITC 38. Revenne recol|nitinn 39. Interest capitalization 40. ChanlLe in accountinR estimate

!

41. Oisciosure of IoniC-term debt 42. Common stock note 43. Tax note 44. Commitments and contingencies 45. Increase in men,facturinR capacity 46. Oiscinsuro of deprecioble lives 47. Relationship between accounts receivable. sales and sales retnrns 48. Ret,rn policy on sales

Fig. 3. E x p e r t l i n e o f

reasoning.

96

PAULE.JOHNSONet aL

ification and disclosure LOR, respectively. He then came back and followed up the issues identiffed in the review. Based on his review, SI noticed a large increase in sales returns ( c u e 17) which was considerably in excess of the increase in sales ( c u e 16). He used the profitability LOR to interpret this cue. Based on his experience, he knew that companies in this industry tend to manipulate income by overstating sales to ~cooperative" distributors and then record sales returns in subsequent periods. This contradicts the assertion provided by the profitability LOR that growth in demand for the company's products is responsible for the income reported. Because the expert has an alter.native explanation for the increase in income, he is able to detect the fraud. As can be seen in Fig. 3, the expert kept all three lines of reasoning "open" until he had completed his review. He used the profitability line of reasoning to correctly interpret the critical sales returns cue. Figure 4 provides a trace of the line of reasoning used by the novice subject ($2). This subject went d o w n the garden path, and issued an unqualified (clean) opinion. He explained away discrepant cues with reference to the growth subcategory. As discussed earlier, the increase in sales returns ( c u e 18) was explained by assuming that the c o m p a n y must have improved its warranty policy. The increase in inventory ( c u e 8) was explained by assuming that the c o m p a n y was stockpiling inventory in anticipation of a major sale soon after year end. The novice did not find any problems in the balance sheet and income statement, so he d r o p p e d the profitability LOR and concentrated mostly on disclosure. This can be seen clearly in Fig. 4, w h e r e the early cues are being interpreted using the profitability LOR. However, instead of looking for areas of risk, the novice is trying to make all the cues fit together. Due to this, the discrepant cues are explained away. While reviewing the notes to the financial statements the novice doesn't notice the decline in research and development expenses and is not able to combine the various income-enhancing cues. Discrepant cues which c o m e at the latter stage of the task (accounting policy changes - - cues 26, 29, 32, and 34) w e r e

evaluated with respect to their effect on disclosure, and the subject was willing to sign an unqualified opinion after s o m e disclosure and financial statement preseotation items w e r e "cleaned up". For the novice, early closure of alternative lines of reasoning led to problem-solving errors (a result consistent with those obtained in medicine, e.g. Elstein e t a l . , 1978). In general, the analysis presented here reveals that the expert subject used a highly efficient problem-solving strategy. Using a combination of three lines of reasoning, the expert did an overall risk analysis and identified areas of concern. He used his knowledge of the industry to identify problems and p e r f o r m e d the calculations required to support his conclusion. His tolerance level for errors was well tuned and he was able to find the problems e m b e d d e d in the case. This result suggests the hypothesis that expert auditors may have knowledge of the types of manipulations that are likely in a particular industry. When a set of critical cues are c o m b i n e d appropriately, they represent a pattern that is recognized as a fraud. The analysis also suggests that novices may not have such knowledge, and are thus far less likely to find frauds, even though they pay attention to the same cues as the expert. The industry-specific nature of this knowledge suggests that industry specialization may lead to the creation of knowledge structures which enable auditors to conduct industryspecific reviews of financial statements. In addition to having this industry knowledge, it may be important that the auditor be able to keep the relevant lines of reasoning "open" so that he/she has an opportunity to recognize problem situations w h e n they occur.

SUMMARY AND CONCLUSIONS This p a p e r has reviewed a "slice" of w o r k in the field of auditing. This slice consisted of research that attempted to understand audit judgment. Beginning with the lens model w o r k of Ashton (1974), and continuing through the heuristic and bias w o r k of Joyce & Biddle ( 1981 ), there was a tendency to rely on "laborat-

97

AUDIT JUDGMENT RESEARCH

CUES

PROFITABILITY

QUALIFICATION

DISCLOSURE

1. Inventory obsolescence 2.

Warranty

3.

Liability

4.

Litilatinn

e-..

S. Disclosures for imblicly-trded company 6.

Increase in accounts receivable

7.

Allowance for accounts receivable

8.

Increase in inventor,/

9.

Soles commitments

10. Increase in long-term debt ll.

Increase in PPE

12. Increase in other assets 13. Increase in accounts poyible 14. Cut-off work

15. Liquidity concern ]6, Issued common stock

t

17. SEC reporting 18. Increase in sales returns 19. Decline in gross margins 20. Income tax credits 21. Consistency of EPS calculation 22. Sources of funds 23. Construction of building 24. Stock split 25. Consolidation policy 26. MoMs and dies

L

27. Oepreciation method 28. Inventory method 29 R & D expense method 10 Inve,tment tax credit method 31 Revenue recoltnltion method 32. Capitalization of interest 33. Foreign currency tronslotlon 34. Change in accounting estimate ]S. Long-term debt footnote 36. Common stock footnote 3?. Income tax footnote 38. Segment data 39. Deferred tax ItWesentatinn " 40. Commitments end contingencies

A

.

41. Future lease rentals 42. Property. plant and equipment note 43. Asset classification

1

44. Foreign operations dS. Warranty policy 46. Offering document

Fig 4. Novice line of reasoning.

98

PAULE.JOHNSON et aL

ory versions" o f audit tasks. Moreover, little dev e l o p m e n t o f c o n s t r u c t s o r t h e o r y s e e m s to exist to e x p l a i n a u d i t o r behavior. Even w h e n research m o v e d t o w a r d d e t e c t i n g a u d i t o r d e c i s i o n processes, researchers addressed the s h o r t - t e r m goal o f l o o k i n g for "rules" to be used in b u i l d i n g d e c i s i o n aids for p r a c t i c i n g auditors rather than the l o n g e r t e r m goal o f d e v e l o p i n g c o n c e p t s a n d p r i n c i p l e s that m i g h t guide s u b s e q u e n t research. T h e u p s h o t o f the "themes" identified in the w o r k e x a m i n e d here is that the field of audit research finds itself looking to o t h e r fields for ideas a n d insights. Rather than c o n t i n u e in this way, it has b e e n suggested that the field of audit j u d g m e n t is a w o r t h y area o f investigation in its o w n right. This m e a n s that the p r o b l e m s auditors e n c o u n t e r a n d the m e t h o d s suitable for addressing t h e m s h o u l d arise from the field itself, rather than from the w o r k of investigators in o t h e r disciplines. As an e x a m p l e o f h o w such an a r g u m e n t might b e made, the role of c o n t e x t and m e a n i n g in audit j u d g m e n t tasks was considered. More specifically, it was p r o p o s e d that a t h e o r y of the task of a u d i t i n g b e d e v e l o p e d as a m e a n s of p r o v i d i n g h y p o t h e s e s regarding the operative k n o w l e d g e u n d e r l y i n g successful task p e r f o r m a n c e . This knowledge, called expertise, is used b y auditors at all levels to give m e a n i n g to the characteristics of the tasks they perform. An e x a m p l e o f w o r k

d e s i g n e d to d e s c r i b e the e x p e r t i s e of practising auditors in the task o f c o n c u r r i n g p a r t n e r reviews a n d fraud d e t e c t i o n was p r e s e n t e d to illustrate the p r o p o s e d m e t h o d o l o g y . T h e r e is n o single best way to do research. A m a j o r s t r e n g t h o f a n y field is the diversity of app r o a c h e s it tolerates a n d the a t m o s p h e r e it pror i d e s for the e x c h a n g e o f ideas o n the results and merits o f each alternative. What is p r o p o s e d h e r e is b u t o n e o f a n u m b e r of d i r e c t i o n s audit research might p r o c e e d from its c u r r e n t position. T h e i m p o r t a n t t h i n g is to r e c o g n i z e that ultimately it is the t h o u g h t s a n d actions of the p r a c t i c i n g a u d i t o r that n e e d to b e explained. For, as Ulric Neisser in his b o o k C o g n i t i o n a n d R e a l i t y notes:

...The prediction and control of behavior is not primarily a psychological matter. What would we have to know to predict how a chess master would move his pieces or his eyes? His moves are based upon information he has picked up from the board so they can only be predicted by someone who has access to the same information, in other words, the aspiring predictor would have to understand the position at least as well as the master does. He would have to be a chess master himseff. If I play chess against a master he will alwayswin precisely because he can predict and control my behavior while ! cannot do the reverse. To change this situation I must improve my knowledge of chess, not my knowledge of psychology... (Neisser, 1976, pp. 182-183).

BIBLIOGRAPHY Ashton, R. H., An Experimental Study of Internal Control Judgments,Journal of Accounting Research (Spring 1974) pp. 143-157. Ashton, IL H. & Kramer, $. 5., Students as Surrogates in Behavioral Research: Some Evidence,Journal of Accounting Research ( Spring 1980) pp. 269-277. Biggs, S. F., Messier, W. F. & Hansen,J. V., A Descrtptit~Analysis of Computer Audt¢ Specialists Decision Making Behavior in Advancea Computer Environmentg Unpublished manuscript ( 1986a). Biggs, S. F. & Mock,T.J., Auditor Information Search Processes in the Evaluation of Internal Controls and Audit Scope Decisions,Journal of Accounttn&Research (Spring 1983) pp. 234--.255. Biggs, $. F., Mock, T.J. & Watldns,P. IL,ADescriptive Study of Auditors Useo f Analytical Review in Audit Program Desigr~ Paper presented at USCAuditJudgment Conference, Los Angeles(1986b). Brunswik, E., The ConceptualFramework of Psychology ( Chicago, IL:Universityof Chicago Press, 1952). Chomsky, N., A Review of Sldrmer'sVerbal Behavior,Language (1959) pp. 26--58.

AUDIT JUDGMENT RESEARCH Cohen, J., A Coefficient of Agreement for Nominal Scales, Educational and Psychological Measuremen~ (1960) pp. 37-46. Dungan, C. W. & Chandler, J. S., Auditor: a Microcomputer-based Expert System to Support Auditors in the Field, Expert Systems (October 1985) pp. 210-221. Ebbesen, E. B. & Konecni, V.J., On the External Validity of Decision Making Research: What do we Know about Decisions in the Real World, in Wallsten, T. S. (ed) Cognitive Processes in Choice and Decision Making (Hillsdale, NJ: Lawrence Erlbaum, 1980). Einhorn, H. J., Expert Judgment: Some Necessary Conditions and an Example, Journal of Applied Psychology (October 1974) pp. 562-571. Einhorn, H. J., Synthesis: Accounting and Behavioral Science. Studies on Human Information Processing in Accounting, Supplement toJournal of Accounting Research ( ! 976) pp. 196-206. Elstein, A. S., Shulman, L. S. & Spralka, S. A., Medical Problem Solving (Cambridge, MA: Harvard University Press, 1978). Felix, W. L. & Kinney, W. R., Research in the Auditors Opinion Formulation Process: State of the Art, Tbe Accounting Review (1982) pp. 245--2171. Hansen, J. v. & Messier, W. F., A Preliminary Investigation of EDP-XPERT, Auditing: a Journal of Practice a n d Theory (Fall 1986) pp 109-123, Jenkins, J., The Challenge to Psychological Theorists, in Dixon T. R. and Dorton D. L. (eds) Verbal Behavior and General Behavior Theory (New York, NY: Prentice-Hall, 1968). Johnson, P.E., Hassebrock, F., Duran, A. S. & Moiler, J, H., Multimethod Study of Clinical Judgment, Organizational Behavior and Human Performance (1982) pp. 201-230. Johnson, P. E. Moen, J. & Thompson, W., Garden Path Errors in Diagnostic Reasoning, in Bole, L and Coombs, M.J. ( eds ) Computer Expert Systems ( Berlin: Springer, in press ). Johnson, P. E., Zualkeruan, I &Garber, S., Specification of Expertise, International Journal of Man. Machine Studies (1987) pp. 161-181. Joyce, E.J., Expert Judgment in Audit Program Planning. Studies in Human Processing in Accounting. Supplement to the Journal of Accounting Research (1976) pp. 29--60. Joyce, E. J. & Biddle, G. C., Anchoring and Adjustment in Probabilistic Inference in Auditing.Journal of AccounttngResearch ( 1981 ) pp. 120-145. Libby, R., Accounting and Human Information Processing: Theory and Applications ( Englewood Cliffs, NJ: Prentice-Hall, 1981 ). Libby, R. & Lewis, B., Human Information Processing Research in Accounting: the State of the Art, Accounting~ Organizations and Society ( 1977 ) pp. 245-268. Libby, R. & Lewis, B., Human Information Processing Research in Accounting: the State of the Art in 1982, Accounting Organizatiorts and Society ( 1982 ) 231-285. Meservy, R. D., Bailey, A. D. Jr, & Johnson, P. E., Internal Control Evaluation: a Computational Model of the Review Process, Auditing: a Journal of Practice and Theory (Fall 1986) pp. 44-74. Mock, T. & Turner, J , Internal Accounting Control Evaluation andAudttorJudgment (Audit Research Monograph No. 3, MCPA, 1981 ). Mohr, L B., Explaining Organizational Behavior (San Francisco: Jossey-Bass, 1982). Nlesser, U., Cognition andRealtty (San Francisco, CA: Freeman, 1976). Newell, A., The Knowledge Level, AIMagazine (Summer 1981 ) pp. 1-20. Newell, A., Reasoning, Problem Solving and Decision Processes: the Problem Space as a Fundamental Category, in Nickerson, R. (ed.)Attention and Performance (Hilisdale, NJ: Lawrence Erlbaum, 1980). Newell, A. & Simon, H. A., Human Problem Solving ( Englewood Cliffs, NJ: Prentice-Hall, 1972 ). NUsson, N.J., Principles of Artificial Intelligence ( Palo Mto, CA: Tioga, 1980 ). Pepper, S. C., World Hypotheses (Berkeley, CA: University of California Press, 1972). Shpilberg, D. & Graham, L E., Developing Expertax: an Expert System foi" Corporate Tax Accrual and Planning, Auditing: A Journal of Practice and Theory (Fall 1986) pp. 75-94. Simon, H. A., The Structure of lll-stroctured Problems, Artificial Intelligence (1973) pp. 181-201. Tversky, A. & gahneman, D., Judgment Under Uncertainty: Heuristics and Biases, Science (1974) pp. 1124-1131.

99