What is Professional Skepticism?

Four Determinants of Auditors Skeptical Disposition and Their Relationship to Analytical Procedures Planning Behavior
Luc Quadackers Vrije Universiteit Amsterdam and Ernst & Young lquadackers@feweb.vu.nl Tom Groot Vrije Universiteit Amsterdam Arnold Wright Northeastern University, Boston

August 25, 2007

Draft version, please do not refer to or quote without permission

We are indebted to the Big Four auditing firms that participated in this study. We would like to thank all people who have helped in developing the research materials, coding and analyzing the results.

Abstract The purpose of this study is to examine the relationships between four determinants of auditors skeptical disposition and to look at what determinants are most closely related to auditors skeptical behaviors. Despite the importance of professional skepticism in the auditing literature, there is a lack of consensus and empirical data on how skeptical disposition is measured and the extent to which such measures map with auditor behaviors. Gaining insight into factors that lead to desired skeptical behaviors forms the basis for auditor training, guidance in audit tools, and can be used as a starting point for future research. Four determinants of skeptical disposition are compared: (1) interpersonal trust; (2) suspension of judgment; (3) locus of control; and (4) a comprehensive professional skepticism scale. Moreover, the influence of strength of management control philosophy on this relationship is studied. An experimental study is conducted to address these issues involving a sample of 376 auditors from offices of the Big Four auditing firms in the Netherlands with experience ranging from staff to partner level. The results show that the strength of the effects of the determinants on skeptical behaviors is different across the determinants. Overall, interpersonal trust appears to have the highest explanatory power in predicting skeptical behaviors.

1.

Introduction Professional skepticism is considered to be an essential ingredient of the financial

statement audit, as reflected in the professional auditing standards (e.g., IFAC, 2007) and the audit methodologies of international audit firms. Furthermore, the academic and professional auditing literatures emphasize the importance of the use of professional skepticism (see e.g., Hurtt et al., 2003a, p.2; Kadous, 2000; Nelson, 2007). In addition, analyses of fraud related SEC cases conclude that a lack of sufficient professional skepticism often resulted in auditors failing to detect material misstatements (e.g., Beasley et al., 2001; POB, 2000; Benston and Hartgraves, 2002, p. 122). Hence, studying professional skepticism is important.

Despite its importance, there exists no clear definition of what professional skepticism is and how auditors skeptical characteristics are related to skeptical behaviors (see e.g., Hurtt et al., 2003a; Nelson, 2007).1 In a review of the philosophical literature, Kurtz (1992, p. 21-22) summarizes:

... skeptikos means to consider, examine; skepsis means inquiry and doubt. ... Skeptics always bid those overwhelmed by Absolute Truth or Special Virtue to pause. They ask, What do you mean?- seeking clarification and definition and Why do you believe what you do?- demanding reasons, evidence, justification, or proof. ... they say, Show me. ... Skeptics wish to examine all sides of a question; and for every argument in favor of a thesis, they can usually find one or more arguments opposed to it.

In a recent working paper, Nelson (2007) summarizes the existing literature on professional skepticism in auditing and derives the following definition:

indicated by auditor judgments and decisions that reflect a heightened assessment of the risk that an assertion is incorrect, conditional on the information available to the auditor.

ISA 200.15 (IFAC, 2006) states that an auditor should plan and perform an audit with an attitude of professional skepticism, while ISA 200.16 (IFAC, 2006) states that [a]n attitude of professional skepticism means the auditor makes a critical assessment, with a questioning mind, of the validity of audit evidence obtained and is alert to audit evidence that contradicts or brings into question the reliability of documents and responses to inquiries and other information obtained from management and those charged with governance. Professional skepticism requires auditors to evaluate the

The detailed genesis of the concept of professional skepticism in the auditing standards is not determinable. The technical managers Audit and Attest Standards of the AICPA checked the archives but were unable to find information on this issue. However, it appears to have been first used in SAS No. 16 (i.e. the predecessor standard to SAS No. 53).

reliability of management assertions and to develop an audit program (Shaub and Lawrence, 1999, p. 62). Most of the auditing research literature that explicitly considers skeptical traits (cf. Nelson, 2007) has equaled trust (or more specifically the complement of trust: suspicion) as a measure for auditors skeptical disposition. Quadackers (2007) summarizes the literature on the use of the complement of trust as a skeptical disposition and recognizes the need for examining the Interpersonal Trust Scale (e.g., Rotter, 1967). The Quadackers (2007) paper uses the same case materials that are used in this paper. The interpersonal trust results of the Quadackers (2007) study are integrated in this paper for reasons of comparison. In particular, this paper adds to the Quadackers (2007) study in that it considers and compares three additional alternative measures of skeptical disposition and in evaluating their power of explaining auditors skeptical behaviors. Auditor skepticism will be studied based on the relationships between four determinants of auditors skeptical disposition, a situational factor (management control philosophy) and auditors skeptical behaviors (measured by planning judgments in an analytical procedures task). Knowing what determinants constitute an auditors skeptical disposition and how these determinants are related to skeptical behavior provides crucial insights for the audit. Finding out the factors that make up an auditors skeptical disposition can lead to a definition of the key attributes of skepticism in audit manuals and training. Ultimately, audit firms may administer personality tests to their personnel in order to assess the skepticism attributes of their auditors (e.g., Nelson, 2007). Alternatively, Nelson (2007, p. 35) suggests that audit firms can stress the importance of professional skepticism traits when hiring new auditors and when marketing the audit profession in the hopes of discouraging people with low professional skepticism traits from applying. Furthermore, the insights gained from this study will also help in structuring future auditing research. Knowing what determinants of skeptical attitude are most closely related to skeptical behavior can further narrow down the search for specific elements of those determinants and how they influence subsequent skeptical behaviors. An analytical procedures setting is selected, since such procedures are important in financial statement auditing and substantially influence audit effectiveness and

efficiency (see e.g., Hirst & Koonce, 1996, p. 461, 464; Peecher, 1996, p. 125-126; Koonce et al., 1995, p. 369). The experimental case, adapted from Peecher (1996), contains an unexpected material increase in gross margin and a related management nonerror explanation concerning the fluctuation. 376 auditors read the case materials and completed questions on how they would respond to the case situation in terms of audit planning. In addition, they completed instruments measuring skeptical personality dispositions. In total, four different determinants are used to capture skeptical disposition: (1) interpersonal trust (as measured by the Interpersonal Trust Scale, Rotter, 1967); (2) suspension of judgment (as measured by the complement of the Need for Closure Scale, Webster and Kruglanski, 1994); (3) locus of control (as measured by the Locus of Control Scale, Rotter, 1966); and (4) a comprehensive professional skepticism scale (i.e. the Hurtt Professional Skepticism Scale, Hurtt, 2003). These scales are widely recognized and cited in the decision science and/or auditing literatures and will be described in depth in the next section. Six variables were used to examine skeptical judgments (i.e. skeptical behaviors): (1) (the complement of) the likelihood that managements change in sales mix explanation accounts for substantially all of the increase in gross margin; (2) the likelihood of fraud; (3) the number of alternative explanations provided; (4) the number of error explanations; (5) the likelihood that error explanations account for substantially all of the increase in gross margin; and (6) the number of budgeted hours. Higher assessments for each of these variables represent greater skeptical behavior by the auditor. As mentioned, an important contribution of this paper is the empirical examination of four determinants of skeptical disposition. This study enables a comparison of the predictive validity of different determinants of skeptical disposition in shaping skeptical behavior. Furthermore, this paper studies the impact of a situational factor (i.e. client risk) on the relationship between the determinants of skeptical disposition and skeptical behavior. The findings show that interpersonal trust appears to have the highest explanatory power in predicting skeptical behaviors. Furthermore, the results show that there exists a

negative correlation between interpersonal trust and locus of control. There appear to be significant interaction effects of client risk on the relationships between determinants of skeptical disposition and skeptical planning behaviors. The remainder of the paper is organized as follows. In Section 2 the theory, literature and hypotheses are described. The research method will be discussed in Section 3 and the results are presented in Section 4. The final section provides a discussion of the findings and their implications for future research.

2.

Theory, literature and hypotheses

Determinants of auditors skeptical disposition As mentioned in the introduction, there exists no universally accepted definition of professional skepticism. Considering various determinants of auditors skeptical disposition (and related measures) enables a careful identification of personality factors that engender desired behaviors, how they are related and what determinants have highest explanatory value. Some authors assume auditors personality traits to be stable when they start audit training and practice (see e.g., Libby and Luft, 1993; Nelson, 2007). However, others have found auditor traits to be alterable (e.g. Carpenter, 2004). Hence, knowing the status quo of the determinants of auditors skeptical disposition may be influential in staffing decisions and/or planning auditors training. This study examines four determinants of skeptical disposition. These measures are based on skeptical characteristics which are of particular importance for auditors. These characteristics and the related measures are discussed in this section, as well as the associated findings of empirical auditing literature.

Reserved interpersonal trust Prior auditing research has considered auditors skeptical disposition as the complement of trust (see e.g., Cushing, 2000; Choo and Tan, 2000; Payne and Ramsay, 2005, p. 324; Quadackers, 2007; Shaub, 1996; Shaub and Lawrence, 1996; Shaub and Lawrence, 1999). For purposes of benchmarking this construct is integrated in the study.

Interpersonal trust can be defined as a generalized expectancy held by an individual or a group that the word, promise, verbal or written statement of another individual or group can be relied upon (Rotter, 1967, p. 651; Rotter, 1980, p. 1). ISA 200.16 states that [w]hen making inquiries and performing other audit procedures, the auditor is not satisfied with less-than persuasive audit evidence based on a belief that management and those charged with governance are honest and have integrity. The basic thought is that if an auditor has a lower level of interpersonal trust he is assumed to be more skeptical (e.g., Shaub, 1996; Hurtt, 2003). To measure trust, Shaub (1996) used the trustworthiness and independence parts of the Wrightsman Philosophies of Human Nature scale (Wrightsman, 1964, 1974) and a self-developed Client Trust scale. Choo and Tan (2000) used a modified version of the Rempel et al. Trust Scale (1985), originally measuring trust in the relationship with a persons life partner. To a limited extent these scales showed significant direct or interaction effects in explaining skeptical behaviors. However, none of the scales yield strong results. The trustworthiness and independence parts of the Wrightsman Philosophies of Human Nature Scale do not lead to significant results. Shaubs Client Trust Scale only shows significant results in two of the 18 regressions tested. The Rempel et al. Trust Scale showed some significant results but Choo and Tan (2000, p. 81) state that the scale is not psychometrically valid. With the exception of a study by Quadackers (2007), the Rotter Interpersonal Trust Scale (1967) has not been used yet as a measure for skeptical disposition in prior auditing research despite the fact that this scale is widely accepted in other fields (see e.g., Hoell, 2004; Johnson-George and Swap, 1982; Stack, 1978; Webb and Worchel, 1986). In nonauditing studies Rotters Interpersonal Trust Scale has been associated more strongly with actual behaviors and other validating evidence than other interpersonal trust scales (Stack, 1978, p. 569; Rotter, 1980, p.2).2 Therefore, the first measure for skeptical disposition considered in this study is the (complement of) the Interpersonal Trust Scale (Rotter, 1967).

In addition, the trustworthiness part of Wrightsmans Philosophies of Human Nature Scale has not shown significant results in the single related previous auditing study (Shaub, 1996).

Suspension of judgment Suspension of judgment is one of the main characteristics of skeptics (e.g., Bunge, 1991; Kurtz, 1992; Hurtt et al., 2003a). Kurtz states that suspension of judgments ... is a necessary ingredient of skeptical inquiry (1992, p. 41). According to Bunge (1991, p. 131) [s]keptics do not accept naively the first things they perceive or think; they are not gullible. Nor are they neophobic. They are just critical; they want to see evidence before believing. Skeptics particularly suspend judgment concerning whatever has not been checked (Bunge, 1991, p. 132). Skeptics keep on gathering evidence until no reasonable person would doubt the claim stated (Kurtz, 1992, p. 132). Also the auditing standards state that the auditor should gather sufficient and appropriate audit evidence to the point that he is able to draw reasonable conclusions on which to base the audit opinion (ISA 200.16).3 Particularly when there is a risk of material misstatement due to fraud, this can affect the auditors professional skepticism in two ways (ISA 240.64). Firstly, it may increase the auditors assessment of the risk of material misstatement and sensitivity in the selection of the nature and extent of documentation. Secondly, it may increase the auditors recognition of the need to corroborate management explanations or representations concerning material matters. This leads to suspension of judgment. Suspension of judgment is assumed to be negatively correlated with the need for cognitive closure. Need for cognitive closure is one of the important dispositional constructs affecting the knowledge acquisition process (e.g., Kruglanski and Ajzen, 1983). If a person has a higher need for closure then it might hinder the persons hypothesis-generation process because conflicting hypotheses would threaten an existing or inherited conclusion. Research evidence suggests that the need for quickly reaching a decision leads to the tendency to seek cognitive closure and to refrain from critical probing of a given seemingly adequate solution to a problem (Kruglanski and Freund, 1983, p. 450).
3

In addition, ISA 200.16 states: [w]hen making inquiries and performing other audit procedures, the auditor is not satisfied with less-than persuasive audit evidence based on a belief that management and those charged with governance are honest and have integrity. Accordingly, representations from management are not a substitute for obtaining sufficient appropriate audit evidence to be able to draw reasonable conclusions on which to base the auditors opinion.

The Need for Cognitive Closure Scale measures the desire to possess some knowledge on a given topic, any definite knowledge as opposed to confusion and ambiguity (Webster and Kruglanski, 1994). The proposition is that a skeptical person will suspend judgment until he or she knows more about an ambiguous situation and hence will have a lower score on this scale. This implies that an auditor with a higher need for closure has a lower ability to suspend judgment when confronted with a management explanation (i.e. he will be more eager to accept the explanation) and hence will behave a less skeptically. Or as Bailey et al. (2006) state: [a] heightened need for closure would inhibit the hypothesis-generation process because conflicting hypotheses would threaten an existing conclusion. Consideration of initial hypotheses wrongly may cause problems since they may steer auditors further evidence evaluation and judgment. The only auditing study to examine need for closure was conducted by Bailey et al. (2006). They report that the need for closure is significantly less for higher ranks than for lower ranks. Furthermore, a series of experiments show that need for closure affects judgment and decision making in professional settings. More specifically, auditors with a higher need for cognitive closure show more stereotyping behavior and spend less time on tasks.

Internal locus of control Locus of control refers to assumed internal states that explain why some people actively, tenaciously, and eagerly try to deal with difficult circumstances, while others get passively stuck in negative emotions (Lefcourt, 1991, p. 413). An external locus of control indicates a pervasive belief that outcomes cannot be influenced by ones personal efforts, while an internal locus of control implies the belief that outcomes are contingent upon personal actions (Lefcourt, 1991, p. 414).4 Persons with a more external locus of

The locus of control concept originates from Social Learning Theory (e.g. Rotter, 1954). According to Lefcourt (1991, pp. 413-414): [i]n social learning terminology, locus of control is a generalized expectancy pertaining to the connection between personal characteristics and/or actions and experienced outcomes. It develops as an abstraction from accumulated specific encounters in which persons perceive the causal sequences occurring in their lives. People with a more external locus of control increment their expectancies after reinforcement less than internals do. This means that

control are referred to as externals and persons with a more internal locus of control are referred to as internals. The importance of locus of control has been widely recognized (Phares, 1978, p. 276). Previous research shows that externals are more readily persuasible, confirming, and accepting of information from others (see e.g. Phares, 1978 for an overview). Internals are associated with a more active pursuit of valued goals, information seeking (e.g. they have better learning skills in identifying and searching for task relevant cues) and autonomous decision making, and are better able to cope with stress (e.g., Lefcourt, 1991, p. 414; Phares, 1978). Internals generally appear to be more competent and personally effective than externals (e.g., Phares, 1978, p. 278) and they approach situations with a more directive and alert posture than externals (Phares, 1978, p. 276). Although acceptance by an internal of the control of another does happen, it will most certainly be thoughtful and analytic rather than blind und unthinking (Phares, 1978, p. 279). The characteristics of internals mentioned above appear to be highly relevant for auditors professional skepticism (cf. Hurtt, 2003). For example, in terms of accepting management assertions, internals will be more critical than externals. Being critical of management assertions is considered to be important by the auditing standards (e.g., ISA 240, par. 42). Furthermore, in auditing surveys, several potential reasons for practicing a low level of professional skepticism in auditing have been identified. Pasewark et al. (1992) find that auditors may not take skeptical action because of intimidation or concerns about clients reactions. In addition, Behn et al. (1997) found a negative relationship between demonstrated professional skepticism and client satisfaction (i.e. the more skeptical the auditor was the unhappier the client became). These situations may result in pleasing the client and reducing skepticism. Allegedly, skeptical behavior in such situations may benefit from auditors with an internal locus of control. The locus of control construct has not been extensively used in prior research on auditor skepticism. However, some studies are of importance. Bernardini (1994)

externals do not link reinforcements to their own behavior and hence they do not expect another reinforcement when they show identical behavior (cf. Rotter, 1966, p. 2; Rotter, 1975, p. 56).

10

examines the link between locus of control and auditors fraud detection. The detection rate for fraud was not significantly higher for internal locus of control auditors (however there was an interaction with experience). Tsui and Gul (1996) find a direct effect of locus of control on auditors behavior in an audit conflict situation (but also report an interaction effect with ethical reasoning). Externals were more likely to ignore unrecorded liabilities as wanted by management. Donnelly et al. (2003) find that externals are more accepting of dysfunctional audit behavior (e.g. premature sign-off, gathering of insufficient evidence).

A comprehensive professional skepticism scale The need for development of a specific professional skepticism scale for auditing has been stressed by several authors (e.g., Choo and Tan, 2000; Hurtt, 2003). Hurtt (2003) developed such an instrument to measure an individuals level of professional skepticism based on skeptical characteristics. Hurtt et al. (2003a) derive three sets of skeptical characteristics from philosophical literature, auditing standards and existing literature on skepticism in auditing which together determine an individuals overall level of professional skepticism: (1) examination of evidence; (2) understanding evidenceproviders; and (3) acting on the evidence. The characteristics related to the examination of evidence consist of a questioning mind, suspension of judgment and search for knowledge. A questioning mind is demonstrated in a requirement for reasons, evidence, justification or proof. Suspension of judgment is a characteristic indicating that a skeptic is slow to form judgments, requiring deliberation and additional supporting information to reach that judgment. Search for knowledge is equated with curiosity. The characteristics related to understanding evidence-providers consist of interpersonal understanding of the motivation and integrity of evidence-providers. The characteristics to act on the evidence comprise self confidence and selfdetermination. Self confidence means the professional courage to act on the evidence that has been obtained. Self-determination is the individuals conclusions regarding the sufficiency of evidential matter.

11

These three sets of characteristics determine the individuals overall level of professional skepticism, which in turn drive skeptical behavior (Hurtt et al., 2003a). There is some empirical evidence that the scores on the Hurtt Professional Skepticism Scale are related to skeptical behavior (Hurtt et al., 2003b; and Popova, 2006). Hurtt et al. (2003b) find that scores on this scale are significantly related to contradictions detected in the working papers (one of two variables) and marginally related to one of three variables measuring expanded information search. Popova (2006) studied the relationship between the scores on the Hurtt Professional Skepticism Scale and the determination of initial hypotheses regarding potential misstatements. Generally, the findings show that the auditors who are less skeptical in terms of the Hurtt scale are guided more by client specific experience when they assess initial hypothesis regarding possible misstatements. Furthermore, more skeptical auditors are more evidence-prone than less skeptical auditors (particularly in the positive client experience setting) and more skeptical auditors judge fraud evidence to be of greater relevance. All four determinants described are hypothesized to influence skeptical disposition. Therefore it is of interest to study the extent to which these determinants are related. Three of the determinants comprise characteristics of skeptical auditors. The fourth determinant is a comprehensive measure of professional skepticism. It is not theoretically clear whether and how the determinants are related. For example, Hurtt (2003, p. 12) found empirical evidence for the fact that the trustworthiness and independence parts of the Wrightsman Philosophies of Human Nature Scale are not significantly correlated with the Hurtt Professional Skepticism Scale. However, she postulates that interpersonal trust aspects are a component of professional skepticism. Therefore, the relationship between the four determinants is stated as a research question.

RQ1: What is the relationship between the four determinants of skeptical disposition?

12

Prior research in psychology has found that personality dispositions influence behavior (e.g., Eagly and Chaiken, 1993; 2005; Ajzen, 2005, pp 34-37).5 In the setting of professional skepticism, Hurtt et al. (2003a; 2003b) theorize that an auditors skeptical characteristics (such as a questioning mind) determine level of professional skepticism, which subsequently drives skeptical behavior. Hence, theory implies that auditors with a more skeptical disposition behave more skeptically (e.g., suspend judgment and engage in more substantive testing) than auditors with a less skeptical disposition. Of particular importance is what determinant is most closely related to the auditors skeptical behaviors. The second research question is formulated as follows.

RQ2: How closely are the four determinants of skeptical disposition related to skeptical behavior and which is most closely related?

The influence of situational factors: client risk Although skeptical behavior is hypothesized to be related to skeptical disposition, behavior is also expected to be related to situational characteristics (see e.g., Eagly and Chaiken, 1993; 2005; Ajzen, 2005; Kee and Knox, 1970). Therefore, it is conceivable that professional skepticism is situation specific (Bhattacharya et al., 1998, p. 461). Auditor skeptical behavior is particularly necessary in high risk situations. For instance, professional standards dictate that engagements with a higher risk of material misstatement due to fraud should be audited with increased professional skepticism (IFAC, 2006, ISA 240.63). Such situations also expose the individual auditor and the firm to increased reputation and other risks (e.g., litigation). Furthermore, clients explanations of unexpected fluctuations should be corroborated by the auditor more fully if the risk related to the areas of explanation is high (cf. Hirst and Koonce, 1996, p. 473). Previous auditing studies suggest that an interaction effect exists between risk and skeptical disposition in explaining behavior (Hurtt et al, 2003b; Popova, 2006). Professional standards also echo this interaction. For instance, ISA 240.63 states that engagements with a higher risk of material misstatement due to fraud should be audited
5

See Quadackers (2007) for a more detailed description.

13

with increased professional skepticism (IFAC, 2006). This may result in all auditors exhibiting skeptical behavior in a high risk situation, regardless of their skeptical disposition. However, it is unclear whether behavior will then differ according to skeptical disposition in the low risk setting. Alternatively, auditors may all show similar levels of low skeptical behavior in the low risk setting (regardless of their skeptical disposition), since they may judge the situation as non risky. However, then skeptical behavior might be dependent on skeptical disposition in high risk settings. One of the most pervasive client risks is the clients control environment (e.g. Haskins, 1987; Bernardi, 1994). Management control philosophy is an important constituent of the control environment. Financial reporting problems of companies are often related to a weak management control philosophy (see e.g., COSO, 1992). Cohen and Hanno (2000) found that audit planning judgments are responsive to management control philosophy. Since the nature of the potential interaction effect between skeptical disposition, skeptical behavior, and client risk is unclear the final issue is addressed as a research question.

RQ3:

How is the relationship between auditors skeptical disposition and skeptical behavior (as reflected in audit planning judgments) dependent on the strength of management control philosophy?

3.

Method

Research setting The study utilizes an experimental case that is embedded in a planning stage analytical procedures setting.6 This setting is chosen because analytical procedures affect audit risk assessments and consequentially the audit work conducted which can have important ramifications on audit efficiency and effectiveness (see e.g., Cohen and Kida, 1989; Hirst and Koonce, 1996, p. 461, 464; Peecher, 1996, p. 125-126; Koonce et al.,
6

The case is adapted from Peecher (1996) with permission.

14

1995, p. 369). For example, missing a material overstatement of earnings that could have been identified by analytical procedures can result in misleading financial statements with the associated misallocation of resources and litigation (see e.g., Asare and Wright, 2001). The case contains an unexpected material increase in gross margin. Since the audit client is the most common source of explanations concerning unexpected fluctuations in planning-stage analytical procedures (see e.g., Hirst and Koonce, 1996, p. 463; Trompeter and Wright, 2007), in the case a client explanation is provided. The CFO of the company gives a non-error explanation concerning the fluctuation stating that the increase in gross margin is caused by a change in the sales-mix. In view of the fact that management may lack independence, auditors should evaluate client explanations with professional skepticism (cf. Bedard and Biggs, 1991, pp. 77-79; Glover et al., 2000, p. 29; ISA 240, par. 42). In the experiment, strength of management control philosophy was manipulated as strong or weak by using vignettes based on Cohen and Hanno (2000).

Research variables Dependent variables Six proxies for auditors skeptical behavior are used as dependent variables. They are presented below. An auditor should reflect on the information that is provided by a client. Particularly for a skeptical auditor it is common to ponder over the incentives a client might have in furnishing that information (e.g., Hurtt et al. 2003a). Auditors who assess a lower probability of accepting managements explanation are assumed to show more skeptical behavior. The variables studied in this respect are the likelihood that managements explanation (a change in sales mix) accounts for substantially all of the increase in gross margin (>85%) and the likelihood of fraud (c.f., Peecher, 1996; Shaub, 1996; Shaub and Lawrence, 1996; Payne and Ramsay, 2005; Choo and Tan, 2000; Knapp and Knapp, 2001). Furthermore, skeptical auditors should build explanations, hypotheses, or scenarios that can function as alternative interpretations for the information that they

15

examine (e.g., Hurtt et al. 2003a). Auditors are assumed to exhibit more skeptical behavior when they: (1) are able to generate a greater number of plausible alternative explanations; (2) provide more error-explanations (since these are opposite to the clients non error explanation); and (3) assess higher probabilities of the accuracy of error explanations. As a result, the variables studied are the number of alternative explanations provided, the number of error explanations provided and the likelihood that the error explanations account for substantially all of the increase in gross margin (>85%) (c.f., Peecher, 1996; McMillan and White, 1993). Finally, an indication of skepticism is the extent to which auditors want to perform further testing (e.g., Hurtt et al. 2003a). The main variable studied is number of budgeted hours (cf., Shaub, 1996; Shaub and Lawrence, 1996; Hurtt et al., 2003b; Popova, 2006). A reference point of 100 hours for last year is given.

Independent variables: four determinants of skeptical disposition The measurement scales for the four determinants of auditors skeptical disposition will now be discussed.7

Interpersonal trust The first determinant of skeptical disposition, interpersonal trust, is measured by (the complement of) the Interpersonal Trust Scale (Rotter, 1967). Rotters Interpersonal Trust Scale consists of 25 items that are scored on a five point Likert Scale (varying from strongly disagree to strongly agree).8 Three typical scale-items are, for example: (1) in

The primary focus of this study is on the general affects of determinants of skeptical disposition on auditor behaviors. Therefore, factor-analyses to examine individual factors within the determinants are outside the scope of this paper. As noted by Quadackers (2007), there are four reasons for selecting this measure: (1) Rotters Interpersonal Trust Scale is widely used and accepted (see e.g., Hoell, 2004; Johnson-George and Swap, 1982; Stack, 1978; Webb and Worchel, 1986) but has not been used yet as a measure for skeptical disposition in prior auditing research; (2) the stability and internal consistency of Rotters Interpersonal Trust Scale are strong (Rotter, 1967); (3) in non-auditing studies, Rotters Interpersonal Trust Scale has been associated more strongly with actual behaviors and other validating evidence than has Wrightsmans Philosophies of Human Nature Scale (Stack, 1978, p. 569; Rotter, 1980, p.2) ; and (4) the trustworthiness part of Wrightsmans Philosophies of Human Nature Scale has not shown significant results in the single related previous auditing study (Shaub, 1996).

16

dealing with strangers one is better off to be cautious until they have provided evidence that they are trustworthy (negatively scored); (2) it is safe to believe that in spite of what people say most people are primarily interested in their own welfare (negatively scored); and (3) most salesmen are honest in describing their products.

Suspension of judgment Suspension of judgment is studied as the second determinant of skeptical disposition and will be measured by the 42-item Need for Closure Scale (Webster and Kruglanski, 1994). The responses to the 42 items are requested on a 6-point Likert scale ranging from 1 (strongly disagree) to 6 (strongly agree). Webster and Kruglanski (1994) rationalized that the need for closure will express itself in a variety of ways. This is in accordance with the underlying theory of lay epistemics (Kruglanski, 1989, 1990a, 1990b). Therefore, they viewed it as a latent variable visible through different facets (cf. Carver, 1989).9 Webster and Kruglanski (1994) identified five major facets that represent the construct: (1) preference for order; (2) preference for predictability; (3) decisiveness; Webster and Kruglanski (1994) reasoned that persons with a high need for closure would possess a desire to reach closure, revealed in a decisiveness of their judgments and choices; (4) discomfort with ambiguity; and (5) closed-mindedness; Webster and Kruglanski (1994) theorize that a person with a high need for closure possesses an unwillingness to have one's knowledge confronted (hence, rendered insecure) by alternative opinions or inconsistent evidence. They found that the Need for Closure Scale is reliable in assessing a single coherent construct with five facets and has high testretest reliability. The Need for Closure Scale has been confirmed and has been validated in numerous studies cross-culturally. Three examples of scale items are: (1) when I am confused about an important issue, I feel very upset; (2) I dislike questions which could be answered in many different ways; and (3) I feel uncomfortable when I dont understand the reason why an event occurred in my life.

Webster and Kruglanski (1994) state that [a]s our theoretical interest was in this latent construct as such, its extent was assessed additively across the different item categories.

17

Locus of control The third determinant of skeptical disposition studied is locus of control and is measured by Rotters (1966) Locus of Control scale. This scale contains 23 question pairs that have to be answered in a forced choice format (i.e. one of the two possibilities per question). For example, one of the question pairs was: (a) without the right breaks one cannot be an effective leader; and (b) capable people who fail to become leaders have not taken advantage of their opportunities. The participants had to choose one of the two options. In this case option b indicated an external locus of control. In this fashion the number of hits on external locus of control statements constitutes the Locus of Control Score. Rotters Locus of Control scale has been proven reliable and valid in numerous samples (see e.g. Lefcourt, 1991).

Comprehensive professional skepticism The fourth comprehensive determinant of skeptical disposition is the Hurtt Professional Skepticism Scale (2003). This is a relatively new scale which has only been examined in a few unpublished papers (e.g. Hurtt et al. 2003b; Popova, 2006; Fullerton and Durtschi, 2005). The scale consists of 30 items to be scored on a 6-point scale, ranging from 1 (strongly disagree) to 6 (strongly agree). The scale is designed to provide a single score of professional skepticism, although the items are organized along six theoretically derived facets: (1) search for knowledge; (2) suspension of judgment; (3) self-determining; (4) interpersonal understanding; (5) self-confidence; and (6) questioning mind (see Hurtt, 2003). Three illustrative items are, for example: (1) I often accept other peoples explanations without further thought (negatively scored); (2) I dont like to decide until Ive looked at all of the readily available information; and (3) It is easy for other people to convince me (negatively scored). Analyses by Hurtt (2003) in a student and professional subjects sample indicate that the scale has adequate inter-item consistency and test-retest reliability. Hurtt et al. (2003b) find some evidence for the predictive validity of the scale.

18

Independent variables: manipulated and control variables Strength of the management control philosophy is manipulated as being strong (coded as 0) or weak (coded as 1).10 In the strong management control philosophy setting, management is conservative in doing business, has modest disagreements with the external auditors, has strict guidelines for following internal control procedures, and obtains reward based on several financial and non-financial performance measures. In the weak management control philosophy setting, management is aggressive in doing business, has frequent disagreements with the external auditors, has less strictly applied internal control procedures if advancement of work suffers from them, and has compensation plans that are mainly based on reaching short-term accounting-based performance measures. Strength of management control philosophy is manipulated in a between subjects design. To examine the effects of the determinants of skeptical disposition and strength of management control philosophy on skeptical behavior, it is necessary to control for the auditors prior task experience by measuring the participants years of experience in conducting analytical procedures and adding that variable to the models.

Research instrument validation and manipulation checks The majority of the participants in the experiment (i.e. 86%) was living in the Netherlands and their mother tongue was Dutch.11 Therefore the case-materials were translated into Dutch. Concerning the measurement scales, Dutch translations were readily available for the Need for Closure scale (Cratylus, 1995) and for the Locus of Control Scale (see e.g. Boone and De Brabander, 1993). Translations were developed for the Interpersonal Trust Scale and for the Hurtt Professional Skepticism Scale. For that purpose a combination of the Parallel Blind Technique and Translation/Back-Translation methods was used (Behling and Law, 2000). Two experienced auditors compared the English and the Dutch versions of the case description and evaluated the contents of the case materials. In addition, four professors provided remarks on the research instrument. Two pilot-tests were conducted

10 11

The case materials of Cohen and Hanno (2000) are used with permission. The remaining 14% of the participants completed the experiment in English.

19

with 19 staff level auditors and minor modifications were made to the instrument on the basis of the comments received. The research instrument contains two parts. The first part includes the case description and the tasks (i.e. the dependent variables).12 The second part contained the scales to measure skeptical disposition, demographic information and debriefing questions about the case. Two questions were asked as manipulation checks concerning the strength of management control philosophy variable (cf. Cohen and Hanno, 2000). The questions pertaining to the effectiveness of the control environment and overall control risk were assessed on a 9-point scale.13 The scale regarding control environment effectiveness was ranging form very ineffective to very effective. The scale concerning overall control risk was ranging from very low risk to very high risk. The means score on the control environment effectiveness was 3.49 in the weak management control philosophy setting and 6.21 in the strong management control philosophy setting. The mean score on the overall control risk was 6.63 in the weak management control philosophy setting and 4.76 in the strong management control philosophy setting. These results are in the expected direction and are highly significant (One-way ANOVAs show a p-value of 0.000). In order to assess case realism and understandability two questions were scored on a 9-point scale. The scale for case realism ranged from highly unrealistic to highly realistic and the scale for case understandability ranged from very unclear to very clear. The mean score for case realism was 6.08 and the mean score for case understandability was 6.73. One-sample t-tests show that the mean scores are significantly above the middle points (i.e. a value of 5) of the scales (p-values are 0.000).

12

13

The only exception is that one of the dependent variables (i.e. the likelihood of fraud) was asked in the second part of the questionnaire to prevent the possibility of revealing the purpose of the study. Before stating the questions, the ISA definitions (IFAC, 2006) for the control environment and control risk were given.

20

Sample All of the Big Four firms cooperated in this study. The participating firms are randomly coded as Firm A, B, C and D.14 Descriptive information on the sample can be found in Table 1.

*** INSERT TABLE 1 ABOUT HERE ***

There were 376 participants in total of which 86 were female (i.e. 23%) and 286 were male (i.e. 77%). All staff levels contained a number of participants that was sufficient for our statistical analyses. However, the number of partners (73) and staff (56) participating was less than managers (116) and seniors (124) taking part. The spread of the auditors across staff levels varied per firm as a result of availability. On average the auditors had about 10 years of general experience and 10 years of experience with conducting analytical procedures. Both general experience and task specific experience suggest that the participants possess the requisite task knowledge. The number of participants was relatively balanced across the two cells of the strength of management control philosophy manipulation. All participants completed only two of the measurement scales. This was done due to limited time availability as well as to avoid information overload and/or fatigue. One-way ANOVAs show that there are no order effects in the administration of scales for skeptical disposition. However, since the scales were offered to the participants in different combinations, order-effects cannot be assessed for the sample as a whole. Table 2 provides information on the sample sizes for the four measures of skeptical disposition studied across firms.

*** INSERT TABLE 2 ABOUT HERE ***

14

The audit firms decided during what sessions the researchers were to collect the data. This was out of the researchers influence. This resulted in the fact that firm correlates with experience of the participants and language of the research instrument. Hence, experience, firm and language effects cannot be disentangled. However, the careful procedures used for translation provide comfort that the meaning of the questions was conveyed in an accurate, clear manner. Furthermore, prior studies on skepticism have not found firm effects, and professional standards are the same across firms with respect to the need for skepticism. Hence, there is no basis to expect firm effects.

21

As can be seen in Table 2 there are three possibilities for mutual comparison of the scales (1) the Hurtt Professional Skepticism Scale and the Need for Closure Scale; (2) the Hurtt Professional Skepticism Scale and the Interpersonal Trust Scale; and (3) the Interpersonal Trust Scale and the Locus of Control Scale.

Administration of the experiment The experiment was conducted during 12 sessions of which 9 sessions were part of the yearly summer courses, two sessions were held during monthly partner meetings and one session was an audit methodology training course for instructors. The first author attended all sessions and provided a brief introduction before the start of the experiment. The single remark concerning the content of the study was that it comprised a case on conducting preliminary analytical procedures. After the introduction the participants received an envelope containing the research materials. A printed instruction regarding completion of the instrument was stuck on the envelope. The different versions of the questionnaire were spread across participants in a fully randomized fashion. After a starting-signal from the instructor the participants opened the envelope. The envelope included two smaller envelopes with parts one and two of the research instrument. The two envelopes containing the parts of the instruments were identified with identical numbers in order to enable ex post matching. The participants were asked to write down their name on the envelope in order to induce a feeling of accountability. Upon completion of part one (risk assessment and audit planning tasks), the participants were instructed to put the first part of the questionnaire into the first envelope and seal it. Then they proceeded with part two (skeptical dispositions and debriefing questions) which had to be put in the second envelope and sealed after completion. Finally, envelopes 1 and 2 had to be put into the large envelope. At the end of the research instrument, participants could leave their email address to indicate that they would like to receive a summary of the findings.

22

Coding In the case description, the CFO provided a non-error explanation for the increase in the gross margin percentage. The participants were requested to think about possible alternative explanations for the increase in the gross margin percentage. Those alternative explanations were coded by the author and an experienced audit manager. The coding encompassed an assessment of the type of explanation and of the plausibility of the explanation (i.e. is the explanation logical and in the right direction). In order to appropriately study error explanations, which are of the highest importance to the study, the following types of explanations were distinguished: non-error explanations, unintentional error explanations, intentional error explanations and ambiguous unintentional/intentional explanations. Cohens Kappa Coefficient regarding the coding of the plausibility of the explanation was 0.795 (p = 0.000) and Cohens Kappa Coefficient for coding the type of explanation was 0.864 (p = 0.000). These levels of the Kappa are strong (e.g., Landis and Koch, 1977). Differences were discussed by the two coders and mutually resolved. Explanations were also coded as a repeated explanation or not (i.e. does the respondent repeat an earlier explanation). Although the Kappa Coefficient was significant (p = 0.000) its value was low (0.193). In the total of 1290 explanations there were disagreements in only 54 cases (4.2%). In most cases one coder considered the explanations to be repeated while the second coder did not. A closer examination of the explanations revealed that there were new additional elements present in explaining what was going on in those cases. For this reason, after discussion the repeated explanations were coded as not repeated explanations.

4.

Results Descriptive statistics regarding the measurements scales used for skeptical

disposition in the study are shown in Table 3. As can be derived from the table, all mean scores are relatively close to the theoretical mid points of the scale, except for the Hurtt Professional Skepticism Scale.

*** INSERT TABLE 3 ABOUT HERE ***

23

The Cronbach alpha values for the scales are all adequate (see e.g. Nunnally, 1978). One-sample Kolmogorov Smirnov tests and an examination of the histograms indicate that the measurement scales studied show normally distributed scores.

Results for Research Question 1 The quintessence of RQ1 is whether there is a relationship between the four measures for skeptical disposition. Table 4 shows Pearson correlations between the measures.

*** INSERT TABLE 4 ABOUT HERE ***

Table 4 shows a significant negative correlation between interpersonal trust and locus of control. This implies that an auditor who is more trusting (i.e. a higher score on interpersonal trust) has a more internal locus of control (i.e. a lower score on locus of control) and vice versa. If the theoretical assumption is correct that more skepticism means lower trust and more internal locus of control, this finding is contrary to expectations. None of the correlations between the other combinations are significant. This could be an indication that the measures capture different aspects of skepticism. Alternatively, this may be a first indication that some of the factors studied are not strong determinants of skeptical disposition. However, it has to be bore in mind that for reasons of brevity of the research instruments the study did not collect samples which enable comparison between interpersonal trust and need for closure, need for closure and locus of control, and locus of control and the Hurtt Professional Skepticism Scale.

Results for Research Question 2 MANOVAs were conducted to assess the impact of the independent variables on the set of dependent variables as a whole.15 Four MANOVAs were done, one per
15

None of the dependent variables are normally distributed (i.e. one-sample Kolmogorov-Smirnov tests are statistically significant in rejecting the null-hypothesis that the distribution equals a normal distribution). Nevertheless, the MANOVA technique is robust to departures from normality, particularly with larger sample sizes (e.g., Hair et al., 2006, p. 410). According to Hair et al. (2006, p. 86) normality might have severe effects in small samples (i.e. less than 50 cases). The sample sizes

24

determinant of skeptical disposition. The dependent variables were: the likelihood that managements change in sales mix explanation accounts for substantially all of the increase in gross margin (>85%), the likelihood of fraud, the number of alternative explanations provided, the number of error explanations provided, the likelihood that the error explanations account for substantially all of the increase in gross margin (>85%), and the number of budgeted hours. As independent variables the models included the respective determinant of skeptical disposition, the management control philosophy manipulation, years of experience with conducting analytical procedures (i.e. task specific experience) and an interaction term between the determinant of skeptical disposition and the management control philosophy manipulation. Results of Levenes tests of equality of error variances lead to rejection of the hypothesis that error variances are equal for: 3 (out of 6) dependent variables for the Interpersonal Trust scale; 2 dependent variables for the Need for Closure scale; 4 dependent variables for the Locus of Control scale; and 1 dependent variable for the Hurtt Professional Skepticism Scale. However, the MANOVA technique is quite robust for violations of the equality of variances assumption when the sample sizes in the cells are roughly equal (see e.g., Hair et al., 2006, 409-410; Finch, 2005). The cells of the manipulated variable for the four professional skepticism measure samples are about equal (i.e., 149 and 142 for the Interpersonal Trust measure, 43 and 42 for the Need for Closure measure, 100 and 95 for the Locus of Control measure, and 92 and 89 for the Hurtt Professional Skepticism Scale measure). Hence, the MANOVA results will be interpreted. An assumption for MANOVA is that a significant correlation exists between the dependent variables. This can be tested by Bartletts test for sphericity, which is significant (p = 0.000) for all four professional skepticism measure samples.

for all professional skepticism measures are considerably above 50 (see Table 2). Moreover, except for the Need for Closure measure (80 observations), all measures are close to or above 200 observations. Hair et al. (2006, p. 86) state that sample sizes of 200 or more successfully moderates the negative impact of non-normality. However, violation of the normality assumption jeopardizes interpretation of the results of the Boxs M test of equality of covariance matrices. For that reason, Boxs M tests are not shown.

25

The multivariate effects are examined (untabulated). For the Interpersonal Trust measure sample, the tests show statistical significance for all independent variables in the model. Hence, all independent variables appear to have a significant influence on the collective set of dependent variables. For the Need for Closure measure sample, the tests show statistical significance for management control philosophy strength and for the interaction of the Need for Closure measure and management control philosophy strength. For the Locus of Control measure sample, the tests show statistical significance for management control philosophy strength and for task specific experience (and marginal significance for the interaction of the Locus of Control measure and management control philosophy strength). For the Hurtt Professional Skepticism Scale sample, none of the tests show statistical significance. However, the level of skepticism is marginally significant. Linear regressions were conducted in order to understand the nature of the impact of the independent variables on the skeptical behaviors. A summary of the regressions is provided in Table 5.

*** INSERT TABLE 5 ABOUT HERE ***

Interpersonal trust is significant in two of the six regression models (i.e. the likelihood that management explanation is right16 and the likelihood of fraud). These effects are in the expected direction. Need for closure is significant in one of the models (i.e. the number of error explanations). However, the effect is opposite to the expected direction. This effect will be examined further when Research Question 3 is discussed. Locus of control is not significant in any of the models. The Hurtt Professional Skepticism Scale is significant for the model regarding the number of error explanations and marginally significant for the model with the number of budgeted hours as a dependent variable. In sum, interpersonal trust and the Hurtt Professional Skepticism Scale show the strongest direct effects of disposition on behaviors. However, both constructs have an
16

Note that this model is marginally significant.

26

impact of different behaviors. Locus of control appears to have no direct influence on skeptical behavior, which suggests that locus of control does not directly determine skeptical behavior. Need for closure showed one effect that was opposite to the expected direction. That effect will be addressed when discussing the interaction effects.

Results for Research Question 3 RQ3 aims at assessing the interaction effects between determinants of auditors skeptical disposition and strength of management control philosophy in explaining auditors skeptical behavior. First, the direct impact of management control philosophy strength is examined. For interpersonal trust, strength of management control philosophy is significant in four of the six significant models (i.e. the likelihood of fraud, the number of alternative explanations, the number of error explanations and the weight of the error explanations). The effects are all in the expected direction. For need for closure, strength of management control philosophy is significant in three of the four significant models (i.e. the likelihood of fraud, the number of error explanations and the weight of the error explanations). The effects are all in the expected direction. For locus of control, strength of management control philosophy is significant in one of the four significant models (i.e. the number of error explanations) and marginally significant in one of the four significant models (i.e. the weight of the error explanations). Both effects are in the expected direction. For the Hurtt Professional Skepticism Scale, strength of management control philosophy is not significant in any of the significant models. All significant direct effects of strength of management control philosophy are in the expected direction. In the models with the number of error explanations and the weight of error explanations as dependents, strength of management control philosophy has the strongest direct effect. There are significant interaction effects between interpersonal trust and the strength of management control philosophy for three of the six significant models (i.e. the assessment of the likelihood of fraud, the number of alternative explanations and the number of error explanations). In addition, there is a marginally significant effect for the weight of the error explanations. These effects were presented in detail by Quadackers

27

(2007). The interaction effects are about equal concerning the effects for the need for closure construct. To be exact, there is a significant interaction effect between need for closure and the strength of management control philosophy concerning the number of error explanations and there are two marginally significant interaction effects for the likelihood of fraud and the weight of the error explanations. The plot of the interaction effect pertaining to the number of error explanations is presented in Figure 1. The figure shows that the number of generated error explanations is about equal for the low and high need for closure groups in the weak management control philosophy setting.17

*** INSERT FIGURE 1 ABOUT HERE ***

Hence, when there is a management control philosophy problem obviously all auditors seem to adequately respond to this, regardless of the skeptical disposition. However, in the strong management control philosophy setting there is a significant difference (post hoc Scheffe test) between the mean number of total error explanations for the low and high need for closure groups. Contrary to intuition, the line is upward sloping which indicates that auditors with a lower need for closure (i.e. they are supposed to possess a higher degree of skeptical disposition) generate less alternative error explanations than auditors with a higher need for closure (i.e. auditors with a supposedly lower degree of skeptical disposition). This seems to be related to the already presented finding that the direct effect of need for closure is opposite to the expected direction for the number of error explanations. This pattern is also suggested by the plot of the interaction effect of the weight of the error explanations (untabulated). However, post hoc Scheffe tests and independent sample t-tests do not show significant differences between the means of the low and high need for closure groups in the weak and the strong management control philosophy settings concerning the weight of the error explanations variable.18 As was concluded
17

Post hoc Scheffe tests and independent sample t-tests show no significant difference between the means for the low and high need for closure groups in the weak setting. 18 As presented in the Quadackers (2007) paper, independent sample t-tests for weight of error explanations showed a significant difference between the means of the low and high interpersonal trust groups concerning the strong management control philosophy setting.

28

concerning the interpersonal trust interaction in the Quadackers (2007) paper, the counterintuitive slope of the line may be explained by the fact that skeptical auditors focus more (i.e. put more weight) on fewer errors which would result in a higher average weight per error. This would result in a different slope if the interaction effect is calculated for the variable weight of the error explanations divided by the number of error explanations. A plot of this variable is presented in Figure 2.

*** INSERT FIGURE 2 ABOUT HERE ***

As can be seen from Figure 2, the slope for the strong control philosophy setting has reversed, as expected. An independent samples t-test shows marginal significance of the difference between the low and high need for closure groups (one-sided p = 0.07).19 This implies that more skeptical auditors focus more (i.e. put more weight) on fewer errors which would result in a higher average weight per error. Apparently, this conception both holds for interpersonal trust and need for closure. The interaction effect concerning the likelihood of fraud is presented in Figure 3.

*** INSERT FIGURE 3 ABOUT HERE ***

The plot of the interaction effect looks similar to the plot of the interpersonal trust variable presented by Quadackers (2007). However, post hoc tests showed a significant difference for the low and high interpersonal trust groups in the weak management control philosophy setting but post hoc tests and independent sample t-tests do not show a significant difference for the need for closure measure (this might be due to the relatively small sample size of the need for closure sample). There is a marginally significant interaction effect between locus of control and the strength of management control philosophy for the likelihood of fraud. The plot of this interaction effect is presented in Figure 4.

*** INSERT FIGURE 4 ABOUT HERE ***

19

Note that the sample size is small.

29

An independent sample t-test shows a marginal significance (two-tailed p = 0.07) of the difference between the low and high locus of control groups in the weak management control philosophy setting. However, the slope of the line is opposite to the slopes for the interpersonal trust and need for closure constructs. This implies that auditors with a more internal locus of control (i.e., a lower score on the locus of control construct), which allegedly means a higher skeptical disposition, assess lower likelihoods of fraud. This seems to suggest locus of control is not capturing behaviors that depict the notion of skepticism as outlined in professional standards. There are no interaction effects for the Hurtt Professional Skepticism Scale. In sum, the results suggest that the effect of skeptical disposition on the audit planning variables depends on the strength of management control philosophy. This in particular holds for the interpersonal trust and need for closure measures, which basically show similar interaction effects.20

5.

Discussion

Summary A summary of the findings concerning the effects of the determinants of skeptical disposition and strength of management control philosophy on skeptical behavior (as well as their interaction effects) are presented in Table 6. The findings show that interpersonal trust, need for closure and the Hurtt Professional Skepticism Scale are directly of influence on skeptical behaviors. Furthermore, there are significant interaction effects for the models concerning interpersonal trust, need for closure and locus of control. Predominantly, these interaction effects are present for the likelihood of fraud, the number of error explanations and the weight of the error explanations. In particular, the number of error explanations seems to be directly explained by three of the four determinants of skeptical disposition.

*** INSERT TABLE 6 ABOUT HERE ***

20

This study controlled for experience. As can be seen in Table 5, task specific experience has a significant impact in several of the models. This variable is studied in more detail in Quadackers (2007).

30

More skeptical behaviors are associated with interpersonal trust than any of the other determinants, suggesting it most closely depicts the skeptical behaviors prescribed in the auditing literature and professional standards. This also confirms the use of trust variables in earlier research on professional skepticism. An examination of the correlations between the determinants of skeptical disposition only revealed a significant negative correlation between interpersonal trust and locus of control. From a skeptical viewpoint this appears to be contrary to expectations. Interestingly, interpersonal trust has the strongest influence on skeptical behaviors while locus of control only has a weak impact. This suggests that although interpersonal trust and locus of control are related, locus of control does not seem to capture skeptical disposition. The importance of skepticism may overpower locus of control such that both internals and externals show skeptical behavior (cf. Phares, 1978, p. 282). Furthermore, none of the correlations between the other combinations of the determinants were found significant.21 This could be an indication that the measures capture different uncorrelated aspects of skepticism. This indication is supported by the fact that need for closure and the Hurtt Professional Skepticism Scale also do have significant (in)direct effects on skeptical behaviors.

Implications and suggestions for further research The fact that interpersonal trust is most closely related to skeptical behaviors confirms the use of trust variables in earlier research on professional skepticism. It warrants a further study of the components of the trust variable (see Quadackers, 2007) in order to more specifically find the factors comprising skeptical disposition. Moreover, the study shows that determinants of skeptical disposition appear to be related to skeptical behaviors. This has not been strongly evidenced by previous studies and it justifies further study in order to develop adequate staffing and training guidance. As mentioned, evidence is unclear about the stability of auditors personality traits (see

21

Keep in mind that not all possible correlations could be tested due to the sample.

31

e.g., Libby and Luft, 1993; Nelson, 2007; Carpenter, 2004). Further research can address this issue. Additionally, alternative measures of auditors skeptical disposition should be examined. Another avenue of research may focus on developing more specific measures of skeptical attitude (see e.g. Ajzen, Lefcourt, 1991, pp. 414-), instead of the so-called global measures.

Literature Ajzen, I., Attitudes, Personality, and Behavior, Open University Press, Berkshire, 2005.

Asare, S.K. and A. Wright, Design Considerations for Research on Analytical Procedures, International Journal of Auditing, Vol. 5, Nr. 3, November 2001 (10), pp. 205-214.

Bailey, C.D., C.M. Daily and T.J. Philips Jr., A Study of Kruglanskis Need for Closure Construct and Its Implications for Judgment and Decision Making in Accounting and Auditing, Working paper, January 2006.

Beasley, M. S., J.V. Carcello, and D.R. Hermanson, Top 10 audit deficiencies, Journal of Accountancy, April 2001, 63-66.

Bedard, J.C. and S.F. Biggs, The Effect of Domain-Specific Experience on Evaluation of Management Representations in Analytical Proceures, Auditing, 1991, Supplement, pp. 77-95.

Behling, O. and K.S. Law, Translating Questionnaires and Other Research Instruments: Problems and Solutions, Sage Publications, 2000.

32

Behn, B.K., J.V. Carcello, D.R. Hermanson and R.H. Hermanson, The Determinants of Audit Client Satisfaction Among Clients of Big 6 Firms, Accounting Horizons, Vol. 11, No. 1, 1997, pp. 7-24.

Benston, G.J. and A.L. Hartgraves, A. L., Enron: What happened and what we can learn from it, Journal of Accounting and Public Policy, 2002, pp. 105-127.

Bernardini, R.A., Fraud Detection: The Effect of Client Integrity and Competence and Auditor Cognitive Style, Auditing: A Journal of Practice & Theory, 1994 (13), Supplement, pp. 68-84.

Bonne, C. and B. de Brabander, Generelized vs. Specific Locus of Control Expectancies of Chief Executive Officers, Strategic Management Journal, 1993, Vol. 14, pp. 619-625.

Bunge, M., A Skeptics Beliefs and Disbeliefs, New Ideas in Psychology, Vol. 9, No. 2, 1991, pp. 131-149.

Carpenter, T.D., Partner Influence, Team Brainstorming, and Fraud Risk Assessment: Some Implications of SAS No. 99, Unpublished Dissertation, Florida State University, 2004.

Carver, C.S., How should multifaceted personality constructs be tested? Issues illustrated by self-monitoring, attributional style, and hardiness, Journal of Personality and Social Psychology, 1989, 56, 577-585.

Choo, F. and K. Tan, Instruction, Skepticism, and Accounting Students Ability to Detect Frauds in Auditing, Journal of Business Education, Fall 2000, pp. 72-87.

Cohen, J. and T. Kida, The Impact of Analytical review Results, Internal Control Reliability, and Experience on Auditors Use of Analytical Review, Journal of Accounting Research, Vol. 27, No. 2, Autumn 1989, pp. 263-276.

33

Cratylus, Een Nederlandse Need for Closure-lijst, Nederlands Tijdschrift voor de Psychology, 1995 (50), pp. 231-232.

Cushing, B.E., Economic Analysis of Skepticism in an Audit Setting. In; I. Solomon and M.E. Peecher (Eds.), 14th Symposium on Auditing Research, 2000, pp. 1-3.

Donnelly, D.P., J.J. Quirin and D. OBryan, Auditor Acceptance of Dysfunctional Audit Behavior: An Explanatory Model Using Auditors Personal Characteristics, Behavioral Research in Accounting, Vol. 15, 2003, pp. 87-110.

Eagly, A.H. and S. Chaiken, The Psychology of Attitudes, Wadsworth, 1993.

Finch, H., Comparison of the Performance of Nonparametric and Parametric MANOVA Test Statistics when Assumptions Are Violated, Methodology, Vol. 1, No. 1, 2005, pp. 27-38.

Glover, S.M., J. Jiambalvo, J. Kennedy, Analytical Procedures and Audit Planning Decisions, Auditing: A Journal of Practice and Theory, Vol. 19, No. 2, Fall 2000, pp. 27-45.

Hair, J.F. Jr., W.C. Black, B.J. Babin, R.E. Anderson and R.L. Tatham, Multivariate Data Analysis, Sixth Edition, Pearson Prentice Hall, New Jersey, 2006.

Hirst, D.E. and L. Koonce, Audit analytical procedures: A field investigation, Contemporary Accounting Research, 1996, Volume 13, Issue 2, pp. 457-486.

Hoell, R.C., The Effect of Interpersonal Trust and Participativeness on Union Member Commitment, Journal of Business and Psychology, Vol. 19, No. 2, Winter 2004.

34

Hurtt, R.K., Development of an Instrument to Measure Professional Skepticism, Working paper, University of Wisconsin-Madison, 2003.

Hurtt, R.K., M. Eining and D. Plumlee, Professional Skepticism: A Model with Implications for Research, Practice and Education, working paper, University of Wisconsin-Madison, 2003a.

Hurtt, R.K., M. Eining and D. Plumlee, An Experimental Examination of Professional Skepticism, working paper, University of Wisconsin-Madison, 2003b.

Hyatt, T.A. and D.F. Prawitt, Does Congruence between Audit Structure and Auditors Locus of Control Affect Job Performance?, The Accounting Review, Vol. 76, No. 2, April 2001, pp. 263-274.

IFAC, Handbook of International Auditing, Assurance and Ethics Pronouncements, 2006.

Johnson-George, C. and W.C. Swap, Measurement of Specific Interpersonal Trust: Construction and Validation of a Scale to Assess Trust in a Specific Other, Journal of Personality and Social Psychology, 1982, Vol. 43, No. 6, pp. 1306-1317.

Kadous, K., The Effects of Audit Quality and Consequence Severity on Juror Evaluations of Auditor Responsibility for Plaintiff Losses, The Accounting Review, 2000 (75), pp. 327-341.

Kee, H.W. and R.E. Knox, Conceptual and Methodological Considerations in The Study of Trust and Suspicion, Journal of Conflict Resolution, 1970 (14), pp. 357-366.

Koonce, L., Explanation and Counterexplanation During Audit Analytical Review, The Accounting Review, Vol. 67, No. 1, January 1992, pp. 59-76.

35

Koonce, L., U. Anderson and G. Marchant, Justification of decisions in auditing, Journal of Accounting Research, 1995, Volume 33, Issue 2, pp. 396-384.

Kruglanski and Ajzen, Bias and Error in Human Judgment, European Journal of Social Psychology, Vol. 13, 1983, pp. 1-44.

Kruglanski, A.W. and T. Freund, The Freezing and Unfreezing of Lay-Inferences: Effects on Impressional Primacy, Ethnic Stereotyping, and Numerical Anchoring, Journal of Experimental Social Psychology, 1983 (19), pp. 448-468.

Kurtz, P., The New Skepticism: Inquiry and Reliable Knowledge, Buffalo/New York, Prometheus Books, 1992.

Lefcourt, H.M., Locus of Control. In: Robinson, J.P., P.R. Shaver and L.S. Wrightsman, Measures of Personality and Social Psychological Attitudes, Volume 1 in Measures of Social Psychological Attitudes Series, 1991, pp. 413-500.

Libby, R. and J. Luft, Determinants of Judgment Performance in Accounting Settings: Ability, Knowledge, Motivation, and Environment, Accounting, Organizations and Society, 1993 (18), Nr. 5,pp. 425-450.

Nelson, M., A Model and Literature Review of Professional Skepticism in Auditing, working paper, Cornell University, July 2007.

Nunnally, J. C., Psychometric theory, second edition, New York: McGraw-Hill, 1978.

Pasewark, W.R., R.A. Shockley and J.E. Wilkerson, Do Senior Auditors Experience Difficulty With Personal Objectivity When Making Inquiries of Client Employees, Ohio CPA Journal, 51, 2, 1992, pp. 14-18.

36

Payne, E.A. and R.J. Ramsay, Fraud Risk Assessments and Auditors Professional Skepticism, Managerial Auditing Journal, Vol. 20, No. 3, 2005, pp. 321-330.

Peecher, M.E., The Influence of Auditors Justification Processes on Their Decisions: A Cognitive Model and Experimental Evidence, Journal of Accounting Research, 1996 (34), pp. 125-140.

Phares, E.J., Locus of Control. In: London, H. and J.E. Exner Jr., Dimensions of Personality, John Wiley & Sons, 1978, pp. 263-304.

Public Oversight Board (POB), The Panel on Audit Effectiveness: Report and Recommendations, 2000.

Quadackers, L.M., The Relationship between Auditors Skeptical Disposition and Planning Judgments in an Analytical Procedures Task, working paper, Vrije Universiteit Amsterdam, April 2007.

Rotter, J.B., Social Learning and Clinical Psychology, Englewood Cliffs: Prentice-Hall, 1954.

Rotter, J.B., Generalized Expectancies for Internal versus External Control of Reinforcement, Psychological Monographs: General and Applied, Vol. 80, No. 1, 1966, pp. 1-28.

Rotter, J.B., A New Scale for the Measurement of Interpersonal Trust, Journal of Personality, 1967 (35), pp. 651-665.

Rotter, J.B., Some Problems and Misconceptions Related to the Construct of Internal Versus External Control of Reinforcement, Journal of Consulting and Clinical Psychology, 1975, Vol. 43, No. 1, pp. 56-67.

37

Rotter, J.B., Interpersonal Trust, Trustworthiness, and Gullibility, American Psychologist, Vol. 35, No. 1, January 1980, pp. 1-7.

Shaub, M.K., Trust and Suspicion: The Effects of Situational and Dispositional Factors on Auditors Trust of Clients, Behavioral Research in Accounting, 1996 (8), pp. 154174.

Shaub, M.K. and J.E. Lawrence, Ethics, Experience, and Professional Skepticism: A Situational Analysis, Behavioral Research in Accounting, 1996, Supplement, pp. 124157.

Shaub, M.K. and J.E. Lawrence, Differences in Auditors' Professional Skepticism Across Career Levels in The Firm, Advances in Accounting Behavioral Research, 1999 (2), pp. 61 -83.

Stack, L.C., Trust. In: London, H. and J.E. Exner Jr., Dimensions of Personality, John Wiley & Sons, 1978.

Trompeter, G. and A. Wright, The World Has Changed; Have Analytical Procedure Practices?, working paper, Boston College, 2007.

Tsui, J.S. and F.A. Gul, Auditors Behaviour in an Audit Conflict Situation: A Research Note on The Role of Locus of Control and Ethical Reasoning, Accounting, Organizations and Society, Vol. 21, No. 1, 1996, pp. 41-51.

Webb, W.M. and P. Worchel, Trust and Distrust. In: Worchel, S. and W.G. Austin, Psychology of Intergroup Relations, Nelson-Hall Publishers, Chicago, 1986, pp. 213-228.

Webster, D.M. and A.W. Kruglanski, Individual Differences in Need for Cognitive Closure, Journal of Personality and Social Psychology, 1994, Vol. 67, No. 6, pp. 10491062.

38

Wrightsman, L.S., Measurement of Philosophies of Human Nature, Psychological Reports, 1964, 14, pp. 743-751.

Wrightsman, L.S., Assumptions about Human Nature: A Social-psychological Analysis, Monterey, CA, Brooks/Cole, 1974.

39

Table 1. Descriptive sample information

Firm A Number of participants Number of sessions Staff level Partners Managers Seniors Staff Male Female Mean years St.dev. Mean years St.dev. Strength of management control philosophy Weak 42 43 85 0 47 49 96 0 68 72 101 39 27 28 41 14 184 192 323 53 Firm B Firm C Firm D Total

85 1 1 50 32 2 59 26 10.11 6.40 9.67 6.11

96 3 25 41 27 0 84 11 15.36 8.95 14.75 9.36

140 6 0 19 65 54 92 47 2.99 2.00 2.75 1.93

55 2 47 6 0 0 51 2 22.16 8.52 20.07 6.86

376 12 73 116 124 56 286 86 10.48 9.37 9.85 8.86

Gender

General experience Experience with analytical review

Strong Language Dutch English

Table 2. Participants responding to measures of skeptical disposition across firms

Firm A Firm B Firm C Firm D Total

Interpersonal Trust Scale Need for Closure Scale Locus of Control Scale Hurtt Professional Skepticism Scale

0 85 0 85

96 0 0 96

140 0 140 0

55 0 55 0

291 85 195 181

40

Table 3. Descriptive statistics concerning the measurement scales

Measurement Scale Interpersonal Trust scale Need for Closure scale Locus of Control scale Hurtt Professional Skepticism Scale 131.66 10.71 30-180 103-158 105 0.821 10.06 3.76 0-23 0-19 11.5 0.713 154.85 15.03 42-252 121-193 147 0.822 72.98 8.76 25-125 45-105 75 0.760 Mean score Standard deviation Theoretical range Actual range Theoretical mid point Cronbach alpha

41

Table 4. Correlations between the professional skepticism determinants

Hurtt Professional Skepticism Scale -,084 ,417 96

Interpersonal trust Interpersonal trust Pearson Correlation Sig. (2-tailed) N Pearson Correlation Sig. (2-tailed) N Pearson Correlation Sig. (2-tailed) N Pearson Correlation 1 291

Need for cognitive closure .(a) . 0

Locus of Control -,367(**) ,000 195

Need for cognitive closure

(a) . 0 -,367(**) ,000 195

1 85 (a) . 0

(a) . 0 1 195

,089 ,416 85 .(a) . 0

Locus of Control

Hurtt Professional Skepticism Scale

** (a)

-,084 ,089 Sig. (2-tailed) ,417 ,416 N 96 85 Correlation is significant at the 0.01 level (2-tailed). Cannot be computed because at least one of the variables is constant.

(a) . 0

1 181

42

Table 5. Regression coefficients for the determinants of auditors skeptical disposition and the models adjusted R-squares
Interpersonal Trust
Regression of the likelihood that management explanation is right Constant Management control philosophy (MCPH) Determinant of disposition (DOD) Task-specific experience (TSE) MCPH*DOD Adj. R-square (N) Regression of the likelihood of fraud Constant Management control philosophy (MCPH) Determinant of disposition (DOD) Task-specific experience (TSE) MCPH*DOD Adj. R-square (N) Regression of the number of alternative explanations Constant Management control philosophy (MCPH) Determinant of disposition (DOD) Task-specific experience (TSE) MCPH*DOD Adj. R-square (N) Regression of the number of error explanations Constant Management control philosophy (MCPH) Determinant of disposition (DOD) Task-specific experience (TSE) MCPH*DOD Adj. R-square (N) Regression of the weight of the error explanations Constant Management control philosophy (MCPH) Determinant of disposition (DOD) Task-specific experience (TSE) MCPH*DOD Adj. R-square (N) Regression of the number of budgeted hours Constant Management control philosophy (MCPH) Determinant of disposition (DOD) Task-specific experience (TSE) MCPH*DOD Adj. R-square (N) 13.446 -.331 .359** -.153 -.045 .015* (N=277)

Need for Closure

45.362 -67.239 -.051 .226 .391 .001 (N=85)

Locus of Control
36.168 .594 .262 -.144 -.279 -.015 (N=188)

Hurtt PS Scale

42.931 41.454 -.031 .066 -.370 .019 (N=175)

40.747*** 71.549*** -.304** .164 -.666** .310*** (N=277)

3.688 128.778*** .122 -.174 -.667* .267*** (N=84)

19.745*** 10.289 -.120 .214 1.404* .277*** (N=186)

-5.032 26.875 .210 -.179 -.027 .236*** (N=176)

1.885 4.235*** .018 .023** -.056** .019* (N=283)

-2.242 9.064 .037 .002 -.057 .028 (N=85)

3.677*** .123 -.049 .056*** -.005 .071*** (N=191)

1.069 4.384 .018 -.019 -.031 .005 (N=178)

-.796 4.852*** .021 .045*** -.057*** .111*** (N=282)

-5.729*** 9.388*** .044*** .015 -.054** .163*** (N=85)

.901** 1.534** -.041 .078*** -.057 .234*** (N=190)

-2.027 3.326 .027** -.010 -.021 .042*** (N=178)

10.275 65.566*** .030 .593*** -.667* .124*** (N=279)

-37.837 143.429** .354 .229 -.765* .165*** (N=85)

10.702 17.664* -.385 .903*** .461 .215*** (N=189)

-7.215 -21.735 .254 -.186 .283 .071*** (N=176)

147.380*** 28.356 -.360 .558** -.287 .022** (N=272)

113.601*** 19.352 .003 .466 -.025 .109** (N=82)

124.131 -2.659 -.360 .068 1.013 -.009 (N=186)

70.269 -79.512 .375* .725** .691 .105*** (N=169)

Significance is indicated by asterisks: *=.10; **=.05; ***=.01. For the directional hypotheses, all significance-levels are one-sided unless the results are in the opposite direction. For the non-significant models no significance is shown for the coefficients.

43

Figure 1.

A plot of the interaction effect between need for closure and the strength of management control philosophy for the number of total error explanations

44

Figure 2.

A plot of the interaction effect between need for closure and the strength of management control philosophy for the average weight per error.

45

Figure 3.

A plot of the interaction effect between need for closure and the strength of management control philosophy for the likelihood of fraud.

46

Figure 4.

A plot of the interaction effect between locus of control and the strength of management control philosophy for the likelihood of fraud.

47

Table 6. Summary of significant and marginally significant effects.

Significant effects determinants skeptical disposition 1tailed pvalue 0.032 0.042 sign Significant effects management control philosophy strength 1tailed pvalue 0.000 0.009 0.002 0.008 0.010 0.005 0.017 0.013 0.055 sign Significant interactions 2tailed pvalue 0.013 0.022 0.009 0.073 0.059 0.021 0.077 0.092

Interpersonal trust

Likelihood mngt expl. is right Likelihood of fraud

+ -

Likelihood of fraud Number of alternative expl. Number of error explanations Weight of error explanations Likelihood of fraud Number of error explanations Weight of error explanations Number of error explanations Weight of error explanations

+ + + + + + + + +

Likelihood of fraud Number of alternative expl. Number of error explanations Weight of error explanations Likelihood of fraud Number of error explanations Weight of error explanations Likelihood of fraud

Need for closure

Number of error explanations

0.002

+*

Locus of control

Hurtt Professional Skepticism Scale

Number of error explanations Number of budgeted hours

0.023 0.085

+ +

* This sign is not in the expected direction.

48