38
THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN FRAUD RISK BRAINSTORMING Michelle McAllister Florida State University Email: [email protected] Allen Blay Associate Professor Florida State University Email: [email protected] Kathryn Kadous McIntyre Term Chair and Professor of Accounting Goizueta Business School Emory University Email: [email protected] Preliminary Draft. Do not cite or circulate without author permission. Special thanks to Kenny Reynolds, Jeremy Douthit, and Bud Fennema for their helpful comments.

THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM

IN FRAUD RISK BRAINSTORMING

Michelle McAllister

Florida State University Email: [email protected]

Allen Blay

Associate Professor Florida State University

Email: [email protected]

Kathryn Kadous McIntyre Term Chair and Professor of Accounting

Goizueta Business School Emory University

Email: [email protected]

Preliminary Draft. Do not cite or circulate without author permission. Special thanks to Kenny Reynolds, Jeremy Douthit, and Bud Fennema for their helpful comments.

Page 2: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

1

THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM

IN FRAUD RISK BRAINSTORMING

Abstract

PCAOB AS 2110 Identifying and Assessing Risks of Material Misstatement provides auditors with guidance on how to plan the audit in order to obtain reasonable assurance about whether the financial statements are free of material misstatement due to fraud. Two key elements addressed in PCAOB AS 2110 are the importance of exercising professional skepticism and the requirement of a fraud brainstorming session. Despite this requirement, little is known about how individual professional skepticism might affect the effectiveness and efficiency of responses to fraud risk indicators in group settings. The findings of this study demonstrate that individual differences in trait professional skepticism among group members in fraud risk brainstorming settings can significantly impact the quality of these sessions. Including members with high levels of trait professional skepticism in the group significantly increases the group’s perceived risk of fraud in both higher and lower risk settings. However, inclusion of these members does not appear to significantly improve the group’s ability to identify relevant underlying fraud hypotheses. This research contributes to the fraud brainstorming and professional skepticism streams of research by demonstrating that the mix of individual differences in professional skepticism among group members can significantly impact the outcomes of fraud risk brainstorming.

Page 3: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

2

INTRODUCTION

The ability to detect fraudulent financial reporting in client financial statements continues to be of

utmost concern to auditors and regulators. PCAOB AS 2110, Identifying and Assessing Risks of Material

Misstatement, provides auditors with guidance on how to plan the audit in order to obtain reasonable

assurance about whether the financial statements are free of material misstatement due to fraud. Two key

elements addressed in PCAOB AS 2110 are the importance of exercising professional skepticism and the

requirement of a discussion among engagement personnel regarding the risks of material misstatement

due to fraud (AS 2110, paragraphs .49-.53). We examine how these elements operate jointly.

Specifically, this study addresses whether individual trait professional skepticism impacts the

effectiveness and efficiency of fraud risk brainstorming. It further examines whether one highly skeptical

member of a brainstorming group is sufficient, or whether more are needed for an impact.

The “discussion” referred to in PCAOB AS 2110 is intended to be operationalized as a group

brainstorming session among the audit team members with the purpose of identifying “how and where

they believe the entity's financial statements might be susceptible to material misstatement due to fraud,

how management could perpetrate and conceal fraudulent financial reporting, and how assets of the entity

could be misappropriated” (AS 2110). Beyond stating the purpose of the group brainstorming session,

and asserting that the discussion “should occur with an attitude that includes a questioning mind”,

consistent with the definition of professional skepticism, little guidance is provided as to how best to

conduct these sessions (AS 2110). Accordingly, accounting research has attempted to fill this gap by

exploring how audit team brainstorming affects fraud risk planning, along with methods for improving

such brainstorming sessions.

Previous accounting research finds evidence that brainstorming audit teams generate more high

quality fraud ideas than individual auditors, and are more effective at modifying standard audit

procedures in response to fraud risk indicators (Carpenter 2007; Brazel, Carpenter, and Jenkins 2010;

Hoffman and Zimbelman 2009). Previous research also indicates that computer-mediated, or electronic

Page 4: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

3

brainstorming can significantly improve fraud brainstorming, yet in practice most brainstorming

conducted is face-to-face open brainstorming (Brazel et al. 2010; Lynch, Murthy, and Engle 2009).

Lastly, Brazel et al. (2010) propose three important elements to modeling brainstorming quality:

attendance and communication, brainstorming structure, and engagement team effort. Within the Brazel

et al. (2010) model, we propose auditors with higher assessed levels of trait professional skepticism who

attend brainstorming sessions may improve overall brainstorming quality by helping the group to more

accurately produce fraud risk assessments and identify fraud hypotheses.

PCAOB AS 2401 asserts that the characteristics of fraud make the auditor’s exercise of

professional skepticism throughout the audit of particular import and defines professional skepticism as

“an attitude that includes a questioning mind and a critical assessment of audit evidence” (AS 2401.13).1,2

In addition, PCAOB AS 2110 explicitly states the brainstorming discussion should occur with an attitude

of professional skepticism whereby team members “set aside any prior beliefs they might have that

management is honest and has integrity” (AS 2110.52). While extensive accounting research has

explored the link between individual professional skepticism and the evaluation of audit evidence and

assessments of fraud risk (Hurtt, Eining, and Plumlee 2008; Quadackers, Groot, and Wright 2009,

Quadackers, Groot, and Wright 2014), researchers have yet to consider how an individual’s level of

professional skepticism can impact the group as a whole. This is an extremely important question given

that audits are not completed by individuals, but rather by teams of auditors.

Recent auditing research identifies two primary components of individual professional

skepticism: skeptical judgment and skeptical action. Skeptical judgments relate to an auditor’s

recognition that a potential issue may exist, whereas skeptical actions refer to changes in auditor behavior

based on skeptical judgments (Hurtt, Brown-Liburd, Earley, Krishnamoorthy 2013). According to the

1 The specific requirements of the fraud brainstorming session are addressed in PCAOB AS 2110 Identifying and Assessing Risks of Material Misstatement. However, PCAOB AS 2401 Consideration of Fraud in a Financial Statement Audit speaks to all other aspects of the auditor’s responsibility as it relates to fraud. 2 PCAOB AS 2401 suggests fraud is difficult to detect because the perpetrators of fraud generally take steps to conceal their actions and it is often difficult to determine management’s intentions.

Page 5: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

4

model developed by Nelson (2009) and extended by Hurtt et al. (2013), one of the four broad inputs to an

auditor’s professional judgments is individual auditor characteristics. Individual auditor characteristics

include the auditor’s trait professional skepticism, experience and training, and motivation. Of particular

importance to this study is the concept of trait professional skepticism. Hurtt (2010) defines a trait as a

relatively stable, enduring aspect of an individual and she posits a link between an auditor’s trait

professional skepticism and skeptical judgments and actions. This link is validated in recent studies that

find that higher levels of inherent skepticism lead to differential evidence assessment (Hurtt, Eining, and

Plumlee 2008; Quadackers et al. 2009). While trait professional skepticism has been shown to affect

skeptical actions on an individual level, it is not clear how individual trait skepticism presents in a group

setting.

We conduct an experiment measuring individual trait professional skepticism using the Hurtt

(2010) professional skepticism scale. We place participants into groups based on their skepticism levels to

create groups that contain no highly skeptical individuals, one highly skeptical individual, and two highly

skeptical individuals. The participants first make an individual risk assessment and then participate in a

group brainstorming session in either a higher or lower fraud risk scenario. The setting is designed to

closely match the requirements of AS 2110.

The results of the study indicate that individual differences in inherent trait professional

skepticism can significantly impact the outcomes of fraud risk brainstorming groups. In both high and

low risk situations, we find evidence that groups that contain at least one member with high trait

professional skepticism evaluate the overall risk of fraud as higher than groups that do not contain

individuals with high levels of trait professional skepticism. Furthermore, we find no evidence of a

significant relationship between the number of members with high levels of trait professional skepticism

and the assessed risk of fraud when at least one member in the group possesses a high level of trait

professional skepticism. Because a significant difference in risk assessments persists across the higher

and lower risk settings even when high trait professional skeptics are in the group, we also conclude that

Page 6: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

5

the targeted inclusion of these individuals does not negatively impact the efficiency of the audit by

causing the group to over-estimate the risk of fraud in low-risk settings. However, we find no evidence

that including individuals with high trait professional skepticism in the group improves the group’s ability

to identify fraud hypotheses. Thus, while the synergistic properties of brainstorming make it possible for

those with high levels of professional skepticism to positively influence the skeptical awareness of the

group, it does not appear to help the group identify the underlying drivers of higher levels of risk.

Further, final post-brainstorming fraud risk assessments of participants with low levels of professional

skepticism were no higher when the individual was in a group with a highly skeptical individuals. In

sum, these findings indicate that including at least one highly skeptical auditor in brainstorming sessions

can increase audit effectiveness at least in terms of perceptions of fraud risk in brainstorming sessions, but

further research is necessary to determine the post-brainstorming effects on audit effectiveness.

The remainder of this paper proceeds as follows. First, we provide a summary of prior research

and develop our hypotheses. Second, we discuss our research design and experimental procedures.

Third, we provide a thorough analysis of our findings. Last, we provide several concluding thoughts,

limitations of the current study, and suggestions for future research.

PRIOR RESEARCH AND HYPOTHESIS DEVELOPMENT

Fraud Brainstorming

The purpose of fraud risk brainstorming as outlined in PCAOB AS 2110 is to help audit team

members consider how and where the entity’s financial statements might be susceptible to material

misstatement due to fraud and to reinforce the importance of an appropriate mindset of professional

skepticism. It requires auditors to complete a fraud brainstorming session as part of the planning stage of

an audit. Aside from outlining the purpose of the brainstorming session, the standard provides little

guidance concerning how to best conduct a proper brainstorming session. Perhaps as a consequence of

this lack of guidance, the PCAOB has noted several instances where auditors have failed to sufficiently

Page 7: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

6

comply with this auditing standard (PCAOB Release No. 2007-001). Indeed, recent accounting research

documents considerable variation in the reporting quality of fraud risk brainstorming across audit

engagements (Brazel et al. 2010). In response to this lack of guidance, accounting researchers, drawing

on previous psychology research, have attempted to determine the positive effects, if any, of fraud risk

brainstorming on audit quality and the best practices for conducting such brainstorming sessions.

The psychological research into brainstorming finds mixed results concerning the positive

outcomes of the practice. While some research in psychology posits that brainstorming can result in

performance improvements from cognitive stimulation and synergy among group members, many other

studies document significant drawbacks to the production quality of group brainstorming.3 Production

blocking, social loafing, and evaluation apprehension are generally the primary reasons cited for poor

group brainstorming performance. Production blocking refers to a cognitive interference mechanism

whereby group participants forget their unique ideas while waiting for an opportunity to speak. Social

loafing implies that some participants may reduce their effort in the group even when they have the ability

to contribute important content to the group. Evaluation apprehension occurs when group members feel

intimidated about expressing an idea for fear that others might not like it (Dugosh, Leggett, Paulus,

Roland, and Yang 2000, Carpenter 2007; Hoffman and Zimbelman 2009). In addition to these commonly

cited drawbacks to group brainstorming, many social psychologists posit social comparison processing

may also play a role in the productivity of group brainstorming. Social comparison processing refers to

the idea that group members reduce their performance to match that of the least productive member of the

group (Dugosh et al. 2000; Camacho and Paulus 1995, Paulus and Dzindolet 1993). As an example,

Camacho and Paulus (1995) find that brainstorming groups in which the participants are mixed between

high socially anxious individuals and low socially anxious individuals perform significantly worse than

groups containing all low socially anxious individuals or high socially anxious individuals working alone.

3 See Carpenter (2007) for a review of the psychology literature concerning brainstorming.

Page 8: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

7

Despite the potential drawbacks cited in the psychology literature, accounting research documents

improvements in fraud risk planning due to group brainstorming. Accounting researchers generally

attribute these differences to the use of participants with specific subject knowledge. Most psychology

research conducted on brainstorming uses novices as subjects and employs tasks that do not require

specific knowledge or experience to complete. In contrast, individuals who participate in fraud risk

brainstorming experiments have experience and training in assessing the risk of fraud (Carpenter 2007;

Hoffman and Zimbelman 2009; Lynch, Murthy, and Engle 2009). Further, additional psychological

research indicates brainstorming quality is significantly improved when participants actively attend to the

ideas generated by other members of the group (Dugosh et al. 2000). It may be that participants in audit

research studies perform better in brainstorming groups relative to novice psychology subjects simply

because they are motivated to actively listen and engage in the discussion. This may be because audit

research participants have specific knowledge and insight to contribute during these sessions.

Carpenter (2007) first considered how fraud risk brainstorming might impact the evaluation of the

likelihood of fraud. In her experiment using hierarchical teams of three auditors, she found evidence that

while the quantity of fraud hypotheses created during fraud risk brainstorming is reduced, the hypotheses

generated are of a higher quality (Carpenter 2007). In a follow up study using audit managers, Hoffman

and Zimbelman (2009) found evidence that group fraud risk brainstorming can also lead to greater

adjustments to the nature, extent, and timing of standard audit procedures in high fraud risk situations.

Finally, using responses to auditor field surveys, Brazel et al. (2010) found additional evidence that high-

quality brainstorming, as measured using a 21 item self-report scale, improves the relation between fraud

risk factors and fraud risk assessments. The results of this study also imply that low-quality

brainstorming can lead to under-auditing (Brazel et al. 2010). Specifically, Brazel et al. (2010) find that

the extent to which testing procedures are increased or changed in high risk situations is significantly

lower for lower quality brainstorming sessions relative to high quality brainstorming sessions. The

implication of these results underscore the importance of identifying techniques that improve

Page 9: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

8

brainstorming quality. Developing a better understanding of how individual differences in trait

professional skepticism affect the effectiveness and efficiency of fraud risk brainstorming represents an

important opportunity to assist in this endeavor.

Brazel et al. (2010) propose a model of fraud brainstorming quality that indicates that the three

primary inputs of brainstorming quality are attendance and communication, brainstorming structure and

timing, and engagement team effort. Attendance and communication addresses who is present in the

brainstorming session and asserts that as more members of the audit team attend and engage in the

brainstorming session, brainstorming quality is improved due to greater diversity of thought.

Brainstorming structure and timing addresses the importance of holding the brainstorming session early in

the planning process in order to maximize improvements to individual auditor fraud judgments by

minimizing the possible negative effects of increased time pressure. Engagement team effort asserts that

greater team member preparation before, and participation during, the brainstorming session can

significantly improve brainstorming quality. Overall, this model suggests brainstorming quality impacts

the identification of fraud risk factors and hypotheses, which in turn causes the auditor to make changes to

the fraud risk assessment, which ultimately leads to changes in audit testing. Within this framework, the

present study seeks to understand how individual differences in trait professional skepticism of the

participants within a fraud brainstorming session can produce differential outcomes on brainstorming

quality. If individual differences in trait professional skepticism within a fraud brainstorming group

impact brainstorming quality, this represents a potential factor associated with the attendance and

communication dimension of fraud brainstorming quality articulated in Brazel et al. (2010).

Brazel et al. (2010) point specifically to the importance of holding the session early in the

planning process when considering the importance of brainstorming structure and timing in determining

overall brainstorming quality. In addition to timing, previous research also demonstrates that the method

used to conduct brainstorming can significantly impact brainstorming quality. For example, previous

research finds that brainstorming teams that are provided with brainstorming discussion guidance

Page 10: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

9

generate more fraud hypotheses than those with no such instruction (Trotman, Simnett, Khalifa 2009).

Furthermore, Lynch et al. (2009) find that overall brainstorming performance can be improved if

conducted using electronic brainstorming rather than a face-to-face configuration. However, Brazel et al.

(2010) document that in practice the overwhelming methodology used to conduct fraud brainstorming is a

face-to-face configuration despite the possible benefits of electronic brainstorming.

Professional Skepticism

Professional skepticism refers to “an attitude that includes a questioning mind and a critical

assessment of audit evidence” (AS 1015). Professional skepticism can be viewed as a lens through which

auditors evaluate evidence and risk throughout the audit process. This questioning attitude and behavior

is “essential to the performance of effective audits” and “is required in every aspect of every audit by

every auditor working on the audit” (Baumann 2012). The consistent application of professional

skepticism throughout the audit process has become a topic of increasing concern to regulators as

“PCAOB inspections have identified numerous audits with deficiencies where auditors did not

consistently and diligently apply professional skepticism” (Franzel 2013). Further, “this issue has been of

such prevalence that [the PCAOB] ha[s] identified the apparent failure to appropriately apply professional

skepticism as a systemic quality control issue in some firms” (Franzel 2013). In response to this concern,

recent accounting research has sought to model the determinants of professional skepticism in order to

develop strategies that positively impact the auditors’ application of skeptical judgments and decisions

throughout the audit process.

Recent research indicates that the determinants of professional skepticism should be split into two

distinct categories: skeptical judgments and skeptical actions (Nelson 2009; Hurtt 2010; Hurtt et al.

2013). Skeptical judgments refer to the auditor’s ability to recognize a potential problem, whereas

skeptical actions are those taken by auditors once a potential problem has been identified. Nelson (2009)

asserts that skeptical judgments and actions are affected by several factors including the auditor’s

incentives, traits, knowledge, audit experience and training, along with the engagement’s evidential input.

Page 11: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

10

Hurtt et al. (2013) expand upon this model by categorizing these inputs into four categories: auditor

characteristics, evidential characteristics, client characteristics, and external environmental characteristics.

In this study we examine the interplay between the auditor characteristic of inherent trait professional

skepticism and the external environmental characteristic of group fraud brainstorming. Most prior

research concerning professional skepticism has focused on attempting to manipulate decision inputs in

order to observe how these inputs affect auditor skeptical actions. For example, research has shown that

auditors who prefer to frame a hypothesis in either a confirmatory or error-framed manner follow

different patterns of subsequent information searches (McMillan and White 1993). Previous research also

finds evidence that varying auditor incentives by manipulating partner emphasis on aspects of the

likelihood of fraud and the importance of professional skepticism produces differential skeptical actions

(Carpenter and Reimers 2013; Harding and Trotman 2011).

In contrast to research that attempts to evaluate auditor professional skepticism through decision

outcomes, recent research has attempted to measure trait professional skepticism directly. As outlined by

Nelson (2009), there are two primary conceptualizations of professional skepticism within the accounting

literature. These divergent conceptualizations have led to two distinct perspectives on how to measure

and evaluate trait professional skepticism. Nelson (2009) supports a presumptive doubt view of

professional skepticism in which an auditor demonstrating high professional skepticism needs relatively

more persuasive evidence to conclude an assertion is correct relative to the norm. In contrast, Hurtt

(2010) conceptualizes a neutral view of professional skepticism in which an auditor suspends judgment

until substantial evidence is obtained to reach a conclusion. She defines professional skepticism as a

multidimensional individual characteristic that can be categorized as both a trait and a state. Under this

conceptualization, Hurtt (2010) draws on prior research on skepticism from auditing, psychology,

philosophy, and consumer behavior literature in order to develop six individual characteristics that shape

an individual’s trait professional skepticism. These characteristics include a questioning mind, a

suspension of judgment, a search for knowledge, interpersonal understanding, self-esteem, and autonomy.

Page 12: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

11

Using these characteristics as a guide, Hurtt (2010) develops a scale to measure an auditor’s trait

professional skepticism. Subsequent studies using Hurtt’s professional skepticism scale find evidence of

a link between an individual’s professional skepticism as measured using the Hurtt Scale and skeptical

judgments and actions (Quadackers et al. 2009; Hurtt et al. 2008).

While there is currently no direct measurement of an individual’s inherent presumptive doubt as

conceptualized by Nelson (2009), recent research operationalizes presumptive doubt as reduced

interpersonal trust and uses the Rotter Interpersonal Trust (RIT) scale from the psychology literature to

measure auditor trait professional skepticism (Quadackers et al. 2014). Results of this study indicate that

both the RIT scale and the Hurtt scale are equally predictive of auditor skeptical decisions in low risk

situations, but in high risk situations the RIT has greater impact on auditor skeptical judgment and

decisions (Quadackers et al. 2014). However that study does not capture the correct level of skeptical

actions the auditors should take in response to the risk present in the high risk case.

Furthermore, no research to date has considered how individuals with higher levels of trait

professional skepticism, as measured using either scale, behave in low risk settings relative to high risk

settings. It may be the case that while those scoring higher on the trait PS scales demonstrate more

skeptical judgments and actions in high risk situations, they fail to adequately adjust downwards their

judgments of the risk of fraud in lower risk scenarios. Thus, it may be the case that what the scales really

capture are those individuals who consistently judge the risk of fraud to be high even when the risk is

really quite low. To this end, these scales may capture auditor effectiveness at the expense of efficiency.

To the extent individual differences in trait professional skepticism produce differential outcomes in

terms of skeptical judgments and actions, the mix of individuals within a fraud brainstorming session may

have a significant impact on the quality of fraud risk assessments and responses generated by the session.

Hypothesis Development

Page 13: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

12

As previously stated, the accounting literature documents evidence of a link between an auditor’s

inherent level of trait professional skepticism (PS) and skeptical judgements and actions. The research

also finds fraud brainstorming can lead to increased audit quality through better identification of fraud

risk factors and hypotheses, and appropriate changes to fraud risk assessments. What is unclear, however,

is whether the trait PS of individuals within a brainstorming group can create differential outcomes in

brainstorming quality. The results of group brainstorming represent a synthesis of the ideas generated by

the individual members of the group. A primary benefit of brainstorming is posited to be the stimulation

and synergy created by the group dynamic. If auditors with elevated levels of trait PS are able to provide

incremental insight into the risk of fraud within a given engagement, then it seems reasonable that

targeted inclusion of these individuals in fraud brainstorming sessions could improve overall fraud

brainstorming quality. Indeed, consistent with this expectation, prior research documents that the

presence of minority viewpoints within a group can improve overall group performance (Wood et al.

1994; McLeod et al. 1997; De Dreu and West 2001).

Because auditors with higher levels of trait PS form more skeptical judgments relative to auditors

with lower levels of the trait (Hurtt et al. 2008, Quadackers et al. 2009), their inputs to brainstorming are

likely to be perceived as more extreme than their less skeptical counterparts in high risk situations. This

increased extremity represents a minority viewpoint. Prior research finds that the expression of minority

opinions within a group can stimulate divergent thinking and widen the scope of a group’s search for

solutions (McLeod et al 1997). Furthermore, the presence of a minority viewpoint can prevent premature

movement towards consensus and promote cognitive complexity (De Dreu & West 2001, Martin et al.

2007). In essence, the presence of a minority viewpoint within a group can encourage the group as a

whole to think more deeply about the matter being discussed. This deeper consideration is thought to not

only increase the amount of time spent contemplating the matter amongst group members, but also to

increase the diversity of factors considered (McLeod et al 1997). Thus, we suspect that brainstorming

groups that contain high trait PS auditors may perform better because their inclusion encourages the

Page 14: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

13

group to develop a more complete set of potential fraud hypotheses. Furthermore, groups which identify

a greater number of fraud hypotheses are likely to evaluate the risk of fraud as higher (Hammersley,

Bamber, and Carpenter 2010). Stated formally, our first set of hypotheses assert:

H1: Brainstorming audit teams that include at least one individual with high trait professional skepticism will produce higher fraud risk assessments than audit teams with no individuals with high trait professional skepticism in high fraud risk situations.

H2: Brainstorming audit teams that include at least one individual with high trait professional skepticism will identify more fraud hypotheses than teams with no individuals with high trait professional skepticism in high fraud risk situations. These hypotheses assume that social comparison processing does not influence the participation

behavior of high trait PS individuals within the group. If participants with high levels of trait PS engage

in social comparison processing, then these individuals may reduce the quality of their participation in the

group to match their less skeptical counterparts. Thus, if social comparison processing negatively affects

the behavior of individuals with high trait PS scores in groups when they are outnumbered by those with

low levels of trait PS, it is also possible that if the majority of the group does not possess high trait PS, the

potential cognitive stimulation gained by including an individual with high trait PS in the group may be

reduced or lost completely. However, if the synergistic properties of group brainstorming allow the

minority views of auditors with high trait PS to raise the skeptical awareness of the group, it is not clear

that every participant within a fraud brainstorming group must have high levels of trait PS in order to

produce the best brainstorming results. Given these alternative possibilities, we explore whether, given

that at least one group member has high trait PS, a higher number of group members with high trait PS

improves the quality of group brainstorming. To examine this relationship, we develop our first research

question:

RQ1: Given that at least on member in the group has a high trait PS score, does a relationship exist between the number of group members with high levels of trait professional skepticism and the quality of group brainstorming outcomes as measured by the assessed risk of fraud and the quantity and quality of fraud hypotheses generated?

Page 15: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

14

While the effectiveness of fraud risk brainstorming is the primary focus of auditing standards,

efficiency is also an important concern for audit firms. Examination of the percentage of firms

investigated by the SEC indicates that the actual risk of fraudulent financial reporting among publicly

traded companies remains quite low, and determining the level of work to be performed during an audit

represents a tradeoff between reaching a sufficient level of assurance and controlling costs (Beasley,

Carcello, Hermanson, and Neal 2010). Given that the risk of fraud is in actually quite low, an auditor

must not only ensure that their audit program reaches a sufficient level of assurance, but must also control

costs in order retain clients.

If the synergistic properties of group brainstorming allow auditors with high trait PS to raise the

skeptical awareness of the group in high fraud risk situations, the question remains as to what, if any,

impact this might have on the group in lower fraud risk situations. If group brainstorming allows a single

individual with higher levels of trait PS to increase group fraud risk assessments in high risk settings, then

it seems possible this would lead to similar effects in lower fraud risk situations. However, auditors with

high trait PS may be able to appropriately differentiate between low and high fraud risk in their decision

making. If this is the case, then there would be no reason to suspect efficiency to be compromised in

lower fraud risk situations. However, little research to date considers the association between audit

efficiency and the level of an individual’s inherent trait PS.

To consider whether efficiency differences exist between groups that have members with high

levels of trait PS versus those that don't, we consider fraud risk brainstorming outcomes in a lower fraud

risk setting. If groups that include members with high trait PS become too skeptical due to the synergistic

nature of brainstorming, then these groups would tend to over-estimate the risk of fraud when fraud risk is

lower. Since it is difficult to predict what effect heightened trait PS will have in a lower fraud risk

situation, we pose a research question with no prediction:

RQ2: What impact does heightened individual trait professional skepticism of members within a brainstorming group have on the efficiency of fraud risk brainstorming in lower fraud risk settings?

Page 16: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

15

Figure 1 presents a graphical representation of the operationalization of our hypotheses and

research questions.

INSERT FIGURE 1 HERE

RESEARCH DESIGN

To test our research hypotheses and questions we employed a 3 (group type) X 2 (case type)

between-subjects design. The three types of brainstorming groups included three-person groups with no

members with high trait PS scores, groups with one member with a high trait PS score, and groups with

two members with high trait PS scores. The two levels of the case type included a higher fraud risk case

and a lower fraud risk case.

The two versions of the case were based on the Helecom Communications case developed by

Ballou and Mueller (2005b). The original case contains a detailed and realistic description of a client

organization with numerous fraud risk indicators related to both the client and the client’s industry. This

case was designed to be used in undergraduate and graduate auditing courses with an emphasis on fraud

risk assessments or assurance.

Participants

The participants were 252 Master of Accounting accounting students in eighty-four groups

consisting of three students per group.4,5 Accounting students are appropriate participants for this

experiment for several reasons. First, the study focuses on individual differences, or traits, and these have

4 The student participants were drawn from multiple advanced auditing courses at three major universities. Sixty-eight percent of the participants reported that they had completed at least one accounting-specific internship. Furthermore, 46% reported completing an auditing specific internship. All students received the same instructions concerning PCAOB AS 2401 and AS 2110 fraud risk brainstorming requirements prior to the experiment. 5 271 students participated in both days of the experiment. Of these participants, 255 students were assigned to a group of 3 on the second day of the experiment. One group of 3 was later determined to be an outlier. The three participants from this outlier group were excluded from all individual-level analyses in addition to the group level analyses.

Page 17: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

16

been shown to be relatively stable across time (Roberts & DelVecchio 2000). Second, while advanced

accounting students may lack auditing experience, they have gained from their academic studies

knowledge of the audit process, fraud risk, and the notion of professional skepticism. Thus, students

should have enough relevant knowledge about, and interest in, the subject matter to productively

participate in group brainstorming, especially given that the case was designed for audit students (Ballou

and Mueller 2005b). Furthermore, previous auditing studies which examine the relative performance of

groups in a fraud brainstorming task have utilized students as subjects using the same case study

employed in the current experiment (Lynch et al. 2009, Hartt 2014).

The experiment was conducted over two class periods with a one day break between classes. On

the first day of the experiment, PCAOB AS 2401 and PCOAB AS 2110 fraud risk brainstorming were

covered. Participants also completed the Hurtt Scale (Hurtt 2010) and the RIT Scale (see Quadackers

(2009)). They were given case materials to read before returning to class to complete the second phase of

the experiment. Participants were randomly assigned to receive one of the two case types (higher versus

lower risk) and were specifically instructed not to discuss the materials with anyone outside of class. The

participants were instructed to review the case materials prior to returning to complete the second phase

of the experiment and were made aware of the fact that the first task they would face upon returning

would be a quick five question quiz designed to test their familiarly with the details of the case. All

participants were offered extra credit for their participation in the study.

We computed participants’ trait PS scores using the Hurtt scale. The standardized mean score was

73.57 (s.d. 7.23), which is not significantly different than the average student score reported in Hurtt

(2010).6 Previous research does not document a scale score that is indicative of high levels of trait PS, so

for the purposes of this study, those scoring at least half a standard deviation above the mean were

designated as having high levels of trait PS. While this represents an arbitrary cut point, it does provides

6 The unstandardized mean score was 132.43 and the standard deviation was 13.01. A two-sample t-test revealed an insignificant difference between the average score obtained in this study and the original mean score obtained in Hurtt (2010). There was also a non-significant difference in mean scores across students from the three universities that comprise our sample population.

Page 18: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

17

a more stringent definition than using a simple median split to separate participants into high and low

professional skepticism, and also enabled an appropriate split of participants into experimental groups.7

Students were then randomly assigned to one of three group types using a stratified sampling technique.

The first group type, the “No PS” group was assigned no participants designated as high in trait PS. The

second group type, the “One-High PS” group was assigned one participant designated as high in trait PS.

The third group type, the “Two-High PS” group was assigned two participants designated as high in trait

PS.8

Case Materials and Experimental Procedures

We constructed two versions of the case study used in this experiment based on the Helecom

Communications case developed by Ballou and Mueller (2005b). For the higher risk condition the

Helecom Communications case was used without any adjustments. However, adjustments were made to

the case for the lower risk condition. For the lower risk case, the narrative description of the firm was

adjusted to remove most of the client-specific indicators of high fraud risk. For example, in the higher

risk case participants were told that this was the first year of the engagement. In contrast, in the lower

risk case participants were informed that this was the fifth year of the engagement. As another example,

in the higher risk case participants were informed that the CEO was also the chairman of the board and

that the board was comprised of friends and family members of the CEO. This is contrasted with the

details of the lower risk case, in which participants were told that the chairman of the board was

completely independent from the CEO and that the board was comprised of several highly respected and

experienced individuals within the industry. Table 1 presents a list of the differences in risk

characteristics between the lower risk case and the higher risk case.

7 An analysis after the fact revealed that the cutoff for placement in the high PS category was in the 70th percentile of the sample as a whole. 8 Because the data was collected over multiple courses, what constituted half a standard deviation above the mean shifted slightly from our initial data collection to our final data collection. This affected the assignment of 3 participants between high and low professional skepticism, which in turn affected the group type for three observations. These three observations were re-classified into the appropriate group type based on the final calculation of our high skepticism cut-off score. Excluding these groups from the group level analyses does not materially affect the interpretation of any of our results.

Page 19: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

18

While the details of the case were altered in the lower risk case, the lengths of the descriptions

between the two versions were not materially different9. Furthermore, the original case also included a

condensed set of comparative financial statements. The financial statements were not altered between the

two cases except for one footnote related to related-party transactions that was excluded from the lower

risk case. Other than the changes described here, the case materials and information provided in both

versions of the case were identical.10 Furthermore, the participants were not made aware of the fact that

there were multiple versions of the case study, but were told that it was important not to discuss the case

between experimental sessions.

INSERT TABLE 1 HERE

271 students completed both phases of the experiment.11 Three participants were assigned to

each group. This resulted in 84 useable group samples. See Table 2 for a breakdown of the number of

groups by experimental condition.

INSERT TABLE 2 HERE

At the beginning of the second phase of the study, participants were given a brief quiz. The quiz

questions addressed simple details of the case and were designed to ensure each participant had reviewed

the case before the beginning of the experiment. The average quiz score was a 97.98, indicating that

participants took the task seriously and had adequately reviewed the case. Stage 1 of the experiment

began immediately following the completion of the quiz.

9 The narrative descriptions of both cases were approximately 4.5 pages long. 10 The experimental instrument is available from the authors upon request. 11 Analyses completed at the individual level pre-brainstorming are based on 267 students who completed both stages of the experiment and were not excluded for being either an outlier (3 observations) or for failing to complete the initial individual fraud risk assessment (1 observation). Students who were not assigned to a 3 person group brainstormed in smaller groups, but these observations were not included in the group level analyses.

Page 20: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

19

In Stage 1, participants were given a new copy of the case along with an initial fraud risk

assessment form. They were given 15 minutes to review the case materials and complete the initial fraud

risk assessment. Participants were asked to separately evaluate the significance of the fraud risk, the

pervasiveness of the fraud risk, and the likelihood of the risk using a series of 10-point Likert scales.12 At

the completion of Stage 1, each participant’s initial fraud risk assessment was collected by the

experimenter, however each participant was allowed to keep his copy of the case materials.

In Stage 2 of the experiment the participants were given their group assignments and asked to

relocate to their designated group area. One member of each group was randomly assigned to act as the

group’s recorder. Each group was instructed to brainstorm as a team to develop a list of fraud hypotheses

and complete a group fraud risk assessment. The groups were given 30 minutes to complete this phase of

the experiment. Following the completion of the brainstorming session, all materials were collected from

each group.

At the conclusion of Stage 2, participants returned to their original seats to provide a final fraud

risk assessment. Participants were told that they should not complete this task based on their initial

individual responses or their group’s responses, but rather they should complete the risk assessment based

on their current personal judgments. Participants concluded the study by completing an exit

questionnaire.

Dependent Measures and Data Coding

We calculated dependent measures at the group level from the group fraud risk assessments and

the fraud hypotheses each group generated. Dependent measures are (1) the assessed risk of fraud, and

(2) the number of unique fraud hypotheses generated. To collect each group’s fraud risk assessment, we

used the same series of 10-point Likert scale questions from the individual initial fraud risk assessments.

To measure the total number of unique fraud hypotheses generated by each group, two coauthors

with prior audit practice experience who were blind to the group condition independently coded each item

12 See PCAOB AS 2110.59 for a discussion of the three attributes of risk.

Page 21: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

20

in both the higher and lower risk case conditions. Coders used a case solution based on the teaching notes

provided by Ballou and Mueller (2005a) as a guide in order to determine whether each item recorded by a

group constituted a fraud hypothesis. Cohen’s kappas for both the higher and lower risk conditions

(kappa = 0.89, 0.94) indicates a high degree of initial inter-rater reliability. Once initial independent

coding was complete, all differences were reconciled between the two coders.

RESULTS

Preliminary Analyses

To determine whether our manipulations succeeded, we first examined whether the case study

condition (higher versus lower risk), and trait PS cutoff (high PS versus low PS) affected the level of

assessed fraud risk in the participants’ initial individual fraud risk assessments. The individual participant

summary statistics are reported in Panel A of Table 3. We utilized a 2X2 full-factorial MANCOVA to

evaluate whether the level of assessed fraud risk for the three attributes of risk (likelihood, significance,

pervasiveness) significantly differed across case type (higher or lower risk) and trait PS level (high versus

low PS).13 The results are reported in Panel B of Table 3 and provide strong evidence that the case study

manipulation (higher versus lower risk) succeeded. Specifically, the main effect of case type was

statistically significant (F3, 259 = 10.67, p < 0.01) and all follow-up univariate ANOVAs for each attribute

of risk were significant at the p <0.05 level. We further observed a significant main effect for PS level

(F3, 259 = 3.85, p = 0.01) and again all follow-up univariate ANOVAs for each attribute of risk were

statistically significant at the p < 0.05 level.14 High trait PS participants evaluated the risk of fraud as

higher than participants with lower levels of trait PS. The interaction was non-significant indicating that

the relationship between the assessed risk of fraud for the three attributes of risk and trait PS level does

not significantly differ across case conditions.

13 School (one of three universities) is included as a covariate in all analyses. 14 The F-stats and p-values reported in the results section for all MANCOVA analyses are based on Pillai-Bartlett Trace as it tends to be the most conservative test statistic.

Page 22: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

21

INSERT TABLE 3 HERE

Primary Analyses

Our first hypothesis predicted that brainstorming audit teams that included at least one individual

with high trait PS would assess the risk of fraud as higher in the higher risk case relative to groups that

included no high trait PS participants. To address this hypothesis we employed a 3X2 full-factorial

MANCOVA to examine the effects of group type (No High PS, One High PS, Two High PS) and case

type (lower versus higher risk) on group risk assessments for the three attributes of risk.15 The results of

this analysis are reported in Panel B of Table 4. An examination of the MANCOVA indicates a

significant main effect for case type (F3,74 = 2.94, p = 0.04), and a significant main effect for group (F6, 150

= 2.31, p = 0.04) with a non-significant interaction. To address H1, we used a planned contrast to

examine whether groups which included at least one high PS participant evaluated the risk of fraud as

higher than groups which did not contain at least one high PS participant. The results of this analysis are

presented in Panel C of Table 4. The multivariate planned contrast was statistically significant (F3, 74 =

3.50, p = 0.02) as were all of the follow-up univariate ANOVAs. Thus, consistent with H1, including at

least one high PS participant in the group significantly increases the group’s perceived risk of fraud.

INSERT TABLE 4 HERE

Our second hypothesis predicted that groups with at least one high trait PS member would

identify more fraud hypotheses, and more relevant hypotheses, relative to groups with no high PS

participants in the higher risk case. We utilized a Poisson binomial regression to address this hypothesis

in which the dependent variable was total number of unique hypotheses generated by the group. The

independent variables of interest was group type (No High PS, At Least One High PS) and case condition

15 The group level summary statistics for the group risk assessments are reported in Panel A of Table 4.

Page 23: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

22

(lower risk, higher risk). The results of these analyses are reported in Table 5. An examination of the

regression in Panel B of Table 5 indicates that while the regression coefficient was in the expected

direction, the result was not statistically significant (Coef. = 0.17, p = 0.22). Inconsistent with H2, groups

that contained at least one high PS participant did not identify more unique fraud hypotheses relative to

groups which contained no high PS participants. These data indicate that while including high PS

participants in the group increases the group’s perceived risk of fraud, it does not translate into an

improved ability to identify specific fraud hypotheses.

INSERT TABLE 5 HERE

Research Question 1 addressed whether a relationship exists between the number of high trait PS

members within a brainstorming group and the outcomes of the group’s brainstorming session. To

address this research question we return to the 3X2 group level MANCOVA presented in Table 4 and use

a planned contrast to examine whether there are significant differences in fraud risk assessments across

groups that contained only one high PS participant and groups that contained two high PS participants.

The results of this analysis are presented in Panel E of Table 4. The results of the multivariate planned

contrast indicate a non-significant difference between these two group types (F3, 74 = 1.15, p = 0.33).

These data indicate that it only requires the targeted inclusion of at least one high PS participant in the

group to significantly increase the skeptical awareness of the group as a whole, and the inclusion of

additional high PS members has no marginal benefit.

Our second research question speaks to whether there are any significant efficiency differences

between groups containing high trait PS members and those that do not in lower risk situations. To

address this research question we returned to our fraud risk assessment 3X2 group level MANCOVA.

Because there was a main effect for both the case condition, and the group condition, but no significant

interaction, we conclude that including high PS participants in the group does not appear to cause the

Page 24: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

23

group to systematically over-estimate the risk of fraud. That is, while groups which contain high PS

participants appear to systematically evaluate the risk of fraud as higher on average, it appears that the

groups are still able distinguish between high and low risk situations.

Supplemental Analyses

Best Member Analysis

Previous research in the group decision-making literature finds consistent evidence that the

group’s ability to identify the best member within the group is a significant predictor of group

performance (Libby, Trotman, & Zimmer, 1987). This suggests that a potential driver of a relationship

between the inclusion of high PS participants and higher levels of brainstorming quality is driven by the

group’s ability to accurately identify the member with the highest level of trait professional skepticism.

To investigate this possibility, we examined the frequency with which participants chose high trait PS

group members as the best member of the group.

In the post-experimental questionnaire, we asked participants to identify the best member of their

group. We then focused on the One-High PS groups and used mean proportion testing to examine

whether the high trait PS person was statistically more likely to be chosen as the best participant. The

high trait PS participant was chosen by group members in the One-High PS group as the best member of

the group 62% of the time, which is significantly higher than what would be expected based on chance

alone (z = 4.98, p < 0.01). We further broke it down by case type to see if there was a significant

difference between the likelihood that the high trait PS person was chosen as the best member of the

group for both the higher and lower risk case. The high PS participant was significantly more likely to be

chosen as the best member of the group regardless of whether the group was in the higher risk case

(z=2.63, p = 0.01) or the lower risk case (z = 4.44, p < 0.01). Lastly, we broke this analysis down by case

type and PS level to examine whether participants with high or low levels of professional skepticism

varied in their assessment of who was the best member of the group. The results indicate that participants

Page 25: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

24

routinely evaluated the high PS participant as the best member of the group (z stats ranged from 3.92-

1.84, p value from <0.01 – 0.07).

Effect of Skeptical Groups on Auditors with Low Trait Skepticism

Another potential benefit of brainstorming is that it may enable auditors with lower levels of PS

to more effectively evaluate risk on an individual basis post-brainstorming. Because we found that group

members identified the more highly skeptical members as the best members, and highly skeptical

members had higher initial risk evaluations, we focus our analysis of post-brainstorming risk assessment

on low PS participants. Table 6 presents the effects of group type on the final fraud risk assessments of

low PS participants.

INSERT TABLE 6 HERE

Table 6 demonstrates that although low PS participants continue to assess the high-risk scenario

as being of greater risk than the low-risk scenario (F3,163 = 6.40, p < 0.01), being grouped with a high PS

participant did not incrementally increase the final risk assessments of low PS participants. As shown in

Table 6, low PS participants who were in a brainstorming group with at least one high PS participant did

not make final fraud risk assessments that were statistically different from low PS participants without a

high PS group member (F3,163 = 1.61, p = 0.20), nor did this vary by case type (F3,163 = 0.82, p = 0.48).

Thus, even though participants identified high PS participants as the best members, and groups with high

PS participants assessed fraud risk as higher, this effect did not carry over for low PS participants post-

brainstrorming.

SUMMARY AND CONCLUSIONS

The purpose of this study was to examine how individual differences in trait professional

skepticism impact group fraud risk brainstorming. The consistent application of professional skepticism

Page 26: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

25

throughout the audit process continues to be of utmost concern to the PCAOB and audit firms. By

providing insight into how individual differences in professional skepticism affect the outcomes of group

brainstorming, this study helps to address these concerns.

The results of this study indicate that individual differences in trait professional skepticism of

group members can significantly impact the outcomes of fraud risk brainstorming at least in terms of the

group’s perceptions of fraud risk. In both high and low risk situations, groups that contain at least one

member with high trait professional skepticism evaluate the overall risk of fraud higher than groups that

do not contain any individuals with high trait professional skepticism. However, a significant difference

in risk perceptions for these groups across high and low risk conditions indicates that the targeted

inclusion of these high PS participants does not negatively impact the efficiency of the audit by causing

the group to systematically over-estimate the risk of fraud. These results suggest that the synergistic

properties of brainstorming make it possible for those with high levels of professional skepticism to

positively influence the skeptical awareness of the group. However, while the targeted inclusion of high

PS participants in the group does appear to systematically raise the skeptical awareness of the group, it

does not improve the group’s ability to identify relevant underlying fraud hypotheses. Further, low PS

participants’ final individual fraud risk assessments were not affected by the presence of high PS

participants in their brainstorming groups. Thus, further research is needed to determine the influence of

high PS auditors on post-brainstorming outcomes.

The findings presented in our study contribute not only the accounting literature, but also have

implications for practice. Regulators, practitioners, and academicians generally focus on the individual

when considering professional skepticism. However, our study examines the possibility that within a

group of auditors, not all auditors must possess high inherent levels of professional skepticism in order for

the group as a whole to make more skeptical judgements.

This study is subject to several limitations. First, we relied on a measured variable to capture

individual levels of trait professional skepticism. While previous research documents the relation

Page 27: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

26

between higher levels of trait PS as measured by the Hurtt scale and higher levels of assessed risk,

previous research has not identified an appropriate cut point for identifying those with high professional

skepticism. As such, we used a half a standard deviation above the mean to identify individuals with high

levels of trait professional skepticism. There is no guarantee that this cut point truly captures individuals

who should be labeled high in trait professional skepticism.

We used students with classroom audit experience as our subjects. Because the case we used was

designed for students and has been used in prior research that addresses fraud risk brainstorming (Lynch

et al. 2009, Hartt 2014), we believe this design choice is appropriate. Furthermore, recent research

indicates that advanced audit students may be a good proxy for first year auditors (Bennett and Hatfield

2013). However, to the extent that the relation between trait skepticism interacts with experience in

influencing fraud risk evaluations, it is possible that the phenomena documented here would not

generalize to a more natural setting. Future research is needed which examines the relationship between

trait skepticism and experience and its potential effect on group-level performance.

Our study represents a first step in examining how individual-level characteristics such as trait

professional skepticism can influence group-level performance during fraud risk brainstorming. Because

we were interested in how individual levels of trait PS might impact the group as a whole, we elected to

assemble brainstorming groups that were not hierarchical in nature. This choice gave us the ability to

examine this relationship in isolation of a group hierarchy. However, additional research is needed in this

area to understand the potential interplay between individual levels of trait PS and group performance in

hierarchical groups. For example, if audit seniors are susceptible to social loafing in electronic

brainstorming groups (Chen et al. 2014); does higher levels of individual trait PS help to counteract this

tendency? Furthermore, what influence does the trait PS level of more senior members in the group have

on the group’s performance? We believe our research provides a foundation for exploring these

important topics in future research.

Page 28: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

27

We adapted a published case study to use as the instrument in this case. Our inferences from the

data concerning the effectiveness and efficiency of these groups during brainstorming are based on the

differences between these groups. However, they do not capture the difference between the group’s

assessments and an expert’s assessment of the risk of fraud. Future research should explore in more detail

the notion of auditor efficiency and the potential downside of consistently focusing on skeptical

judgments and actions.

Page 29: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

28

REFERENCES

“AS 2110: Identifying and Assessing Risks of Material Misstatement.” 2010. Public Company Accounting Oversight Board. https://pcaobus.org/Standards/Auditing/Pages/AS2110.aspx.

“AS 2401: Consideration of Fraud in a Financial Statement Audit.” 2002. Public Company Accounting Oversight Board. https://pcaobus.org/Standards/Auditing/Pages/AS2401.aspx.

“AS 1015: Due Professional Care in the Performance of Work.” 1997. Public Company Accounting Oversight Board. https://pcaobus.org/Standards/Auditing/Pages/AS1015.aspx.

Ballou, Brian, and Jennifer M. Mueller. 2005a. Teaching Notes: Helecom Communications: Considering Fraud Risk on an Audit Engagement before and after Analyzing a Key Business Process. Issues in Accounting Education 20 (1).

———. 2005b. Helecom Communications: Considering Fraud Risk on an Engagement before and after Analyzing a Key Business Process. Issues in Accounting Education 20 (1): 99–118. doi:10.2308/iace.2005.20.1.99.

Baumann, Martin, Chief Auditor. 2012. Remarks Concerning PCAOB Developments presented at the AICPA Conference on Current SEC and PCAOB Developments, December 4, Washington, DC. http://pcaobus.org/News/Speech/Pages/12042012_AICPA.aspx.

Beasley, Mark, Joseph Carcello, Dana Hermanson, and Terry Neal. 2010. Fraudulent Financial Reporting 1998-2007: An analysis of U.S. Public Companies. Committee of Sponsoring Organizations of the Treadway Commission.

Bennett, Bradley, and Richard Hatfield. 2012. The Effect of the Social Mismatch Between Staff Auditors and Client Management on the Collection of Audit Evidence. The Accounting Review 88 (1): 31-50.

Brazel, Joseph F., Tina D. Carpenter, and J. Gregory Jenkins. 2010. Auditors’ Use of Brainstorming in the Consideration of Fraud: Reports from the Field. The Accounting Review 85 (4): 1273–1301.

Camacho, L. Mabel, and Paul B. Paulus. 1995. The Role of Social Anxiousness in Group Brainstorming. Journal of Personality and Social Psychology 68 (6): 1071–80.

Carpenter, Tina D. 2007. Audit Team Brainstorming, Fraud Risk Identification, and Fraud Risk Assessment: Implications of SAS No. 99. The Accounting Review 82 (5): 1119–40.

Carpenter, Tina D., and Jane L. Reimers. 2013. Professional Skepticism: The Effects of a Partner’s Influence and the Presence of Fraud on Auditors’ Fraud Judgments and Actions. Behavioral Research in Accounting 25 (2): 45-69.

Cicchetti, Domenic V. 1994. Guidelines, Criteria, and Rules of Thumb for Evaluating Normed and Standardized Assessment Instruments in Psychology. Psychological Assessment 6 (4): 284–90.

De Dreu Carsten K. W., Michael A. West. 2001. Minority Dissent and Team Innovation: The Importance of Participation in Decision Making. Journal of Applied Psychology 86 (6): 1191-1201.

Dugosh, Karen Leggett, Paul B. Paulus, Evelyn J. Roland, and Huei-Chuan Yang. 2000. Cognitive Stimulation in Brainstorming. Journal of Personality and Social Psychology 79 (5): 722–35.

Franzel, Jeanentte. 2013. Protecting Investors by Seizing the Opportunity to Strengthen Audit Quality presented at the American Accounting Association Midyear Conference and Doctoral Consortium, January 18, New Orleans, LA.

Page 30: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

29

Hallgren, Kevin. 2012. Computing Inter-Rater Reliability for Observational Data: An Overview and Tutorial. Tutor Quant Methods Psychology 8 (1): 23–34.

Hammersley, Jacqueline S., E. Michael Bamber, Tina D. Carpenter. 2010. The Influence of Documentation Specificity and Priming on Auditors’ Fraud Risk Assessments and Evidence Evaluation Decisions. The Accounting Review 85 (2): 547-571.

Hammersley, Jacqueline S., Karla Johnstone, and Kathryn Kadous. 2011. A Review and Model of Auditor Judgments in Fraud-related Planning Tasks. Auditing: A Journal of Practice & Theory 30.4: 101-128.

Hammersley, Jacqueline S. 2011. How do Audit Seniors Respond to Heightened Fraud Risk? Auditing: A Journal of Practice and Theory 30 (3): 81-101.

Harding, Noel, and Ken Trotman. 2011. Enhancing Professional Skepticism via the Fraud Brainstorming Discussion Outcomes. Working Paper.

Hartt, Allen. 2014. The Impact of Collective Intelligence on the Fraud Brainstorming Effectiveness of Traditional and Virtual Audit Groups. Working Paper.

Hoffman, Vicky B., and Mark F. Zimbelman. 2009. Do Strategic Reasoning and Brainstorming Help Auditors Change Their Standard Audit Procedures in Response to Fraud Risk? The Accounting Review 84 (3): 811–37.

Hooghe, Marc, Sofie Marien, and Thomas de Vroome. 2012. The Cognitive Basis of Trust. The Relation Between Education, Cognitive ability, and Generalized and Political Trust. Intelligence 40 (6): 604-613.

Hurtt, Kathy. 2010. Development of a Scale to Measure Professional Skepticism. AUDITING: A Journal of Practice & Theory 29 (1): 149–71.

Hurtt, Kathy, Martha Eining, and David Plumlee. 2008. An Experimental Examination of Professional Skepticism. Working Paper.

Hurtt, R. Kathy, Helen L. Brown-Liburd, Christine E. Earley, and Ganesh Krishnamoorthy. 2013. Research on Auditor Professional Skepticism- Literature Synthesis and Opportunities for Future Research. AUDITING: A Journal of Practice & Theory 32 (Sp. 1): 45-97.

Laidra, Kaia, Helle Pullmann, Juri Allik. 2007. Personality and Intelligence as Predictors of Academic Achievement: A Cross-sectional Study from Elementary to Secondary School. Personality and Individual Differences 42: 441-451.

Libby, R., Trotman, K. T., & Zimmer, I. 1987. Member Variation, Recognition of Expertise, and Group Performance. Journal of Applied Psychology, 72 (1): 81–87.

Lounsbury, John, Eric Sundstrom, James Loveland, and Lucy Gibson. 2003. Intelligence, "Big Five" Personality Traits, and Work Drive as Predictors of Course Grade. Personality and Individual Differences 35: 1231-1299.

Lynch, Antoinette L., Uday S. Murthy, and Terry J. Engle. 2009. Fraud Brainstorming Using Computer‐Mediated Communication: The Effects of Brainstorming Technique and Facilitation. The Accounting Review 84 (4): 1209–32.

Martin, Robin, Pearl Y. Martin, Joanne R. Smith, and Miles Hewstone. 2007. Majority versus Minority Influence and Prediction of Behavioral Intentions and Behavior. Journal of Experimental Social Psychology 43: 763-771.

McGraw, Kenneth O., and S. P. Wong. 1996. Forming Inferences about Some Intraclass Correlation Coefficients. Psychological Methods 1 (1): 30–46.

Page 31: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

30

McLeod, Poppy L., Robert S. Baron, Mollie Weighner Marti, and Kuh Yoon. 1997. The Eyes Have It: Minority Influence in Face-to-Face and Computer Mediated Group Discussion. Journal of Applied Psychology 82 (5):706-718.

McMillan, Jeffrey J., and Richard A. White. 1993. Auditors’ Belief Revisions and Evidence Search: The Effect of Hypothesis Frame, Confirmation Bias, and Professional Skepticism. The Accounting Review 68 (3): 443–65.

Morris, CPA, PhD, Jan, and William Thomas, CPA, PhD. 2011. Clarified Auditing Standards: The Quiet Revolution. Journal of Accountancy, June. http://www.journalofaccountancy.com/issues/2011/jun/20113792.

Nelson, Mark W. 2009. A Model and Literature Review of Professional Skepticism in Auditing. AUDITING: A Journal of Practice & Theory 28 (2): 1–34.

Observations on Auditors’ Implementation of PCAOB Standards Relating to Auditors’ Responsibilities with Respect to Fraud. 2007. PCAOB Release No. 2007-001.

Paulus, Paul B., and Mary T. Dzindolet. 1993. Social Influence Processes in Group Brainstorming. Journal of Personality and Social Psychology 64 (4): 575-586.

Peecher, Mark E., and Ira Solomon. 2001. Theory and Experimentation in Studies of Audit Judgments and Decisions: Avoiding Common Research Traps. International Journal of Auditing 5 (3): 193–203.

Quadackers, Luc, Tom Groot, and Arnold Wright. 2009. Auditor’s Skeptical Characteristics and Their Relationship to Skeptical Judgments and Decisions. Working Paper.

———. 2014. Auditors’ Professional Skepticism: Neutrality versus Presumptive Doubt. Forthcoming in Contemporary Accounting Review.

Roberts, Brent, and Wendy DelVecchio. 2000. The Rank-order Consistency of Personality Traits from Childhood to Old Age: a Quantitative Review of Longitudinal Studies. Psychological bulletin 126 (1).

Sturgis, Patrick, Sanna Read, and Nick Allum. 2010. Does Intelligence Foster Generalized Trust? An Empirical Test using the UK Birth Cohort Studies. Intelligence 38 (1) 45-54.

Trotman, Ken T., Roger Simnett, and Amna Khalifa. 2009. Impact of the Type of Audit Team Discussions on Auditors' Generation of Material Frauds. Contemporary Accounting Research 26 (4): 1115-1142.

Wood, Wendy, Sharon Lundgren, Judith A. Oullette, Shelly Busceme, and Tamela Blackstone. 1994. Minority Influence: A Meta-analytic Review of Social Influence Processes. Psychological Bulletin 115 (3): 323-345.

Page 32: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

31

Table 1 Differences in Fraud Risk Indicators

High Risk Case Lower Risk Case Family owned company until the public offering, and little to no change in the culture after becoming a public company.

Significant change in the culture after going public as the CEO worked hard to ensure investors were comfortable with the company's oversight.

CEO is also the chairman of the board. Chairman of the board is independent of the CEO position.

Family members/managers have little regard for other shareholders, believing that the company belongs to them.

Significant focus on ensuring investors are well informed and company decisions are in the best interests of investors.

Board of directors consists of management and friends of the family, reducing the ability of independence of mental attitudes.

Board of directors is comprised of highly respected and experienced individuals from the industry.

Voting power is concentrated among family members.

Equal voting power between all investors.

Privately owned businesses established by Helecom CEO to subsidize income, each of which generate primary revenue streams from Helecom, based on below-market prices charged to Helecom.

No related party transactions.

Purchase of Neo was justified as a project for George and structured so that Neo is not a subsidiary of Helecom

Purchase of Neo wireless was approved by the board and was structured as a subsidiary of Helecom.

The accounting firm is providing a first-year audit This is the 5th year the accounting firm has provided the audit.

Table 2 Observations by Cell

Low Fraud Risk High Fraud Risk Low PS Group 16 15 Mid PS Group 13 14 High PS Group 15 11

Page 33: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

32

1 1

3 2 3 2

4

4

1 1

3 2 3 2

4 4

H1:Brainstorming audit teams that include at least one individual with high trait professional skepticism will produce higher fraud risk assessments than audit teams with no individuals with high trait professional skepticism in high fraud risk situations.

H2: Brainstorming audit teams which include at least one individual with high trait professional skepticism will identify more fraud hypotheses, and more relevant hypotheses, than audit teams with no individuals with high trait professional skepticism in high fraud risk situations.

INDEPENDENT DEPENDENT

Operational Definition Operational Definitions

INDEPENDENT DEPENDENT

Concept Definion Concept Definition

Influence of individuals with high levels of professional skepticism

Brainstorming quality

1) Total number of unique fraud hypotheses identified by the group2) Total number of relevant fraud hypotheses identified by the group

RQ1: Given that at least one member in the group has a high trait PS score, does a relationship exist between the number of group members with high levels of trait professional skepticism and the quality of group brainstorming outcomes as measured by the assessed risk of fraud and the quantity and quality of fraud hypotheses generated?

INDEPENDENT DEPENDENT

Concept Definion Concept Definition

Influence of individuals with high levels of professional

skepticismBrainstorming quality

Operational Definition Operational Definitions

Inclusion of at least one member rated high in PS versus groups

with no members rated high in PS

Risk assessment score:* Likelihood of the risk* Significance of the risk* Pervasiveness of the risk

Influence of individuals with high levels of professional skepticism

Brainstorming quality

Operational Definition Operational Definitions

Inclusion of at least one member rated high in PS versus groups with no members rated

high in PS

1) Inclusion of at least one member rated high in trait professional skepticism2) Number of members rated high in PS included in the group

Risk assessment score: * Likelihood of the risk * Significance of the risk * Pervasiveness of the risk

Figure 1: Graphical Representation of Hypotheses and Research Questions

Concept Definion Concept Definition

Influence of individuals with high levels of professional

skepticismBrainstorming efficiency

Operational Definition Operational Definitions

Number of members rated high in PS included in the group

1) Risk assessment score: * Likelihood of the risk * Significance of the risk * Pervasiveness of the risk2) Total number of unique fraud hypotheses identified by the group3) Total number of relevant fraud hypotheses identified by the group

RQ2: What impact does heightened individual trait professional skepticism of members within a brainstorming group have on the efficiency of fraud risk brainstorming in lower fraud risk settings?

INDEPENDENT DEPENDENT

Concept Definion Concept Definition

Page 34: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

33

TABLE 3 Manipulation Check – Initial Individual Participant Analyses

Panel A: Descriptive Statistics: Marginal Means, Standard Error, Number of Observationsa High Professional Skepticism Low Professional Skepticism Mean Std. Er. Obs. Mean Std. Er. Obs. Higher Risk Likelihood of Risk 7.74 0.25 37 6.95 0.17 90 Significance of Risk 7.69 0.26 37 7.06 0.18 90 Pervasiveness of Risk 7.34 0.25 37 6.71 0.17 90 Lower Risk Likelihood of Risk 6.40 0.23 45 6.19 0.16 95 Significance of Risk 7.18 0.24 45 6.60 0.17 95 Pervasiveness of Risk 6.54 0.23 45 6.22 0.16 95 Panel B: MANCOVA Attributes of Riskb, c

Source Value df Error df F p

School 0.05 6 520 2.43 0.03 Case Type 0.11 3 259 10.67 <0.01 Professional Skepticism 0.04 3 259 3.85 0.01 Case Type X Professional Skepticism 0.01 3 259 0.98 0.40 Panel C: Simple Effect of Case Conditiond

Source Cont. Est.

Std. Er. df Error

df F p

Likelihood of the Risk 1.06 0.19 1 261 31.12 <0.01 Significance of the Risk 0.49 0.20 1 261 6.02 0.02 Pervasiveness of the Risk 0.65 0.19 1 261 11.32 <0.01 Panel D: Simple Effect of Professional Skepticism Levele

Source Cont. Est.

Std. Er. df Error

df F p

Likelihood of the Risk 0.50 0.19 1 261 6.94 0.01 Significance of the Risk 0.61 0.20 1 261 9.31 <0.01 Pervasiveness of the Risk 0.48 0.19 1 261 6.19 0.01 These data represent the average assessed risk of fraud reported by participants in their initial fraud risk assessments in Stage 1 of the experiment on day 2. All fraud risk assessments were measured using a 10-point Likert scale. The significance of the risk, the pervasiveness of the risk, and the likelihood of the risk represent attributes of risk as outlined in PCAOB AU Sec. 316, paragraph 40. a. These data represent the marginal means for high and low PS participants separated by case type. b. This analysis presents a 2X2 full factorial MANCOVA where the outcome variables are the three attributes of risk and the predictor variables are PS level and Case Type. School is included in the analysis as a covariate. c. The reported test statistics are based on Pillai-Bartlett trace as it tends to be the most conservative. d. This analysis represents the simple effect of case and the test statistics are based on univariate ANOVAs. The reference group in this analysis is the low risk case. e. This analysis represents the simple effect of professional skepticism level and the test statistics are based on univariate ANOVAs. The reference group in this analysis is low professional skepticism.

Page 35: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

34

TABLE 4 Group Level Analyses for Attributes of Risk

Panel A: Descriptive Statistics: Marginal Means, Standard Error, Observationsa No High PS Groups One High PS Groups Two High PS Groups Mean SE Obs. Mean SE Obs. Mean SE Obs. Higher Risk Likelihood of Risk 7.43 0.29 15 7.77 0.31 14 8.50 0.34 11 Significance of Risk 7.86 0.27 15 8.07 0.29 14 8.29 0.31 11 Pervasiveness of Risk 7.91 0.30 15 8.32 0.32 14 8.13 0.34 11 Lower Risk Likelihood of Risk 6.92 0.28 16 7.62 0.32 13 7.45 0.29 15 Significance of Risk 7.52 0.25 16 8.21 0.29 13 7.93 0.27 15 Pervasiveness of Risk 7.10 0.28 16 8.18 0.33 13 7.58 0.30 15 Panel B: MANCOVA Attributes of Riskb, c

Source Value df Error df F p

School 0.19 6 150 2.63 0.02 Case Type 0.11 3 74 2.94 0.04 Group Type 0.17 6 150 2.31 0.04 Case Type X Group Type 0.05 6 150 0.64 0.70 Panel C: Simple Effect of Case Typed

Source Cont. Est.

Std. Er. df Error

df F p

Likelihood of the Risk 0.58 0.23 1 76 6.22 0.02 Significance of the Risk 0.19 0.21 1 76 0.76 0.39 Pervasiveness of the Risk 0.50 0.24 1 76 4.44 0.04 Panel D: Planned Contrast No High PS Groups vs Groups with High PS Participantse Multivariate Planned Contrast

Source Value df Error df F p

No High PS Groups vs Other Groups 0.12 3 74 3.50 0.02 Follow-up Univariate ANOVAs

Source Cont. Est.

Std. Er. df Error

df F p

Likelihood of the Risk 0.66 0.24 1 76 7.43 0.01 Significance of the Risk 0.44 0.22 1 76 3.84 0.05 Pervasiveness of the Risk 0.55 0.25 1 76 4.86 0.03 Panel E: Planned Contrast One High PS Groups vs Two High PS Groupsf Multivariate Planned Contrast

Source Value df Error df F p

One High PS vs Two High PS Groups 0.05 3 74 1.15 0.33 Follow-up Univariate ANOVAs

Source Cont. Est.

Std. Er. df Error

df F P

Likelihood of the Risk 0.29 0.34 1 76 0.92 0.34 Significance of the Risk -0.03 0.27 1 76 0.01 0.92 Pervasiveness of the Risk -0.39 0.30 1 76 1.67 0.20 These data represent the average assessed risk of fraud reported by groups during the brainstorming session. All fraud risk assessments were measured using a 10-point Likert scale. The significance of the risk, the

Page 36: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

35

pervasiveness of the risk, and the likelihood of the risk represent attributes of risk as outlined in PCAOB AU Sec. 316, paragraph 40. a. These data represent the marginal means for groups separated by case type. b. This analysis presents a 3X2 full factorial MANCOVA where the outcome variables are the three attributes of risk and the predictor variables are Group Type and Case Type. School is included in the analysis as a covariate. c. The reported test statistics are based on Pillai-Bartlett trace as it tends to be the most conservative. d. This analysis presents the simple effect of Case Type using follow-up univariate ANOVAs. The outcome variables for this analysis are the three attributes of risk. The low risk case is the comparison group in this analysis, positive contrast estimates indicate groups in the high risk case evaluated the risk of fraud to be higher than groups in the low risk case. e. This analysis presents the planned contrast of the No High PS Groups versus groups that contain at least one high PS participant. The outcome variables for this analysis are the three attributes of risk. The multivariate planned contrast is presented first, followed by univariate ANOVAs for each attribute of risk. The No High PS group is the comparison group in this analysis, positive contrast values indicate that groups with at least one high PS participant evaluated the fraud risk to be higher than No High PS groups. f. This analysis presents the planned contrast of One High PS Groups versus Two High PS Groups. The outcome variables for this analysis are the three attributes of risk. The multivariate planned contrast is presented first, followed by univariate ANOVAs for each attribute of risk. The One High PS group is the comparison group in this analysis, a positive contrast value would indicate that groups with two high PS participants evaluated the fraud risk to be higher than groups with one high PS participant.

Page 37: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

36

TABLE 5 Group Level Analyses for Fraud Hypotheses

Panel A: Descriptive Statistics: Means, Standard Error, Observationsa No High PS Groups At least One High PS Groups Mean. Std. Er. Obs. Mean Std. Er. Obs. Higher Risk 2.80 2.18 15 2.32 1.65 25 Lower Risk 2.50 1.67 16 2.66 1.90 29 Panel B: Regression Unique Fraud Hypothesesb Model Statistics Wald Chi2(4 df) = 13.47, p = 0.01 Pseudo R2 = 0.04 Source Coef. Std. Er. z P Intercept 0.54 0.18 3.11 <0.01 School 0.32 0.09 3.64 <0.01 Case Type -0.14 0.17 -0.81 0.42 No High PS Group -0.16 0.20 -0.79 0.43 Case Type X No High PS Group 0.29 0.28 1.03 0.30 These data represent the average number of unique fraud hypotheses identified by groups in both the high and the low risk case. A fraud hypothesis is defined as a group idea that identifies both a method management might use to commit fraud, and the corresponding accounts or classes of transactions that would be affected. a. These data represent the means for groups separated by groups with no high PS participants and groups with at least one high PS participant in the group. b. This analysis presents a Poisson regression where the outcome variable is the total number of unique fraud hypotheses identified by the group and the predictor variables are case type and No High PS Group. School is included in the analysis as a covariate. A Poisson regression was used instead of a negative binomial regression because the LR test of alpha = 0, was not statistically significant (Chi-bar2(1 df) = 0.58, p = 0.22).

Page 38: THE ROLE OF INDIVIDUAL PROFESSIONAL SKEPTICISM IN … · 2018-10-17 · professional skepticism throughout the audit of particular import and defines professional skepticism as “an

37

TABLE 6 Effects of Group Membership on Final Fraud Risk Assessments for Low PS Participants

Panel A: Descriptive Statistics: Marginal Means, Standard Error, Number of Observationsa At Least One High PS Groups No High PS Groups Mean Std. Er. Obs. Mean Std. Er. Obs. Higher Risk Likelihood of Risk 7.86 0.20 45 7.74 0.18 39 Significance of Risk 8.10 0.20 45 8.00 0.18 39 Pervasiveness of Risk 8.06 0.22 45 7.84 0.19 39 Lower Risk Likelihood of Risk 7.44 0.20 39 6.87 0.17 48 Significance of Risk 8.05 0.20 39 7.53 0.17 48 Pervasiveness of Risk 7.49 0.22 39 7.30 0.19 48 Panel B: MANCOVA Attributes of Riskb, c

Source Value df Error df F p

School 0.21 6 328 6.29 <0.01 Case Type 0.11 3 163 6.40 <0.01 No High PS Group 0.03 3 163 1.61 0.20 Case Type X No High PS Group 0.02 3 163 0.82 0.48 Panel C: Simple Effect of Case Conditiond

Source Cont. Est.

Std. Er. df Error

df F p

Likelihood of the Risk 0.65 0.17 1 165 14.23 <0.01 Significance of the Risk 0.26 0.17 1 165 2.35 0.13 Pervasiveness of the Risk 0.56 0.19 1 165 9.05 <0.01 These data represent the average assessed risk of fraud reported by low professional skepticism participants in their final fraud risk assessments following brainstorming. All fraud risk assessments were measured using a 10-point Likert scale. The significance of the risk, the pervasiveness of the risk, and the likelihood of the risk represent attributes of risk as outlined in PCAOB AU Sec. 316, paragraph 40. a. These data represent the marginal means for low PS participants separated by case type and group type. b. This analysis presents a 2X2 full factorial MANCOVA where the outcome variables are the three attributes of risk and the predictor variables are Group Type and Case Type. School is included in the analysis as a covariate. c. The reported test statistics are based on Pillai-Bartlett trace as it tends to be the most conservative. d. This analysis represents the simple effect of case and the test statistics are based on univariate ANOVAs. The reference group in this analysis is the low risk case.