J Jaccpubpol 2015 05 006

J. Account.
Public Policy xxx (2015) xxxxxx
Contents lists available at ScienceDirect
J. Account. Public Policy

journal homepage: www.elsevier.com/locate/jaccpubpol
The effect of alternative fraud model use

on auditors fraud risk judgments
Douglas M. Boyle a, F. Todd DeZoort b, Dana R. Hermanson c,
a
University of Scranton, United States
b
The University of Alabama, United States
c
Kennesaw State University, United States
a b s t r a c t
This study evaluates how the use of alternative fraud model

practice aids affects external auditors fraud risk judgments. The
extant fraud literature, including professional audit standards,
focuses almost exclusively on the fraud triangle model (Cressey,
1973) and its components (i.e., pressure, opportunity, and rational-
ization) to explain the occurrence of fraud. However, the fraud
triangle has not been tested empirically relative to other proposed
fraud models designed to address the triangles limitations. For
example, Wolfe and Hermanson (2004) proposed a fraud diamond
model that adds individual capability to the fraud triangle as a
fourth component reecting the personal skills and attributes
needed to recognize and act on fraud opportunities in an organiza-
tion. A sample of 89 auditors in public accounting participated in a
2 2 experiment with fraud model practice aid type (fraud trian-
gle or fraud diamond) and CEO risk level (low or high) manipulated
randomly between subjects in a fraud risk assessment task. The
results indicate a signicant fraud model type effect, with auditors
evaluating fraud risk factors based on a fraud diamond practice aid
providing signicantly higher (more conservative) fraud risk
assessments than auditors evaluating fraud risk factors based on
a fraud triangle practice aid. Subsequent factor analysis provides
initial support for a revised fraud triangle that includes pressure,
Corresponding author.
E-mail addresses: boyled2@scranton.edu (D.M. Boyle), tdezoort@cba.ua.edu (F.T. DeZoort), dhermans@kennesaw.edu
(D.R. Hermanson).
http://dx.doi.org/10.1016/j.jaccpubpol.2015.05.006
0278-4254/ 2015 Elsevier Inc. All rights reserved.
Please cite this article in press as: Boyle, D.M., et al. The effect of alternative fraud model use on auditors fraud
risk judgments. J. Account. Public Policy (2015), http://dx.doi.org/10.1016/j.jaccpubpol.2015.05.006
2 D.M. Boyle et al. / J. Account. Public Policy xxx (2015) xxxxxx
opportunity, and a broader capability dimension (that includes

rationalization/attitude) as the three formative components.
2015 Elsevier Inc. All rights reserved.
1. Introduction
Stakeholders rely greatly on auditors to serve in a public watchdog role (U.S. vs. Arthur Young,
1984) and help to manage fraud risk and protect the investing public (McEnroe and Martens, 2001;
Abdel-Khalik, 2002; Ping, 2007; Arel et al., 2012). Fraud risk assessment is a critical part of the audit
process and one of the toughest challenges facing auditors (Carpenter, 2007; Carcello and Hermanson,
2008; Hogan et al., 2008; Beasley et al., 2010; Hammersley et al., 2011; Johnson et al., 2013;
Trompeter et al., 2013). The extant fraud literature in accounting focuses heavily on the fraud triangle
(Cressey, 1973) as a theoretical framework for understanding why fraud occurs and evaluating fraud
risk factors that impact nancial reporting. The fraud triangle posits that the occurrence of fraud
depends on the existence of pressure to commit fraud, an opportunity to commit fraud, and the ability
to rationalize the wrongdoing. Current auditing standards (e.g., Statement on Auditing Standard (SAS)
No. 99 (AICPA, 2002) and International Standard on Auditing (ISA) 240 (IFAC, 2010)) use the fraud tri-
angle as a basis for prescribing reasonable assurance detection responsibility and identifying fraud
risk factors for auditors assessing fraud risk in nancial statement audits.
Beyond authoritative standards, the fraud triangle also dominates fraud-related training and guid-
ance provided by professional associations (e.g., IIA et al., 2008; AICPA, 2009; CAQ, 2010), public
accounting rms (e.g., PwC, 2009; Deloitte, 2010), and textbooks (e.g., Albrecht et al., 2012). Finally,
although the empirical fraud literature reects strong interest in understanding and improving audi-
tors fraud-related judgments (e.g., Asare and Wright, 2004; Wilks and Zimbelman, 2004; Carpenter,
2007; Hogan et al., 2008; Kochetova-Kozloski et al., 2011; Johnson et al., 2013; Trompeter et al.,
2013), it reects an almost exclusive focus on the fraud triangle as a basis for evaluating auditor fraud
detection performance.
Despite the fraud triangles dominant position in the literature, questions exist about whether
the original framework can be improved for use by auditors and others involved in fraud risk man-
agement. The fraud literature includes a number of alternative theoretical models proposed to pro-
vide a more comprehensive analysis of the fraud problem (e.g., Albrecht et al., 1984; Wolfe and
Hermanson, 2004; Crowe Horwath, 2010, 2012; Dorminey et al., 2010, 2011, 2012; Kassem and
Higson, 2012). Dorminey et al. (2011, 1920) reviewed extant proposed fraud models and con-
cluded that professionals and academics have offered important insights that have gone beyond
the fraud triangle. For example, Wolfe and Hermanson (2004) introduced the fraud diamond
to extend Cresseys fraud triangle to include a capability component that prompts consideration
of individual capabilities to act on fraud opportunities and commit fraud. This capability dimension
includes such elements as expertise needed to exploit fraud opportunities, ability to coerce others
to commit or conceal fraud, and ability to lie effectively. Thus, while opportunity reects weakness
in an organizations governance and controls (characteristics of the fraud target), capability reects
the personal traits needed to identify and exploit the opportunity (characteristics of the fraud
source).
Although proposed alternatives to the fraud triangle have the potential to improve understanding
of the fraud problem, the literature lacks comparative testing among competing models. This study
provides an initial empirical test of the effects of alternative fraud model practice aids on auditors
fraud risk judgments. We are motivated by suggestion that the traditional fraud triangle model is
incomplete in its specication and that this possible misspecication could affect auditors evaluation
of critical fraud risk factors and subsequent fraud-related audit judgments. For example, Hammersley
et al. (2011) highlight continuing concern about auditors abilities to identify relevant fraud risks dur-
ing engagements in the triangle-based SAS No. 99 (AICPA, 2002) environment.
D.M. Boyle et al. / J. Account. Public Policy xxx (2015) xxxxxx 3
We compare auditor fraud risk assessments developed using a practice aid based on the fraud tri-
angle to assessments developed using a practice aid based on the fraud diamond (Wolfe and
Hermanson, 2004). The auditing literature (e.g., Eining et al., 1997; Alon and Dwyer, 2010) provides
strong evidence of the importance of fraud-related practice aid use, although comparative analysis
of competing aids is lacking. We use the fraud diamond as the comparative model in the present study
because (1) it represents a direct extension of the fraud triangle rather than a completely different
model, and (2) its capability element captures many of the individual traits and abilities found in other
alternative fraud models (see Dorminey et al., 2010, 2011, 2012; Kassem and Higson, 2012). The focus
on a fraud risk assessment task reects the importance of this judgment in the audit process and its
central role in fraud-related audit standards.
A sample of 89 auditors in public accounting participated in a 2 2 fraud risk assessment exper-
iment with fraud model practice aid type (fraud triangle or fraud diamond) and CEO risk level (low or
high) manipulated randomly between subjects. Given the importance of top management fraud risk
(Bell and Carcello, 2000; Wilks and Zimbelman, 2004; Beasley et al., 2010; Norman et al., 2010), we
manipulate CEO risk level (high or low) to evaluate the extent that fraud model effects depend on
the level of top management risk at the client.
The results indicate that fraud model practice aid type signicantly affects auditors fraud risk judg-
ments. Specically, auditors in the fraud diamond group provide signicantly higher fraud risk assess-
ments than auditors in the fraud triangle group. In addition, when CEO risk is high, auditors in the
fraud diamond group are more likely to list capability items as important risk factors. We nd that
participants fraud risk assessments are positively related to auditor trait skepticism (Hurtt, 2010).
Subsequent factor analysis indicates a three-factor solution that represents a revised fraud triangle
model with pressure, opportunity, and a broader capability dimension (that includes rationalization)
as the three formative components.
This study contributes to the literature in a number of ways. From a research perspective, the study
responds to calls to test the fraud triangle and its components (e.g., Hogan et al., 2008; Kassem and
Higson, 2012; Johnson et al., 2013; Trompeter et al., 2013). The results provide initial empirical evi-
dence that the fraud model type in a practice aid affects auditor fraud risk judgments. The results also
provide support for extending the fraud triangles rationalization component to consider additional
individual capability factors. Finally, although further research is needed to establish an optimal
fraud model to guide practice and policy, our results provide initial empirical evidence in support
of a revised fraud triangle model that includes pressure, opportunity, and a broader capability dimen-
sion as the formative factors. Specically, factor analysis results suggest a three-factor solution that
distinguishes fraud target-based opportunity from source-based capability and includes rational-
ization/attitude within the broader capability component.
The remainder of the paper is organized as follows. The next section provides background informa-
tion and develops our studys hypotheses, and the third section describes the studys method. The
fourth section presents the results of hypothesis testing and supplemental analyses. The nal section
summarizes the ndings and discusses the studys limitations and implications for research, policy,
and practice.
2. Background and hypothesis development
2.1. The role of the fraud triangle in auditor fraud detection
In the fraud triangle (Cressey, 1973), the occurrence of fraud is a function of the pressure to com-
mit, opportunity to commit, and ability to rationalize. The pressure (sometimes referred to as incen-
tive) component refers to the range of nancial and nonnancial factors that motivate a person to
commit fraud. The opportunity component focuses on weaknesses within a fraud target (e.g., poor
governance, control deciencies) that a motivated individual can exploit to commit fraud. The
rationalization component indicates that an individual must be able to nd a morally acceptable
justication for committing fraud. Cressey stated (1973, 30):
Trusted persons become trust violators when they conceive of themselves as having a nancial
problem which is non-shareable, are aware that this problem can be secretly resolved by violation
of the position of nancial trust, and are able to apply to their own conduct in that situation ver-
balizations which enable them to adjust their conceptions of themselves as trusted persons with
their conceptions of themselves as users of the entrusted funds or property.
Professional audit standards and related guidance have evolved over the years using the fraud
triangle as a guide. For example, SAS No. 99 (AICPA, 2002) is based on the fraud triangle, and the
standards Appendix A provides examples of the types of fraud risk factors that are relevant to
external auditors. Similarly, ISA 240 (IFAC, 2010) describes fraud risk factors consistent with the fraud
triangle. Beyond authoritative standards, the Center for Audit Quality (2010), Committee of
Sponsoring Organizations (COSO) (Beasley et al., 2010), and the Public Company Accounting
Oversight Board (PCAOB, 2007) provide additional fraud-related guidance or insight based on the
fraud triangle.
Although a number of studies provide general support for the fraud triangle and its use in auditing
standards (see Hogan et al. (2008) for a review), the literature (e.g., Wilks and Zimbelman, 2004;
Hogan et al., 2008; Johnson et al., 2013; Trompeter et al., 2013) emphasizes that more comprehensive
study of the fraud triangle is needed. The literature also contains very limited research on rationaliza-
tion, which Wilks and Zimbelman (2004) nd to be the most important fraud triangle component,
according to auditors. Others assert that rationalization is difcult to observe and incomplete when
evaluating individual characteristics needed for fraud to occur (Dorminey et al., 2012). For example,
SAS No. 99s use of the fraud triangle extends the rationalization component to require formal assess-
ment of management attitudes, which may be more observable than rationalization.
2.2. Extensions or enhancements of the fraud triangle
Despite the fraud triangles dominant position in the fraud literature, several alternative models
highlight the frameworks limitations in explaining why fraud occurs (Dorminey et al., 2012).
Specically, the fraud literature has focused specically on concern that Cresseys initial focus on
opportunity and rationalization is incomplete when considering characteristics of a person commit-
ting fraud. For example, Wolfe and Hermanson (2004) introduce the fraud diamond model that adds
capability as a fourth component of the fraud triangle model, recognizing that there is overlap
among the components of the triangle/diamond.1 Capability in this context reects characteristics/-
traits of the potential fraudster (the source of the fraud) that allow the person to identify, understand,
and exploit weaknesses, rather than the opportunity components focus on characteristics/traits of the
fraud target, such as weak internal controls, or the rationalization components focus on the fraudsters
thought process. The fraud diamond model describes six capability dimensions, including an individuals
position or function within the entity, intelligence/expertise, condence/ego, ability to coerce others,
ability to lie, and ability to manage stress.
Albrecht et al. (1984) evaluate the fraud problem and develop a fraud scale that complements tra-
ditional pressure and opportunity components with a personal integrity component that extends the
fraud triangle consideration of rationalization. Crowe Horwaths (2010) fraud pentagon extends
Cresseys fraud triangle to also include explicit consideration of arrogance and competence.2 Finally,
Kassem and Higson (2012) introduce a revised fraud triangle model that adds capability from the fraud
diamond and personal integrity from the fraud scale to the traditional motivation and opportunity
components.
The variety of alternative, untested fraud models and formative factors has led to calls for research
to further evaluate extensions of Cresseys original fraud triangle. For example, Johnson et al. (2013)
1
Also, see Dorminey et al. (2012, 560) for discussion of overlapping characteristics that help to explain fraud, as well as
Ramamoorti (2008, 525), who indicates that all three elements of the fraud triangle are inuenced by the fraud perpetrators
psychology.
2
Crowe (2010, 5) denes arrogance (or lack of conscience) as an attitude of superiority and entitlement or greed on the part of
a person who believes that internal controls simply do not personally apply. Competence reects an employees ability to
override internal controls, develop a sophisticated concealment strategy, and control the social situation to his or her advantage.
call for additional research in the area after nding that auditors fraud risk assessments are positively
affected by client manager narcissism. Trompeter et al. (2013) describe interest in learning more about
fraud model factors and encourage further research into individual propensities to commit fraud.
Beasley et al. (2010) also call for additional research on individual characteristics that ultimately con-
tribute to fraudulent nancial reporting, after nding that nearly 90% of accounting fraud cases in U.S
public companies involve the CEO and/or the CFO (72% involve the CEO).
2.3. Fraud risk assessment process
We evaluate whether extending the existing fraud triangle to include the fraud diamonds capabil-
ity element will affect auditors fraud risk assessments. Hammersley (2011) highlights the central role
of fraud risk assessment in the audit process and how auditors fraud risk assessments impact the nat-
ure and extent of audit testing. Specically, she develops a model of auditors fraud-related planning
tasks that begins with auditor characteristics (i.e., motivation, experience, knowledge, ability) and
fraud risk factor characteristics (general and specic) that affect auditors fraud hypothesis generation,
fraud risk assessments, and ultimately audit programs. We focus in this study on how variations in
fraud risk practice aid (based on alternative fraud frameworks) affect auditors fraud risk assessments.
Specically, we use auditors with varying characteristics (e.g., experience level, trait skepticism) to
consider the extent that alternative practice aids (based on the fraud triangle or fraud diamond) affect
their consideration of fraud risk factors and, in turn, fraud risk assessments.3
2.4. Fraud model practice aids
Our study of alternative fraud framework effects focuses on auditor use of fraud-related
practice aids given the common use of such aids by auditors in practice (Nelson and Tan, 2005).
The extant literature (e.g., Dowling and Leech, 2007; Nelson and Tan, 2005; Nelson, 2009) describes
a variety of benets associated with audit practice aid use. Bonner et al. (1996) characterize
much of the focus on audit practice aid development as designed to mitigate mismatches between
audit judgment tasks and the way that knowledge is organized in auditors minds. Dowling and
Leech (2007) interviewed audit partners and managers and found that practice aids improve audit
quality by promoting compliance with standards and rm methodologies, increasing audit efciency
and consistency across clients, improving risk management and documentation, and controlling junior
staff.
The literature also highlights a number of questions about practice aid use among auditors. For
example, Dowling and Leech (2007) highlight a number of perceived costs of practice aid use, includ-
ing the potential for overreliance and mechanistic behavior (i.e., ticking boxes rather than using judg-
ment), training time and effort, and technology instability. Hogan et al. (2008) review the
fraud-related practice aid literature and conclude that there is only limited evidence that fraud check-
lists result in better fraud risk assessments.4 Similarly, Hammersley (2011) points out that practice aids
highlighting various fraud red ags do not promote the development of specic hypotheses about how
fraud is committed. Instead, such practice aids tend to focus on more general fraud indicators, rather
than specic red ags. Ultimately, the auditing literature highlights the frequent use of practice aids dur-
ing audits, but comparative analysis of competing aids is lacking.
We posit that the use of a practice aid based on the fraud diamond will result in a more compre-
hensive evaluation of capability to commit nancial statement fraud than a traditional
triangle-based aid focused only on the personal characteristics of rationalization and attitude. In a
CEO context, the extant literature describes frequent CEO involvement in accounting fraud (Beasley
et al., 2010) and CEO pressure on subordinates (e.g., CFOs) to commit fraud (Feng et al., 2011).
Accordingly, evaluating a CEOs fraud-related capabilities should help auditors to recognize and
3
As discussed in the limitations section, we do not focus on fraud hypothesis generation or test for audit program modication
effects in this study.
4
Nelson (2009) synthesizes the literature on professional skepticism and indicates that fraud checklists may inhibit skepticism
depending on how negative items are handled in the checklist.
respond to top executive risk because capability factors should be more salient and have greater
weight when auditors consider and assess fraud risk (see Hammersley, 2011).5 Alternatively, when
such capability factors are not included in the practice aid (as when the fraud triangle framework is
used), auditors should be less likely to fully consider capability-related fraud risk factors, resulting in
lower fraud risk assessments. Stated formally:
H1. Auditors using a practice aid based on the fraud diamond will provide higher (more conservative) fraud
risk assessments than auditors using a practice aid based on the fraud triangle.
2.5. CEO risk level
We also evaluate whether fraud model effects are contingent on CEO risk level. The leadership lit-
erature in social and organizational psychology (e.g., Paulhus and Williams, 2002; Rosenthal and
Pittinsky, 2006; Eagly and Carli, 2007; Harvey et al., 2007; Chandler, 2009; Kish-Gephart et al.,
2010) describes the risks of dysfunctional and unethical leader behavior (sometimes called dark
or destructive leadership). For example, Chandler (2009) provides a theoretical framework for con-
sidering the unethical leadership risk in nancial reporting settings. This framework addresses the
dark leadership-nancial fraud risk link by describing the unethical behavior of leaders as occurring
within a complex interaction of dynamics (p. 72) affected by the leader, followers, and situational
characteristics. Leader characteristics include a variety of intrapersonal characteristics (e.g., narcis-
sism, hubris, self-deception) and interpersonal characteristics (e.g., previous abuse of power, lack of
accountability). Leaders who adopt an agentic leadership style tend to be aggressive, dominant, and
focused on personal gains (e.g., wealth, power) in a way that dominates their behavior and facilitates
dysfunctional characteristics (Paulhus and Williams, 2002). Alternatively, leaders who have a commu-
nal leadership style focus less on themselves and instead take a more collaborative, supportive, and
considerate approach that is focused on the organizations and other stakeholders interests and
well-being.
The audit literature (e.g., Bell and Carcello, 2000; Apostolou et al., 2001; AICPA, 2002; Trompeter
et al., 2013) complements the dark leadership literature by using extant fraud models to recognize
that top managements characteristics represent a major fraud risk factor. For example, top manage-
ment pressure, opportunity, and rationalization/attitude are frequently studied elements of fraud risk,
with a large body of literature nding linkages between these fraud triangle elements and fraudulent
nancial reporting (e.g., Bell and Carcello, 2000; Hogan et al., 2008; Trompeter et al., 2013). Research
also links manager narcissism with a need for recognition and amorality (Rosenthal and Pittinsky,
2006), and also with auditors fraud risk assessments (Johnson et al., 2013). Hammersleys (2011)
model of auditor judgments in fraud-related tasks suggests that concerns about unethical leadership
should affect auditors mental representations and fraud hypothesis generation, which are antece-
dents to fraud risk assessment.
To the extent that the fraud diamond model explicitly highlights top management capability to
commit fraud, we predict that fraud model type and practice aid effects will vary with the specic
level of CEO risk in the setting. For clients with high CEO risk (e.g., the CEO has a relatively agentic
leadership style that involves aggressively pushing employees to meet performance expectations, a
history of disputing proposed audit adjustments, and a high level of expertise/interest in account-
ing/auditing issues), the fraud diamonds specic focus on top management capability should increase
the salience of CEO risk and, consequently, increase auditors fraud risk assessments compared to risk
assessments based on the fraud triangle. Conversely, for clients with lower CEO risk (e.g., the CEO has a
relatively communal leadership style that involves not pushing aggressively to meet performance
expectations, a history of supporting proposed audit adjustments, and maintaining some accounting
expertise but deferring to accounting personnel for key nancial reporting decisions), explicit consid-
eration of the fraud diamond capability factors should not affect auditors fraud risk assessments as
5
As noted earlier, Johnson et al. (2013) nd that auditors fraud risk assessments are higher when client management is
narcissistic. Narcissism is related to ego, one of the elements of capability as dened in Wolfe and Hermanson (2004).
much because the capability information would not suggest a high risk of unethical CEO behavior.
Thus, we expect the effect of the fraud diamond to be greater when CEO risk is high than when
CEO risk is low.6 Stated formally:
H2. The effect of fraud model type on auditors fraud risk assessments will be greater when CEO risk is high
than when CEO risk is low.
3. Method
Our experimental design includes manipulation of fraud model type and CEO risk level randomly
between-subjects in a fraud risk assessment setting adapted from Wilks and Zimbelman (2004) and
Norman et al. (2010). The fraud model type manipulation involves providing participants with a fraud
risk factor practice aid based on the fraud triangle or the fraud diamond. CEO risk is manipulated at
high or low levels by providing alternative descriptions of the CEO, as described below.
3.1. Participants, procedure, and instrument
The studys participants are 95 practicing auditors working in two Big 4 and ve non-Big 4 public
accounting rms (reduced to 89, as discussed below). Contacts at the participating rms reviewed the
research materials, approved the study, and requested participation from colleagues.7 A majority (71%)
of the 89 participants completed an online version of the research instrument developed in Qualtrics,
while the other participants (29%) completed the instrument on paper, rather than online, at the request
of the accounting rm contact.8 All participants completed an informed consent form and then read gen-
eral instructions that told them to assume they are working on a hypothetical audit client in their current
position and rm.9 The instrument, adapted from Wilks and Zimbelman (2004) and Norman et al. (2010),
provided information about the clients background, industry, competition, management and leadership,
nancial performance, and internal controls.
3.1.1. CEO risk manipulation

In the Management and Leadership section of the materials, we manipulated CEO risk level
(CEORISK) by providing alternative descriptions of the companys CEO. In the high risk group, partic-
ipants were told:
The Companys leader is an extremely condent and persuasive CEO with a strong ego and a high
level of interest in maintaining the Companys earnings and stock price. The CEO has strong
accounting expertise and a high level of interest in accounting and auditing issues at the company,
including selection of accounting principles and determination of accounting estimates. Discussion
with client personnel reveals that the CEO has a history of aggressively pushing employees to meet
performance expectations. The CEO also has disputed proposed audit adjustments and challenged a
6
Clearly, we expect fraud risk assessments to be higher when CEO risk is higher (a main effect for CEO risk), but we do not pose
this as a formal hypothesis.
7
Similar to Cohen et al. (2011) and Hateld et al. (2010), we cannot determine the specic number of requests sent or the
response rate for the study because data collection was facilitated by contacts within the accounting rms.
8
The transition from online to paper instrument resulted in some formatting differences. For example, response scales in
Qualtrics use sliding bars on numbered scales. When creating our paper version, we followed Bierstaker et al. (2012) and asked
participants to place a slash on a scale without numbers (these were subsequently coded to numerical values based on the
placement of the slash). We nd no evidence that the two different experimental formats have any effect on the ANCOVA results or
on case understandability. For example, administration method was insignicant (p = 0.78) when added as a variable to the
ANCOVA model. Case understandability also did not differ signicantly between the online and paper groups (p = 0.95). Further, if
we delete the 26 paper instrument participants from the sample, the results are similar to those in Table 2 (Fraud Model has
p = 0.05 despite the much smaller sample size, n = 63).
9
The study and all research materials were approved by the Institutional Review Boards at the researchers universities. The
research instrument was developed in consultation with groups of auditors and audit researchers to develop case realism and
understandability. The instrument also was pretested with 52 graduate-level accounting students. The development and pretest
procedures led to minor modications to the instrument.
variety of audit issues, including scope of work, timing, and materiality. The CEO hides stress well
and has always been able to tell a strong, positive story about the Company to stakeholders (e.g.,
board of directors, shareholders) even in the face of adversity.
This CEO risk treatment reects components of extant fraud models (i.e., pressure, opportunity,
rationalization/attitude, and capability) and the interpersonal and intrapersonal characteristics
described in Chandlers (2009) conceptual framework for unethical (dark/destructive) leadership.10
For example, the CEOs extreme condence, persuasiveness, ego, and willingness to pressure others in
the organization suggest narcissism and capability to commit fraud, as well as elements of pressure
and rationalization/attitude. Capability is also represented by the CEOs expertise in accounting and
interest in accounting and auditing (needed to identify, understand, and exploit accounting fraud
opportunities), the ability to hide stress well, and the ability to tell a strong, positive story in the face
of adversity. The focus on accounting principles and estimates, the history of disputes with auditors,
and the high level of interest in maintaining the companys earnings and stock price primarily represent
the rationalization/attitude component.11 The opportunity component is implied by the CEOs leadership
(power) status within the organization, which suggests potential domination by one person.
By contrast, participants in the low CEO risk group were told:
The Companys leader is a fairly condent and persuasive CEO with a realistic interest in maintain-
ing the Companys earnings and stock price. The CEO has some accounting expertise, but defers to
accounting personnel for key nancial reporting decisions, including selection of accounting prin-
ciples and determination of accounting estimates. Discussion with client personnel reveals that the
CEO does not have a history of aggressively pushing employees to meet performance expectations.
The CEO also has supported acceptance of proposed audit adjustments and has not disputed
audit-related issues, including scope of work, timing, and materiality. The CEO acknowledges the
stress he faces and has always been focused on telling a truthful, realistic story about the
Company to stakeholders (e.g., board of directors, shareholders) even in the face of adversity.
Although this treatment also reects the components of extant fraud models, the milder descrip-
tions suggest less risk of unethical (dark) leadership and lower overall CEO fraud risk.
3.1.2. Fraud model manipulation

After evaluating client information, the participants evaluated one of two manipulated fraud model
practice aids (triangle and diamond practice aids) that asked them to assess a list of possible nancial
statement fraud risk factors for the client. The order of the risk factors was randomized for all groups,
and items were assessed using a scale ranging from No Fraud Risk (=0) to Very High Fraud Risk
(=100).12 Appendix A provides the specic factors evaluated by the two fraud model groups.
Participants in the fraud triangle group received a practice aid that included 18 fraud risk factors
focused on pressure, opportunity, and rationalization/attitude. The specic risk factors were devel-
oped from SAS No. 99 and related fraud risk studies (Wilks and Zimbelman, 2004; Norman et al.,
2010). The pressure factors focused on company growth and protability, competition, customer
demand, performance expectations, compensation, and management nancial investment in the com-
pany. The opportunity factors dealt with accounting and organizational complexity, internal control
policies and procedures, board and audit committee quality, accounting information system quality,
and management and director turnover. The rationalization/attitude factors included history of legal
10
Some of the language in the treatment is adapted from SAS No. 99 (AICPA, 2002).
11
We acknowledge that some CEO risk factors may reect dimensions of more than one fraud triangle/diamond component (see
Wolfe and Hermanson (2004) and Dorminey et al. (2012) for discussion of overlap in fraud determinants). For example, a history
of aggressively pushing employees to meet performance expectations reects both rationalization/attitude and the existence of
pressure. Also, as discussed later, we use strong accounting expertise and a high level of interest in accounting and auditing
issues as capability factors, for they relate to the knowledge of and focus on accounting that is needed to identify, understand, and
exploit accounting fraud opportunities. SAS No. 99 (AICPA, 2002) cites a related, more specic risk, nonnancial managements
excessive participation in or preoccupation with the selection of accounting principles or the determination of signicant
estimates, as a rationalization/attitude factor because management is unduly focused heavily on specic aspects of accounting.
12
For participants using the online instrument, the risk factors were uniquely randomized for each participant. For those
completing the paper version of the instrument, there was one random order used for all such participants.
and regulatory violations, top managements attitude toward achieving forecasts, excessive demands
on the auditor, history of disputes with the auditor, managements response to internal control prob-
lems, and managements ability to rationalize wrongdoing.
The fraud triangle participants also evaluated a set of six additional items that were relatively irrel-
evant to fraud risk at the client (producing 24 total items to match the length of the fraud diamond
practice aid discussed below).13 The additional items included the companys product mix, location,
number of employees, customer prole, service record, and marketing strategy. Pretest results indicated
that the six additional ller items were signicantly less relevant than the fraud triangle and fraud dia-
mond models fraud risk items.
Participants in the fraud diamond group were given a practice aid that included the 18 fraud trian-
gle risk factors and six additional factors based on the capability component introduced by Wolfe and
Hermanson (2004). The fraud diamond models capability traits include being in the right position/-
function to commit fraud, having the expertise to exploit fraud opportunities, having condence in
the ability to avoid or manage detection, having the ability to coerce others to commit or conceal
fraud, being able to lie effectively, and being able to manage stress. Our focus on the CEO in the sce-
nario operationalizes the position/function trait. We measure the expertise trait with two items, ex-
pertise to exploit fraud opportunities and interest in accounting and auditing issues, which reect
the knowledge of and focus on accounting and auditing that is needed to identify, understand, and
exploit accounting fraud opportunities.14 The measures for the four remaining capability traits include
ability to deal with stress, ability to deceive others, condence in ability to avoid fraud detection,
and ability to coerce others to commit fraud. These four measures each map directly to one of the ele-
ments of capability.
3.1.3. Additional measures

After evaluating the practice aids specic fraud risk factors, we asked participants to assess the
overall risk of nancial statement fraud (FSFRisk) for the client on a scale anchored Very Low Risk
(=0) and Very High Risk (=100). We next asked for a list of the three most important factors under-
lying the overall fraud risk assessment. Participants then indicated how much condence they have in
their fraud risk assessment (using a scale anchored No Condence (=0) and Complete Condence
(=100)).
In addition, we asked participants their perceptions of the extent that management pressure,
opportunity, and attitude/capability contribute to the clients nancial statement fraud risk.
Following a manipulation check question about CEO risk, participants completed the 30-item Hurtt
Professional Skepticism Scale (Hurtt, 2010) designed to measure individuals trait professional skepti-
cism.15 We include skepticism as a covariate in the model to control for differences in participants trait
professional skepticism, given its potential to affect fraud-related judgments (Hurtt, 2010; Hurtt et al.,
2013). We also include as a covariate the mean of each participants ratings of the 18 specic fraud risk
factors in the practice aid (pressure, opportunity, and rationalization/attitude items that all participants
13
In addition to the triangle and diamond groups, we also administered the experiment to a third group (n = 43 after two
manipulation check failures are excluded) that considered only the 18 triangle items in a shorter practice aid (versus 24 items in
each practice aid for the triangle and diamond groups in the main analysis). ANCOVA results indicate that this third reduced
triangle group has fraud risk assessments signicantly lower than the diamond group (p = 0.01) and not signicantly different
from the triangle group (p = 0.75). This third group is not considered in other analyses.
14
Wolfe and Hermanson (2004) discuss capability as being able to identify, understand, and then exploit weaknesses. They
provide an example where the CEO must focus on and understand the sales system well enough to know its weaknesses/vul-
nerabilities before backdating of sales contracts in the system will occur. Accordingly, we focus on both expertise and interest in
accounting and auditing issues as elements of capability.
15
Although the literature (e.g., Nelson, 2009; Hurtt et al., 2013) describes difculty in precisely dening professional skepticism
in an auditing context, the construct is generally considered multidimensional and related to an individuals propensity to
maintain a questioning mind and critically assess evidence. The literature also distinguishes between trait professional skepticism
as a relatively stable, enduring individual characteristic, and state professional skepticism, which reects a temporary response to
situation-specic matters. The Hurtt (2010) scale measures six dimensions of trait skepticism, including questioning mind,
suspension of judgment, search for knowledge, interpersonal understanding, autonomy, and self-esteem.
evaluated) to further control for differences across participants in the level of skepticism or
conservatism.16
After completing the Professional Skepticism Scale, the participants assessed case realism and
understandability. The results indicate that they found the research scenario to be both realistic (over-
all mean = 22.58 on a scale where Very Unrealistic = 50 and Very Realistic = 50) and understand-
able (overall mean = 27.69 on a scale where Very Difcult to Understand = 50 and Very Easy to
Understand = 50).17 The instrument concluded with classication questions related to age, gender, edu-
cation, and experience. The average completion time for the online instrument was approximately
20 min.18
4. Results
4.1. Manipulation check
To assess the effectiveness of our CEO risk manipulation, we asked the participants to identify
whether the CEO in their case was described as fairly condent and persuasive, but realistic in push-
ing employees and pursuing performance expectations or as extremely condent and persuasive,
and very aggressive in pushing employees and pursuing performance expectations. The question
had a 94% pass rate, leaving 89 of the 95 original participants for analysis after excluding individuals
who answered the question incorrectly.19
4.2. Participant characteristics
As shown in Table 1, the participants are primarily females (53.9%) in their 20s (87.6%). Consistent
with prior fraud-related studies (e.g., Carpenter, 2007; Lynch et al., 2009; Johnson et al., 2013), our
participants have varied experience, with 66.3% having less than three years of audit experience.20
The participants are primarily staff (58.4%) or seniors (30.3%), and less than half (46.6%) have a CPA
license.21 Finally, the participants typically possess a bachelors degree (or equivalent) as their highest
level of education (59.1%), and the majority of the participants work for a Big 4 rm (52.8%).22
4.3. ANCOVA results
We used ANCOVA to evaluate our hypotheses focused on the effects of fraud model type and CEO
risk on auditors fraud risk assessments. The results in Panel A of Table 2 show a signicant fraud
model type main effect (F = 8.57, p < 0.01). As predicted in H1, fraud risk assessments were higher
16
We also asked participants to assess their responsibility for detecting nancial statement fraud at this client using a scale from
0 = No Responsibility to 100 = Complete Responsibility (DeZoort et al., 2012). When added to the ANCOVA model, this variable
is not signicant (p = 0.10), and the other results are similar to those in Table 2.
17
Perceived case realism and understandability are insignicant as control variables in the multivariate analysis (p P 0.07).
Other results are similar.
18
For participants using the online instrument, the time to complete the experiment did not differ signicantly between the
triangle (mean of 19.2 min) and diamond (mean of 21.1 min), p = 0.21.
19
Three of the participants who missed the manipulation check question were in the fraud triangle group, and three were in the
fraud diamond group. The ANCOVA results are similar when including the participants who failed the manipulation check (n = 95),
and Fraud Model is signicant at p = 0.02.
20
Johnson et al. (2013, 209) note, discussions with partners and managers. . .conrmed that current practice is to include audit
team members from all experience levels in the fraud risk assessment process. Carpenter (2007) used staff, seniors, and managers
in her fraud brainstorming study. Lynch et al. (2009) used undergraduate and graduate students in their fraud brainstorming task.
In addition, Pincus (1989) participants averaged 18 months of experience.
21
Similarly, 50% of the participants in Hammersley et al. (2011) were CPAs.
22
Participant experience, age, gender, CPA status, and rm size (Big 4 or non-Big 4) are insignicant in the model (p > 0.34). A
variable for Masters degree is signicant (p = 0.03; n = 87 due to missing data and exclusion of one participant with an other
degree) in the model, with the other results similar.
Table 1
Demographics (n = 89).
Gender
Female 53.9%
Male 46.1%
Age
20s 87.6%
30s+ 12.4%
Audit experience
0 years (less than one year) 28.1%
12 years 38.2%
35 years 24.7%
610 years 5.6%
Over 10 years 3.4%
Title
Staff 58.4%
Senior 30.3%
Manager 6.8%
Partner/other 4.5%
Professional License (n = 88)
CPA 46.6%
None/other 53.4%
Highest education
Undergraduate 59.1%
Masters/other 40.9%
Firm size
Big 4 52.8%
Other 47.2%
Table 2
ANCOVA results (DV = FSFRisk).
df MS F p-Value
Panel A: Analysis of covariance
Fraud model 1 909.4 8.57 <0.01
CEO risk 1 5250.5 49.50 <0.01
Fraud model CEO risk 1 130.5 1.23 0.27
Risk factor mean 1 8159.4 76.92 <0.01
Skepticism 1 1251.6 11.80 <0.01
Fraud Triangle Fraud Diamond Total
Panel B: Descriptive statistics by group (FSFRisk)
Low CEO risk Mean 35.42 38.86 37.02
s.d. 17.1 14.11 15.7
n 24 21 45
High CEO risk Mean 58.75 67.25 63.39
s.d. 14.73 10.77 13.3
n 20 24 44
Total Mean 46.02 54 50.06
s.d. 19.76 18.88 19.6
n 44 45 89
Model F = 47.28, p < 0.001; Adj R-sq = 0.725.

Notes: FSFRisk represents perceived nancial statement fraud risk at the client measured on a 0100 scale anchored Very Low
Risk and Very High Risk. Fraud Model is coded 0 for fraud triangle and 1 for fraud diamond. CEO Risk is coded 0 for low CEO
risk and 1 for high CEO risk. Risk Factor Mean measures the participants mean response to the 18 practice aid risk factors
evaluated by all participants (pressure, opportunity, and rationalization/attitude); each item is on a scale anchored No Fraud
Risk (coded = 0) and Very High Fraud Risk (coded = 100). Skepticism represents trait professional skepticism using the Hurtt
(2010) scale.
in the fraud diamond group. The means of FSFRisk in Panel B are 46.02 for the triangle group (range of
887) and 54.00 for the fraud diamond group (range of 1591).23
Panel A of Table 2 also reveals a signicant CEO risk main effect (F = 49.50, p < 0.01), with partici-
pants in the high CEO risk group providing a signicantly higher nancial statement fraud risk assess-
ment (mean = 63.39) than participants in the low CEO risk group (mean = 37.02). However, the results
do not support H2 because the insignicant interaction between fraud model type and CEO risk level
(p = 0.27) indicates that fraud model type effects are not contingent on the level of CEO risk.24 The
covariates, Risk Factor Mean and Skepticism, are both signicant at p < 0.01 and positively related to fraud
risk assessment.25,26 The signicance of Skepticism complements an emerging body of research nding
that trait skepticism affects audit judgments in a variety of contexts (see Hurtt et al. (2013) for a review
of this literature). In this case, auditors with higher trait professional skepticism appear to view the com-
pany through a more critical lens, thus arriving at higher fraud risk assessments, even after controlling
for Risk Factor Mean.
4.4. Fraud risk factor rankings
To better understand the fraud model type results, we asked participants in the fraud triangle
and fraud diamond groups to list the three most important fraud risk factors underlying their over-
all fraud risk assessments. Three graduate students who were not otherwise associated with the
study coded the participants responses independently using a coding scheme based on the fraud
risk factors in the study. Inter-rater reliability among the three individuals was 0.93, and the
researchers reconciled all coding differences. After the coding process, we weighted the fraud risk
factors with three points assigned to the most important fraud risk factor, two points to the
second most important fraud risk factor, and one point to the third most important fraud risk
factor. These weighted scores were designed to capture the ordering of the items listed by the
participants.
Table 3 provides the weighted score results for the triangle and diamond groups. In the high CEO
risk condition, the fraud triangle participants focused heavily on management aggressiveness/attitude
and performance expectations as the most important factors underlying their fraud risk assessment.
Management interest in accounting and auditing was the only capability factor that emerged among
the top factors. In contrast, although the fraud diamond participants in the high CEO risk group also
cited management aggressiveness/attitude and performance expectations as the top two factors, three
capability factors (i.e., management interest in accounting and auditing, management coercion/per-
suasion, and management ego) also appear as top factors, suggesting that use of the fraud diamond
heightens focus on a capability-related issues.
In the low CEO risk condition, the fraud triangle participants cited performance expectations, inter-
nal control quality, and competition as the three most important fraud risk factors, while the fraud
diamond group participants cited competition, performance expectations, and management compen-
sation. Neither group included any capability factors in the top ve risk factors when CEO risk was low
(all participants had information about capability items in the case materials, but only the diamond
group was prompted to consider this information).
23
A t-test (univariate test) indicates that the mean of 54.00 is greater than the mean of 46.02 (p = 0.03 one-tailed). In addition, if
we convert the dependent variable to ranks, the results are similar to those in Table 2, with Fraud Model signicant at p = 0.02
(Shirley, 1981; Conover and Iman, 1982).
24
While the interaction term is not signicant, (a) we do nd that the difference in FSFRisk means between the triangle and
diamond groups is signicant for the high CEO Risk group (p = 0.02 one-tailed) and not for the low risk group (p = 0.23 one-tailed),
and (b) if we run a reduced ANCOVA on the high CEO risk and low CEO risk groups separately, Fraud Model is signicant in the high
risk group (p < 0.01) and not in the low risk group (p = 0.21).
25
We also considered sub-measures of Risk Factor Mean, replacing this variable with Risk Factor Mean Pressure, Risk Factor Mean
Opportunity, and then Risk Factor Mean Rationalization/Attitude. Each of these three alternate variables is signicant at p < 0.01, and
the three sub-measures are highly correlated (r > 0.50).
26
If we delete the covariates from the model, CEO Risk is signicant at p < 0.01 and Fraud Model at p = 0.05.
Table 3
Fraud risk factors listed.
High CEO risk Score Low CEO risk Score

Panel A: Fraud triangle group
Management aggressiveness/attitude 41 Performance expectations 25
Performance expectations 34 Internal control quality 25
Competition 11 Competition 24
History of problems 11 Complexity 12
Internal control quality 9 Management aggressiveness/attitude 11
Mgt. interest in accounting/auditing 9
Panel B: Fraud diamond group
Management aggressiveness/attitude 41 Competition 17
Performance expectations 26 Performance expectations 17
Mgt. interest in accounting/auditing 25 Management compensation 15
Management coercion/persuasion 16 History of problems 10
Management ego 10 Internal control quality 8
Internal control quality 10 Management aggressiveness/attitude 8
Note 1: Participants were asked to list the three most important fraud risk factors underlying their overall fraud risk assess-
ments in descending order. We weighted the fraud risk factors with three points assigned to the most important fraud risk
factor, two points to the second most important fraud risk factor, and one point to the third most important fraud risk
factor. These weightings were designed to capture the ordering of the items listed by the participants. Score represents the
total weighted score for each risk factor.
Note 2: Items in italics represent capability-related factors.
4.5. Supplemental analyses
We conduct a number of supplemental analyses to better understand the studys primary results.
First, we assess the impact of fraud model type and CEO risk on auditors condence in their fraud risk
judgments (scale from 0 = No Condence to 100 = Complete Condence). The accounting and psy-
chology literature (e.g., Shanteau and Peters, 1989; Bonner, 2008; Parker and Fogarty, 2012) recognize
the importance of evaluating condence in individual judgment and decision-making (JDM) in tasks
that lack objective criteria. The ANCOVA results (not tabulated) for Condence provide limited evi-
dence that fraud model type affects auditor judgment condence (F = 3.17, p = 0.08), with participants
in the diamond group (mean = 71.04) marginally more condent than those in the triangle group
(mean = 63.50). Other than Skepticism (p = 0.03), the variables in the ANCOVA are not signicant,
and the overall model is not signicant (p = 0.12).27
Second, we asked participants to indicate the extent management pressure, opportunities, and atti-
tude/capability each contributed to FSFRisk for this client (each 0100 scale anchored Very Little to
A Great Deal). We regressed FSFRisk on these component assessments and found (untabulated,
n = 89) that the overall model is signicant (F = 14.96, p < 0.01), with an adjusted R2 of 32.3%.28 The
coefcient for attitude/capability is positive and signicant (p < 0.01), indicating that participants per-
ceptions of the importance of top management attitude/capability were related directly to their overall
nancial statement fraud risk assessments. The coefcients for pressure and opportunities are both
insignicant (p > 0.80 in both cases).
Third, given this signicant attitude/capability result, we next evaluated the effects of fraud model
type, CEO risk, and the covariates on participants attitude/capability assessments. ANCOVA results
(untabulated) indicate that fraud model type (F = 11.79, p < 0.01) is signicant as a predictor of the
assessed importance of attitude/capability as a contributor to FSFRisk.29 The results also indicate a
27
The BreuschPagan test indicates the presence of heteroskedasticity, and the ANCOVA models condence variable is
converted to ranks (Shirley, 1981; Conover and Iman, 1982). In addition, FSFRisk and Condence are correlated (r = 0.21, p = 0.04). If
we include FSFRisk and Condence as dependent variables in a MANCOVA, Fraud Model, CEO Risk, Risk Factor Mean, and Skepticism all
are signicant at p < 0.01.
28
The variance ination factors for the variables in the regression model are all less than 1.7, indicating that multicollinearity is
not a problem.
29
The BreuschPagan test indicates the presence of heteroskedasticity. The ANCOVA models dependent variable is converted to
ranks.
Table 4
Exploratory factor analysis Fraud diamond group.
Fraud risk factor* Loadings

Pressure Opport. Capability
Pressure
Competition in Industry .813
Changes in Customer Demand .719
Performance Expectations .549
Management Financial Investment in Company .498
Management Compensation Plan .458
Opportunity
Organizational Complexity .876
Accounting Complexity .851
AIS Quality .628
Board & Audit Committee Monitoring Quality .426
Control Policies & Procedures .420
Capability/Rationalization/Attitude
Top Managements Attitude Toward Achieving Aggressive Forecasts .878
History of Legal or Regulatory Violations .853
Top Managements Ability to Coerce Others to Commit Fraud .853
Top Managements Interest in Accounting & Auditing Issues .819
Top Managements Expertise to Exploit Fraud Opportunities .812
Top Managements Ability to Deal with Stress .786
Top Managements Ability to Deceive Others .763
Top Managements Condence in Ability to Avoid Fraud Detection .750
Top Managements Ability to Rationalize Wrongdoing .642
History of Disputes with Auditor .636
Excessive Demands on Auditor .522
Top Managements Response to Internal Control Problems .469
All items are measured using a scale anchored No Fraud Risk (coded = 0) and Very High Fraud Risk (coded = 100).
Loadings > 0.50 are in bold.
*
Two items are excluded from the table, as their highest loading is <0.40: Extreme Company Growth or Protability has a
loading of 0.359 for Pressure, and Turnover of Top Management & Directors has a loading of 0.360 for Opportunity.
signicant CEO risk level main effect (p < 0.01), as well as signicant covariate effects (p < 0.05 for each
covariate).
Finally, we used exploratory factor analysis to evaluate fraud model structure and whether
rationalization/attitude represents a distinct fraud model construct (as suggested in the fraud triangle
and fraud diamond), or whether it reects a dimension of a larger capability construct. Using principal
axis factoring with direct oblimin rotation (due to the expected correlation among factors; see Wolfe
and Hermanson (2004), Ramamoorti (2008), and Dorminey et al. (2012)), the results in Table 4 for the
fraud diamond participants indicate that the items focused on SAS No. 99 rationalization/attitude
issues (based on the fraud triangle) and the items focused on the Wolfe and Hermanson (2004) capa-
bility issues (based on the fraud diamond) load on one factor with almost all item loadings exceeding
0.50.30 The three-factor solution explains 57.4% of the total variance, and the Cronbachs alpha of 0.907
indicates a high level of internal consistency among all of the items. The results also indicate that the
pressure and opportunity items load reasonably on their theorized factors. Collectively, these results sug-
gest a revised three-component fraud model with pressure, opportunity, and a broader capability ele-
ment as the components, with capability including the fraud triangles rationalization/attitude factor
and the six capability-related factors based on Wolfe and Hermanson (2004).
We also considered how the most important factors presented in Table 3 compare to the factor
loadings in Table 4 (only for the diamond group). If we combine the high and low CEO risk groups,
30
This analysis focuses only on the fraud diamond group (n = 45) because participants in this group were the only ones to
evaluate both rationalization/attitude items and capability items. Factor analysis of the fraud triangle group produced similar
factor loadings for the pressure, opportunity, and rationalization/attitude items. In addition, use of varimax rotation produced
similar results.
the overall most important risk factors from Table 3 for the diamond group are management aggres-
siveness and attitude (49 points), performance expectations (43), management interest in accounting
and auditing (27), competition (24), and management coercion/persuasion (23). The highest factor
loadings on the broader capability factor in Table 4 include three of these highest-rated items from
Table 3: management aggressiveness and attitude, management coercion/persuasion, and manage-
ment interest in accounting and auditing. Therefore, there is some consistency between Tables 3
and 4 analyses for the diamond groups focus on capability items.
5. Conclusion
The fraud literature provides a variety of models for understanding fraud risk and why fraud occurs
(Albrecht et al., 1984; Crowe Horwath, 2010, 2012; Dorminey et al., 2010, 2012; Kassem and Higson,
2012). However, authoritative standards and the empirical literature focus almost exclusively on the
fraud triangle (Hogan et al., 2008; Trompeter et al., 2013). This study provides an initial empirical test
of whether auditor fraud risk assessments are affected by the specic type of fraud model practice aid
used. The results indicate that fraud model type signicantly affects auditors fraud risk judgments.
Specically, fraud risk assessments are higher with the use of a practice aid based on the fraud dia-
mond than with a practice aid based on the traditional fraud triangle. In addition, supplemental factor
analysis provides evidence in support of a revised three-factor fraud model that includes pressure,
opportunity, and a new expanded capability factor that includes rationalization/attitude and the addi-
tional capability factors.
The studys ndings have a number of research, policy, and practice implications. From a research
perspective, the study provides an initial response to calls to test the fraud triangle on an overall basis
(e.g., Hogan et al., 2008) and provides empirical evidence that fraud model type in a practice aid affects
auditor judgments. Specically, we nd that extending Cresseys traditional fraud triangle beyond
consideration of rationalization to include broader consideration of individual capability to commit
fraud impacts fraud risk judgments. Further, our nding that rationalization/attitude and capability
items (suggested in Crowe Horwath (2010) and Wolfe and Hermanson (2004)) load on one factor that
is distinct from fraud target-based opportunity suggests that there is room for additional theoretical
renement of extant fraud models. Finally, our study contributes to the practice aid literature and to
calls for additional research on how practice aids can help auditor performance in areas most vulner-
able to audit failure (Nelson and Tan, 2005).
From policy and practice perspectives, the results highlight the impact of explicitly considering the
capability component when assessing the risk of fraudulent nancial reporting. Although all partici-
pants received background information about client capability to commit fraud, the participants using
a practice aid that explicitly prompted evaluation of capability items (e.g., interest in accounting and
auditing issues, ability to deal with stress, ability to deceive others, and ability to coerce others to com-
mit fraud) provided higher (more conservative) assessments of fraud risk. In addition, when CEO risk
was high, participants in the fraud diamond group ranked capability items as more important than did
participants in the fraud triangle group. As this area of research develops further, we encourage audit
standard setters and policymakers in accounting rms to consider the value of further developing
fraud-related practice aids and specically recognizing management capability in their guidance for
external auditors, particularly given ndings that CEOs and CFOs are implicated in the vast majority
of public company accounting fraud cases (Beasley et al., 2010).
The signicant relation between trait skepticism (Hurtt, 2010) and fraud risk assessments extends
an emerging line of research evaluating the link between professional skepticism and auditor judg-
ment and decision-making (see Hurtt et al., 2013). In the present study, trait skepticism is signicant
even after controlling for Risk Factor Mean, suggesting that skepticism affects fraud risk assessment
beyond any effects of skepticism on the assessment of the individual fraud risk indicators in the prac-
tice aid. Although this nding suggests that trait skepticism contributes to higher fraud risk assess-
ments, additional research is needed in the area to evaluate that extent that trait skepticism
impacts hypothesis generation, audit procedure choice, and fraud detection. If trait skepticism is con-
sistently found to enhance auditors evaluation of fraud risk, then audit rms may consider measuring
their personnel on this dimension and using this information in audit stafng decisions (see Farag and
Elias, 2012).
The studys implications should be considered in the context of several limitations. For example,
this study tests only two extant fraud models (i.e., the fraud triangle and the fraud diamond), and
others have indicated that fraud determinants overlap to some extent (Wolfe and Hermanson,
2004; Dorminey et al., 2012). Although the literature includes a number of other proposed fraud mod-
els (e.g., the fraud scale, fraud pentagon) that might affect auditors judgments differently than our
focal models, we selected the fraud diamond for comparative testing in this study because it
represents a relatively comprehensive model that captures the alternative factors proposed in other
models. Nevertheless, additional research is needed to test for alternative fraud model effects on audi-
tor judgments and fraud detection effectiveness.
In terms of the specic task used, the possibility exists that the capability effects reect a salience
or novelty effect resulting from participants encountering new capability-related fraud risk factors not
currently included in fraud standards and not typically present in practice aids. Future research is
needed to further evaluate whether the capability effects found in this study are sustainable after
repeated use. We do note, however, that in Table 3, the triangle and diamond groups (high CEO risk
conditions) have the same two non-capability factors as the most important fraud risk factors under-
lying their overall fraud risk assessments. Also, the capability items cited by the fraud diamond par-
ticipants are the third, fourth, and tied for fth most important risk factors. Thus, it does not appear
that the capability items, despite their novelty, dominated the more traditional fraud triangle ele-
ments. In addition, the irrelevant items evaluated by the triangle group also were novel, but they
do not appear in Table 3 at all. Overall, there does not appear to be strong evidence of a novelty effect,
but we call for additional research into this issue.
We also recognize the possibility that the six irrelevant items in the triangle group could dilute the
impact of the other items and affect the groups fraud risk assessments. However, previous research
nding dilution effects (Hackenbrack, 1992; Hoffman and Patton, 1997) used large amounts of irrel-
evant information.31 We believe our use of 18 relevant items and six irrelevant items in the triangle
group minimizes the risk of a systemic dilution effect. In addition, as discussed previously, the results
indicate that the triangle groups fraud risk assessments were similar to a group using the reduced
triangle 18-item practice aid (without the irrelevant items). Therefore, there is evidence in this study
indicating that the irrelevant items assessed by the triangle group did not dilute the relevant items.
Finally, the studys focus on relatively junior auditors limits generalizability. Although the results
do not indicate an experience effect, additional fraud model testing is needed with more experienced
auditors to extend understanding in the area. In addition, despite nding that fraud model type affects
auditors fraud risk judgments, the lack of objective judgment criteria or subsequent audit
performance measures (e.g., audit program development, testing) limits our ability to comment on
judgment quality. For example, the audit literature (e.g., Hammersley, 2011; Hammersley et al.,
2011) highlights that higher auditor fraud risk assessments can lead to audit programs modications
that reect hyper-conservatism that leads to over-auditing and inefciency. Future research should
extend study of linkages among fraud model type, fraud risk assessments, hypothesis generation,
and audit program modication to help develop an optimal fraud model that can guide research, prac-
tice, and policymaking.
Acknowledgements
We gratefully acknowledge the help we received from two anonymous reviewers, the participating
auditors, their respective rms, Chris Agoglia, Bright Asante-appiah, Erin Barry, Jim Bierstaker, Jim
Boyle, Frank Cade, Joe Carcello, Cori Crews, Caroline Hayek, Travis Holt, Scot Justice, Sri
Ramamoorti, Jason Rasso, Dick Riley, Jonathan Stanley, Daniel Street, Jeff Wilks, Mark Zimbelman,
and MAcc students at Kennesaw State University and The University of Alabama.
31
For example, Hackenbrack (1992) used one relevant excerpt and several pages of irrelevant information. Similarly, Hoffman
and Patton (1997) used two relevant cues and four irrelevant cues.
Appendix A
Fraud risk factors.*
Triangle Diamond
Pressure
p p
1. Extreme Company Growth or Protability
p p
2. Competition in Industry
p p
3. Changes in Customer Demand
p p
4. Performance Expectations
p p
5. Management Compensation Plan
p p
6. Management Financial Investment in Company
Opportunity
p p
7. Accounting Complexity
p p
8. Organizational Complexity
p p
9. Control Policies & Procedures
p p
10. Board & Audit Committee Monitoring Quality
p p
11. AIS Quality
p p
12. Turnover of Top Management & Directors
Rationalization/Attitude
p p
13. History of Legal or Regulatory Violations
p p
14. Top Managements Attitude Toward Achieving Aggressive Forecasts
p p
15. Excessive Demands on Auditor
p p
16. Top Managements Response to Internal Control Problems
p p
17. History of Disputes with Auditor
p p
18. Top Managements Ability to Rationalize Wrongdoing
Capability
p
19. Top Managements Interest in Accounting & Auditing Issues
p
20. Top Managements Expertise to Exploit Fraud Opportunities
p
21. Top Managements Ability to Deal with Stress
p
22. Top Managements Ability to Deceive Others
p
23. Top Managements Condence in Ability to Avoid Fraud Detection
p
24. Top Managements Ability to Coerce Others to Commit Fraud
Additional Items (Irrelevant)

p
25. Company Product Mix
p
26. Company Location
p
27. Company Customer Service Record
p
28. Company Customer Prole
p
29. Companys Marketing Strategy
p
30. Number of Employees in Company
Participants were asked to assess the possible fraud risk factors using a scale anchored No Fraud Risk (=0) and Very High
Fraud Risk (=100).
References
Abdel-Khalik, A., 2002. Reforming corporate governance post Enron: Shareholders board of trustees and the auditor. J. Account.
Public Policy 21 (2), 97103.
American Institute of Certied Public Accountants (AICPA), 2002. Statement on Auditing Standards No. 99: Consideration of
Fraud in a Financial Statement Audit. AICPA, New York, NY.
American Institute of Certied Public Accountants (AICPA), 2009. The Guide to Investigating Business Fraud. AICPA, New York,
NY.
Albrecht, W.S., Howe, K.R., Romney, M.B., 1984. Deterring Fraud: The Internal Auditors Perspective. The Institute of Internal
Auditors Research Foundation, Altamonte Springs, FL.
Albrecht, W.S., Albrecht, C.O., Albrecht, C.C., Zimbelman, M.F., 2012. Fraud Examination, fourth ed. South-Western Cengage
Learning, Mason, OH.
Alon, A., Dwyer, P., 2010. The impact of groups and decision aid reliance on fraud risk assessment. Manage. Res. Rev. 33 (3), 240
256.
Apostolou, B., Hassell, J., Webber, S., 2001. The relative importance of management fraud risk factors. Behavioral Res. Account.
13, 124.
Arel, B., Jennings, M., Pany, K., Reckers, P., 2012. Auditor liability: a comparison of judge and juror verdicts. J. Account. Public
Policy 31 (5), 516532.
Asare, S.K., Wright, A.M., 2004. The effectiveness of alternative risk assessment and program planning tools in a fraud setting.
Contemp. Account. Res. 21 (2), 325352.
Beasley, M.S., Carcello, J.V., Hermanson, D.R., Neal, T., 2010. Fraudulent Financial Reporting: 19982007, An Analysis of U.S.
Public Companies. Committee of Sponsoring Organizations of the Treadway Commission, Durham, NC. <http://www.coso.
org>.
Bell, T.B., Carcello, J.V., 2000. A decision aid for assessing the likelihood of fraudulent nancial reporting. Audit.: J. Practice
Theory 19 (1), 169184.
Bierstaker, J.L., Cohen, J.R., DeZoort, F.T., Hermanson, D.R., 2012. Audit committee compensation, fairness, and the resolution of
accounting disagreements. Audit.: J. Practice Theory 31 (2), 131150.
Bonner, S., 2008. Judgment and Decision Making in Accounting. Pearson Prentice Hall, Upper Saddle River, NJ.
Bonner, S., Libby, R., Nelson, M., 1996. Using decision aids to improve auditors conditional probability judgments. Account. Rev.
71 (2), 221240.
Center for Audit Quality (CAQ), 2010. Deterring and Detecting Financial Reporting Fraud. CAQ, Washington, DC.
Carcello, J.V., Hermanson, D.R., 2008. Fraudulent Financial Reporting: How Do We Close the Knowledge Gap? Institute for Fraud
Prevention, Morgantown, WV. <http://www.theifp.org/research-grants/IFP-Whitepaper-1.pdf>.
Carpenter, T., 2007. Audit team brainstorming, fraud risk identication, and fraud risk assessment: implications of SAS No. 99.
Account. Rev. 82 (5), 11191140.
Chandler, D.J., 2009. The perfect storm of leaders unethical behavior: a conceptual framework. Int. J. Leadership Stud. 5 (1), 69
93.
Cohen, J.R., Gaynor, L.M., Krishnamoorthy, G., Wright, A.M., 2011. The impact of auditor judgments of CEO inuence on audit
committee independence. Audit.: J. Practice Theory 30 (4), 129147.
Conover, W.J., Iman, R., 1982. Analysis of covariance using the rank transformation. Biometrics 38, 715724.
Cressey, D., 1973. Other Peoples Money: A Study in the Social Psychology of Embezzlement. Patterson Smith, Montclair, NJ.
Crowe Horwath, 2010. Playing Offense in a High-Risk Environment. Crowe Horwath, New York, NY.
Crowe Horwath, 2012. Putting the Freud in Fraud: Why the Fraud Triangle is No Longer Enough. Crowe Horwath, New York, NY.
Deloitte, 2010. Procurement Fraud: Investigative Techniques to Help Mitigate Risk. Deloitte Development LLC, New York, NY.
DeZoort, F.T., Harrison, P.D., Schnee, E.J., 2012. Tax professionals responsibility for fraud detection: the effects of engagement
type and audit status. Account. Horizons 26 (2), 289306.
Dorminey, J., Fleming, A., Kranacher, M., Riley Jr., R., 2010. Beyond the fraud triangle: enhancing deterrence of economic crimes.
CPA J., 1723 (July)
Dorminey, J., Fleming, A., Kranacher, M., Riley Jr., R., 2011. Beyond the fraud triangle: enhancing deterrence of economic crimes.
Fraud Magazine (September/October), 1825.
Dorminey, J., Fleming, A., Kranacher, M., Riley Jr., R., 2012. The evolution of fraud theory. Issues Account. Educ. 27 (2), 555579.
Dowling, C., Leech, S., 2007. Audit support systems and decision aids: current practice and opportunities for future research. Int.
J. Account. Inform. Syst. 8, 92116.
Eagly, A., Carli, L., 2007. Through the Labyrinth: The Truth About How Women Become Leaders. Harvard Business School Press,
Boston, MA.
Eining, M., Jones, D., Loebbecke, J., 1997. Reliance on decision aids: an examination of auditors assessment of management
fraud. Audit.: J. Practice Theory 16 (2), 119.
Farag, M.S., Elias, R.Z., 2012. The impact of accounting students professional skepticism on their ethical perception of earnings
management. Res. Prof. Responsib. Ethics Account. 16, 185200.
Feng, M., Ge, W., Luo, S., Shevlin, T., 2011. Why do CFOs become involved in material accounting manipulations? J. Account.
Econ. 51 (1/2), 2136.
Hackenbrack, K., 1992. Implications of seemingly irrelevant evidence in audit judgment. J. Account. Res. 30 (1), 126136.
Hammersley, J.S., 2011. A review and model of auditor judgments in fraud-related planning tasks. Audit.: J. Practice Theory 30
(4), 101128.
Hammersley, J.S., Johnstone, K.M., Kadous, K., 2011. How do audit seniors respond to heightened fraud risk. Audit.: J. Practice
Theory 30 (3), 81101.
Harvey, M., Buckley, R., Heames, J., Zinko, R., Brouer, R., Ferris, G., 2007. A bully as an archetypal destructive leader. J. Leadership
Organiz. Stud. 14 (2), 117129.
Hateld, R.C., Houston, R.W., Stefaniak, C., Usrey, S., 2010. The effect of magnitude of audit difference and prior client
concessions on negotiations of proposed adjustments. Account. Rev. 85 (5), 16471668.
Hoffman, V.B., Patton, J.M., 1997. Accountability, the dilution effect, and conservatism in auditors fraud judgments. J. Account.
Res. 35 (2), 227237.
Hogan, C.E., Rezaee, Z., Riley Jr., R., Velury, U.K., 2008. Financial statement fraud: insights from the academic literature. Audit.: J.
Practice Theory 27 (2), 231252.
Hurtt, K., 2010. Development of a scale to measure professional skepticism. Audit.: J. Practice Theory 29 (1), 149171.
Hurtt, K., Brown-Liburd, H., Earley, C., Krishnamoorthy, G., 2013. Research on auditor professional skepticism: literature
synthesis and opportunities for future research. Audit.: J. Practice Theory 32 (Supplement 1), 4597.
International Federation of Accountants (IFAC), 2010. International Standard on Auditing No. 240: The Auditors Responsibilities
Relating to Fraud in an Audit of Financial Statements. IFAC, New York, NY.
Institute of Internal Auditors (IIA), AICPA, and ACFE, 2008. Managing the Business Risk of Fraud: A Practical Guide. The IIA,
Altamonte Springs, FL.
Johnson, E., Kuhn Jr., J., Apostolou, B., Hassell, J., 2013. Auditor perceptions of client narcissism as a fraud attitude risk factor.
Audit.: J. Practice Theory 32 (1), 203219.
Kassem, R., Higson, A., 2012. The new fraud triangle model. J. Emerg. Trends Econ. Manage. Sci. 3 (3), 191195.
Kish-Gephart, J.J., Harrison, D.A., Trevino, L.K., 2010. Bad apples, bad cases, and bad barrels: meta-analytic evidence about
sources of unethical decisions at work. J. Appl. Psychol. 95 (1), 131.
Kochetova-Kozloski, N., Messier, W.F., Eilifsen, A., 2011. Improving auditors fraud judgments using a frequency response mode.
Contemp. Account. Res. 28 (3), 837858.
Lynch, A.L., Murthy, U.S., Engle, T.J., 2009. Fraud brainstorming using computer-mediated communication: the effects of
brainstorming technique and facilitation. Account. Rev. 84 (4), 12091232.
McEnroe, J.E., Martens, S.C., 2001. Auditors and investors perceptions of the expectation gap. Account. Horizons 15 (4), 345
358.
Nelson, M., 2009. A model and literature review of professional skepticism in auditing. Audit.: J. Practice Theory 28 (2), 134.
Nelson, M., Tan, H., 2005. Judgment and decision-making research in auditing: a task, person, and interpersonal interaction
perspective. Audit.: J. Practice Theory 24, 4171.
Norman, C., Rose, A., Rose, J., 2010. Internal audit reporting lines, fraud risk decomposition, and assessments of fraud risk. Acc.
Organ. Soc. 35 (5), 546557.
Parker, T., Fogarty, T., 2012. Seeing what you want to see: perceptual biases of auditors. J. Manage. Policy Practice 13 (2), 1125.
Paulhus, D., Williams, K., 2002. The dark triad of personality: narcissism, machiavellianism, and psychopathy. J. Res. Pers. 36,
556563.
Public Company Accounting Oversight Board (PCAOB), 2007. Observations on Auditors Implementation of PCAOB Standards
Relating to Auditors Responsibilities with Respect to Fraud. PCAOB Release No. 2007-001 (January 22). PCAOB, Washington,
DC. <http://pcaobus.org/Inspections/Documents/2007_01-22_Release_2007-001.pdf>.
Pincus, K.V., 1989. The efcacy of a red ags questionnaire for assessing the possibility of fraud. Acc. Organ. Soc. 14 (1/2), 153
163.
Ping, Z., 2007. The impact of the publics expectations of auditors on audit quality and auditing standards compliance. Contemp.
Account. Res. 24 (2), 631654.
PricewaterhouseCoopers (PwC), 2009. Fraud in a Downturn: A Review of How Fraud and Other Integrity Risks Affect Business.
PwC, New York, NY.
Ramamoorti, S., 2008. The psychology and sociology of fraud: Integrating the behavioral sciences component into fraud and
forensic accounting curricula. Issues Account. Educ. 23 (4), 521533.
Rosenthal, S.A., Pittinsky, T.L., 2006. Narcissistic leadership. Leadership Quart. 17, 617633.
Shanteau, J., Peters, J., 1989. The 3 Cs of Expert Audit Judgment: Creativity, Condence, and Communication. Paper Presented at
the February 1989 USC Audit Judgment Symposium. <http://www.k-state.edu/psych/cws/pdf/3C89.pdf>.
Shirley, E.A.C., 1981. A distribution-free method for analysis of covariance based on ranked data. Appl. Stat. 30 (2), 158162.
Trompeter, G., Carpenter, T., Desai, N., Jones, K., Riley Jr., R., 2013. A synthesis of fraud related research. Audit.: J. Practice Theory
32 (suppl. 1), 287321.
United States v. Arthur Young & Co, 1984. 465 U.S. 805.
Wilks, J., Zimbelman, M., 2004. Decomposition of fraud-risk assessments and auditors sensitivity to fraud cues. Contemp.
Account. Res. 21 (3), 719745.
Wolfe, D., Hermanson, D.R., 2004. The fraud diamond: considering four elements of fraud. CPA J., 3842 (December).

J Jaccpubpol 2015 05 006

Diunggah oleh

Informasi Dokumen

Judul Asli

Hak Cipta

Format Tersedia

Bagikan dokumen Ini

Bagikan atau Tanam Dokumen

Opsi Berbagi

Apakah menurut Anda dokumen ini bermanfaat?

Apakah konten ini tidak pantas?

Hak Cipta:

Format Tersedia

J Jaccpubpol 2015 05 006

Diunggah oleh

Hak Cipta:

Format Tersedia

J. Account.

Public Policy xxx (2015) xxxxxx

Contents lists available at ScienceDirect

J. Account. Public Policy

The effect of alternative fraud model use

This study evaluates how the use of alternative fraud model

opportunity, and a broader capability dimension (that includes

2. Background and hypothesis development

2.1. The role of the fraud triangle in auditor fraud detection

2.2. Extensions or enhancements of the fraud triangle

2.3. Fraud risk assessment process

2.4. Fraud model practice aids

2.5. CEO risk level

3.1. Participants, procedure, and instrument

3.1.1. CEO risk manipulation

3.1.2. Fraud model manipulation

3.1.3. Additional measures

4.1. Manipulation check

4.2. Participant characteristics

4.3. ANCOVA results

Model F = 47.28, p < 0.001; Adj R-sq = 0.725.

4.4. Fraud risk factor rankings

High CEO risk Score Low CEO risk Score

4.5. Supplemental analyses

Fraud risk factor* Loadings

Fraud risk factors.*

Additional Items (Irrelevant)

Anda mungkin juga menyukai