J. Account.

Public Policy 29 (2010) 439–458

Contents lists available at ScienceDirect

J. Account. Public Policy

journal homepage: www.elsevier.com/locate/jaccpubpol

Exploring Sarbanes–Oxley’s effect on attitudes, perceptions

of norms, and intentions to commit financial statement fraud
from a general deterrence perspective
Joseph C. Ugrin a,*, Marcus D. Odom b,1
a
College of Business Administration, Kansas State University, Manhattan, KS 66506, USA
b
College of Business, School of Accountancy, Southern Illinois University Carbondale, Carbondale, IL 62901, USA

a r t i c l e i n f o a b s t r a c t

Article history: This paper uses an experiment to examine how deterrence mecha-
nisms within the Sarbanes–Oxley Act’s (SOX) Sections 404 and 906
influence the fraudulent financial reporting behavior of individuals.
The results indicate that the threat of potential jail time can be an
effective mechanism for reducing financial statement fraud, but
its effectiveness is limited and influenced by a wide range of social,
environmental, and demographic factors. The findings show that
the incremental increase in potential jail time imposed by SOX
creates little deterrence beyond mechanisms that were in place
pre-SOX. The findings also reveal that the effect of jail time is
primarily a function of economic consequences, such as lost career
opportunities that are created from serving just a minimal amount
of time in jail. The results should be of interest to regulators and
practitioners wanting to understand how SOX-based deterrence
mechanisms can influence individual behavior. The results contribute
to the general deterrence theory literature by showing how the effect
of deterrence mechanisms on illicit behavior can be influenced by
social, environmental, and demographic factors.
Ó 2010 Elsevier Inc. All rights reserved.

1. Introduction

Financial statement fraud has received considerable attention from the popular press (Black, 2007),
the corporate and financial community, and regulators, due in large part to the collapse of Enron and

* Corresponding author. Tel.: +1 785 532 5897; fax: +1 785 532 5959.
E-mail addresses: jugrin@ksu.edu (J.C. Ugrin), modom@cba.siu.edu (M.D. Odom).
1
Tel.: +1 618 453 1408; fax: 1 618 453 1411.

0278-4254/$ - see front matter Ó 2010 Elsevier Inc. All rights reserved.
doi:10.1016/j.jaccpubpol.2010.06.006
440 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458

the fraudulent activities at other public companies such as Adelphia, Worldcom, and Tyco. These cases
alone have resulted in nearly $500 billion in losses to investors. Annually, financial statement fraud
costs organizations and investors nearly 6% of revenues. (Association of Certified Fraud Examiners,
2006; Rezaee, 2003). In addition to the cost, financial statement fraud is difficult to control since it
is typically a crime committed by managers and executives (Singleton et al., 2003).
The Sarbanes–Oxley Act (SOX) was passed into law on July 30, 2002, in response to public pressure
over acts of corporate malfeasance. At its core, SOX aims to restore investor confidence in US capital
markets by improving corporate governance, the quality of audits, and the strength of internal con-
trols in part by increasing the consequences management can encounter if they are caught committing
financial statement fraud. SOX Section 906 increases the severity of sanctions that can potentially be
leveled against management for inaccurate financial reporting, and Section 404 imposes regulations
that require increased assessments of internal controls by auditors. However, SOX has not come with-
out debate. Researchers have suggested that sanctions and prosecutions may not significantly reduce
financial statement fraud enough, considering SOX is costly to implement and maintain (Ronen, 2002;
Nyberg, 2003; Engle et al., 2004).
Due to the high costs associated with both financial statement fraud and SOX, it is important to
develop a better understanding about why fraud is committed and critically evaluate SOX’s effective-
ness at reducing it. A wide array of research has examined fraud at the organizational (Skousen and
Wright, 2008; Persons, 1995; Kaminski et al., 2004; Cullinan et al., 2006) and individual levels
(Johnson et al., 2006; Healy, 1985; Dechow et al., 1996; Bebchuk and Fried, 2003; Goldman and Slezak,
2006; Carpenter and Reimers, 2005), yet only recently has research started to address the effect of SOX
on financial statement fraud and the role that management, audit committees, and auditors have in
the financial reporting process. For example, audit committee members feel they have more power
in the post-SOX era (DeZoort et al., 2008), yet they still take a passive role in resolving financial report-
ing disputes between management and auditors, and management is still the driving force behind
auditor appointments and terminations (Cohen et al., 2008). This suggests that management still
directs financial reporting decisions even though audit committees are more active and possess
greater expertise in the post-SOX era (Cohen et al., 2008). Thus, it is important to understand how
SOX influences management specifically.
Little research on management influences exists except for McEnroe’s (2006) survey that found
that financial executives do not perceive that SOX will reduce earnings management practices.
McEnroe’s (2006) study gives some insight into perceptions about the effectiveness of SOX on various
financial reporting issues (including the deferral of expenses which is used in our experiment), but it
does not examine ‘‘why” SOX may or may not influence financial reporting. We assert that an
explanation can be found in the criminal justice literature on deterring illicit behavior.
Several components of SOX may function as deterrence mechanisms as defined by general deter-
rence theory (GDT), a widely used utility based criminological theory. This theory states that punish-
ment can deter illicit behavior to the extent that punishment is likely to be enforced. GDT ties closely
with SOX since deterrence mechanisms imposed by SOX increase potential sanctions (jail time) and
strengthen mechanisms that can catch fraudulent activity (internal controls) making potential sanc-
tions more likely. However, it is unclear if these mechanisms will achieve their desired effect of deter-
ring fraudulent financial reporting either in general or beyond existing mechanisms. For instance,
prior to SOX, the Securities and Exchange Act of 1934 (1934 Act) threatened jail time for fraudulent
financial reporting, albeit less than that imposed by SOX. Increasing potential jail time for illicit activ-
ities has had mixed results in the criminal justice literature due to several factors, including the dis-
utility of each additional year of jail time. In addition, the effectiveness of jail time is influenced by
general perceptions about the likelihood that it will be imposed, whether or not the time is served
under severe conditions, and other ramifications that arise from spending time in jail, like lost future
opportunities that result from being a felon.
The motivation to investigate whether the SOX-based deterrence mechanisms mentioned affect
the attitudes and intentions of managers and executives to fraudulently report financial data is
grounded in this criminal justice literature. In addition, the study examines whether or not outcomes
are influenced by social, environmental, and situational factors such as perceptions about the general
effectiveness of internal controls, the severity of sanctions, awareness of recent prosecutions for
 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458 441

similar crimes, perceptions of future ramifications that may arise if one is caught fraudulently report-
ing financial data, and demographic factors such as experience. This is done by testing how differing
levels of potential jail time for committing financial statement fraud combined with differing levels of
internal controls shape attitudes and perceptions about normative behaviors related to fraudulent
financial reporting, which have been shown to relate significantly with fraudulent financial reporting
intentions (Carpenter and Reimers, 2005).
Our results indicate that the threat of potential jail time can be an effective mechanism for reducing
financial statement fraud, but its effectiveness is limited and influenced by a wide range of social,
environmental, and demographic factors. We find that the incremental increase in potential jail time
imposed by SOX creates little deterrence beyond mechanisms that were in place pre-SOX. The results
also reveal that the effect of jail time is primarily a function of economic consequences that are created
from serving even a minimal amount of time in jail.
These findings should be of interest to regulators and practitioners wanting to understand how
SOX-based deterrence mechanisms can influence individual behavior. They will also contribute to
the academic literature by showing how the effect of deterrence mechanisms on illicit behavior can
be influenced by social, environmental, and demographic factors.

2. Hypotheses development

2.1. Fraud deterrence

Tyler and Blader (2005) note that there is a great deal of literature on how incentives motivate
employees to commit fraud, but very little on the effect that sanctions have on deterring fraud. This
interesting observation about the study of fraud also is the case with financial statement fraud, where
the empirical research related to factors that perpetuate it, like incentive based compensation, is
abundant compared to research related to how deterrence mechanisms affect it.
General deterrence theory (GDT) emphasizes that punishment-inclusive regulations that are
imposed on individuals by authorities (such as organizations, regulators, or lawmakers) modify
individual actions. The key premise behind this model is that individuals make rational decisions to
benefit themselves. When confronted with opportunities, individuals weigh the costs and rewards
of taking advantage of an opportunity against other options (Beccaria, 1963; Blair and Stout, 2001;
Klepper and Nagin, 1989; Simpson and Koper, 1992; Tyler and Blader, 2005; Williams and Hawkins,
1986). GDT is like other economic models, it:
. . .starts from the simple premise that individuals are willing to commit crimes if the expected ben-
efits of the crime exceed the expected benefits of engaging in lawful activity. In other words, the
decision to commit a crime is simply another type of decision-making under conditions of uncer-
tainty, the same kind of calculus that a rational utility-maximizing individual will apply to the deci-
sion to engage in any activity. In this model, penalties are necessary for prohibited activities so that
individuals internalize the cost of those activities (Perino, 2002, p. 675).

GDT suggests the more likely that improper behavior will be detected, and the greater the severity
of the punishment for the behavior, the more effectual a deterrent will be (Beccaria, 1963; Becker,
1968; Williams and Hawkins, 1986). In addition, deterrence mechanisms are more likely to have a
greater effect on white collar crimes such as financial statement fraud, than other forms of crime
because white collar crimes are typically not ‘‘. . .crimes of passion; they are not spontaneous or
emotional, but calculated risks taken by rational actors. As such, they should be more amenable to
control by policies based on the utilitarian assumptions of the deterrence doctrine,” (Braithwaite
and Geis, 1982, p. 300).
A majority of the deterrence literature supports the notion that stronger deterrence mecha-
nisms reduce illicit behavior, yet accounting literature suggests that executives do not perceive
that SOX will be an effective deterrent (McEnroe, 2006). In addition, research on the effect of
deterrence mechanisms on financial statement fraud is lacking. Given the pervasiveness of finan-
cial statement fraud over the last decade, understanding the effect of deterrence mechanisms is
442 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458

imperative. GDT is a well tested deterrence theory that may bring some insight to this research,
and to date, GDT has not been used as a lens to examine the effectiveness of financial fraud
deterrence.

2.2. Sox Section 906, increasing the severity of potential sanctions

Considering GDT’s assumption that individuals are rational actors who weigh the perceived costs of
punishment with the perceived benefits of action, ramifications for illicit behavior must be severe
enough to deter. As such, it is critical that regulatory and enforcement agencies introduce sanctions
that are meaningful to potential perpetrators. SOX attempts to do this by stiffening criminal and civil
sanctions for financial statement fraud.
Section 906 of SOX covers corporate responsibility for financial reports, and states that manage-
ment’s false certification of financial statements carries a prison sentence of up to 20 years in jail,
an increase from the 10 years threatened by the 1934 Act for essentially the same offense. However,
the mere threat of sanctions yields uncertain results without the consideration of other deterrence
factors as not all types of sanctions have the same effect on all types of criminals. For example, the
threat of jail time has been shown to have a particularly strong influence on deterring white collar
crime in general. White collar criminals are particularly influenced by jail time because they are typ-
ically individuals who have more to lose than other types of criminals (Braithwaite, 1985).
Despite Braithwaite’s position, other literature suggests, but does not empirically test, that criminal
sanctions imposed by SOX will provide little deterrence beyond existing sanctions already in effect in
the 1934 Act. Perino (2002) argues that managers do not expect maximum sentences will be handed
down so they discount the penalties imposed by SOX. As a result, they will not incorporate the long jail
sentences threatened into their decision process. However, this effect may have been describe better
by Polinsky and Shavell (1999) who argue, when examining the effect of potential jail time the mar-
ginal effects (or marginal deterrence) of potential jail time will diminish as the potential number of
years in jail is increased. They suggest that much of this is due to the social stigma1 that is attached
to individuals who go to jail. The cost of social stigma is internalized and incorporated into one’s beliefs
about the cost versus benefit of committing fraud, and thus influences attitudes and beliefs about how
others important to the individual would think he or she should behave. Social stigma changes relatively
little when the time spent in jail is increased. One is stigmatized whether he or she has spent a long time
or relatively little time in jail, and the incremental change in stigmatization from increases in jail time is
relatively small. Individuals will likely have already incorporated social stigmas into their attitudes2 and
perceptions of subjective norms3 when faced with a potential 10-year sentence (imposed by the 1934
Act), so the change in those attitudes and perceptions when faced with the threat of a potential 20-year
sentence (imposed by SOX) will be minimal. Incorporating potential jail time into a behavioral model
aimed at understanding financial statement fraud, we form the following hypotheses4:

H1. Longer potential jail times will result in more negative attitudes about committing financial
statement fraud relative to shorter jail times.

H1a. The incremental change in attitude will be lower as the length of potential jail time is increased.

H2. Longer potential jail times will result in more negative perceptions of subjective norms about
committing financial statement fraud relative to shorter jail times.

1
‘‘Social Stigma” is defined here as the perception that even a minimal amount of time in jail will have a large consequence that
manifests itself through lost future opportunities that one would have otherwise had a high probability of obtaining.
2
‘‘Attitude” is defined here as ‘‘a person’s general feeling of favorableness or unfavorableness for that behavior” (Ajzen and
Fishbein, 1980, p. 6). Attitude toward an act or behavior is characterized as the sum of an individual’s feelings and beliefs about
costs versus benefits of an act (Ajzen, 1991).
3
‘‘Subjective norms” is defined here as a person’s perception that individuals who are important to them think the behavior in
question should be carried out (Ajzen, 1991).
4
A link between attitudes, subjective norms, and fraud intentions have been found in existing literature (Carpenter and Reimers,
2005). This paper attempts to show how those attitudes and perceptions are influenced.
 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458 443

H2a. The incremental change in perceptions of subjective norms will be lower as the length of poten-
tial jail time is increased.

2.3. Sox Section 404, increasing the certainty of potential sanctions

GDT proposes that deterrence is enhanced when punishment is more likely. When faced with deci-
sions to perform criminal or unethical behavior, individuals weigh perceived benefits against per-
ceived consequences. Consequences that are perceived to be more likely will have greater
deterrence. In other words, there must be a strong chance of being caught for deterrence factors to
be effective (Williams and Hawkins, 1986). Thus, deterrence mechanisms that increase the likelihood
of punishment through threat of detection (e.g. monitoring mechanisms) should theoretically result in
a lower amount of illicit behavior. Literature has demonstrated that various types of monitoring can
help reduce the potential for financial statement fraud, particularly misleading disclosures (Latham
and Jacobs, 2000). However, these studies do not examine the extent of internal monitoring activities
like internal controls, but rather they examine the amount of monitoring an organization receives
from institutional investors, which is an external control. Nonetheless, Latham and Jacobs (2000)
found that firms that had a larger institutional following had fewer instances of misleading disclo-
sures. The reason may be that greater institutional ownership in a firm results in more information
activities by analysts (Potter, 1992; O’Brien and Bhushan, 1990). We conjecture from Latham and
Jacob’s (2000) study that the greater the amount of information available to be examined by institu-
tional watchdogs, the greater the threat of detection and reduction of instances of misleading
disclosures. Again, this does not cover ‘internal’ control mechanisms, but it does illustrate that mon-
itoring control mechanisms can be effective at reducing financial statement fraud. Other related
research has found that an increased probability of detection has an influence on other types of illicit
behavior such as Braithwaite and Makkai’s (1991) study of a deterrence model on nursing home reg-
ulation compliance and Klepper and Nagin’s (1989) study of the impact of deterrence measures on tax
fraud intentions of graduate students.
Detection alone may not have a direct effect on behavior or attitudes. Rather, detection may have
an indirect effect on attitudes and subjective norms, by moderating the impact of sanctions on atti-
tudes and perceptions of subjective norms. The studies cited do not examine this interaction.
SOX Section 404 increases the internal control responsibility placed on management and public
accountants causing increased attention to internal control and fraud detection. In short, public orga-
nizations must develop internal control strategies that are deemed effective by an external auditor
and SOX attempts to make the audit assessment function more effective by providing a framework
for internal control assessments and requiring a separate attestation of internal controls. In a recent
study, DeZoort et al. (2008) found audit committee members perceive that improving internal controls
will improve financial reporting and fraud detection. Is this the same for managers who are the pri-
mary drivers of financial reports and play an active role with the audit team? By indirectly increasing
the strength of the internal control function5 and indirectly increasing the likelihood that fraud will be
detected by internal controls, SOX Section 404 strengthens the detection component of the GDT
framework.
Most deterrence studies test the direct effect of the likelihood of sanctions with no regard for sever-
ity; however, the certainty of sanctions will theoretically have an effect on how salient sanctions are to
potential perpetrators. The utility-based model of deterrence suggests the cost of sanctions is a func-
tion of both severity and likelihood working together (Nagin and Pogarsky, 2001). If individuals feel
the likelihood of detection is low, they will theoretically place a lower weight on sanctions. We posit
then, the threat of jail time has a more pronounced effect when effective internal controls are present.

H3. The effects of the length of jail time on attitudes about committing financial statement fraud will
be more pronounced when internal controls are strong relative to when internal controls are weak.

5
For example, Gordon et al. (2006) find indirect evidence that a result of SOX is stronger internal control activities.
444 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458

H4. The effects of the length of jail time on perceptions of subjective norms about committing finan-
cial statement fraud will be more pronounced when internal controls are strong relative to when
internal controls are weak.
The Nomological model illustrating the unobserved constructs and observable measures is pre-
sented in Fig. 1.

3. Methodology

The study used an experiment with a 3  2 between subjects factorial design. Participants were
given a scenario, similar to the one used by Carpenter and Reimers (2005), that contained information
about a financial reporting situation. Participants were randomly assigned information pertaining to
one of three levels of sanctions and one of two levels of internal control. They then answered question-
naire items related to the dependent variable, control variables, and debriefing items. This type of
scenario based methodology is widely used in ethics research (Weber, 1992).

Unobserved Constructs

Attitudes
Punishment About
Severity Financial
Statement
Fraud Financial
Statement
Fraud
Subjective Intention
Certainty of
Norms
Punishment
(Detection)
________________________________________________________________

Observable Measures

Attitudes
About
Length of Potential Financial
H1 (H1a) Financial
Jail Time Statement
(Short/Medium/Long) Statement
Fraud (3 item
scale per Fraud
Carpenter and Intention (1
Reimers item
H2 (2005)) measuring
(H2a) individual
H3 intentions and
Perceptions of 1 item
H4 Subjective measuring the
intentions of
Norms About a referent
Financial other)
Internal Control
Statement
Structure
(Weak/Strong) Fraud (1 item
per Carpenter and
Reimers (2005))

Fig. 1. Nomological map of the full research model.

 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458 445

3.1. Participants

Two sets of participants were utilized. An initial data set was collected and analyzed from 130 Mas-
ters of Accounting students and a second data set was collected from a sample of 43 corporate exec-
utives to validate and add generalizability to the results. In the student data set, responses from seven
individuals were removed due to their failure to accurately respond to manipulation checks (discussed
in the next section), resulting in 123 usable sets of responses. In the executive data set, responses from
four individuals were removed for failure to respond to all items. Overall, the participants reflect a
wide range of demographics. Since the student data was used in the primary tests of the model, stu-
dent demographic data is presented in Table 1.
To test for randomization of the demographic variables across treatment conditions, a Sheffe’s test
of pairwise differences was performed for each variable, and none were significantly different across
treatment conditions (all p values > 0.270).

3.2. The scenario

Carpenter and Reimers (2005) were able to examine how different levels of attitudes and percep-
tions of subjective norms affect intentions to commit financial statement fraud and their scenario is
well suited for use in this study with only slight modification. The scenario provided an ethical dilem-
ma to participants who, if they did not intentionally commit a GAAP violation, would not get a bonus.
This gave participants a reason to rationalize making a GAAP departure. The scenario we use states:
Assume that you are the controller for a small-sized public company. Your bonus as controller is
calculated on your company’s net income targets that you must meet. This year that target is $1.5 mil-
lion. In your position, you are authorized to sign off on any decisions made regarding financial report-
ing. You are faced with the following situation:
On December 15, your firm ordered $150,000 worth of supplies in anticipation for the seasonal
rush. These supplies were delivered on the evening of December 27, 2007, and you expect to use all
of the supplies by the end of the year 2007. If you record that supplies expense before year end,
net income will be $1.45 million and you will not meet the target and will therefore not receive your
bonus of $25,000 that you had worked hard for. It is a violation of generally accepted accounting
principles (GAAP) not to record the expense upon use. You expect to use all of the supplies before
December 31, 2007, but if you record the expense for the year ended December 31, 2007, you will
not receive a bonus.

Table 1
Demographic information.

1. Gender Female: 63 Male: 59 No Response: 1

2. Age 25.45 (6.04)
3. Ethnicity White: 106 Black: 6
Asian: 7 Indian/Middle Eastern: 3
4. Years of total work experience 5.29 (5.84)
5. Have served as a manager or Yes: 57 No: 64 No response: 2
supervisor
6. Years as a manager or 1.38 (2.41)
supervisor
7. Currently employed Yes: 85 No: 35 No response: 2
8. Current occupation Grad assist (#): 34 Mgmt/exec (#): 3
Accountant or auditor (#): 29 Laborer (#): 2
Other admin (#): 12 Sales (#): 5
Not employed (#): 35 No response (#): 3
9. Annual income Under $15,000: 52 $15,000–$29,999: 12
$30,000–$44,999: 7 $45,0000–$59,999: 11
Over $59,999: 1 Not employed: 35
No response: 4
446 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458

3.3. Independent variables

The independent variables were operationalized by directing participants to consider a combination

of jail times and internal controls. The length of potential jail time was operationalized by declaring the
law related to fraudulent financial reporting and disclosure states that if convicted, ‘‘You may be jailed for
up to 1 year” (a low degree of potential sanction); ‘‘You may be jailed for up to 10 years” (a mid-level
degree of potential sanction); or ‘‘You may be jailed for up to 20 years” (a high degree of potential
sanction). Recall, SOX threatens up to 20 years in jail for fraudulent financial reporting, whereas the
Securities and Exchange Act of 1934 threatens up to 10 years in jail. The strength of internal control
measures was operationalized by stating that the participant’s company’s internal audit department had
reviewed the company’s internal control structure and found the control system to be ‘‘weak” or ‘‘strong”.

3.4. Dependent variables

The dependent variables used to test the hypotheses were participants’ attitudes (H1, H1a, and H3)
about making the GAAP departure, measured by a three-item summated scale, and participants’ per-
ceptions of subjective norms (H2, H2a, and H4) about making the GAAP departure, using one item. The
attitude scale and item used to measure subjective norms were previously utilized by Carpenter and
Reimers (2005), and were developed and validated by Chang (1998),6 which in turn were fashioned
after Ajzen and Fishbein’s Theory of Planned Behavior (TPB) (1980). The measurement of attitude used
in this study asked participants how they felt about delaying the supply expense, whether it would be
bad or good, harmful or beneficial, wise or foolish. The measurement of subjective norms asked partic-
ipants if most people important to them felt they should delay the supply expense. These items were
measured on seven-item Likert type scales.

3.5. Covariates

We expect individuals’ inherent levels of self-control (individuals low in self-control are less likely
to control the urge to engage in illicit behavior) and responsibility denial (individuals that rate high in
responsibility denial tend to depersonalize illicit acts and place responsibility on others) could system-
atically influence the outcomes, so we control for these variables. Self-control was measured by a 24-
item, five-point Likert type scale developed by Grasmick et al. (1993) and further validated by Nagin
and Paternoster (1993). Responsibility denial was measured by a 28-item, five-point Likert type scale
developed by Schwartz (1973, 1977) and additionally validated by Harrington (1996).
Information related to two other potential covariates that we expected to have a direct relationship
with the outcome variables was also collected. First, participants could be affected by the utility they
place on the bonus presented in the scenario, thus, they were asked if a bonus of $25,000 represented a
large amount of money to them. Second, they could be more apprehensive about fraud in general if
they recognize its affect on others. To measure that perception, participants were asked if financial
statement fraud harms others.

3.6. Manipulation checks

We used several questions to check if participants understood the scenario and the manipulations.
If they did not understand, they were removed from further analyses. To check if participants under-
stood the ramifications of reporting the supply expense, participants were asked if they would get a
bonus if they deferred the supply expense into 2008. This resulted in four students and no executives
being removed for failure to respond accurately. To check if the participants understood their obliga-
tion under generally accepted accounting principles (GAAP), they were asked if GAAP required them to
record the supply expense in 2007. This resulted in two students and no executives being removed for

6
Chang (1998) performed a confirmatory factor analysis to determine if the measured variables reliably predict the latent
constructs of attitude, subjective norms and perceived behavioral control. Their three factor solution had a fit index of 0.973,
indicating a good fit.
 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458 447

failure to respond accurately. To check if participants understood the potential ramifications of not
recording the supply expense correctly, they were asked how many years (1/10/20) they could poten-
tially be put in jail if they violated GAAP. This resulted in one student and no executives being
removed for failure to respond accurately. Finally, to check if participants understood the strength
of the internal controls, they were asked if the internal control structure was weak or strong. No
participants were removed based on responses to this check.

3.7. Debriefing items

Several debriefing items were used to measure the perceptions of the participants. Outcomes can be
influenced by participants’ perceptions of various factors related to sanctions and controls. To measure
the participants perceived severity of the sanctions, they were asked, ‘‘If you are convicted of financial
statement fraud, do you consider the potential jail time to be long?” Outcomes can also be affected by
the perceived likelihood that the internal control structure would result in a perception that sanctions
are more certain. This was assessed by asking participants, ‘‘In the scenario previously described, would
it be likely that you would get caught if you did not report the supply expense correctly?”
Outcomes can also be influenced by participants’ knowledge of enforcement activities. In other
words, deterrence efforts are influenced by an awareness of whether or not action has been taken
against those who have been caught. At the firm level, Simpson and Koper (1992) support the impor-
tance of awareness of enforcement in their findings that past guilty verdicts had an impact on reduc-
ing antitrust violations by corporations, but it has not been examined at the individual level. The two
debriefing questions to address this issue were, ‘‘Are you aware of the prosecution of executives at
Enron and World Com?” and ‘‘How aware are you of the ‘details’ regarding the sentences handed
down to executives at Enron and World Com?”
Being stigmatized as a felon may have as large of an impact on a person who has spent one day in
jail as it does a person who has spent many years (Polinsky and Shavell, 1999). Since perceptions of
social stigmas could influence the results participants were asked if: (1) their future job prospects
would be reduced, (2) their social status would be adversely affected, and (3) their career would be
over if they spent time in jail.
Lastly, researchers have suggested perpetrators do not expect that sanctions will be fully enforced, and
time served by white collar criminals in low security prisons is not as severe as that served by perpetrators
of other types of crimes (Braithwaite, 1985). Both of these factors have been posited to minimize the effect
of sanctions. Thus, participants were asked, ‘‘If they thought they would serve a full sentence if caught and
convicted?” and ‘‘If convicted, would they serve their jail time under harsh conditions?”

5. Results

5.1. Preliminary tests

Tests of correlations were used in a preliminary analysis of demographic factors and potential
covariates across the outcome variables. In the student population, attitudes correlated significantly
with years of work experience, feelings that financial statement fraud harms others, self-control,
and responsibility denial. However, none of the variables had a significant relationship with percep-
tions of subjective norms. Experience, self-control, responsibility denial, and perceptions that fraud
harms others were included as control variables in subsequent tests of hypotheses. A Sheffe’s test
on the demographics and potential covariates was performed, and none of the variables were signif-
icantly different across treatment conditions. Consequently, none of the variables would confound the
experimental results (all p > 0.260).

5.2. Validating existing research

Based on feedback from a pilot study and a focus group of 12 academics, slight modifications to
Carpenter and Reimer’s (2005) scenario were made to add clarity in the context of this study. We mea-
448 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458

sured fraud intentions in two ways: (1) asking participants their intentions directly and (2) asking par-
ticipants what they felt the intentions of their peers (a referent other) would be. We elicited partici-
pants’ perceptions about a referent other because responses to ethical decisions may be biased by a
‘‘halo effect” or social projection. Social projection theory suggests when individuals are asked how
a referent other would respond to a situation (rather than asking the individual about his or her inten-
tions directly), they project their true intentions on the referent other (Beeler and Hunton, 2002;
Clement and Krueger, 2000; Cohen et al., 1993, 2001; Mikulineer and Horesh, 1999; Ruvolo and Fabin,
1999; Smith, 1997).
Two regressions were run to test Carpenter and Reimer’s (2005) model using both measures of in-
tent. The first using direct intentions as the dependent variable shows a statistically significant rela-
tionship between participants’ attitudes and intentions to commit fraud (p < 0.050) and participants’
perceptions of subjective norms and intentions to commit fraud (p < 0.100). The coefficient of deter-
mination for the model is r2 = 0.360. The results of the second using referent intentions show a statis-
tically significant relationship between attitude and the perceived referent intentions (p < 0.050) and
subjective norms and the perceived referent intentions (p < 0.050). The coefficient of determination
for this model is r2 = 0.209. Our results concur with Carpenter and Reimer’s (2005) findings and sug-
gest that attitudes and perceptions of subjective norms relate to fraud intentions, supporting the use
of the TPB as an underlying model of the decision process and the appropriateness for testing the ef-
fect of deterrence mechanisms on attitudes and perceptions of norms.
The reliability of the multi-item scales used in the tests of attitude and perceived behavioral control
were checked. The alpha coefficient for student’s attitude is 0.751, and the alpha coefficient for per-
ceived behavioral control is 0.768. Both results indicate acceptable reliability of the scales utilized.

5.3. Hypotheses tests

To test the hypotheses, two ANCOVA analyses were performed. The first analysis tested the effects
of jail time and internal control structures on participants’ attitudes, while controlling for participants’
level of responsibility denial, self-control7, work experience, and perceptions that financial statement
fraud harms others. Table 2, Panel A, shows means and standard deviations for attitudes across treat-
ment conditions. Table 2, Panel B, provides the results of the ANCOVA model. The results show that jail
time is statistically significant (F = 16.865, p < 0.000), confirming H1, but the interaction term for jail
time and internal controls is not statistically significant (F = 1.595, p > 0.100), failing to support H3.
The results also show a significant main effect for internal controls (F = 4.310, p < 0.050), although this
was not hypothesized based on the model. Fig. 2 presents a graphical representation of the mean atti-
tudes across treatment conditions. The pattern of means in Table 2, Panel A, indicates that attitudes
are more positive for short jail times (1 year) (7.219) than for medium (10 years) (4.754) and long
(20 years) (4.511) jail times, and attitudes are more negative for stronger internal controls (5.060) ver-
sus weak internal controls (5.929).
Additional pairwise comparisons using Tukey’s HSD were performed on jail time to test H1a. Table
2, Panel C shows the results of the pairwise comparisons for the various levels of jail time. The results
indicate that attitudes for 1-year jail terms are statistically different than attitudes for 10- and 20-year
jail terms, while attitudes for 10- and 20-year jail terms are not different from one another; hence, H1a
is supported8.
The second ANCOVA analysis tested the effects of jail time and internal control structures on par-
ticipants’ perceptions of subjective norms including the same control variables used in the first anal-
ysis. None of the hypotheses related to subjective norms are significant. The effect of jail time on

7
Reliability tests for the self-control and responsibility denial scale had Cronbach alpha values of 0.846 and 0.871, respectively.
Both of these variables were significantly correlated with participants’ attitudes and thus included in the model as a covariate in
the test of attitudes.
8
Two additional tests were performed substituting intentions and intentions of a referent other as the dependent variables, the
results were similar to those using attitude as the dependent variable. Intentions were significantly related to jail time (p < 0.050)
yet the interaction term between jail time and internal controls was not (p > 0.100). The perceived intentions of a referent other
were significantly related to jail time (p < 0.050) yet the interaction term between jail time and internal controls was not
(p > 0.100).
 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458 449

Table 2
Effect of Jail Time and Internal Control on Attitude.

Jail time Internal control Main effect (jail time)

Weak Strong
Panel A: mean (standard error) {sample size} across treatment conditions
Short (1 year) 8.078 6.359 7.219
(.507) (.506) (.359)
{21} {21} {42}
Medium (10 years) 5.247 4.261 4.754
(.506) (.534) (.363)
{22} {19} {41}
Long (20 years) 4.460 4.561 4.511
(.512) (.520) (.365)
{20} {20} {40}
Main effect (internal control) 5.929 5.060 5.494
(.290) (.298) (.206)
{63} {60} {123}

Source Sum of squares df Mean square F Sig. Hypothesis

Panel B: partial ANOVA table (dependent variable: attitude)
Jail_time 176.137 2 88.069 16.865 0.000 H1
Internal cont 22.507 1 22.507 4.310 0.040
Jail_time  internal 16.658 2 8.329 1.595 0.207 H3
Jail timei Jail timej Mean diff. (i–j) Std. error Sig. 95% Confidence
Upper bound Lower bound
Panel C: pairwise comparisons for jail time (H3)
1 Year 10 Years 2.9384a 0.545 0.000 1.6438 4.2331
20 Years 3.1238a 0.549 0.000 1.8210 4.4266
10 Years 1 Year 2.9384a 0.545 0.000 4.2331 1.6438
20 Years 0.1854 0.552 0.940 1.1252 1.4959
20 Years 1 Year 3.1238a 0.549 0.000 4.4266 1.8210
10 Years 0.1854 0.552 0.940 1.4959 1.1252

Covariates appearing in the model are evaluated at the following values: sc = 53.3333, rd = 66.5528, experience = 5.1707,
harm = 6.4878.
R squared = 0.423 (adjusted R squared = 0.377).
P values with out covariates: jail_time p < 0.050; internal control p < 0.100; jail_time  internal control p > 0.100.
Without the covariates, the P values do not change.
a
The mean difference is significant at the 0.050 level using Tukey HSD.

subjective norms is not statistically significant (F = 0.580, p > 0.10), failing to support H2. The interac-
tion term for jail time and internal controls also is not statistically significant (F = 2.112, p > 0.10), fail-
ing to support H4. The pattern of mean perceptions of subjective norms are consistent with short
(1 year) (3.0000), medium (10 years) (2.5122), and long (20 years) (2.7480) jail terms, and our results
indicate that perceptions of subjective norms for 1-, 10-, and 20-year jail terms are not statistically
different than one another, failing to support H2a. In addition, there is no significant main effect for
the relationship between internal controls and perceptions of subjective norms9.

5.4. Supplemental analysis

Various debriefing items were used to assess other factors expected to influence results. A correla-
tion analysis revealed how these additional variables correlate with intent, a referent other’s intent,

9
Tabulated results for tests that used subjective norms as the dependent variable are not presented because none of the results
were significant.
450 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458

Fig. 2. Mean attitudes across treatment conditions.

Likelihood of getting
Caught (Z2)
P21 (1.984) (p < .000) P32 (-.467) (p < .000)

Internal
Controls (Z1) Attitude(Z3)
P31 (-.771) (p < .140)
P31’ (.280) (p < .720)

Fig. 3. Mediating effect of perceptions of the likelihood of getting caught.

attitude, and subjective norms. Our analysis revealed that nearly all of these other factors are signif-
icantly related to intent, a referent other’s intent, and attitude (the complete correlation matrix is pre-
sented in Appendix A), but a Sheffe’s test does not show any of these variables to be significant across
treatment conditions (all p values > 0.153). Nonetheless, since these variables are significant predic-
tors of attitudes, any insight into how these variables can be influenced and how they interact with
the main variables tested in the study would provide valuable knowledge for those aiming to reduce
fraud and may provide a direction for future research.
Additional tests of moderation and mediation (significant results presented in Figs. 3–5) were per-
formed on all variables. Discussion of these additional analyses follows (see Table 3).

5.4.1. Awareness of enforcement

We suggested that strong sanctions and increased likelihood of detection may not be effective if
participants are not aware of the sanctions being enforced. One way participants might have been
made aware of potential sanctions is if recent prosecutions and details about the sentences handed
down following convictions are publicized. Awareness of prosecutions and sentences imposed may
influence perceptions about internal controls’ effectiveness at detecting fraud, perceptions about

Career Over(Z2)
P21 (.337) (p < .050) P32 (-.754) (p < .010)

Jail Time(Z1) Attitude(Z3)

P31 (-1.573) (p < .010)
P31’ (-1.319) (p < .010)

Fig. 4. Mediating effect of perceptions that participants’ career would be over due to serving time in jail.
 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458 451

Jail Time is Long (Z2)

P21 (.592) (p < .010) P32 (-.362) (p < .010)

Jail Time(Z1) Attitude(Z3)

P31 (-1.573) (p < .010)
P31’ (-1.369) (p < .010)

Fig. 5. Mediating effect of perceptions that the jail time potentially imposed is long.

Table 3
Effect of jail time and internal control on attitude for executives.

Jail time Internal control Main effect (jail time)

Weak Strong
Short (1 year) 4.800 4.889 4.845
(.702) (.523) (.438)
{5} {9} {14}
Medium (10 years) 3.750 3.625 3.688
(.555) (.555) (.392)
{8} {8} {16}
Long (20 years) 3.600 3.500 3.550
(.702) (.785) (.526)
{5} {4} {9}
Main effect (internal control) 4.050 4.005 4.027
(.379) (.365) (.263)
{18} {21} {39}

Mean (standard error) {sample size} across treatment conditions.

whether or not individuals will or will not serve a full sentence, and whether or not they will serve
jail time under harsh conditions, and so on. In addition, awareness of prosecutions and sentences
imposed may help shape other perceptions that influence individuals’ attitudes and intentions, such
as their perceptions that they will be caught and their future job prospects and social status will be
reduced.
Awareness of enforcement was measured by asking participants how aware they were of the pros-
ecution of and the details regarding the sentences handed down to executives at Enron and World
Com. Only 6 of the 123 participants who offered usable responses were unaware of these prosecutions
in general. As a result, only the second measure, awareness of the details of the prosecutions, was used
in the analyses. A correlation analysis shows significant relationships between awareness of enforce-
ment and perceptions that internal controls will catch fraudulent activity (p < 0.050), perceptions that
future job prospects are reduced by serving time in jail (p < 0.010), perceptions that social status is
reduced by serving time in jail (p < 0.050), perceptions that individuals will serve a full sentence if
convicted (p < 0.050), and perceptions that one’s career will be over if he/she spends time in jail
(p < 0.100), suggesting that these perceptions can be influenced by activities that raise awareness.
From an additional regression test, we find a statistically significant three-way interaction between
jail time, internal controls, and awareness of recent prosecutions, with attitude as the dependent var-
iable (p < 0.100). The three independent variables are negatively related to attitudes and enhance one
another. This finding suggests that controls enhance jail time to the extent that individuals are aware
of other incidences where individuals have been caught and sent to jail.

5.4.2. Other factors affecting internal controls

We suggested that internal controls will be effective only to the point where participants sense an
increased likelihood that they will get caught. Hence, we used a debriefing question that asked partic-
452 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458

ipants if they thought it likely they would get caught if they did not report the supplies expense cor-
rectly. This perception correlated significantly with participant intent, their perceptions of a referent
other’s intent, and their attitudes (all p < 0.050). However, the participants’ perceptions did not vary
across treatment conditions. Nonetheless, the additional analysis implies that this perception is an
essential component in understanding the effects of internal controls. Recall that the internal control
manipulation did not enhance the effect of jail time, but based on theory, it was expected that internal
control changes would change perceptions of the likelihood of getting caught, which should in turn
influence attitudes. A test of mediation, following the steps outlined in Baron and Kenny (1986), re-
veals that internal controls have an influence on attitudes, via the participants’ perception that the
internal controls are likely to catch their behavior (Fig. 3). When including the likelihood of getting
caught, the unstandardized regression coefficient for the relationship between internal controls and
attitude changed significantly (from 0.771 to 0.280) and the insignificance of the relationship be-
tween internal control and attitude suggests full mediation. Mediation was also confirmed by a Sobel
test (p < 0.010).
Along with the mediation test, the regression that tests the interaction between jail time and
perceived likelihood of getting caught (rather than internal controls) shows a significant interac-
tion on participant’s attitudes (p < 0.050). This lends additional support to the importance of the
perception that internal controls will result in participants getting caught. In addition, the lack of
interaction between jail time and internal controls in the initial tests of hypotheses may be due
to two other factors: (1) jail time was not effective beyond 10 years, suggesting that the interac-
tion between jail time and internal controls may only exist when jail times are shorter and (2)
experience. Although evidence is mixed, decision-making research in accounting finds that
experience has an influence on auditors’ control risk assessments (Ashton, 1974; Bonner, 1990;
Trotman and Wood, 1991). Experience might also extend beyond auditors to employees. Employees
with different experience levels may make different assessments about the effectiveness of inter-
nal controls.
To further examine the interaction between jail time and internal controls, a test of the hypothe-
sized model was performed over a subset of the original sample. The subset included only the
responses of participants that were given jail times of 1 and 10 years. The results do not change
substantially from the original hypotheses tests. There are significant main effects for jail time and
internal controls, but still no significant interaction in the test of the subset.
To further examine the effect of experience, we split the student data set into high- and
low-experience groups, and the original ANCOVA model used to test H1 and H3 was reana-
lyzed for each group. The results show that jail time (F = 8.551, P < 0.010) and internal con-
trols (F = 8.662, P < 0.010) are significant for the low-experience group, and the interaction
between jail time and internal control is also significant for the low-experience group
(F = 2.883, P < 0.100). Only jail time is significant for the high-experience group (F = 6.694,
P < 0.010).

5.4.3. Other factors affecting jail time

We also suggested that short jail sentences are effective when individuals feel merely going to jail
will affect future opportunities, which we labeled ‘‘social stigma”. Participants were presented with
three debriefing questions that attempted to capture various facets of social stigma by asking them
if their future job prospects would be reduced, if their social status would be adversely affected,
and if their career would be over if they spent time in jail. These items correlate significantly with par-
ticipants’ attitudes. Additional tests were performed to assess if these variables moderated or medi-
ated the influence of jail time on participants’ attitudes. Three regressions were run to test
moderation, and only the interaction term between jail time and social status is significant
(p < 0.050). Additionally, three mediation tests were performed on these items, but only perceptions
of one’s career being over partially mediates the effect of jail time on attitude as shown by a significant
change in the path between jail time and attitude (from 1.573 to 1.319). This was confirmed by a
Sobel test (p < 0.050) (Fig. 4).
 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458 453

The next group of debriefing questions assessed general perceptions about jail time. We proposed
that jail time will only be effective when participants feel the amount of potential jail time is long, will
actually be imposed, and will be served under harsh conditions. In regression tests of moderation, the
interaction terms between all three debriefing items and jail time on participant’s attitudes are signif-
icant (p < 0.050). Three more mediation tests were performed on these items but only perceptions of
jail time being long partially mediates the effect of jail time on attitude as shown by a significant
change in the path between jail time and attitude (from 1.573 to 1.319). This was confirmed by
a Sobel test (p < 0.050) (Fig. 5).
Finally, to test their affects in the mediation model relative to one another, a mediation analysis
was performed with perceptions of participants’ career being over (a proxy for economic effects)
and perceptions of jail time being long (a proxy for the severity of incarceration) entered in the regres-
sion model at the same time. When entered together, the path coefficient between jail time and atti-
tude reduces to 1.209 (p < 0.010) but only perceptions of one’s career being over remains significant
(p < 0.010) suggesting it is the more salient of the two factors.10

5.4.4. Analysis of executive data

Because of the difference in the results for high and low experienced student groups, a post
hoc test of data from 39 corporate financial executives was performed to confirm results. The
results nearly mirror those of the high experience student group. Only attitudes and jail time
have a significant effect, but to a diminishing degree. Our findings reveal mean attitudes of
4.84 for 1-year jail times, 3.69 for 10 year jail times, and 3.55 for 20 year jail times resulting
in a difference in attitudes of 1.15 when jail time is increased from 1 to 10 years but only a
mean difference in attitude of 0.14 when jail time is increased from 10 to 20 years. Our results
show a mean attitude of 4.05 when internal controls are weak and 4.00 when they are strong
resulting in a mean difference of 0.05, suggesting internal controls have no effect on executives.
Finally, attitudes are affected to a diminishing degree as jail time is increased for both the weak
and strong internal control groups thus failing to show that internal controls enhance the effec-
tiveness of jail time.

6. Discussion, limitations, and conclusion

The results of this study show that jail time can influence participants’ attitudes about financial
statement fraud. The mean attitude about committing financial statement fraud changed significantly
when jail time was increased from 1 to 10 years but not when jail time was increased from 10 to
20 years, suggesting that the incremental effect of more jail time diminishes as jail time is increased.
The results also indicate that jail time is effective because of lost future opportunities and social stig-
mas that arise from going to jail. Taken as a whole, our findings suggest that the effectiveness of jail
time on reducing attitudes about committing financial statement fraud is a function of the length of
jail time and perceptions that merely spending time in jail will have negative ramifications on one’s
career. Beyond its main effect, lengthening potential jail time influences participants perceptions of
both the physical and economic consequences associated with incarceration. This may explain why
jail time has a diminishing effect since the economic consequences, as we measured them, are strong
regardless of the length of time served.
SOX has essentially lengthened the potential jail time for committing financial statement
fraud to 20 years, from the 10 years imposed by the 1934 Act. Our results indicate the increased
jail time imposed by SOX may not offer any additional deterrence beyond the potential jail time
that was in place pre-SOX, while social and job market forces (e.g. the end of an executive’s
career) may offer additional deterrence. We assert that regulatory actions that reduce future
job opportunities may be more effective and less costly than sending individuals to jail for long-
er terms.

10
The perception that jail time is long remained marginally significant (p < 0.100).
454 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458

The lack of a significant interaction between jail time and the strength of internal control
structures was unexpected. We believe this may be explained by understanding participants’
experience and perceptions that their indiscretions will be detected by the internal controls.
GDT suggests that mechanisms that increase the likelihood of being detected can enhance the
effectiveness of sanctions. Following this logic, participants’ perceptions of the likelihood their
actions would be detected mediated the direct effect of jail time on attitudes, indicating that
perceptions about the likelihood of being detected are important factors in the role of internal
controls. In addition, internal control mechanisms can be overridden and the findings suggest
they are a useful deterrent to the extent that individuals feel such mechanisms are likely to un-
cover their illicit activities. This has important implications for practice, since the PCAOB, through
Audit Standard Five, has reduced requirements for assessments of internal controls, potentially
changing perceptions about the likelihood of internal controls uncovering financial statement
fraud, potentially weakening their effectiveness.
Although, the direct effect of internal controls was significant and a significant interaction existed
between internal controls and jail time for less experienced participants, this study has not fully
answered the question, ‘‘Why are less experienced workers deterred by control mechanisms to a greater
degree?” We conjecture that the difference is caused by experienced workers’ better understanding of
the short comings of internal control mechanisms, but this should be tested further. Nevertheless, we
believe the failure of internal controls to influence experienced students and executives is important
for two reasons: (1) financial statement fraud is typically perpetrated by managers and executives and
(2) increased internal control requirements are one of the most costly components of SOX. If the pri-
mary objective is fraud deterrence, then we must question whether or not money spent on internal
controls passes the cost-benefit test.
We speculate that jail time and internal controls could be enhanced by making the public aware of
prosecutions, which was suggested by the significant three-way interaction between jail time, internal
controls, and participants’ awareness of recent prosecutions. The three-way interaction implies that,
although the threat of punishment is enhanced by control mechanisms that increase people’s percep-
tions of the likelihood they will get caught, the threat is further enhanced if participants have seen
these mechanisms enforced.
Heightened awareness was also shown to relate significantly with perceptions that internal con-
trols would catch fraudulent financial reporting and various perceptions that may limit the effective-
ness of jail time. From this, we posit that financial statement fraud could be reduced further if these
perceptions could be changed. Making individuals aware of effective internal controls and severe pros-
ecutions might change perceptions, which could eventually result in changes in behavior. A conse-
quence of practical significance in our findings is that deterrence can be enhanced by publicizing
prosecutions, assuming those prosecutions are severe.
Our results are supportive of the editorial position stated by Ronen (2002, p. 283), ‘‘Prosecution
and punishment may not adequately deter wrongdoing, as intentional misrepresentation is difficult
to discover or prove. Overhauling the regulatory structure and adding layers of supervision and
monitoring by the government would be inefficient and socially wasteful.” Similar to Ronen’s posi-
tion, our study suggests that control mechanisms are discounted by experienced individuals, and
perceptions about jail time are a function of social and economic factors. Changing perceptions
through heightened awareness may make punishment and control mechanisms more effective.
This might be true even if the mechanisms are not enhanced themselves; rather they are merely
discounted less. Without changes in perceptions, the deterrence mechanisms within SOX may not
reduce fraud as hoped and may indeed be wasteful. Ronen asserts that market mechanisms such
as financial statement insurance would provide greater fraud deterrence. Financial statement insur-
ance serves as an economic sanction on auditors, the ‘‘external gatekeepers” of financial data, but
also has a potential to serve as an economic sanction on firms through increased insurance premi-
ums and a greater cost of capital when those costs are passed along. Our results support Ronan’s
idea that economic motivation and punishment may be the best place to focus deterrence efforts.
 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458 455

Hence, we offer that those convicted of fraud should be prohibited from serving in future executive
capacities.
Our study has a variety of limitations. First, we used a scenario. Although the design is quite com-
mon in ethics literature, there is still the possibility individuals will respond differently in a real sit-
uation. As sufficient data becomes available, future researchers could test the propositions made
here with archival studies. A combination of experimental and archival analyses would likely result
in a rich understanding of financial statement fraud deterrence.
Second, our findings in the supplemental analysis were measured through debriefing questions
that attempt to measure perceptions. Interesting information was revealed through this analysis,
but much of it should be studied further. For instance, social stigmas are important factors when
measured in lost future career opportunities, but social stigmas can manifest in a number of ways.
It would be interesting to test the concept of social stigma with a validated measure incorporating
many facets. We are unable to fully explain why the level of internal controls only affects less
experienced workers. This needs further analysis and although our study sheds some initial light
on the issue of internal controls, many questions remain. In addition, future research could also
experimentally manipulate the perception variables—such as the perception of the likelihood of
serving a full sentence, the perception of the harshness of time served, the awareness of enforce-
ment, and perceptions about whether or not internal control systems will catch fraudulent finan-
cial reporting—in order to test them more robustly. Future research could also examine the relative
salience of specific internal control mechanisms. Additionally, the lack of a nontreatment condition
limited the analysis of many of the debriefing items, and additional data collection for a nontreat-
ment group would allow analysis of those variables with or without the presence of jail time or
internal controls.
Finally, future research should incorporate factors that can further explain how attitudes are
formed. Attitudes are formulated through an assessment of an individual’s beliefs about the conse-
quences associated with the outcome of a behavior (including outcomes that could result in the indi-
vidual getting caught and going to jail), and the findings related to awareness of recent prosecutions
suggest an outcome effect where attitudes are shaped by observations of the action outcomes of oth-
ers who took part in similar behaviors. In addition, committing fraud does not come without risk to
the perpetrator, regardless of the outcome. Reputational capital is at risk, and examining fraud in
the same manner as other types of decisions made under conditions of uncertainty through use of a
framing process (Tversky and Kahneman, 1974, 1981) may add further insight into how attitudes
are shaped.
In conclusion, fraud deterrence can be approached from a behavioral perspective by imple-
menting mechanisms that attempt to influence individual attitudes and perceptions about nor-
mative behaviors related to committing fraud. SOX attempts to do just that by adding to
potential sanctions for committing financial statement fraud and increasing the requirements
for internal controls. Taken as a whole, we find that the threat of potential jail time can be
an effective mechanism for reducing financial statement fraud, but its effectiveness is influenced
by a wide range of social, environmental, and demographic factors. In the future, when evalu-
ating the overall effectiveness of SOX, researchers, practitioners, and regulators should take
these factors into account, because their influence may hold the key to the ultimate success
or failure of SOX.

Acknowledgments

Thanks to John Pearson, Scott McClurg, Jacob Rose, and Ania Rose for their guidance. Thanks to Bill
Pasewark for feedback on the experiment. Thanks to Dann Fisher and John Morris for your comments.
We also appreciate the departments of accounting at Kansas State University, Southern Illinois Univer-
sity Carbondale and Missouri State University and the editor and reviewers at the Journal of Account-
ing and Public Policy for their invaluable assistance.
 456
Appendix A. Correlation matrix

Jail Internal Likely Job Career Social Harsh Full Jail time Awareness Intent Ref. Attitude Subjective
time control caught prospects over status conditions sentence is long enforcement other’s norms
reduced intent

Jail time Pearson Corr. 1

Sig. (2-tailed)

Internal control Pearson Corr. 0.000 1

Sig. (2-tailed) 0.996

Likely caught Pearson Corr. 0.130 0.505** 1

J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458

Sig. (2-tailed) 0.152 0.000

Job prospects Pearson Corr. 0.160 0.076 0.173 1

reduced
Sig. (2-tailed) 0.077 0.401 0.056

Career over Pearson Corr. 0.200* 0.020 0.176 0.445** 1

Sig. (2-tailed) 0.027 0.825 0.051 0.000

Social status Pearson Corr. 0.064 0.097 0.051 0.563** 0.487** 1

Sig. (2-tailed) 0.480 0.288 0.574 0.000 0.000

Harsh conditions Pearson Corr. 0.145 0.091 0.172 0.122 0.202* 0.232** 1
Sig. (2-tailed) 0.109 0.315 0.057 0.178 0.025 0.010

Full sentence Pearson Corr. 0.061 0.227* 0.191* 0.141 0.190* 0.172 0.451** 1
Sig. (2-tailed) 0.503 0.011 0.034 0.120 0.035 0.057 0.000

Jail time is long Pearson Corr. 0.277** 0.071 0.190* 0.362** 0.290** 0.291** 0.196* 0.163 1
Sig. (2-tailed) 0.002 0.432 0.036 0.000 0.001 0.001 0.029 0.072

Awareness Pearson Corr. 0.098 0.102 0.213* 0.360** 0.166 0.194* 0.149 0.218* 0.097 1
enforcement
Sig. (2-tailed) 0.283 0.264 0.018 0.000 0.066 0.031 0.099 0.015 0.287

Intent Pearson Corr. 0.302** 0.101 0.192* 0.385** 0.305** 0.269** 0.197* 0.251 0.288** 0.189* 1
Sig. (2-tailed) 0.001 0.267 0.033 0.000 0.001 0.003 0.029 0.005 0.001 0.036

Ref. other’s intent Pearson Corr. 0.189* 0.146 0.198* 0.228* 0.198* 0.183** 0.098 0.296** 0.222* 0.224* 0.433** 1
Sig. (2-tailed) 0.037 0.107 0.028 0.011 0.028 0.043 0.281 0.001 0.014 0.013 0.000

Attitude Pearson Corr. 0.445** 0.134 0.318** 0.296** 0.435** 0.252** 0.289** 0.268** 0.326** 0.098 0.528** 0.378** 1
Sig. (2-tailed) 0.000 0.140 0.000 0.001 0.000 0.005 0.003 0.003 0.000 0.279 0.000 0.000

Subjective norms Pearson Corr. 0.068 0.001 0.165 0.173 0.250** 0.105 0.095 0.309** 0.106 0.067 0.309** 0.321** 0.340** 1
Sig. (2-tailed) 0.454 0.990 0.067 0.056 0.005 0.248 0.295 0.001 0.245 0.458 0.001 0.000 0.000

N = 123.
*
Correlation is significant at the 0.05 level (2-tailed).
**
Correlation is significant at the 0.01 level (2-tailed).
 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458 457

References

Ajzen, I., 1991. The theory of planned behavior. Organizational Behavior and Human Decision Processes 50 (2), 179–211.
Ajzen, I., Fishbein, M., 1980. Understanding Attitudes and Predicting Social Behavior. Prentice-Hall, Englewood Cliffs, NJ.
Ashton, R., 1974. An experimental study of internal control judgments. Journal of Accounting Research Spring 143, 57.
Association of Certified Fraud Examiners, 2006. Report to the Nation on Occupational Fraud and Abuse.
Baron, R., Kenny, A., 1986. The moderator–mediator variable distinction in social psychological research: conceptual, strategic,
and statistical considerations. Journal of Personality and Social Psychology 51 (6), 1173–1182.
Bebchuk, L., Fried, J., 2003. Executive compensation as an agency problem. Journal of Economic Perspectives 17, 71–92.
Beccaria, C., 1963. On Crime and Punishment. Bobbs Merril, Indianapolis, IN.
Becker, G., 1968. Crime and punishment: an economic approach. Journal of Political Economy 76, 169–217.
Beeler, J., Hunton, J., 2002. Contingent economic rents: insidious threats to audit independence. Advances in Accounting
Behavioral Research 5, 21–50.
Black, W., 2007. Worldwide: looters have a foothold. Journal of Accountancy January 32, 38.
Blair, M., Stout, L., 2001. Trust, trustworthiness, and the behavioral foundations of corporate law. University of Pennsylvania
Law Review 149, 1735–1810.
Bonner, S., 1990. Experience effects in auditing: the role of task-specific knowledge. The Accounting Review 65 (1), 72–92.
Braithwaite, J., 1985. White collar crime. Annual Review of Sociology 11, 1–25.
Braithwaite, J., Geis, G., 1982. On theory and action for corporate crime control. Crime and Delinquency 28 (2), 292–314.
Braithwaite, J., Makkai, T., 1991. Testing and expected utility model of corporate deterrence. Law and Society Review 25 (1), 7–
40.
Carpenter, T., Reimers, J., 2005. Unethical and fraudulent financial reporting: applying the theory of planned behavior. The
Journal of Business Ethics 60, 115–129.
Chang, K., 1998. Predicting unethical behavior: a comparison of the theory of reasoned action and the theory of planned
behavior. Journal of Business Ethics 17, 1825–1834.
Clement, R., Krueger, J., 2000. The primacy of self-referent information in perceptions of social consensus. British Journal of
Social Psychology 39 (2), 279–299.
Cohen, J., Pant, L., Sharp, D., 1993. Validation and extension of a multidivisional ethics scale. Journal of Business Ethics 12, 13–
26.
Cohen, J., Pant, L., Sharp, D., 2001. An examination of the differences in ethical decision-making between Canadian business
students and accounting professionals. Journal of Business Ethics 30, 319–336.
Cohen, J., Krishnamoorthy, G., Wright, A., 2008. Corporate governance in the post Sarbanes–Oxley era: auditors’ experiences.
Working paper, Boston College.
Cullinan, C., Du, H., Wright, G., 2006. A test of the loan prohibition of the Sarbanes–Oxley Act: are firms that grant loans to
executives more likely to misstate their financial results? Journal of Accounting and Public Policy 25 (4), 485–497.
Dechow, P., Sloan, R., Sweeney, A., 1996. Causes and consequences of earnings manipulation, an analysis of firm subject to
enforcement actions by the SEC. Contemporary Accounting Research 13, 1–36.
DeZoort, F., Hermanson, D., Houston, R., 2008. Audit committee member support for proposed audit adjustments: pre-SOX
versus post-SOX judgments. Auditing: A Journal of Practice and Theory 27 (1), 85–104.
Engle, E., Hayes, R., Wang, X., 2004. The Sarbanes–Oxley act and firms going private decisions. <http://
accounting.wharton.upenn.edu/seminar/ehw-gp04.pdf>.
Goldman, E., Slezak, S., 2006. An equilibrium model of incentive contracts in the presence of information manipulation. Journal
of Financial Economics 80, 603–626.
Gordon, L., Loeb, M., Lucyshyn, W., Sohail, T., 2006. The impact of the Sarbanes–Oxly Act on corporate disclosures of information
security activities. Journal of Accounting and Public Policy 25 (5), 503–530.
Grasmick, H., Tittle, C., Bursik, R., Arneklev, B., 1993. Testing the core implications of Gottfredson and Hirschi’s general theory of
crime. Journal of Research in Crime and Delinquency 30 (5), 5–29.
Harrington, S., 1996. The effect of code of ethics and personal denial of responsibility on computer abuse judgments and
intentions. MIS Quarterly 20 (3), 257–278.
Healy, P., 1985. The effect of bonus schemes on accounting decisions. Journal of Accounting and Economics 33, 85–107.
Johnson, S., Ryan, H., Tian, Y., 2006. Managerial incentives and corporate fraud: the sources of incentives matter. EFA 2006
Zurich Meetings <http://ssrn.com/abstract=395960>.
Kaminski, K., Wetzel, T., Guan, L., 2004. Can financial ratios detect fraudulent financial reporting? Managerial Auditing Journal
19 (1), 15–28.
Klepper, S., Nagin, D., 1989. Tax compliance and perceptions of the risks of detection and criminal prosecution. Law and Society
Review 23, 209–240.
Latham, C., Jacobs, F., 2000. Monitoring and incentive effects influencing misleading disclosures. Journal of Managerial Issues 12,
169–187.
McEnroe, J., 2006. Perceptions of the effect of Sarbanes–Oxley on earnings management practices. Research in Accounting
Regulation 19, 137–157.
Mikulineer, M., Horesh, N., 1999. Adult attachment style and the perception of others: the role of projective mechanisms.
Journal of Personality and Social Psychology 79 (6), 1022–1034.
Nagin, D., Paternoster, R., 1993. Enduring individual differences in rational choice theories of crime. Law and Society Review 27
(3), 467–496.
Nagin, D., Pogarsky, G., 2001. Integrating celerity, impulsivity, and extralegal sanction threats into a model of general
deterrence: theory and evidence. Criminology 39 (4), 865–891.
Nyberg, A., 2003. Sticker shock: the true cost of Sarbanes–Oxley compliance. CFO September 51, 62.
O’Brien, P., Bhushan, R., 1990. Analyst following and institutional ownership. Journal of Accounting Research Supplement 55, 73.
Perino, M., 2002. Enron’s legislative aftermath: some reflections on the deterrence aspects of the Sarbanes–Oxley Act of 2002. St.
John’s Law Review 76 (4), 671–698.
458 J.C. Ugrin, M.D. Odom / J. Account. Public Policy 29 (2010) 439–458

Persons, O., 1995. Using financial statement data to identify factors associated with fraudulent financial reporting. Journal of
Applied Business Research 11 (3), 38–46.
Polinsky, A., Shavell, S., 1999. On the disutility and discounting of imprisonment and the theory of deterrence. Journal of Legal
Studies 28 (1), 1–16.
Potter, G., 1992. Accounting earnings announcements, institutional investor concentration, and common stock returns. Journal
of Accounting Research 30 (1), 146–155.
Rezaee, Z., 2003. Causes, consequences, and deterrence of financial statement fraud. Critical Perspectives in Accounting 16 (3),
277–298.
Ronen, P., 2002. Policy reforms in the aftermath of accounting scandals. Journal of Accounting and Public Policy 21 (4–5), 281–
286.
Ruvolo, A., Fabin, J., 1999. Two of a kind: perceptions of own and partner’s attachment characteristics. Personal Relationships 6
(1), 57–79.
Schwartz, S., 1973. Normative explanations of helping behavior: a critique, proposal, and empirical test. Journal of Experimental
Social Psychology 9, 349–364.
Schwartz, S., 1977. Normative influences on altruism. In: Berkowitz, L. (Ed.), Advances in Social Psychology. Academic Press,
New York, pp. 221–279.
Simpson, S., Koper, C., 1992. Deterring corporate crime. Criminology 30 (3), 347–375.
Singleton, T., King, B., Messina, F., Turpen, R., 2003. Pro-ethics activities: do they really reduce fraud? The Journal of Corporate
Accounting and Finance 14 (6), 85–94.
Skousen, C., Wright, C., 2008. Contemporaneous risk factors and the prediction of financial statement fraud. Journal of Forensic
Accounting 9, 37–62.
Smith, E., 1997. Private selves and shared meanings: or forgive us for our projections as we forgive those who project onto us.
Psychodynamic Counseling 3 (2), 117–131.
Trotman, K., Wood, R., 1991. A meta-analysis of studies on internal control judgments. Journal of Accounting Research 29 (1),
180–192.
Tversky, A., Kahneman, D., 1974. Judgment under uncertainty: heuristics and biases. Science 185 (4157), 1124–1131.
Tversky, A., Kahneman, D., 1981. The framing of decisions and the psychology of choice. Science 211 (4481), 452–458.
Tyler, T., Blader, S., 2005. Can businesses effectively regulate employee conduct? The antecedents of rule following in work
settings. Academy of Management Journal 48 (6), 1143–1158.
Weber, J., 1992. The empirical quest for normative meaning: empirical methodologies for the study of business ethics. Business
Ethics Quarterly 2 (2), 137–160.
Williams, K., Hawkins, R., 1986. Perceptual research on general deterrence: a critical review. Law and Society Review 20 (4),
545–572.