Description
This study examines an aspect of earnings management that we refer to as audit management.
We define audit management as a client’s strategic use of diversions to decrease the
likelihood that auditors will discover earnings management during the audit. Specifically,
we examine whether diverting auditors’ attention to either clean financial statement
accounts or accounts that contain other errors affect an auditor’s ability to uncover earnings
management. Auditors performed analytical review, searching financial statements
for unusual fluctuations suggestive of errors. Following prior studies, we seeded an intentional
accounting error which created an unusual fluctuation that allowed the client to
meet an earnings target.
Managing audits to manage earnings: The impact of diversions
on an auditor’s detection of earnings management
Benjamin L. Luippold
a,?
, Thomas Kida
b
, M. David Piercey
b
, James F. Smith
b
a
Babson College, United States
b
University of Massachusetts Amherst, Isenberg School of Management, United States
a b s t r a c t
This study examines an aspect of earnings management that we refer to as audit manage-
ment. We de?ne audit management as a client’s strategic use of diversions to decrease the
likelihood that auditors will discover earnings management during the audit. Speci?cally,
we examine whether diverting auditors’ attention to either clean ?nancial statement
accounts or accounts that contain other errors affect an auditor’s ability to uncover earn-
ings management. Auditors performed analytical review, searching ?nancial statements
for unusual ?uctuations suggestive of errors. Following prior studies, we seeded an inten-
tional accounting error which created an unusual ?uctuation that allowed the client to
meet an earnings target. We manipulated whether management provided a diversionary
statement that explicitly identi?ed risk in other areas of the audit, and whether those areas
were clean or contained other detected errors that had no impact on earnings. We ?nd that
auditors’ earnings management detection is worst when they are diverted to clean
accounts and best when auditors are diverted to accounts that contain other errors. Our
results suggest that managers can potentially exploit an audit management tactic as simple
as a diversion to a clean area to reduce auditors’ effectiveness at detecting earnings man-
agement. The implications of these ?ndings for audit and decision making research are
discussed.
Ó 2014 Elsevier Ltd. All rights reserved.
Introduction
Earnings management has been a topic of great interest
in both the popular press and academic literature (e.g.,
Chen, Kelly, & Salterio, 2011; Dechow, Hutton, Kim, &
Sloan, 2012; Guerrera, 2012). In fact, attempts to manipu-
late ?nancial performance have become so widespread
that books have been written on earnings management
strategies (e.g., Giroux, 2003; McKee, 2005). This study dis-
cusses an aspect of earnings management that we refer to
as audit management. We de?ne audit management as a
client’s strategic use of diversions to decrease the likeli-
hood of auditors discovering managed earnings during
the audit. Evidence suggests that managers strategically
attempt to conceal earnings management (e.g., Beasley,
Carcello, Hermanson, & Neal, 2010; Bowlin, Hobson, &
Piercey, 2014; Knapp, 2010).
Our study investigates whether managers who manipu-
late earnings can successfully employ diversions to in?u-
ence auditors’ detection of unusual ?uctuations during
analytical review. That is, we investigate whether diversion-
ary statements made by the client (i.e., identifying areas of
riskinthe ?nancial statements tolurethe auditor awayfrom
managed earnings) affect an auditor’s detection of managed
earnings containedelsewhere inthe ?nancial statements. Inhttp://dx.doi.org/10.1016/j.aos.2014.07.005
0361-3682/Ó 2014 Elsevier Ltd. All rights reserved.
?
Corresponding author. Address: Babson College, Accounting & Law
Division, Babson Park, MA 02457, United States. Tel.: +1 (781) 239 5995;
fax: +1 (781) 239 5930.
E-mail addresses: [email protected] (B.L. Luippold), tkida@
isenberg.umass.edu (T. Kida), [email protected] (M.D. Piercey),
[email protected] (J.F. Smith).
Accounting, Organizations and Society 41 (2015) 39–54
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
an experiment, we seeded an earnings management error
(i.e., an intentional misstatement used to meet an earnings
target) into?nancial statements, creating anunusual ?uctu-
ationfor auditors to detect.
1
We test whether auditors detect
earnings management when they are diverted to: (1)
accounts that contain no errors, or (2) accounts that do con-
tain errors, but have no impact on earnings.
In the study, auditors completed analytical review pro-
cedures on the ?nancial statements of a hypothetical client
in order to determine if any unusual ?uctuations (sugges-
tive of errors) were present. In all conditions, an earnings
management error (which reduced compensation expense
and accruals) was embedded into the ?nancial statements,
creating an unusual ?uctuation in that account that
resulted in the client meeting analysts’ forecasted earn-
ings. We manipulated whether or not management pro-
vided a diversionary statement that informed the auditor
of a personnel change in the department responsible for
non-current assets. This statement was designed to elevate
the perceived misstatement risk in that area and lure the
auditor away from the earnings manipulation. We also
manipulated whether the accounts in this other area iden-
ti?ed by the diversionary statement were clean or con-
tained other errors that offset, and therefore had no
impact on earnings.
Managers may be motivated to divert auditors to areas
that contain, or do not contain, other errors. For example, if
managers point auditors to ostensibly risky areas that are
clean, auditors may conclude that the client’s accounts
are likely to be accurate in other areas as well. This would
most likely result in a strong diversion effect. Conversely,
management may want to direct auditors to areas that
contain other errors, thinking that these other errors may
occupy their attention, leading auditors to feel satis?ed
that they are detecting misstatements and ‘‘doing their
job,’’ resulting in auditors feeling less compelled to dis-
cover other errors. However, auditors are also trained to
practice professional skepticism (Nelson, 2009;
Quadackers, Groot, & Wright, 2014), and the diversion to
the other errors should elevate their sensitivity to the risk
of material misstatement in the remainder of the ?nancial
statements, resulting in greater overall audit effort and a
greater likelihood that they would ?nd the earnings
manipulation. We therefore investigate the impact of man-
agement intentionally directing auditors to both clean
accounts and accounts containing errors.
Our ?ndings are consistent with our predictions. Specif-
ically, we ?nd that auditors’ detection of earnings manage-
ment was worst when they were diverted to clean ?nancial
statement accounts, and best when they were diverted to
accounts containing other errors, with earnings manage-
ment detection in between these levels when no diversions
were used (whether other errors were present or not).
Overall, these results suggest that if management directs
auditors to accounts that contain errors, the discovery of
those errors heightens their sensitivity to errors in other
areas of the audit. However, if auditors are directed to
clean accounts, the use of diversionary statements can
deter auditors from ?nding earnings management. These
?ndings have signi?cant implications for auditors and
decision researchers in general.
Testing diversions to accounts both with and without
other errors allows us to demonstrate signi?cantly differ-
ent reactions from auditors to two basic strategies that
management could use to divert auditors from an area
used to manage earnings. The different effects also provide
possible insight into why diversions to clean areas may be
an effective means of concealing earnings management. On
one hand, even though the other errors do not impact earn-
ings, diverting auditors toward them increases auditors’
sensitivity to the risk of material misstatement in other
areas of the audit. In contrast, diverting auditors to osten-
sibly risky areas that turn out to be clean appears to make
those auditors less vigilant in their search for errors else-
where in the ?nancial statements. These results suggest
that managers can potentially exploit an audit manage-
ment tactic as simple as a diversion to a clean area because
such a diversion reduces auditors’ effectiveness at detect-
ing earnings management elsewhere in the ?nancial state-
ments. This extends the literature on auditor skepticism by
identifying one type of claim that managers could make to
mislead auditors without raising red ?ags (Nelson, 2009;
Quadackers et al., 2014). Additionally, it contributes to
the accounting literature on auditing and earnings man-
agement with evidence of how managers can conceal
material misstatements from auditors in order to manage
earnings (Beasley et al., 2010; Beasley, Carcello,
Hermanson, & Lapides, 2000; Boone, Khurana, & Raman,
2012; Caramanis & Lennox, 2008; Chen et al., 2011).
Our study also contributes to decision making research
(in both psychology and auditing) that investigates the
effectiveness of diversionary tactics. As we explain in Sec-
tion ‘Background literature’, psychology literature on dis-
traction suggests that diversions to other errors would be
an effective means of concealing earnings management.
In contrast, our ?nding that earnings management detec-
tion is greatest when auditors are diverted to accounts that
contain errors suggests that our context provides a bound-
ary condition to the predictions of the psychology litera-
ture on distractions, based on the task-speci?c
experience of auditors and their reaction to the other
errors. Furthermore, our ?ndings contribute to the
accounting and psychology literature on information pur-
suit effects (Bastardi & Sha?r, 1998, 2000; Nelson &
Tayler, 2007; Redelmeier, Sha?r, & Aujla, 2001), by demon-
strating how management diverting auditors to search
other accounts can amplify auditors’ reactions to what is
(or to what is not) in those other accounts. Thus, we not
only contribute to the accounting literature on earnings
management and auditing, we also contribute to the gen-
eral judgment and decision making literature on diversion,
distraction, and information pursuit effects.
Diversions can have important practical implications
beyond the setting of deliberate earnings management.
For example, even when managers are not deliberately
managing earnings with a particular account, they may
still prefer that auditors pay more attention to areas where
1
While auditing standards typically characterize unintentional mis-
statements as errors and intentional misstatements as fraud, we use the
term error in its more generic sense to refer to any departure from
accuracy.
40 B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54
they are less likely to ?nd problems. As a result, they may
divert auditors from higher risk areas to accounts that they
believe are not at risk of misstatement. Our ?ndings sug-
gest that auditors would be susceptible to these diversions
as well. However, to the extent that errors can occur any-
where, managers might mistakenly direct auditors’ atten-
tion to an area with errors, which could back?re on
managers. Even more broadly, managers may inadver-
tently direct auditors’ attention from an account that
materially misstates earnings to another account that does
not. No matter the cause of the diversion, our results dem-
onstrate that such diversions signi?cantly in?uence an
auditor’s detection of a material misstatement of earnings
elsewhere in the ?nancial statements.
In the next section, we introduce background literature
related to both earnings and audit management, and dis-
cuss relevant theories related to our research. After the
background literature, we describe the method, while the
?nal two sections present the results and concluding
remarks, respectively.
Background literature
Earnings management
Earnings management refers to ?nancial reporting
practices designed to achieve desired or favorable ?nancial
results (e.g., smoothing earnings, meeting earnings targets)
(Bouillon, 2007; Jackson & Pitman, 2001; McKee, 2005;
Millstein, 2005). Management faces several pressures, such
as meeting analysts’ forecasts, which may prompt them to
resort to such practices (Duncan, 2001). Evidence suggests
that these short-term pressures can take priority over
long-term economic growth, as research has found that
executives sometimes sacri?ce economic value to smooth
earnings or hit an earnings target (Bhojraj & Libby, 2005;
Graham, Harvey, & Rajgopal, 2005).
Archival research provides substantial evidence that
earnings management occurs (e.g., Dechow et al., 2012).
For example, several studies have examined speci?c
accrual accounts that clients use to manage earnings
(e.g., Dhaliwal, Gleason, & Mills, 2004; Marquardt &
Wiedman, 2004). More striking, however, is evidence sug-
gesting that managers sometimes resort to fraudulent
measures to manage earnings (e.g., Beasley, Carcello, &
Hermanson, 1999; Beasley et al., 2000, 2010; Farber,
2005; Jones, Krishnan, & Melendrez, 2008).
2
However, in
order for management to successfully report over-aggressive
or fraudulent earnings, auditors must fail to discover how
and where income is being manipulated. Yet, how managers
can successfully divert auditors’ attention from earnings
manipulations remains an important unanswered question
for researchers (Bell, Peecher, & Solomon, 2005; Peecher,
Schwartz, & Solomon, 2007).
Audit management
We de?ne audit management as a client’s strategic use
of techniques to reduce the likelihood that auditors will
identify or recognize managed earnings during the audit.
Managing the audit may include a variety of methods.
For instance, managers may frame evidence in certain
ways to manipulate the level of perceived risk (e.g.,
Jamal, Johnson, & Berryman, 1995; Johnson, Grazioli, &
Jamal, 1993). They may provide the auditor with incom-
plete or incorrect information to cover-up questionable
accounting practices, or they may use diversions to prevent
the auditor from uncovering earnings management, which
is the primary focus of this study.
Diversions are methods designed to direct an auditor’s
attention away from a certain audit area. For example,
management may use diversionary statements which
identify speci?c areas of risk in other areas of the ?nancial
statements in an effort to lure the auditor away from the
accounts used to manage earnings. In this study, we inves-
tigate the impact of diversions to areas that do not contain
errors, or to areas where errors exist, but have no impact
on earnings.
Managers could identify a clean area of the ?nancial
statements as high risk, hoping that auditors would then
conclude that the rest of the ?nancial statements are likely
to be error-free as well. On the other hand, managers could
direct auditors to areas containing errors for several rea-
sons. The other errors may occupy auditors’ attention,
and more attention paid to one area of the audit may result
in less attention paid to other areas. In addition, auditors
may feel satis?ed that they have found errors, making
them less likely to search for earnings management else-
where in the ?nancial statements. Finally, management
may feel that pointing out areas that lead to error discov-
ery may increase the trust that auditors have in them,
resulting in auditors performing less work in areas that
management suggests are problem free.
3
However, diver-
sions to other errors may also have the opposite effect. If
management diverts an auditor to another account by iden-
tifying it as high risk, and yet apparently did nothing to
remove errors there, the diversion to those errors would
send an especially negative signal about management’s
internal controls, which could lead auditors to search more
extensively for errors in other areas of the audit. As a result,
we investigate auditors’ detection of earnings management
when they are diverted to areas that contain and do not con-
tain errors.
While clients may attempt to ‘‘manage’’ many different
aspects of the audit, an area that is of particular interest is
analytical review. Analytical review is used to determine
the extent of required detailed testing in different audit
areas, and sometimes is the only audit procedure used to
test certain accrual based accounts (AICPA, 2012;
2
Dyck, Morse, and Zingales (2010) estimate archivally that at least one
?nancial reporting fraud is ongoing at any time in at least 11.2–13.2% of
public companies with more than $750 million in assets, and that managers
successfully conceal a large majority of these frauds for some time from
auditors, SEC enforcement, and other governance mechanisms.
3
Our conversations with practitioners suggest that this and similar
tactics occur in practice. For example, a former manager of a technology
company indicated that, when auditors found error corrections that would
reduce earnings, he would direct them toward other error corrections that
would increase earnings. Similarly, an audit partner indicated that man-
agers may indeed see the audit as a diversionary game.
B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54 41
Ricchiute, 2006) and income statement accounts (FRC,
2011a). For example, unless speci?c risks are identi?ed
that warrant detailed testing, compensation accruals are
typically audited using only some variant of analytical
review (Ricchiute, 2006).
4
Thus, if such an account contains
errors that go undetected during analytical review, there
may not be subsequent procedures in the audit plan to
detect them. Failure to highlight the highest-risk accounts
during analytical review impacts planning, risk assessment
and substantive testing throughout the remainder of the
audit (FRC, 2011a; Messier, Simon, & Smith, 2013; PCAOB,
2010b).
Regulators indicate that analytical review is a recurring
problem in their inspections of audit ?rms (e.g., FRC,
2011a, 2011b, 2012b; IFIAR, 2012; PCAOB, 2008b).
5
These
regulatory ?ndings include auditors’ tendency to perform
analytical procedures in insuf?cient depth, as well as their
tendency to rely on and accept managers’ claims during ana-
lytical review without corroboration (e.g., FRC, 2011a,
2011b, 2012a; PCAOB, 2008a). Trompeter and Wright
(2010) report that auditors are increasingly using analytical
review to reduce the amount of detailed audit tests, using
less experienced auditors to perform analytical review, and
increasingly relying on their clients’ uncorroborated claims
for guidance during analytical review. These problems sug-
gest that auditors could be vulnerable to tactics used by
managers during analytical reviewto divert themaway from
managed earnings, especially if done on the pretext of
directing them to other risky areas. If successful, such tactics
could lead auditors to signi?cantly reduce or even eliminate
further detailed tests of the accounts from which they were
diverted (FRC, 2011a). Thus, consistent with prior research,
we focus on analytical review, since failure to detect mis-
statements during analytical review can in?uence the
planned detailed tests and trajectory of the rest of the audit
(e.g., Asare, Trompeter, & Wright, 2000; Asare & Wright,
2003; Brewster, 2011; Ismail & Trotman, 1995; Knapp &
Knapp, 2001; Knechel, Salterio, & Kochetova-Kozloski,
2010; Luippold & Kida, 2012; Messier et al., 2013; Moreno,
Bhattacharjee, & Brandon, 2007; Peecher, Piercey, Rich, &
Tubbs, 2010; Trompeter & Wright, 2010; Trotman &
Wright, 2012; Yip-Ow & Tan, 2000).
Given that managers may attempt to manage the audit,
the question therefore arises, can managers employ diver-
sionary tactics that allow them to effectively manage earn-
ings? We ?rst review relevant theories and research to
develop our predictions. On one hand, prior psychological
research on distraction suggests that diversions to both
clean accounts and to accounts containing other errors
should be effective. On the other hand, audit practice and
research suggests that auditors display professional skepti-
cism and are likely to react to discovering errors (even
those with no impact on earnings) as a red ?ag (Nelson,
2009; Quadackers et al., 2014), which may heighten their
search for additional errors in the ?nancial statements.
Hypotheses development
Psychological research on distractions and diversions
suggests that diversions inhibit performance. Studies on
persuasion have found that diversions make individuals
more susceptible to agreeing with the arguments of others,
as they detrimentally affect comprehension (Baron, Baron,
& Miller, 1973; Festinger & Maccoby, 1964; Petty, Wells, &
Brock, 1976; Watts & Holt, 1979; Zimbardo, Snyder,
Thomas, Gold, & Gurwitz, 1970). Similarly, research exam-
ining the Elaboration Likelihood Model (for attitude forma-
tion) indicates that diversions make cognitive processing
more dif?cult, resulting in more peripheral (shallow) infor-
mation processing (Petty & Cacioppo, 1986; Street,
Douglas, Geiger, & Martinko, 2001). Further, cognitive
research suggests that diversions consume an individual’s
attention, and since attention is limited, less is available
to process important information (Kahneman, 1973;
Sagarin, Britt, Heider, Wood, & Lynch, 2003). In a classic
study on inattentional blindness, participants viewed a
video of individuals passing around a basketball, and were
instructed to count the number of passes (Simons &
Chabris, 1999). In the video, a person in a gorilla suit
walked through the group, stopped, beat their chest and
exited. Notably, over half of the participants never saw
the gorilla, as they were too distracted by the task-at-hand.
Diversions also leave individuals feeling less compelled to
discover other issues, due to their focus on what they have
discovered. For example, in successful illusions, magicians
divert attention to speci?c distracting items (e.g., smoke,
noise, and ?ashes of light) to occupy an audience’s atten-
tion and inhibit their ability to uncover the ‘‘tell’’ of the
trick (e.g., Freudenburg & Alario, 2007; Kuhn & Tatler,
2005; Kuhn, Tatler, Findlay, & Cole, 2008).
On the other hand, research on information pursuit
effects in both psychology and accounting suggests that
the effect of diversions on the performance of professional
auditors may either inhibit or enhance performance
depending upon the context considered. Speci?cally, the
information pursuit literature suggests that individuals’
beliefs are based on not just the information that supports
that belief, but also on whether they explicitly and actively
searched a speci?c area for that information. As a result,
searching a highlighted area for information ampli?es indi-
viduals’ reactions to the information that they ?nd there,
above and beyond what their reactions would be based on
the information alone (Bastardi & Sha?r, 1998, 2000;
Redelmeier et al., 2001). This ampli?ed reaction to informa-
tion sought out in a speci?c area is an information pursuit
effect. In our auditing context, this suggests that diverting
auditors directly toward a particular clean area of the audit
will amplify their reaction to the lack of errors found there.
Without a diversion from management that highlights a
speci?c area as high risk, any information pursuit effects
that may occur would be diluted across accounts, lessening
the impact of any single ?nancial statement account. Incon-
trast, if management directs auditors to search a speci?c,
4
In fact, when pretesting this study’s experimental materials, a Big Four
audit manager commented on how an error in compensation would
probably go undetected if not uncovered at this stage.
5
For example, we examined all regulatory inspection reports for the ?ve
largest accounting ?rms in the United States, dating from 2004 to 2007, and
found that thirteen of the 19 reports (68%) identi?ed de?ciencies in
analytical review (PCAOB, 2008a). Similarly, the IFIAR (2012) reports that
problems with analytical review is one of the top recurring themes in
regulatory inspection ?ndings internationally.
42 B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54
allegedly higher risk audit area, and the search reveals that
the accounts are clean, auditors would tend to react more
strongly to their ?ndings in those speci?c accounts than
they would if they had not been highlighted. The auditors
may thenbe more likely to believe that other ?nancial state-
ment accounts are clean as well, which would inhibit audi-
tors’ performance in detecting earnings management
elsewhere in the ?nancial statements.
Nelson and Tayler (2007) and Smith, Tayler, and Prawitt
(2011) provide evidence that information pursuit effects
occur in accounting and auditing settings. For example,
Smith et al. (2011) show that auditors are more persuaded
by audit evidence uncovered as a result of an active search
in a speci?c, highlighted area. Nelson and Tayler (2007)
investigate possible causes of information pursuit effects,
and ?nd evidence that a process they refer to as reconcili-
ation drives the effects. During reconciliation, decision
makers who proactively pursued speci?c information will
then process its potential implications within the context
and react more to that information. This reconciliation
‘‘potentially renders the pursued information more salient,
so the information may demand more attention in the
judgment process’’ (p. 739). Related psychology research
on salience similarly suggests that managers’ highlighting
a speci?c area as high-risk would tend to amplify auditors’
reactions to a lack of errors found there (cf. Fiske, 1982;
Madan & Spetch, 2012; Taylor & Fiske, 1978). Finally, if
an ostensibly high-risk area of the audit turns out to be
clean, auditors may feel more trusting of management
regarding areas of the ?nancial statements not identi?ed
as risky (cf. Bowlin, Hales, & Kachelmeier, 2009; Bowlin
et al., 2014; King, 2002) and search for errors in the rest
of the ?nancial statements less vigilantly.
6
In contrast, a diversion to an ostensibly risky area in
which errors are discovered by the auditor is likely to have
a very different effect on auditor behavior. While prior psy-
chology research on distraction suggests that such diver-
sions would serve to inhibit the auditors’ effectiveness at
uncovering earnings management, the information pursuit
literature and other accounting research suggests that
diversions to areas in which auditors uncover errors would
have the opposite effect. Speci?cally, just as a purposeful
pursuit of evidence in a clean audit area that was speci?-
cally highlighted as high risk could amplify auditors’ reac-
tions to the lack of errors there, a similar diversion from
management that highlights speci?c accounts with errors
would similarly amplify auditors’ reactions to the errors
that they do ?nd there (cf. Nelson & Tayler, 2007;
Redelmeier et al., 2001). Even though the other errors have
no impact on earnings, they are not neutral with respect to
the audit. By increasing auditors’ reactions to the other
errors, the diversions to the other accounts would increase
the likelihood that the auditors would subsequently
uncover the earnings management. Management’s diver-
sion makes those speci?c accounts more salient, which
would tend to amplify auditors’ reactions to what they ?nd
there (cf. Fiske, 1982; Madan & Spetch, 2012; Taylor &
Fiske, 1978).
7
Combining the information pursuit literature with the
auditing literature suggests speci?c ways that diverting
auditors to other errors would amplify auditors’ reactions
to them. In particular, Nelson and Tayler (2007) predict
and ?nd that proactively pursuing information leads deci-
sion makers to consider potential implications of the infor-
mation in contextually speci?c ways, and therefore react to
the pursued information more strongly. The auditing liter-
ature suggests contextually speci?c ways in which these
reactions might occur within our setting. For example, if
management speci?cally directs auditors to accounts that
they believe are high risk, and yet apparently did little or
nothing to detect and remove any errors there, this diver-
sion sends an especially negative signal about manage-
ment’s internal controls over its ?nancial statement
preparation (COSO, 2012; PCAOB, 2007), elevating the
auditors’ beliefs about the risk of material misstatement
in the rest of the ?nancial statements (IFAC, 2009;
PCAOB, 2010a). Auditors are trained to exercise profes-
sional skepticism (Nelson, 2009; Quadackers et al., 2014;
Smith & Kida, 1991), suggesting that auditors would react
to a diversion from management to errors that the client
did not detect and remove as a signi?cant red ?ag for con-
trols over the ?nancial statements as a whole. If auditors
react to the discovery of other errors as a red ?ag, then
having been directed to a speci?c audit area to search for
such errors will likely strengthen their reaction to ?nding
them (cf. Nelson & Tayler, 2007; Redelmeier et al., 2001).
And, given that the reaction prescribed by audit standards
to uncovering such errors is to heighten concerns regard-
ing the risk of errors elsewhere in the ?nancial statements,
it is likely that auditors will be more vigilant in their search
for errors in the rest of the ?nancial statements.
Thus, while prior psychological research on distractions
suggests that diversions would be generally effective at
concealing earnings management, the information pursuit
literature in psychology and accounting, as well as other
accounting research and practice, suggests that the effect
of diversions would be quite different depending on
whether auditors are diverted to clean accounts or to
accounts containing other errors. Speci?cally, the diversion
frommanagement to other accounts would tend to amplify
auditors reactions to what is (or is not) in those other
accounts. Without a diversion to the other accounts, audi-
tors would be (if anything) more likely to detect earnings
management elsewhere in the ?nancial statements when
6
In fact, there is evidence of managers attempting to conceal informa-
tion by using diversions to clean areas. For example, to perpetuate its fraud,
the ZZZZ Best Company management repeatedly used diversionary state-
ments that directed auditors to accounts in its legitimate carpet-cleaning
subsidiary in order to divert attention away from its fraudulent restoration
business (Knapp, 2010). In addition, managers of HealthSouth and Crazy
Eddie allegedly concealed fraud from auditors with overt efforts to divert
auditors’ attention away from the fraud (Knapp, 2010).
7
Theory does not require that the diversions impact the rate at which
the other errors are discovered in order to in?uence auditors’ reactions to
them. Even if the other errors are discovered relatively easily and at just as
high of a rate without the diversion there, information pursuit theory still
predicts a stronger reaction to what is in those other accounts when they
are speci?cally directed to search there. Further, when no other errors are
present, diversions also cannot impact the rate at which the other errors are
discovered (since there are none to be discovered). Yet information pursuit
theory still predicts a stronger reaction to the lack of errors with a diversion
to those accounts.
B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54 43
other errors are present. However, this difference would be
more likely to be signi?cant when management speci?-
cally directs auditors to search those other accounts, based
upon information pursuit effects in accounting and psy-
chology, as well as other attributes of the auditing setting
discussed above. This does not preclude the possibility of
a smaller reaction to the other errors without management
highlighting the accounts that they are in. However, the
strongest a priori case for a reaction to the other errors
(or to a lack of other errors) in the ?nancial statements
would be when management explicitly directs auditors
to search those speci?c accounts. The different reactions
to the errors (or lack of errors) in those other accounts
would then affect the vigilance with which auditors search
the rest of the ?nancial statements, and, therefore, the like-
lihood of uncovering the earnings management error.
Taken together, this discussion suggests that auditors
would be the least likely to uncover earnings management
when clients divert auditors’ attention to accounts that are
clean, and most likely to uncover earnings management
when clients divert auditors’ attention to accounts that
contain errors. Auditors’ earnings management detection
would be between these highest and lowest levels when
there are no diversions to the other errors, as well as when
there are no diversions and no other errors. This suggests
the following hypothesis:
H1. An auditor’s detection of earnings management will
be lowest when diverted to clean accounts, highest when
diverted to accounts containing other errors, and in
between these levels either when there are no diversions
to clean accounts, or when there are no diversions to
accounts containing other errors.
Method
Participants
A representative from each of the Big Four and other
audit ?rms identi?ed auditors with suf?cient knowledge
to perform the task. Seventy-six auditors, with an average
of four years of audit experience, took part in the study.
8
Neither experience nor rank signi?cantly in?uences our
results. Our participants’ experience is consistent with prior
research on analytical review (see Messier et al., 2013 for a
review).
Overview of the study
The experiment required that auditors complete analyt-
ical review procedures on the ?nancials statements of a
hypothetical client. Following prior studies (e.g., Asare
et al., 2000; Asare & Wright, 2003; Bedard & Biggs,
1991a, 1991b; Bhattacharjee, Kida, & Hanno, 1999; Cohen
& Kida, 1989; Luippold & Kida, 2012; Moreno et al.,
2007), we adapted a set of ?nancial statements with stable
account balances over time, and then seeded an accounting
error into the accrual of compensation expense that would
understate current-year expenses and accruals by approx-
imately $450,000. The error caused unusual and material
?uctuations in these accounts for auditors to detect during
their analytical review.
9
The compensation expense error
was divided evenly into administrative compensation
expense and sales compensation expense, understating both
by approximately $225,000. The accrual entry was divided
evenly between current and non-current accrued compensa-
tion.
10
Background data, provided prior to beginning the
analytical review, revealed that the company beat analysts’
forecasted EPS by approximately $0.025/share (net income
was about $8.45 million). If the compensation error was dis-
covered, the company would miss its earnings target.
The study employed a 2 Â 2 experimental design. The
?rst independent variable manipulated whether manage-
ment provided a diversionary statement. The diversionary
statement involved management explicitly identifying risk
elsewhere in the ?nancial statements in an attempt to lure
the auditor away from managed earnings. In the diversion-
ary statement conditions, the client’s background informa-
tion indicated that the individual responsible for
maintaining non-current assets (i.e., property, plant and
equipment, intangibles and other non-current assets) left
the company about six months ago. It also stated that
her replacement transferred in from the manufacturing
?oor and has very little accounting experience. Aside from
that change, the auditors were told that there was no other
turnover with any of the accounting personnel responsible
for ?nancial reporting. In the other conditions, no speci?c
area of risk was identi?ed.
The second independent variable was manipulated by
seeding two other, offsetting errors in non-current assets
that had no impact on earnings but created unusual ?uctu-
ations for auditors to detect. One error concerned the com-
pany failing to record a portion of depreciation expense for
furniture and ?xtures, resulting in an understatement of
depreciation expense and accumulated depreciation by
approximately $450,000. The other error overstated
8
The original participant pool contained 77 auditors; however, one was
removed due to a software error. Approximately 66% of participants were
employed by Big Four ?rms, while 28% were employed by regional ?rms.
Firm type does not signi?cantly in?uence our results. Approximately 22%
were associate or staff auditors (averaging 1.4 years of audit experience),
61% were senior or in-charge auditors (averaging 3.5 years of experience),
7% were managers or senior managers (averaging 6.1 years of experience),
4% were partners (averaging 18.7 years of experience), and 7% did not
specify an auditor rank (averaging 3.7 years of experience).
9
Note that while errors can cause unusual ?uctuations, ?nancial
statement ?uctuations can also be the result of non-error causes. As a
result, our study more broadly tests diversions from unusual ?uctuations as
much as it tests diversions from earnings management. Because auditors
rely heavily on detection of unusual ?uctuations during analytical review
as a major source of substantive testing, failure to detect unusual
?uctuations elevates the risk of undetected misstatements (Messier,
Glover, & Prawitt, 2012). We are particularly concerned here with the
implications of our ?ndings for audits when management deliberately
conceals earnings manipulations. As Bell et al. (2005) note, those audits
carry the largest social costs, and the effectiveness of various tactics that
managers could use is not well understood by prior research.
10
The error was divided into different accounts so that it was more
dif?cult to uncover. While compensation is often a current accrued liability,
non-current accrued compensation can relate to post-retirement bene?ts,
deferred incentive compensation, pension bene?ts, non-expiring vacation/
sick time, etc.
44 B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54
amortization expense, and hence understated the net value
of goodwill by approximately $450,000.
11
These other
errors were embedded into the same area that the diversion-
ary statement pointed to (i.e., non-current assets). When the
other errors were not present, none of the accounts related
to depreciation and amortization re?ected any material ?uc-
tuations from previous years.
These manipulations resulted in four conditions (i.e.,
whether or not there were diversions to ?xed assets, and
whether or not there were other errors in ?xed assets).
The 2 Â 2 fully crossed design allows us to test the effects
of diversionary statements with or without other errors in
the diversionary accounts.
Procedures
The study was administered through a computer pro-
gram. After agreeing to participate, the auditors received
an email with the relevant information to access the pro-
gram (see Fig. 1 for the timeline of the experiment).
12
Upon
starting the study, the participants were randomly assigned
to a condition and navigated through a set of instructions
and background information about the company, its indus-
try, its position in the market, and details about its audit his-
tory. Participants were told that they were the senior-in-
charge on an audit of a manufacturing company. The com-
pany had consistently met analysts’ earnings forecasts, and
analysts had recently forecasted income to remain at $8.2
million (or $0.82 per share), which was the same as the pre-
vious two years. In addition, the materiality threshold for
the audit was explicitly stated to be $100,000, which made
material the ?uctuations for all of the accounts affected by
the earnings management and the other errors.
After reading the instructions and background, the
auditors began the analytical review. At this stage, partici-
pants were exposed to information from the client that
compared the unaudited ?nancial balances of the current
year to the audited balances of the previous two years.
13
Navigation buttons allowed the participants to access all of
the ?nancial details. In total, twenty pages of information
presented the balance sheet, income statement, statement
of cash ?ows, and additional detailed information which fur-
ther described the balances on the ?nancial statements.
14
For example, one page provided details for accounts receiv-
able, such as gross and net accounts receivable balances,
allowance for doubtful accounts, bad debt expense, an aging
analysis and key ?nancial ratios. The navigation buttons
were always present on the left side of the screen during
the analytical review, so that participants could move freely
to any piece of information in any order they preferred.
A button on each screen labeled ‘‘Record Judgment’’
brought the auditors to a page where they could record
any unusual ?uctuations that they identi?ed in a free
response text box. They could return to the ‘‘Record Judg-
ment’’ page as often as they wanted to add new judgments.
Download
and Install
Program
Enter Pin
Code and
Name
Read
Instructions
Read
Background
Information
Conduct
Analytical
Review
Answer Post
Experimental
Questions
Email
Results
File
Record
Error
Judgments
Fig. 1. Timeline of experimental procedures. The experiment was conducted through a computer program that participants downloaded and completed at
their convenience. Participating auditors received an email with download instructions and a pin-number to grant them access to the program. After
downloading the program, they entered their name and pin number. They then read instructions and background information before beginning the
analytical review. During the review, auditors searched through several pages of information related to the client’s ?nancial statements. If participants
identi?ed an error, they could navigate to a page to record their judgment as part of their evaluation. Participants could make as many error judgments as
needed throughout the review, and they could revert back to the background information as necessary. Upon completing the analytical review, auditors
answered post experimental questions. At this stage, they could not go back to the analytical review. After completing the study, participants emailed the
?le back to the experimenters.
11
To ensure that the earnings management error was more dif?cult to
detect than the other errors, ?ve individuals with a mean of over three
years of audit experience rated the three errors on a 10-point scale, which
ranged from very easy (one) to very dif?cult (ten). Signi?cant differences
were found between the earnings management error (6.4) and the other
errors (amortization = 2.6, p = 0.002 and depreciation = 3.4, p = 0.008),
suggesting that the earnings management was more dif?cult to uncover
than the other errors.
12
There was no time limit to the task, and we ?nd no signi?cant effects of
time on our results.
13
The ?nancial statements were created from several other accounting
studies using similar analytical review procedures, including Bhattacharjee
et al. (1999), Cohen (1994), Cohen and Kida (1989), and Moreno et al.
(2007).
14
In addition to the balance sheet, income statement and cash ?ow
statement, participants could access details for the following accounts:
marketable securities, accounts receivable, inventory, cost of goods man-
ufactured/sold, property, plant and equipment, prepaid expenses, accumu-
lated depreciation, intangibles, other non-current assets, accounts payable,
other current liabilities, debt, other non-current liabilities, equity, sales,
selling expenses and administrative expenses. In addition, a navigation
button allowed participants to revisit the background information.
B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54 45
Each time they returned, all of their previous entries were
listed numerically in the order they were entered. Another
button allowed them to ?nish the exercise, and brought
them to supplemental questions (i.e., demographic infor-
mation, questions pertaining to professional skepticism,
etc.). Participants emailed the results ?le back when they
completed the study.
Results
We ?rst test our hypotheses about the effects of diver-
sions (to accounts with or without other errors) on an
auditor’s identi?cation of an earnings management error.
As a supplemental analysis, we then examine whether
auditor skepticism affects an auditor’s detection of man-
aged earnings.
Dependent variables
We use two dependent variables to proxy for the
identi?cation of earnings management. The purpose of
analytical review is to highlight accounts with unusual
?uctuations in order to focus the audit team’s subsequent
attention and effort where the risk of material misstate-
ment is highest. Since the experiment involved a realistic
analytical review task, which allowed the auditors to list
as many potential errors as they felt necessary, partici-
pants could simply list many areas that they believe
may contain an error (i.e., take a ‘‘shotgun’’ approach).
While identifying a large number of accounts for investi-
gation should, in theory, increase the likelihood of detect-
ing the earnings management error, ?agging additional
irrelevant accounts ultimately diverts the audit team’s
subsequent attention away from the accounts that con-
tain the highest risk of misstatement to other accounts
that do not. As such, each additional account identi?ed
dilutes the amount of resources the audit team can
commit to locating the earnings manipulation (i.e., the
compensation error). This could affect not only audit
ef?ciency, but also effectiveness, because it ultimately
reduces the audit team’s attention to the area containing
the managed earnings.
15
Because of this effect, we use two alternative dependent
variables for our tests of H1. Our ?rst dependent variable
(Task Performance) places identi?cation of the earnings
management account into the context of how many irrele-
vant accounts were also identi?ed. It is calculated as
follows:
Task Performance ¼
Identification of Earnings Management
1þNumber of Irrelevant Judgments
ð1Þ
The numerator (Identi?cation of Earnings Management) is
coded correct (1) if the auditor identi?ed one or more of
the affected compensation accounts (or listed compensa-
tion or an appropriate synonym), and incorrect (0) other-
wise. The number of irrelevant judgments re?ects the
number of accounts (or areas) identi?ed by the auditor as
potentially containing an error where no unusual ?uctua-
tion actually existed. For instance, if a participant listed
inventory, accounts receivable and leases as accounts pos-
sibly containing errors, then three irrelevant judgments
would be recorded.
16
Our second dependent variable is the percentage of
auditors who identi?ed one or more of the affected com-
pensation accounts (or listed compensation or an appropri-
ate synonym) in their responses. This dependent variable
shows the proportion of auditors who named the earnings
management account, but does not account for whether
they also highlighted a large number of other, irrelevant
accounts.
17
0.19
0.04
0.17
0.27
0.00
0.05
0.10
0.15
0.20
0.25
0.30
No Diversionary Statement Diversionary Statement
No Other Errors Other Errors
Fig. 2. The effect of diversions to accounts with or without other errors
on task performance. Task Performance ¼
Identification of Earnings Management
1þNumber of Irrelevant Judgments
.
15
Our approach is consistent with studies on brainstorming (e.g.,
Bellovary & Johnstone, 2007; Carpenter, 2007; Hammersley, 2011;
Trotman, Simnett, & Khal?a, 2009). It is also consistent with our interviews
of audit partners, who indicated that good performance in analytical review
entails not merely highlighting high risk ?uctuations, but doing so without
?agging other accounts that do not contain unusual ?uctuations.
16
This task performance measure scales detection of the compensation
error by the number of erroneous judgments listed, so that the measure
re?ects both audit effectiveness and ef?ciency. Identifying a larger number
of additional irrelevant accounts diverts subsequent audit testing away
from higher risk accounts, risking both audit effectiveness and ef?ciency
(Messier et al., 2012). For instance, an auditor who identi?ed the
compensation error without any irrelevant judgments would receive a
score of 1.00. An auditor who listed three irrelevant judgments along with
the compensation error would receive a score of 0.25. An auditor who failed
to indicate the compensation error would receive a score of 0.00, regardless
of the number of irrelevant judgments listed.
17
Neither of our dependent variables (Task Performance or the percentage
of auditors identifying earnings management) includes identi?cation of the
other errors in their calculation. Libby and Frederick (1990, 360) and Libby
(1985, 660–661) design their dependent variables this way for similar
reasons. Speci?cally, had we also included detection of the other errors in
the calculation of our dependent variables, results consistent with H1 could
have then been merely attributed to participants in the ‘‘no other errors’’
conditions having fewer errors available for detection to begin with, giving
an unfair advantage in the dependent variable to those in the ‘‘other errors’’
conditions. Instead, our dependent variables are consistent with our focus
on testing the effects of diversions on the detection of earnings manage-
ment elsewhere in the ?nancial statements. Roediger, Stellon, and Tulving
(1977) discuss the methodological importance of this approach.
46 B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54
Hypothesis tests: task performance
Fig. 2 shows the means of Task Performance by experi-
mental condition. H1 predicts that auditors’ detection of
earnings management will be lowest when auditors are
diverted to clean accounts, highest when auditors are
diverted to other accounts that contain errors, and in
between these levels when auditors are not diverted to
the other errors, as well as when there are no diversions
and no other errors.
Predictions of this type are tested by a Jonckheere-
Terpstra test (e.g., Abarbanell, 1991; Fanning, Agoglia, &
Piercey, 2015; Frederickson, Hodge, & Pratt, 2006; Libby
& Lipe, 1992; Libby & Trotman, 1993; Phua, Abernathy, &
Lillis, 2011; Sedor, 2002; Tan & Jamal, 2006). As Fig. 2
shows, the mean Task Performance score of auditors in
the Diversion, No Other Errors cell is the lowest at 0.04, fol-
lowed by the two No Diversion cells, with and without
errors (0.17 and 0.19, respectively), followed by the Diver-
sion, Other Errors cell (0.27). As Table 1 shows, a Jonckhe-
ere-Terpstra test shows that the predicted rank-orderings
are statistically signi?cant. Speci?cally, task performance
in detecting earnings management was lowest when audi-
tors were diverted to clean areas, highest when auditors
were diverted to accounts that contained other errors,
and in between these highest and lowest levels when audi-
tors were not diverted to the other errors (J = 1.96,
p = 0.025), as well as when they were not diverted and
there were no other errors (J = 1.91, p = 0.028). These
results support H1.
18
We perform supplementary analyses of our Task
Performance variable using ANOVA. Table 1 shows a
Diversion  Other Errors interaction signi?cant at
p = 0.074. Consistent with information pursuit effects,
the interaction shown in Fig. 2 suggests that diverting
auditors to the other accounts ampli?es their reaction
to what is (or, when the other errors are not present,
to what is not) in those other accounts. Speci?cally,
when no diversions are used, auditors’ did not react to
the presence or absence of the other errors (0.17 vs.
0.19, respectively, t = À0.24, p = 0.810). In contrast, when
diverted to the other accounts, auditors’ reacted signi?-
cantly to the presence or absence of the other errors
(0.27 vs. 0.04, respectively, t = 2.31, p = 0.012). Auditors’
reactions to the errors (or lack of errors) in the other
accounts were signi?cantly larger with the diversion to
those accounts than without (t = 1.81, p = 0.037).
19
These
supplementary results are consistent with our theory and
hypothesis.
As an additional analysis, we also tested whether the
task performance score in each condition was signi?cantly
different from zero. The scores in the two no diversion con-
ditions, as well as in the diversions to accounts with errors
condition, were all signi?cantly different from zero
(p’s 6 0.010). On the other hand, the diversionary state-
ment without other errors was not different from zero
(p = 0.549). These ?ndings suggest that when auditors are
diverted to areas that do not contain errors, they are not
likely to uncover earnings management elsewhere in the
?nancial statements, compared to auditors in the other
experimental conditions.
20
These results suggest that if management alerts audi-
tors to risk in accounts that are ultimately clean, this
diversionary tactic appears to be effective at diverting
the auditor away from managed earnings. That is,
Table 1
The effect of diversions to accounts with or without other errors on task
performance.
a
Diversion
No Yes
Panel A: Means, (standard deviations), sample size
Other errors
No 0.19 (0.37) 0.04 (0.09)
n = 19 n = 21
Yes 0.17 (0.32) 0.27 (0.38)
n = 19 n = 17
Source SS df MS F p
Panel B: ANOVA
Other errors 0.21 1 0.21 2.16 0.146
Diversion 0.01 1 0.01 0.11 0.739
Diversion  other errors 0.31 1 0.31 3.28 0.074
Error 6.84 72 0.09
Total 7.37 75
Test
statistic
p
b
Panel C: Test of hypotheses
Diversion to clean accounts lowest, diversion
to other errors highest, and no diversion
to the other errors in between
J = 1.96 0.025
Diversion to clean accounts lowest, diversion
to other errors highest, and no diversion
with no other errors in between
J = 1.91 0.028
a
Task Performance ¼
Identification of Earnings Management
1þNumber of Irrelevant Judgments
.
b
These tests have directional predictions, and their p-values are one-
tailed.
18
In addition to their use of the Jonckheere-Terpstra test, Fanning et al.
(2015), Sedor (2002) and Arel, Jennings, Pany, and Reckers (2012) also used
cell contrast weights with a rank-ordering re?ecting their expectations. We
use this approach as well, as additional robustness tests of H1. For example,
contrast weights of +0.5, +0.3, +0.2, À1 could be assigned, (respectively) to
the diversion to other errors condition, the no diversion to other errors
condition, the no diversion and no other errors condition, and the diversion
to clean accounts condition, as an alternative test of our theory and
hypothesis. Compared to contrast weights, the Jonckheere-Terpstra test has
the advantage of being a direct test of the rank-ordering of the cell means
without also requiring arbitrary decisions about the relative magnitude of
the contrast weights (Fanning et al., 2015). Because contrast weights
require arbitrary decisions about the magnitude, they should be conducted
as sensitivity analyses, using different sets of weights that all re?ect the
basic rank-ordering to be tested (e.g., Arel et al., 2012; Sedor, 2002).
Consequently, we tested H1 using contrast weights of (+0.5, +0.3, +0.2, À1),
(+1, À0.2, À0.3, À0.5), (+0.75, +0.13, +0.12, À1), and (+1, À0.12, À0.13,
À0.75). In all cases, results are supportive of H1 (all p’s 6 0.032).
19
This test (t = 1.811, p = 0.037) is the one-tailed t-test associated with
the F-test of the Diversion  Other Error interaction (F = 3.28, p = 0.074) in
Table 1 (e.g., Kachelmeier & Williamson, 2010).
20
In addition, the 0.04 task performance score in the diversionary
statement without errors condition is signi?cantly less than the average
score of 0.21 for the rest of the conditions (t = À2.16, p = 0.017). Also, the
0.27 score for auditors diverted to accounts with errors is higher than the
0.13 average score of the remaining conditions (t = 1.64, p = 0.053) (see
Kadous, Koonce, & Towry, 2005).
B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54 47
auditors appear to be relatively ineffective at detecting
managed earnings in other areas of the audit when
diverted to a clean account. However, if management
overtly leads auditors to an area containing errors, audi-
tors perform better at discovering managed earnings else-
where in the ?nancial statements. Overall, our results
appear to be consistent with information pursuit theory.
That is, directing auditors to search other accounts
appears to amplify their reactions to what either is or is
not in those other accounts.
21
Hypothesis tests: percentage of auditors identifying earnings
management
We also test our hypotheses on the percent of audi-
tors who detected the unusual ?uctuations created by
managed earnings.
22
To limit the effect of auditors
identifying managed earnings as a result of simply taking
a shotgun style approach (i.e., by listing many accounts),
we ?rst analyzed the judgments of auditors who identi?ed
fewer than ?ve irrelevant errors. We use this cutoff
because, as we previously discussed, large numbers of
irrelevant judgments imply that auditors are inef?cient,
as they may pursue false leads. In addition, large numbers
of irrelevant judgments suggest that the auditor may not
actually have detected a likely instance of earnings man-
agement, but simply included the relevant accounts by
chance while generating a laundry list of accounts to
examine. A valid measurement of earnings management
detection would not count these chance inclusions of the
relevant accounts. We selected the moderately low cutoff
of fewer than ?ve irrelevant items so as to clearly distin-
guish those including many irrelevant items in their
responses (and thus likely using a shotgun approach) from
those including very few (and thus less likely to be using
one).
As can be seen in Fig. 3, only 6.7% of auditors who
were diverted to clean accounts uncovered managed
earnings, while 43.8% of auditors uncovered earnings
management when they were diverted to accounts that
contained the other errors. In contrast, when auditors
received no diversions to ?xed assets (either with or
without the other errors), 29.4% detected the earnings
management. Jonckheere-Terpstra tests of H1 indicate
that auditors’ earnings management detection was low-
est when diverted to clean accounts, highest when
diverted to the other errors, and in between when no
diversion was used, with or without the other errors
(in both tests, J = 2.29, p = 0.011; Table 2). These results
support H1.
29.4%
6.7%
29.4%
43.8%
0%
10%
20%
30%
40%
50%
No Diversionary Statement Diversionary Statement
No Other Errors Other Errors
Fig. 3. The effect of diversions to accounts with or without other errors
on the percent of auditors identifying earnings management. This
analysis includes auditors who listed fewer than ?ve irrelevant judg-
ments to limit the impact of taking a shotgun approach. Responses were
coded correct if any of the recorded judgments identi?ed the affected
compensation accounts, or listed the word compensation or an appro-
priate synonym, and incorrect otherwise.
Table 2
The effect of diversions to accounts with or without other errors on the
percent of auditors identifying managed earnings.
a
Diversion
No Yes
Panel A: Percentages, sample sizes
Other errors
No 29.4% 6.7%
n = 17 n = 15
Yes 29.4% 43.8%
n = 17 n = 16
Test
statistic
p
b
Panel B: Hypothesis tests
Diversion to clean accounts lowest, diversion
to other errors highest, and no diversion
to the other errors in between
J = 2.29 0.011
Diversion to clean accounts lowest, diversion
to other errors highest, and no diversion
with no other errors in between
J = 2.29 0.011
a
This analysis includes auditors who listed fewer than ?ve irrelevant
judgments to limit the impact of taking a shotgun approach. Responses
were coded correct if any of the recorded judgments identi?ed the
affected compensation accounts, or listed the word compensation or an
appropriate synonym, and incorrect otherwise.
b
These tests have directional predictions, and their p-values are one-
tailed.
21
Note that, even if auditors discovered the other errors at just as high of
a rate without being diverted there, information pursuit theory still
suggests that diversions would amplify auditors’ reactions to the other
errors (or to the lack of errors) in those accounts. In fact, our ?ndings are
consistent with this. When the other errors were present, auditors detected
the other errors there just as much without the diversion as they did with it
(auditors found on average 1.4 vs. 1.5 of the other errors, t = 0.39, p = 0.70).
When the other errors were not present, diversions also did not impact
auditors’ detection of the other errors (since there were none to be found).
Yet, our tests of H1 (and the supplemental interaction test) suggest that
auditors still reacted more to the other errors (or to the lack of errors) in the
other accounts when they were diverted there by management (Table 1,
Panels B and C), consistent with information pursuit theory. In the
information pursuit literature, Nelson and Tayler (2007) and Smith et al.
(2011) discuss the methodological importance that participants acquire
information at approximately the same rate when testing whether the
explicit information pursuit (e.g., the diversions to other accounts)
ampli?es their reactions to it.
22
Again, earnings management was coded as correct if the auditor
identi?ed any of the affected compensation accounts (or listed compensa-
tion or an appropriate synonym), and incorrect otherwise.
48 B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54
Fig. 3 suggests an interaction in this dependent vari-
able similar to the interaction that we observed in the
supplementary analysis of our Task Performance depen-
dent variable. Speci?cally, Fig. 3 suggests that diverting
auditors to the other accounts appears to amplify their
reaction to the errors (or to the lack of errors) in those
other accounts, consistent with information pursuit
effects. To test this interaction, we employ a binomial
logistic regression model with Diversion, Other Errors,
and their interaction as independent variables. We ?nd
that the interaction shown in Fig. 3 is statistically signif-
icant, as expected. Speci?cally, auditors’ detection of the
earnings management is more in?uenced by the presence
or absence of errors in the other accounts when diverted
to those accounts by management (v
2
= 3.55, p = 0.030).
23
Overall, our ?ndings suggest that directing auditors to
clean accounts can be an effective audit management tool,
but directing auditors to accounts containing other errors
may not only be ineffective, it may actually have the oppo-
site effect on management.
Further, we performed a sensitivity analysis to deter-
mine how the number of irrelevant judgments impacts
the percent of auditors uncovering earnings manage-
ment. As shown in Table 3, the results for H1 replicate
when we include auditors who are not simply using a
shotgun approach (i.e., naming a large number of irrele-
vant accounts), and become more signi?cant as the
number of irrelevant items included in the analyses
decreases.
24
Thus, it appears that the effect of diversions (to accounts
with and without other errors) on the proportion of audi-
tors uncovering earnings management are robust across
the number of irrelevant judgments.
Auditor skepticism
We also investigated the impact that skepticism may
have on an auditor’s detection of earnings management.
As part of the post-experimental questionnaire, auditors
were asked to indicate their level of agreement with four
statements (shown in Table 4) re?ecting a skeptical out-
look toward management on six-point scales (ranging
from strongly disagree to strongly agree).
These measures of skepticism capture auditors’ ten-
dency to adopt a presumptive doubt perspective on man-
agers’ representations (Bowlin et al., 2014; Nelson, 2009;
Quadackers et al., 2014). Compared to a more neutral
perspective of professional skepticism (e.g., Hurtt, 2010),
a presumptive doubt view focuses speci?cally on a
heightened assessment of the risk that management is
misleading (Nelson, 2009). Because a presumptive doubt
view would tend to indicate an elevated alertness to
red ?ags, Quadackers et al. (2014) suggest that auditors’
Table 3
The effect of diversions to accounts with or without other errors on the percent of auditors identifying managed earnings: sensitivity to number of irrelevant
judgments.
a
Irrelevant judgments
included
b
N No other errors Other errors Tests of H1
Diversion (A) (%) No diversion (B) (%) No diversion (C) (%) Diversion (D) (%) A < B < D
(p-value)
c
A < C < D
(p-value)
c
All 76 23.8 31.6 31.6 47.1 0.070 0.070
This study examines an aspect of earnings management that we refer to as audit management.
We define audit management as a client’s strategic use of diversions to decrease the
likelihood that auditors will discover earnings management during the audit. Specifically,
we examine whether diverting auditors’ attention to either clean financial statement
accounts or accounts that contain other errors affect an auditor’s ability to uncover earnings
management. Auditors performed analytical review, searching financial statements
for unusual fluctuations suggestive of errors. Following prior studies, we seeded an intentional
accounting error which created an unusual fluctuation that allowed the client to
meet an earnings target.
Managing audits to manage earnings: The impact of diversions
on an auditor’s detection of earnings management
Benjamin L. Luippold
a,?
, Thomas Kida
b
, M. David Piercey
b
, James F. Smith
b
a
Babson College, United States
b
University of Massachusetts Amherst, Isenberg School of Management, United States
a b s t r a c t
This study examines an aspect of earnings management that we refer to as audit manage-
ment. We de?ne audit management as a client’s strategic use of diversions to decrease the
likelihood that auditors will discover earnings management during the audit. Speci?cally,
we examine whether diverting auditors’ attention to either clean ?nancial statement
accounts or accounts that contain other errors affect an auditor’s ability to uncover earn-
ings management. Auditors performed analytical review, searching ?nancial statements
for unusual ?uctuations suggestive of errors. Following prior studies, we seeded an inten-
tional accounting error which created an unusual ?uctuation that allowed the client to
meet an earnings target. We manipulated whether management provided a diversionary
statement that explicitly identi?ed risk in other areas of the audit, and whether those areas
were clean or contained other detected errors that had no impact on earnings. We ?nd that
auditors’ earnings management detection is worst when they are diverted to clean
accounts and best when auditors are diverted to accounts that contain other errors. Our
results suggest that managers can potentially exploit an audit management tactic as simple
as a diversion to a clean area to reduce auditors’ effectiveness at detecting earnings man-
agement. The implications of these ?ndings for audit and decision making research are
discussed.
Ó 2014 Elsevier Ltd. All rights reserved.
Introduction
Earnings management has been a topic of great interest
in both the popular press and academic literature (e.g.,
Chen, Kelly, & Salterio, 2011; Dechow, Hutton, Kim, &
Sloan, 2012; Guerrera, 2012). In fact, attempts to manipu-
late ?nancial performance have become so widespread
that books have been written on earnings management
strategies (e.g., Giroux, 2003; McKee, 2005). This study dis-
cusses an aspect of earnings management that we refer to
as audit management. We de?ne audit management as a
client’s strategic use of diversions to decrease the likeli-
hood of auditors discovering managed earnings during
the audit. Evidence suggests that managers strategically
attempt to conceal earnings management (e.g., Beasley,
Carcello, Hermanson, & Neal, 2010; Bowlin, Hobson, &
Piercey, 2014; Knapp, 2010).
Our study investigates whether managers who manipu-
late earnings can successfully employ diversions to in?u-
ence auditors’ detection of unusual ?uctuations during
analytical review. That is, we investigate whether diversion-
ary statements made by the client (i.e., identifying areas of
riskinthe ?nancial statements tolurethe auditor awayfrom
managed earnings) affect an auditor’s detection of managed
earnings containedelsewhere inthe ?nancial statements. Inhttp://dx.doi.org/10.1016/j.aos.2014.07.005
0361-3682/Ó 2014 Elsevier Ltd. All rights reserved.
?
Corresponding author. Address: Babson College, Accounting & Law
Division, Babson Park, MA 02457, United States. Tel.: +1 (781) 239 5995;
fax: +1 (781) 239 5930.
E-mail addresses: [email protected] (B.L. Luippold), tkida@
isenberg.umass.edu (T. Kida), [email protected] (M.D. Piercey),
[email protected] (J.F. Smith).
Accounting, Organizations and Society 41 (2015) 39–54
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j our nal homepage: www. el sevi er. com/ l ocat e/ aos
an experiment, we seeded an earnings management error
(i.e., an intentional misstatement used to meet an earnings
target) into?nancial statements, creating anunusual ?uctu-
ationfor auditors to detect.
1
We test whether auditors detect
earnings management when they are diverted to: (1)
accounts that contain no errors, or (2) accounts that do con-
tain errors, but have no impact on earnings.
In the study, auditors completed analytical review pro-
cedures on the ?nancial statements of a hypothetical client
in order to determine if any unusual ?uctuations (sugges-
tive of errors) were present. In all conditions, an earnings
management error (which reduced compensation expense
and accruals) was embedded into the ?nancial statements,
creating an unusual ?uctuation in that account that
resulted in the client meeting analysts’ forecasted earn-
ings. We manipulated whether or not management pro-
vided a diversionary statement that informed the auditor
of a personnel change in the department responsible for
non-current assets. This statement was designed to elevate
the perceived misstatement risk in that area and lure the
auditor away from the earnings manipulation. We also
manipulated whether the accounts in this other area iden-
ti?ed by the diversionary statement were clean or con-
tained other errors that offset, and therefore had no
impact on earnings.
Managers may be motivated to divert auditors to areas
that contain, or do not contain, other errors. For example, if
managers point auditors to ostensibly risky areas that are
clean, auditors may conclude that the client’s accounts
are likely to be accurate in other areas as well. This would
most likely result in a strong diversion effect. Conversely,
management may want to direct auditors to areas that
contain other errors, thinking that these other errors may
occupy their attention, leading auditors to feel satis?ed
that they are detecting misstatements and ‘‘doing their
job,’’ resulting in auditors feeling less compelled to dis-
cover other errors. However, auditors are also trained to
practice professional skepticism (Nelson, 2009;
Quadackers, Groot, & Wright, 2014), and the diversion to
the other errors should elevate their sensitivity to the risk
of material misstatement in the remainder of the ?nancial
statements, resulting in greater overall audit effort and a
greater likelihood that they would ?nd the earnings
manipulation. We therefore investigate the impact of man-
agement intentionally directing auditors to both clean
accounts and accounts containing errors.
Our ?ndings are consistent with our predictions. Specif-
ically, we ?nd that auditors’ detection of earnings manage-
ment was worst when they were diverted to clean ?nancial
statement accounts, and best when they were diverted to
accounts containing other errors, with earnings manage-
ment detection in between these levels when no diversions
were used (whether other errors were present or not).
Overall, these results suggest that if management directs
auditors to accounts that contain errors, the discovery of
those errors heightens their sensitivity to errors in other
areas of the audit. However, if auditors are directed to
clean accounts, the use of diversionary statements can
deter auditors from ?nding earnings management. These
?ndings have signi?cant implications for auditors and
decision researchers in general.
Testing diversions to accounts both with and without
other errors allows us to demonstrate signi?cantly differ-
ent reactions from auditors to two basic strategies that
management could use to divert auditors from an area
used to manage earnings. The different effects also provide
possible insight into why diversions to clean areas may be
an effective means of concealing earnings management. On
one hand, even though the other errors do not impact earn-
ings, diverting auditors toward them increases auditors’
sensitivity to the risk of material misstatement in other
areas of the audit. In contrast, diverting auditors to osten-
sibly risky areas that turn out to be clean appears to make
those auditors less vigilant in their search for errors else-
where in the ?nancial statements. These results suggest
that managers can potentially exploit an audit manage-
ment tactic as simple as a diversion to a clean area because
such a diversion reduces auditors’ effectiveness at detect-
ing earnings management elsewhere in the ?nancial state-
ments. This extends the literature on auditor skepticism by
identifying one type of claim that managers could make to
mislead auditors without raising red ?ags (Nelson, 2009;
Quadackers et al., 2014). Additionally, it contributes to
the accounting literature on auditing and earnings man-
agement with evidence of how managers can conceal
material misstatements from auditors in order to manage
earnings (Beasley et al., 2010; Beasley, Carcello,
Hermanson, & Lapides, 2000; Boone, Khurana, & Raman,
2012; Caramanis & Lennox, 2008; Chen et al., 2011).
Our study also contributes to decision making research
(in both psychology and auditing) that investigates the
effectiveness of diversionary tactics. As we explain in Sec-
tion ‘Background literature’, psychology literature on dis-
traction suggests that diversions to other errors would be
an effective means of concealing earnings management.
In contrast, our ?nding that earnings management detec-
tion is greatest when auditors are diverted to accounts that
contain errors suggests that our context provides a bound-
ary condition to the predictions of the psychology litera-
ture on distractions, based on the task-speci?c
experience of auditors and their reaction to the other
errors. Furthermore, our ?ndings contribute to the
accounting and psychology literature on information pur-
suit effects (Bastardi & Sha?r, 1998, 2000; Nelson &
Tayler, 2007; Redelmeier, Sha?r, & Aujla, 2001), by demon-
strating how management diverting auditors to search
other accounts can amplify auditors’ reactions to what is
(or to what is not) in those other accounts. Thus, we not
only contribute to the accounting literature on earnings
management and auditing, we also contribute to the gen-
eral judgment and decision making literature on diversion,
distraction, and information pursuit effects.
Diversions can have important practical implications
beyond the setting of deliberate earnings management.
For example, even when managers are not deliberately
managing earnings with a particular account, they may
still prefer that auditors pay more attention to areas where
1
While auditing standards typically characterize unintentional mis-
statements as errors and intentional misstatements as fraud, we use the
term error in its more generic sense to refer to any departure from
accuracy.
40 B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54
they are less likely to ?nd problems. As a result, they may
divert auditors from higher risk areas to accounts that they
believe are not at risk of misstatement. Our ?ndings sug-
gest that auditors would be susceptible to these diversions
as well. However, to the extent that errors can occur any-
where, managers might mistakenly direct auditors’ atten-
tion to an area with errors, which could back?re on
managers. Even more broadly, managers may inadver-
tently direct auditors’ attention from an account that
materially misstates earnings to another account that does
not. No matter the cause of the diversion, our results dem-
onstrate that such diversions signi?cantly in?uence an
auditor’s detection of a material misstatement of earnings
elsewhere in the ?nancial statements.
In the next section, we introduce background literature
related to both earnings and audit management, and dis-
cuss relevant theories related to our research. After the
background literature, we describe the method, while the
?nal two sections present the results and concluding
remarks, respectively.
Background literature
Earnings management
Earnings management refers to ?nancial reporting
practices designed to achieve desired or favorable ?nancial
results (e.g., smoothing earnings, meeting earnings targets)
(Bouillon, 2007; Jackson & Pitman, 2001; McKee, 2005;
Millstein, 2005). Management faces several pressures, such
as meeting analysts’ forecasts, which may prompt them to
resort to such practices (Duncan, 2001). Evidence suggests
that these short-term pressures can take priority over
long-term economic growth, as research has found that
executives sometimes sacri?ce economic value to smooth
earnings or hit an earnings target (Bhojraj & Libby, 2005;
Graham, Harvey, & Rajgopal, 2005).
Archival research provides substantial evidence that
earnings management occurs (e.g., Dechow et al., 2012).
For example, several studies have examined speci?c
accrual accounts that clients use to manage earnings
(e.g., Dhaliwal, Gleason, & Mills, 2004; Marquardt &
Wiedman, 2004). More striking, however, is evidence sug-
gesting that managers sometimes resort to fraudulent
measures to manage earnings (e.g., Beasley, Carcello, &
Hermanson, 1999; Beasley et al., 2000, 2010; Farber,
2005; Jones, Krishnan, & Melendrez, 2008).
2
However, in
order for management to successfully report over-aggressive
or fraudulent earnings, auditors must fail to discover how
and where income is being manipulated. Yet, how managers
can successfully divert auditors’ attention from earnings
manipulations remains an important unanswered question
for researchers (Bell, Peecher, & Solomon, 2005; Peecher,
Schwartz, & Solomon, 2007).
Audit management
We de?ne audit management as a client’s strategic use
of techniques to reduce the likelihood that auditors will
identify or recognize managed earnings during the audit.
Managing the audit may include a variety of methods.
For instance, managers may frame evidence in certain
ways to manipulate the level of perceived risk (e.g.,
Jamal, Johnson, & Berryman, 1995; Johnson, Grazioli, &
Jamal, 1993). They may provide the auditor with incom-
plete or incorrect information to cover-up questionable
accounting practices, or they may use diversions to prevent
the auditor from uncovering earnings management, which
is the primary focus of this study.
Diversions are methods designed to direct an auditor’s
attention away from a certain audit area. For example,
management may use diversionary statements which
identify speci?c areas of risk in other areas of the ?nancial
statements in an effort to lure the auditor away from the
accounts used to manage earnings. In this study, we inves-
tigate the impact of diversions to areas that do not contain
errors, or to areas where errors exist, but have no impact
on earnings.
Managers could identify a clean area of the ?nancial
statements as high risk, hoping that auditors would then
conclude that the rest of the ?nancial statements are likely
to be error-free as well. On the other hand, managers could
direct auditors to areas containing errors for several rea-
sons. The other errors may occupy auditors’ attention,
and more attention paid to one area of the audit may result
in less attention paid to other areas. In addition, auditors
may feel satis?ed that they have found errors, making
them less likely to search for earnings management else-
where in the ?nancial statements. Finally, management
may feel that pointing out areas that lead to error discov-
ery may increase the trust that auditors have in them,
resulting in auditors performing less work in areas that
management suggests are problem free.
3
However, diver-
sions to other errors may also have the opposite effect. If
management diverts an auditor to another account by iden-
tifying it as high risk, and yet apparently did nothing to
remove errors there, the diversion to those errors would
send an especially negative signal about management’s
internal controls, which could lead auditors to search more
extensively for errors in other areas of the audit. As a result,
we investigate auditors’ detection of earnings management
when they are diverted to areas that contain and do not con-
tain errors.
While clients may attempt to ‘‘manage’’ many different
aspects of the audit, an area that is of particular interest is
analytical review. Analytical review is used to determine
the extent of required detailed testing in different audit
areas, and sometimes is the only audit procedure used to
test certain accrual based accounts (AICPA, 2012;
2
Dyck, Morse, and Zingales (2010) estimate archivally that at least one
?nancial reporting fraud is ongoing at any time in at least 11.2–13.2% of
public companies with more than $750 million in assets, and that managers
successfully conceal a large majority of these frauds for some time from
auditors, SEC enforcement, and other governance mechanisms.
3
Our conversations with practitioners suggest that this and similar
tactics occur in practice. For example, a former manager of a technology
company indicated that, when auditors found error corrections that would
reduce earnings, he would direct them toward other error corrections that
would increase earnings. Similarly, an audit partner indicated that man-
agers may indeed see the audit as a diversionary game.
B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54 41
Ricchiute, 2006) and income statement accounts (FRC,
2011a). For example, unless speci?c risks are identi?ed
that warrant detailed testing, compensation accruals are
typically audited using only some variant of analytical
review (Ricchiute, 2006).
4
Thus, if such an account contains
errors that go undetected during analytical review, there
may not be subsequent procedures in the audit plan to
detect them. Failure to highlight the highest-risk accounts
during analytical review impacts planning, risk assessment
and substantive testing throughout the remainder of the
audit (FRC, 2011a; Messier, Simon, & Smith, 2013; PCAOB,
2010b).
Regulators indicate that analytical review is a recurring
problem in their inspections of audit ?rms (e.g., FRC,
2011a, 2011b, 2012b; IFIAR, 2012; PCAOB, 2008b).
5
These
regulatory ?ndings include auditors’ tendency to perform
analytical procedures in insuf?cient depth, as well as their
tendency to rely on and accept managers’ claims during ana-
lytical review without corroboration (e.g., FRC, 2011a,
2011b, 2012a; PCAOB, 2008a). Trompeter and Wright
(2010) report that auditors are increasingly using analytical
review to reduce the amount of detailed audit tests, using
less experienced auditors to perform analytical review, and
increasingly relying on their clients’ uncorroborated claims
for guidance during analytical review. These problems sug-
gest that auditors could be vulnerable to tactics used by
managers during analytical reviewto divert themaway from
managed earnings, especially if done on the pretext of
directing them to other risky areas. If successful, such tactics
could lead auditors to signi?cantly reduce or even eliminate
further detailed tests of the accounts from which they were
diverted (FRC, 2011a). Thus, consistent with prior research,
we focus on analytical review, since failure to detect mis-
statements during analytical review can in?uence the
planned detailed tests and trajectory of the rest of the audit
(e.g., Asare, Trompeter, & Wright, 2000; Asare & Wright,
2003; Brewster, 2011; Ismail & Trotman, 1995; Knapp &
Knapp, 2001; Knechel, Salterio, & Kochetova-Kozloski,
2010; Luippold & Kida, 2012; Messier et al., 2013; Moreno,
Bhattacharjee, & Brandon, 2007; Peecher, Piercey, Rich, &
Tubbs, 2010; Trompeter & Wright, 2010; Trotman &
Wright, 2012; Yip-Ow & Tan, 2000).
Given that managers may attempt to manage the audit,
the question therefore arises, can managers employ diver-
sionary tactics that allow them to effectively manage earn-
ings? We ?rst review relevant theories and research to
develop our predictions. On one hand, prior psychological
research on distraction suggests that diversions to both
clean accounts and to accounts containing other errors
should be effective. On the other hand, audit practice and
research suggests that auditors display professional skepti-
cism and are likely to react to discovering errors (even
those with no impact on earnings) as a red ?ag (Nelson,
2009; Quadackers et al., 2014), which may heighten their
search for additional errors in the ?nancial statements.
Hypotheses development
Psychological research on distractions and diversions
suggests that diversions inhibit performance. Studies on
persuasion have found that diversions make individuals
more susceptible to agreeing with the arguments of others,
as they detrimentally affect comprehension (Baron, Baron,
& Miller, 1973; Festinger & Maccoby, 1964; Petty, Wells, &
Brock, 1976; Watts & Holt, 1979; Zimbardo, Snyder,
Thomas, Gold, & Gurwitz, 1970). Similarly, research exam-
ining the Elaboration Likelihood Model (for attitude forma-
tion) indicates that diversions make cognitive processing
more dif?cult, resulting in more peripheral (shallow) infor-
mation processing (Petty & Cacioppo, 1986; Street,
Douglas, Geiger, & Martinko, 2001). Further, cognitive
research suggests that diversions consume an individual’s
attention, and since attention is limited, less is available
to process important information (Kahneman, 1973;
Sagarin, Britt, Heider, Wood, & Lynch, 2003). In a classic
study on inattentional blindness, participants viewed a
video of individuals passing around a basketball, and were
instructed to count the number of passes (Simons &
Chabris, 1999). In the video, a person in a gorilla suit
walked through the group, stopped, beat their chest and
exited. Notably, over half of the participants never saw
the gorilla, as they were too distracted by the task-at-hand.
Diversions also leave individuals feeling less compelled to
discover other issues, due to their focus on what they have
discovered. For example, in successful illusions, magicians
divert attention to speci?c distracting items (e.g., smoke,
noise, and ?ashes of light) to occupy an audience’s atten-
tion and inhibit their ability to uncover the ‘‘tell’’ of the
trick (e.g., Freudenburg & Alario, 2007; Kuhn & Tatler,
2005; Kuhn, Tatler, Findlay, & Cole, 2008).
On the other hand, research on information pursuit
effects in both psychology and accounting suggests that
the effect of diversions on the performance of professional
auditors may either inhibit or enhance performance
depending upon the context considered. Speci?cally, the
information pursuit literature suggests that individuals’
beliefs are based on not just the information that supports
that belief, but also on whether they explicitly and actively
searched a speci?c area for that information. As a result,
searching a highlighted area for information ampli?es indi-
viduals’ reactions to the information that they ?nd there,
above and beyond what their reactions would be based on
the information alone (Bastardi & Sha?r, 1998, 2000;
Redelmeier et al., 2001). This ampli?ed reaction to informa-
tion sought out in a speci?c area is an information pursuit
effect. In our auditing context, this suggests that diverting
auditors directly toward a particular clean area of the audit
will amplify their reaction to the lack of errors found there.
Without a diversion from management that highlights a
speci?c area as high risk, any information pursuit effects
that may occur would be diluted across accounts, lessening
the impact of any single ?nancial statement account. Incon-
trast, if management directs auditors to search a speci?c,
4
In fact, when pretesting this study’s experimental materials, a Big Four
audit manager commented on how an error in compensation would
probably go undetected if not uncovered at this stage.
5
For example, we examined all regulatory inspection reports for the ?ve
largest accounting ?rms in the United States, dating from 2004 to 2007, and
found that thirteen of the 19 reports (68%) identi?ed de?ciencies in
analytical review (PCAOB, 2008a). Similarly, the IFIAR (2012) reports that
problems with analytical review is one of the top recurring themes in
regulatory inspection ?ndings internationally.
42 B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54
allegedly higher risk audit area, and the search reveals that
the accounts are clean, auditors would tend to react more
strongly to their ?ndings in those speci?c accounts than
they would if they had not been highlighted. The auditors
may thenbe more likely to believe that other ?nancial state-
ment accounts are clean as well, which would inhibit audi-
tors’ performance in detecting earnings management
elsewhere in the ?nancial statements.
Nelson and Tayler (2007) and Smith, Tayler, and Prawitt
(2011) provide evidence that information pursuit effects
occur in accounting and auditing settings. For example,
Smith et al. (2011) show that auditors are more persuaded
by audit evidence uncovered as a result of an active search
in a speci?c, highlighted area. Nelson and Tayler (2007)
investigate possible causes of information pursuit effects,
and ?nd evidence that a process they refer to as reconcili-
ation drives the effects. During reconciliation, decision
makers who proactively pursued speci?c information will
then process its potential implications within the context
and react more to that information. This reconciliation
‘‘potentially renders the pursued information more salient,
so the information may demand more attention in the
judgment process’’ (p. 739). Related psychology research
on salience similarly suggests that managers’ highlighting
a speci?c area as high-risk would tend to amplify auditors’
reactions to a lack of errors found there (cf. Fiske, 1982;
Madan & Spetch, 2012; Taylor & Fiske, 1978). Finally, if
an ostensibly high-risk area of the audit turns out to be
clean, auditors may feel more trusting of management
regarding areas of the ?nancial statements not identi?ed
as risky (cf. Bowlin, Hales, & Kachelmeier, 2009; Bowlin
et al., 2014; King, 2002) and search for errors in the rest
of the ?nancial statements less vigilantly.
6
In contrast, a diversion to an ostensibly risky area in
which errors are discovered by the auditor is likely to have
a very different effect on auditor behavior. While prior psy-
chology research on distraction suggests that such diver-
sions would serve to inhibit the auditors’ effectiveness at
uncovering earnings management, the information pursuit
literature and other accounting research suggests that
diversions to areas in which auditors uncover errors would
have the opposite effect. Speci?cally, just as a purposeful
pursuit of evidence in a clean audit area that was speci?-
cally highlighted as high risk could amplify auditors’ reac-
tions to the lack of errors there, a similar diversion from
management that highlights speci?c accounts with errors
would similarly amplify auditors’ reactions to the errors
that they do ?nd there (cf. Nelson & Tayler, 2007;
Redelmeier et al., 2001). Even though the other errors have
no impact on earnings, they are not neutral with respect to
the audit. By increasing auditors’ reactions to the other
errors, the diversions to the other accounts would increase
the likelihood that the auditors would subsequently
uncover the earnings management. Management’s diver-
sion makes those speci?c accounts more salient, which
would tend to amplify auditors’ reactions to what they ?nd
there (cf. Fiske, 1982; Madan & Spetch, 2012; Taylor &
Fiske, 1978).
7
Combining the information pursuit literature with the
auditing literature suggests speci?c ways that diverting
auditors to other errors would amplify auditors’ reactions
to them. In particular, Nelson and Tayler (2007) predict
and ?nd that proactively pursuing information leads deci-
sion makers to consider potential implications of the infor-
mation in contextually speci?c ways, and therefore react to
the pursued information more strongly. The auditing liter-
ature suggests contextually speci?c ways in which these
reactions might occur within our setting. For example, if
management speci?cally directs auditors to accounts that
they believe are high risk, and yet apparently did little or
nothing to detect and remove any errors there, this diver-
sion sends an especially negative signal about manage-
ment’s internal controls over its ?nancial statement
preparation (COSO, 2012; PCAOB, 2007), elevating the
auditors’ beliefs about the risk of material misstatement
in the rest of the ?nancial statements (IFAC, 2009;
PCAOB, 2010a). Auditors are trained to exercise profes-
sional skepticism (Nelson, 2009; Quadackers et al., 2014;
Smith & Kida, 1991), suggesting that auditors would react
to a diversion from management to errors that the client
did not detect and remove as a signi?cant red ?ag for con-
trols over the ?nancial statements as a whole. If auditors
react to the discovery of other errors as a red ?ag, then
having been directed to a speci?c audit area to search for
such errors will likely strengthen their reaction to ?nding
them (cf. Nelson & Tayler, 2007; Redelmeier et al., 2001).
And, given that the reaction prescribed by audit standards
to uncovering such errors is to heighten concerns regard-
ing the risk of errors elsewhere in the ?nancial statements,
it is likely that auditors will be more vigilant in their search
for errors in the rest of the ?nancial statements.
Thus, while prior psychological research on distractions
suggests that diversions would be generally effective at
concealing earnings management, the information pursuit
literature in psychology and accounting, as well as other
accounting research and practice, suggests that the effect
of diversions would be quite different depending on
whether auditors are diverted to clean accounts or to
accounts containing other errors. Speci?cally, the diversion
frommanagement to other accounts would tend to amplify
auditors reactions to what is (or is not) in those other
accounts. Without a diversion to the other accounts, audi-
tors would be (if anything) more likely to detect earnings
management elsewhere in the ?nancial statements when
6
In fact, there is evidence of managers attempting to conceal informa-
tion by using diversions to clean areas. For example, to perpetuate its fraud,
the ZZZZ Best Company management repeatedly used diversionary state-
ments that directed auditors to accounts in its legitimate carpet-cleaning
subsidiary in order to divert attention away from its fraudulent restoration
business (Knapp, 2010). In addition, managers of HealthSouth and Crazy
Eddie allegedly concealed fraud from auditors with overt efforts to divert
auditors’ attention away from the fraud (Knapp, 2010).
7
Theory does not require that the diversions impact the rate at which
the other errors are discovered in order to in?uence auditors’ reactions to
them. Even if the other errors are discovered relatively easily and at just as
high of a rate without the diversion there, information pursuit theory still
predicts a stronger reaction to what is in those other accounts when they
are speci?cally directed to search there. Further, when no other errors are
present, diversions also cannot impact the rate at which the other errors are
discovered (since there are none to be discovered). Yet information pursuit
theory still predicts a stronger reaction to the lack of errors with a diversion
to those accounts.
B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54 43
other errors are present. However, this difference would be
more likely to be signi?cant when management speci?-
cally directs auditors to search those other accounts, based
upon information pursuit effects in accounting and psy-
chology, as well as other attributes of the auditing setting
discussed above. This does not preclude the possibility of
a smaller reaction to the other errors without management
highlighting the accounts that they are in. However, the
strongest a priori case for a reaction to the other errors
(or to a lack of other errors) in the ?nancial statements
would be when management explicitly directs auditors
to search those speci?c accounts. The different reactions
to the errors (or lack of errors) in those other accounts
would then affect the vigilance with which auditors search
the rest of the ?nancial statements, and, therefore, the like-
lihood of uncovering the earnings management error.
Taken together, this discussion suggests that auditors
would be the least likely to uncover earnings management
when clients divert auditors’ attention to accounts that are
clean, and most likely to uncover earnings management
when clients divert auditors’ attention to accounts that
contain errors. Auditors’ earnings management detection
would be between these highest and lowest levels when
there are no diversions to the other errors, as well as when
there are no diversions and no other errors. This suggests
the following hypothesis:
H1. An auditor’s detection of earnings management will
be lowest when diverted to clean accounts, highest when
diverted to accounts containing other errors, and in
between these levels either when there are no diversions
to clean accounts, or when there are no diversions to
accounts containing other errors.
Method
Participants
A representative from each of the Big Four and other
audit ?rms identi?ed auditors with suf?cient knowledge
to perform the task. Seventy-six auditors, with an average
of four years of audit experience, took part in the study.
8
Neither experience nor rank signi?cantly in?uences our
results. Our participants’ experience is consistent with prior
research on analytical review (see Messier et al., 2013 for a
review).
Overview of the study
The experiment required that auditors complete analyt-
ical review procedures on the ?nancials statements of a
hypothetical client. Following prior studies (e.g., Asare
et al., 2000; Asare & Wright, 2003; Bedard & Biggs,
1991a, 1991b; Bhattacharjee, Kida, & Hanno, 1999; Cohen
& Kida, 1989; Luippold & Kida, 2012; Moreno et al.,
2007), we adapted a set of ?nancial statements with stable
account balances over time, and then seeded an accounting
error into the accrual of compensation expense that would
understate current-year expenses and accruals by approx-
imately $450,000. The error caused unusual and material
?uctuations in these accounts for auditors to detect during
their analytical review.
9
The compensation expense error
was divided evenly into administrative compensation
expense and sales compensation expense, understating both
by approximately $225,000. The accrual entry was divided
evenly between current and non-current accrued compensa-
tion.
10
Background data, provided prior to beginning the
analytical review, revealed that the company beat analysts’
forecasted EPS by approximately $0.025/share (net income
was about $8.45 million). If the compensation error was dis-
covered, the company would miss its earnings target.
The study employed a 2 Â 2 experimental design. The
?rst independent variable manipulated whether manage-
ment provided a diversionary statement. The diversionary
statement involved management explicitly identifying risk
elsewhere in the ?nancial statements in an attempt to lure
the auditor away from managed earnings. In the diversion-
ary statement conditions, the client’s background informa-
tion indicated that the individual responsible for
maintaining non-current assets (i.e., property, plant and
equipment, intangibles and other non-current assets) left
the company about six months ago. It also stated that
her replacement transferred in from the manufacturing
?oor and has very little accounting experience. Aside from
that change, the auditors were told that there was no other
turnover with any of the accounting personnel responsible
for ?nancial reporting. In the other conditions, no speci?c
area of risk was identi?ed.
The second independent variable was manipulated by
seeding two other, offsetting errors in non-current assets
that had no impact on earnings but created unusual ?uctu-
ations for auditors to detect. One error concerned the com-
pany failing to record a portion of depreciation expense for
furniture and ?xtures, resulting in an understatement of
depreciation expense and accumulated depreciation by
approximately $450,000. The other error overstated
8
The original participant pool contained 77 auditors; however, one was
removed due to a software error. Approximately 66% of participants were
employed by Big Four ?rms, while 28% were employed by regional ?rms.
Firm type does not signi?cantly in?uence our results. Approximately 22%
were associate or staff auditors (averaging 1.4 years of audit experience),
61% were senior or in-charge auditors (averaging 3.5 years of experience),
7% were managers or senior managers (averaging 6.1 years of experience),
4% were partners (averaging 18.7 years of experience), and 7% did not
specify an auditor rank (averaging 3.7 years of experience).
9
Note that while errors can cause unusual ?uctuations, ?nancial
statement ?uctuations can also be the result of non-error causes. As a
result, our study more broadly tests diversions from unusual ?uctuations as
much as it tests diversions from earnings management. Because auditors
rely heavily on detection of unusual ?uctuations during analytical review
as a major source of substantive testing, failure to detect unusual
?uctuations elevates the risk of undetected misstatements (Messier,
Glover, & Prawitt, 2012). We are particularly concerned here with the
implications of our ?ndings for audits when management deliberately
conceals earnings manipulations. As Bell et al. (2005) note, those audits
carry the largest social costs, and the effectiveness of various tactics that
managers could use is not well understood by prior research.
10
The error was divided into different accounts so that it was more
dif?cult to uncover. While compensation is often a current accrued liability,
non-current accrued compensation can relate to post-retirement bene?ts,
deferred incentive compensation, pension bene?ts, non-expiring vacation/
sick time, etc.
44 B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54
amortization expense, and hence understated the net value
of goodwill by approximately $450,000.
11
These other
errors were embedded into the same area that the diversion-
ary statement pointed to (i.e., non-current assets). When the
other errors were not present, none of the accounts related
to depreciation and amortization re?ected any material ?uc-
tuations from previous years.
These manipulations resulted in four conditions (i.e.,
whether or not there were diversions to ?xed assets, and
whether or not there were other errors in ?xed assets).
The 2 Â 2 fully crossed design allows us to test the effects
of diversionary statements with or without other errors in
the diversionary accounts.
Procedures
The study was administered through a computer pro-
gram. After agreeing to participate, the auditors received
an email with the relevant information to access the pro-
gram (see Fig. 1 for the timeline of the experiment).
12
Upon
starting the study, the participants were randomly assigned
to a condition and navigated through a set of instructions
and background information about the company, its indus-
try, its position in the market, and details about its audit his-
tory. Participants were told that they were the senior-in-
charge on an audit of a manufacturing company. The com-
pany had consistently met analysts’ earnings forecasts, and
analysts had recently forecasted income to remain at $8.2
million (or $0.82 per share), which was the same as the pre-
vious two years. In addition, the materiality threshold for
the audit was explicitly stated to be $100,000, which made
material the ?uctuations for all of the accounts affected by
the earnings management and the other errors.
After reading the instructions and background, the
auditors began the analytical review. At this stage, partici-
pants were exposed to information from the client that
compared the unaudited ?nancial balances of the current
year to the audited balances of the previous two years.
13
Navigation buttons allowed the participants to access all of
the ?nancial details. In total, twenty pages of information
presented the balance sheet, income statement, statement
of cash ?ows, and additional detailed information which fur-
ther described the balances on the ?nancial statements.
14
For example, one page provided details for accounts receiv-
able, such as gross and net accounts receivable balances,
allowance for doubtful accounts, bad debt expense, an aging
analysis and key ?nancial ratios. The navigation buttons
were always present on the left side of the screen during
the analytical review, so that participants could move freely
to any piece of information in any order they preferred.
A button on each screen labeled ‘‘Record Judgment’’
brought the auditors to a page where they could record
any unusual ?uctuations that they identi?ed in a free
response text box. They could return to the ‘‘Record Judg-
ment’’ page as often as they wanted to add new judgments.
Download
and Install
Program
Enter Pin
Code and
Name
Read
Instructions
Read
Background
Information
Conduct
Analytical
Review
Answer Post
Experimental
Questions
Results
File
Record
Error
Judgments
Fig. 1. Timeline of experimental procedures. The experiment was conducted through a computer program that participants downloaded and completed at
their convenience. Participating auditors received an email with download instructions and a pin-number to grant them access to the program. After
downloading the program, they entered their name and pin number. They then read instructions and background information before beginning the
analytical review. During the review, auditors searched through several pages of information related to the client’s ?nancial statements. If participants
identi?ed an error, they could navigate to a page to record their judgment as part of their evaluation. Participants could make as many error judgments as
needed throughout the review, and they could revert back to the background information as necessary. Upon completing the analytical review, auditors
answered post experimental questions. At this stage, they could not go back to the analytical review. After completing the study, participants emailed the
?le back to the experimenters.
11
To ensure that the earnings management error was more dif?cult to
detect than the other errors, ?ve individuals with a mean of over three
years of audit experience rated the three errors on a 10-point scale, which
ranged from very easy (one) to very dif?cult (ten). Signi?cant differences
were found between the earnings management error (6.4) and the other
errors (amortization = 2.6, p = 0.002 and depreciation = 3.4, p = 0.008),
suggesting that the earnings management was more dif?cult to uncover
than the other errors.
12
There was no time limit to the task, and we ?nd no signi?cant effects of
time on our results.
13
The ?nancial statements were created from several other accounting
studies using similar analytical review procedures, including Bhattacharjee
et al. (1999), Cohen (1994), Cohen and Kida (1989), and Moreno et al.
(2007).
14
In addition to the balance sheet, income statement and cash ?ow
statement, participants could access details for the following accounts:
marketable securities, accounts receivable, inventory, cost of goods man-
ufactured/sold, property, plant and equipment, prepaid expenses, accumu-
lated depreciation, intangibles, other non-current assets, accounts payable,
other current liabilities, debt, other non-current liabilities, equity, sales,
selling expenses and administrative expenses. In addition, a navigation
button allowed participants to revisit the background information.
B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54 45
Each time they returned, all of their previous entries were
listed numerically in the order they were entered. Another
button allowed them to ?nish the exercise, and brought
them to supplemental questions (i.e., demographic infor-
mation, questions pertaining to professional skepticism,
etc.). Participants emailed the results ?le back when they
completed the study.
Results
We ?rst test our hypotheses about the effects of diver-
sions (to accounts with or without other errors) on an
auditor’s identi?cation of an earnings management error.
As a supplemental analysis, we then examine whether
auditor skepticism affects an auditor’s detection of man-
aged earnings.
Dependent variables
We use two dependent variables to proxy for the
identi?cation of earnings management. The purpose of
analytical review is to highlight accounts with unusual
?uctuations in order to focus the audit team’s subsequent
attention and effort where the risk of material misstate-
ment is highest. Since the experiment involved a realistic
analytical review task, which allowed the auditors to list
as many potential errors as they felt necessary, partici-
pants could simply list many areas that they believe
may contain an error (i.e., take a ‘‘shotgun’’ approach).
While identifying a large number of accounts for investi-
gation should, in theory, increase the likelihood of detect-
ing the earnings management error, ?agging additional
irrelevant accounts ultimately diverts the audit team’s
subsequent attention away from the accounts that con-
tain the highest risk of misstatement to other accounts
that do not. As such, each additional account identi?ed
dilutes the amount of resources the audit team can
commit to locating the earnings manipulation (i.e., the
compensation error). This could affect not only audit
ef?ciency, but also effectiveness, because it ultimately
reduces the audit team’s attention to the area containing
the managed earnings.
15
Because of this effect, we use two alternative dependent
variables for our tests of H1. Our ?rst dependent variable
(Task Performance) places identi?cation of the earnings
management account into the context of how many irrele-
vant accounts were also identi?ed. It is calculated as
follows:
Task Performance ¼
Identification of Earnings Management
1þNumber of Irrelevant Judgments
ð1Þ
The numerator (Identi?cation of Earnings Management) is
coded correct (1) if the auditor identi?ed one or more of
the affected compensation accounts (or listed compensa-
tion or an appropriate synonym), and incorrect (0) other-
wise. The number of irrelevant judgments re?ects the
number of accounts (or areas) identi?ed by the auditor as
potentially containing an error where no unusual ?uctua-
tion actually existed. For instance, if a participant listed
inventory, accounts receivable and leases as accounts pos-
sibly containing errors, then three irrelevant judgments
would be recorded.
16
Our second dependent variable is the percentage of
auditors who identi?ed one or more of the affected com-
pensation accounts (or listed compensation or an appropri-
ate synonym) in their responses. This dependent variable
shows the proportion of auditors who named the earnings
management account, but does not account for whether
they also highlighted a large number of other, irrelevant
accounts.
17
0.19
0.04
0.17
0.27
0.00
0.05
0.10
0.15
0.20
0.25
0.30
No Diversionary Statement Diversionary Statement
No Other Errors Other Errors
Fig. 2. The effect of diversions to accounts with or without other errors
on task performance. Task Performance ¼
Identification of Earnings Management
1þNumber of Irrelevant Judgments
.
15
Our approach is consistent with studies on brainstorming (e.g.,
Bellovary & Johnstone, 2007; Carpenter, 2007; Hammersley, 2011;
Trotman, Simnett, & Khal?a, 2009). It is also consistent with our interviews
of audit partners, who indicated that good performance in analytical review
entails not merely highlighting high risk ?uctuations, but doing so without
?agging other accounts that do not contain unusual ?uctuations.
16
This task performance measure scales detection of the compensation
error by the number of erroneous judgments listed, so that the measure
re?ects both audit effectiveness and ef?ciency. Identifying a larger number
of additional irrelevant accounts diverts subsequent audit testing away
from higher risk accounts, risking both audit effectiveness and ef?ciency
(Messier et al., 2012). For instance, an auditor who identi?ed the
compensation error without any irrelevant judgments would receive a
score of 1.00. An auditor who listed three irrelevant judgments along with
the compensation error would receive a score of 0.25. An auditor who failed
to indicate the compensation error would receive a score of 0.00, regardless
of the number of irrelevant judgments listed.
17
Neither of our dependent variables (Task Performance or the percentage
of auditors identifying earnings management) includes identi?cation of the
other errors in their calculation. Libby and Frederick (1990, 360) and Libby
(1985, 660–661) design their dependent variables this way for similar
reasons. Speci?cally, had we also included detection of the other errors in
the calculation of our dependent variables, results consistent with H1 could
have then been merely attributed to participants in the ‘‘no other errors’’
conditions having fewer errors available for detection to begin with, giving
an unfair advantage in the dependent variable to those in the ‘‘other errors’’
conditions. Instead, our dependent variables are consistent with our focus
on testing the effects of diversions on the detection of earnings manage-
ment elsewhere in the ?nancial statements. Roediger, Stellon, and Tulving
(1977) discuss the methodological importance of this approach.
46 B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54
Hypothesis tests: task performance
Fig. 2 shows the means of Task Performance by experi-
mental condition. H1 predicts that auditors’ detection of
earnings management will be lowest when auditors are
diverted to clean accounts, highest when auditors are
diverted to other accounts that contain errors, and in
between these levels when auditors are not diverted to
the other errors, as well as when there are no diversions
and no other errors.
Predictions of this type are tested by a Jonckheere-
Terpstra test (e.g., Abarbanell, 1991; Fanning, Agoglia, &
Piercey, 2015; Frederickson, Hodge, & Pratt, 2006; Libby
& Lipe, 1992; Libby & Trotman, 1993; Phua, Abernathy, &
Lillis, 2011; Sedor, 2002; Tan & Jamal, 2006). As Fig. 2
shows, the mean Task Performance score of auditors in
the Diversion, No Other Errors cell is the lowest at 0.04, fol-
lowed by the two No Diversion cells, with and without
errors (0.17 and 0.19, respectively), followed by the Diver-
sion, Other Errors cell (0.27). As Table 1 shows, a Jonckhe-
ere-Terpstra test shows that the predicted rank-orderings
are statistically signi?cant. Speci?cally, task performance
in detecting earnings management was lowest when audi-
tors were diverted to clean areas, highest when auditors
were diverted to accounts that contained other errors,
and in between these highest and lowest levels when audi-
tors were not diverted to the other errors (J = 1.96,
p = 0.025), as well as when they were not diverted and
there were no other errors (J = 1.91, p = 0.028). These
results support H1.
18
We perform supplementary analyses of our Task
Performance variable using ANOVA. Table 1 shows a
Diversion  Other Errors interaction signi?cant at
p = 0.074. Consistent with information pursuit effects,
the interaction shown in Fig. 2 suggests that diverting
auditors to the other accounts ampli?es their reaction
to what is (or, when the other errors are not present,
to what is not) in those other accounts. Speci?cally,
when no diversions are used, auditors’ did not react to
the presence or absence of the other errors (0.17 vs.
0.19, respectively, t = À0.24, p = 0.810). In contrast, when
diverted to the other accounts, auditors’ reacted signi?-
cantly to the presence or absence of the other errors
(0.27 vs. 0.04, respectively, t = 2.31, p = 0.012). Auditors’
reactions to the errors (or lack of errors) in the other
accounts were signi?cantly larger with the diversion to
those accounts than without (t = 1.81, p = 0.037).
19
These
supplementary results are consistent with our theory and
hypothesis.
As an additional analysis, we also tested whether the
task performance score in each condition was signi?cantly
different from zero. The scores in the two no diversion con-
ditions, as well as in the diversions to accounts with errors
condition, were all signi?cantly different from zero
(p’s 6 0.010). On the other hand, the diversionary state-
ment without other errors was not different from zero
(p = 0.549). These ?ndings suggest that when auditors are
diverted to areas that do not contain errors, they are not
likely to uncover earnings management elsewhere in the
?nancial statements, compared to auditors in the other
experimental conditions.
20
These results suggest that if management alerts audi-
tors to risk in accounts that are ultimately clean, this
diversionary tactic appears to be effective at diverting
the auditor away from managed earnings. That is,
Table 1
The effect of diversions to accounts with or without other errors on task
performance.
a
Diversion
No Yes
Panel A: Means, (standard deviations), sample size
Other errors
No 0.19 (0.37) 0.04 (0.09)
n = 19 n = 21
Yes 0.17 (0.32) 0.27 (0.38)
n = 19 n = 17
Source SS df MS F p
Panel B: ANOVA
Other errors 0.21 1 0.21 2.16 0.146
Diversion 0.01 1 0.01 0.11 0.739
Diversion  other errors 0.31 1 0.31 3.28 0.074
Error 6.84 72 0.09
Total 7.37 75
Test
statistic
p
b
Panel C: Test of hypotheses
Diversion to clean accounts lowest, diversion
to other errors highest, and no diversion
to the other errors in between
J = 1.96 0.025
Diversion to clean accounts lowest, diversion
to other errors highest, and no diversion
with no other errors in between
J = 1.91 0.028
a
Task Performance ¼
Identification of Earnings Management
1þNumber of Irrelevant Judgments
.
b
These tests have directional predictions, and their p-values are one-
tailed.
18
In addition to their use of the Jonckheere-Terpstra test, Fanning et al.
(2015), Sedor (2002) and Arel, Jennings, Pany, and Reckers (2012) also used
cell contrast weights with a rank-ordering re?ecting their expectations. We
use this approach as well, as additional robustness tests of H1. For example,
contrast weights of +0.5, +0.3, +0.2, À1 could be assigned, (respectively) to
the diversion to other errors condition, the no diversion to other errors
condition, the no diversion and no other errors condition, and the diversion
to clean accounts condition, as an alternative test of our theory and
hypothesis. Compared to contrast weights, the Jonckheere-Terpstra test has
the advantage of being a direct test of the rank-ordering of the cell means
without also requiring arbitrary decisions about the relative magnitude of
the contrast weights (Fanning et al., 2015). Because contrast weights
require arbitrary decisions about the magnitude, they should be conducted
as sensitivity analyses, using different sets of weights that all re?ect the
basic rank-ordering to be tested (e.g., Arel et al., 2012; Sedor, 2002).
Consequently, we tested H1 using contrast weights of (+0.5, +0.3, +0.2, À1),
(+1, À0.2, À0.3, À0.5), (+0.75, +0.13, +0.12, À1), and (+1, À0.12, À0.13,
À0.75). In all cases, results are supportive of H1 (all p’s 6 0.032).
19
This test (t = 1.811, p = 0.037) is the one-tailed t-test associated with
the F-test of the Diversion  Other Error interaction (F = 3.28, p = 0.074) in
Table 1 (e.g., Kachelmeier & Williamson, 2010).
20
In addition, the 0.04 task performance score in the diversionary
statement without errors condition is signi?cantly less than the average
score of 0.21 for the rest of the conditions (t = À2.16, p = 0.017). Also, the
0.27 score for auditors diverted to accounts with errors is higher than the
0.13 average score of the remaining conditions (t = 1.64, p = 0.053) (see
Kadous, Koonce, & Towry, 2005).
B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54 47
auditors appear to be relatively ineffective at detecting
managed earnings in other areas of the audit when
diverted to a clean account. However, if management
overtly leads auditors to an area containing errors, audi-
tors perform better at discovering managed earnings else-
where in the ?nancial statements. Overall, our results
appear to be consistent with information pursuit theory.
That is, directing auditors to search other accounts
appears to amplify their reactions to what either is or is
not in those other accounts.
21
Hypothesis tests: percentage of auditors identifying earnings
management
We also test our hypotheses on the percent of audi-
tors who detected the unusual ?uctuations created by
managed earnings.
22
To limit the effect of auditors
identifying managed earnings as a result of simply taking
a shotgun style approach (i.e., by listing many accounts),
we ?rst analyzed the judgments of auditors who identi?ed
fewer than ?ve irrelevant errors. We use this cutoff
because, as we previously discussed, large numbers of
irrelevant judgments imply that auditors are inef?cient,
as they may pursue false leads. In addition, large numbers
of irrelevant judgments suggest that the auditor may not
actually have detected a likely instance of earnings man-
agement, but simply included the relevant accounts by
chance while generating a laundry list of accounts to
examine. A valid measurement of earnings management
detection would not count these chance inclusions of the
relevant accounts. We selected the moderately low cutoff
of fewer than ?ve irrelevant items so as to clearly distin-
guish those including many irrelevant items in their
responses (and thus likely using a shotgun approach) from
those including very few (and thus less likely to be using
one).
As can be seen in Fig. 3, only 6.7% of auditors who
were diverted to clean accounts uncovered managed
earnings, while 43.8% of auditors uncovered earnings
management when they were diverted to accounts that
contained the other errors. In contrast, when auditors
received no diversions to ?xed assets (either with or
without the other errors), 29.4% detected the earnings
management. Jonckheere-Terpstra tests of H1 indicate
that auditors’ earnings management detection was low-
est when diverted to clean accounts, highest when
diverted to the other errors, and in between when no
diversion was used, with or without the other errors
(in both tests, J = 2.29, p = 0.011; Table 2). These results
support H1.
29.4%
6.7%
29.4%
43.8%
0%
10%
20%
30%
40%
50%
No Diversionary Statement Diversionary Statement
No Other Errors Other Errors
Fig. 3. The effect of diversions to accounts with or without other errors
on the percent of auditors identifying earnings management. This
analysis includes auditors who listed fewer than ?ve irrelevant judg-
ments to limit the impact of taking a shotgun approach. Responses were
coded correct if any of the recorded judgments identi?ed the affected
compensation accounts, or listed the word compensation or an appro-
priate synonym, and incorrect otherwise.
Table 2
The effect of diversions to accounts with or without other errors on the
percent of auditors identifying managed earnings.
a
Diversion
No Yes
Panel A: Percentages, sample sizes
Other errors
No 29.4% 6.7%
n = 17 n = 15
Yes 29.4% 43.8%
n = 17 n = 16
Test
statistic
p
b
Panel B: Hypothesis tests
Diversion to clean accounts lowest, diversion
to other errors highest, and no diversion
to the other errors in between
J = 2.29 0.011
Diversion to clean accounts lowest, diversion
to other errors highest, and no diversion
with no other errors in between
J = 2.29 0.011
a
This analysis includes auditors who listed fewer than ?ve irrelevant
judgments to limit the impact of taking a shotgun approach. Responses
were coded correct if any of the recorded judgments identi?ed the
affected compensation accounts, or listed the word compensation or an
appropriate synonym, and incorrect otherwise.
b
These tests have directional predictions, and their p-values are one-
tailed.
21
Note that, even if auditors discovered the other errors at just as high of
a rate without being diverted there, information pursuit theory still
suggests that diversions would amplify auditors’ reactions to the other
errors (or to the lack of errors) in those accounts. In fact, our ?ndings are
consistent with this. When the other errors were present, auditors detected
the other errors there just as much without the diversion as they did with it
(auditors found on average 1.4 vs. 1.5 of the other errors, t = 0.39, p = 0.70).
When the other errors were not present, diversions also did not impact
auditors’ detection of the other errors (since there were none to be found).
Yet, our tests of H1 (and the supplemental interaction test) suggest that
auditors still reacted more to the other errors (or to the lack of errors) in the
other accounts when they were diverted there by management (Table 1,
Panels B and C), consistent with information pursuit theory. In the
information pursuit literature, Nelson and Tayler (2007) and Smith et al.
(2011) discuss the methodological importance that participants acquire
information at approximately the same rate when testing whether the
explicit information pursuit (e.g., the diversions to other accounts)
ampli?es their reactions to it.
22
Again, earnings management was coded as correct if the auditor
identi?ed any of the affected compensation accounts (or listed compensa-
tion or an appropriate synonym), and incorrect otherwise.
48 B.L. Luippold et al. / Accounting, Organizations and Society 41 (2015) 39–54
Fig. 3 suggests an interaction in this dependent vari-
able similar to the interaction that we observed in the
supplementary analysis of our Task Performance depen-
dent variable. Speci?cally, Fig. 3 suggests that diverting
auditors to the other accounts appears to amplify their
reaction to the errors (or to the lack of errors) in those
other accounts, consistent with information pursuit
effects. To test this interaction, we employ a binomial
logistic regression model with Diversion, Other Errors,
and their interaction as independent variables. We ?nd
that the interaction shown in Fig. 3 is statistically signif-
icant, as expected. Speci?cally, auditors’ detection of the
earnings management is more in?uenced by the presence
or absence of errors in the other accounts when diverted
to those accounts by management (v
2
= 3.55, p = 0.030).
23
Overall, our ?ndings suggest that directing auditors to
clean accounts can be an effective audit management tool,
but directing auditors to accounts containing other errors
may not only be ineffective, it may actually have the oppo-
site effect on management.
Further, we performed a sensitivity analysis to deter-
mine how the number of irrelevant judgments impacts
the percent of auditors uncovering earnings manage-
ment. As shown in Table 3, the results for H1 replicate
when we include auditors who are not simply using a
shotgun approach (i.e., naming a large number of irrele-
vant accounts), and become more signi?cant as the
number of irrelevant items included in the analyses
decreases.
24
Thus, it appears that the effect of diversions (to accounts
with and without other errors) on the proportion of audi-
tors uncovering earnings management are robust across
the number of irrelevant judgments.
Auditor skepticism
We also investigated the impact that skepticism may
have on an auditor’s detection of earnings management.
As part of the post-experimental questionnaire, auditors
were asked to indicate their level of agreement with four
statements (shown in Table 4) re?ecting a skeptical out-
look toward management on six-point scales (ranging
from strongly disagree to strongly agree).
These measures of skepticism capture auditors’ ten-
dency to adopt a presumptive doubt perspective on man-
agers’ representations (Bowlin et al., 2014; Nelson, 2009;
Quadackers et al., 2014). Compared to a more neutral
perspective of professional skepticism (e.g., Hurtt, 2010),
a presumptive doubt view focuses speci?cally on a
heightened assessment of the risk that management is
misleading (Nelson, 2009). Because a presumptive doubt
view would tend to indicate an elevated alertness to
red ?ags, Quadackers et al. (2014) suggest that auditors’
Table 3
The effect of diversions to accounts with or without other errors on the percent of auditors identifying managed earnings: sensitivity to number of irrelevant
judgments.
a
Irrelevant judgments
included
b
N No other errors Other errors Tests of H1
Diversion (A) (%) No diversion (B) (%) No diversion (C) (%) Diversion (D) (%) A < B < D
(p-value)
c
A < C < D
(p-value)
c
All 76 23.8 31.6 31.6 47.1 0.070 0.070