Description
We examine the research literature on audit groups/teams focusing on three main areas: the hierarchical
review process, brainstorming as part of the fraud detection planning process, and consultation within
firms. We restrict our discussion of these three literatures to judgment and decision making (JDM) experiments.
We consider research where two or more individuals within the audit firm interact with one
another face-to-face, electronically, or where one person prepares/reviews working papers for another.
Group judgment and decision making in auditing: Past and future
research
Ken T. Trotman
a, *
, Tim D. Bauer
b
, Kerry A. Humphreys
a
a
School of Accounting, UNSW Australia, Australia
b
University of Illinois at UrbanaeChampaign, USA
a r t i c l e i n f o
Article history:
Received 25 August 2015
Accepted 11 September 2015
Available online 9 October 2015
Keywords:
Review process
Brainstorming
Consultation
a b s t r a c t
We examine the research literature on audit groups/teams focusing on three main areas: the hierarchical
review process, brainstorming as part of the fraud detection planning process, and consultation within
?rms. We restrict our discussion of these three literatures to judgment and decision making (JDM) ex-
periments. We consider research where two or more individuals within the audit ?rm interact with one
another face-to-face, electronically, or where one person prepares/reviews working papers for another.
We outline future research within each of the above areas, as well as considering other areas of future
research involving within-?rm group interactions related to audit teams in context, shared mental
models, and audit team diversity (including sustainability assurance), as well as interactions with groups
outside the audit ?rm, particularly audit committees.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
The judgment and decision making (JDM) literature in auditing
emerged in the early 1970s, with Accounting, Organizations and
Society (AOS) playing a signi?cant role in the early development of
this ?eld (Libby & Lewis, 1977, 1982). While most JDM audit liter-
ature concentrates on individual auditor judgments (Nelson &Tan,
2005), audits are generally conducted by groups. They often involve
a hierarchical team where preparers may stylize audit tests and
working papers depending on the reviewer, and the reviewprocess
will be affected by a range of preparer, task and environmental
factors (Rich, Solomon, &Trotman, 1997a, 1997b). Concentrating on
single-person studies is problematic, as psychology research shows
the results of these studies often do not generalize to multi-person
settings (Solomon, 1987).
Audit group research draws not only on the wider audit JDM
experimental literature for both theory and methods, but also on
the social psychology/organizational behavior literature. The latter
literature has a long history (see Lorge, Fox, Davitz, &Brenner, 1958,
for an early review). Some evidence of the extent of group research
is its inclusion as one of three key parts of the Annual Review of
Psychology in 1961 (McGrath & Kravitz, 1982) leading to regular
review papers of new research being published in the Annual Re-
view of Psychology from 1962 to 1991, approximately every three
years. With the increased impetus on group research, the reviews
in the Annual Review of Psychology from 1996 then began to be split
into more specialized subsections of group/team research.
1
1.1. Scope of this review
Our working de?nition of research on audit groups/teams in-
cludes research where two or more individuals within the audit
?rm interact with one another. This interaction can be face-to-face,
electronic, or one person preparing/reviewing working papers for
another. It implies that the persons are mutually aware of one
another and take into account the actions of other group members
(McGrath & Kravitz, 1982). Our de?nition is broad enough to
include group interactions between an auditor and other internal
* Corresponding author. School of Accounting, UNSW Business School, UNSW,
Sydney, NSW 2052, Australia.
E-mail address: [email protected] (K.T. Trotman).
1
Guzzo and Dickson (1996) and Ilgen et al. (2005) distinguish their reviews from
previous research based on their focus on team performance in organizational
contexts, where work groups and teams are normally embedded in organizations
and exist over a period in time. In comparison, Kerr and Tindale (2004) restrict
themselves to cover group performance and decision making with a focus on the
groups' outputs (e.g., solutions, decisions) and the processes employed by groups.
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j ournal homepage: www. el sevi er. com/ l ocat e/ aoshttp://dx.doi.org/10.1016/j.aos.2015.09.004
0361-3682/© 2015 Elsevier Ltd. All rights reserved.
Accounting, Organizations and Society 47 (2015) 56e72
parties, for example studies where the auditor consults with others
in the ?rm (both auditors and other specialists).
Auditors also interact with other professions outside the audit
?rm including others involved in corporate governance such as
client management and the audit committee, and external spe-
cialists especially for complex estimates. We exclude the literature
on auditor-client negotiations given its coverage in recent review
papers (e.g., Brown & Wright, 2008). We include interactions be-
tween auditors and audit committees in future research, given the
enhanced role of audit committee members (ACMs) for effectively
overseeing the ?nancial reporting and auditing process (CAQ, 2014;
FEE, 2014; FRC, 2013; Peecher, Solomon, & Trotman, 2013) and
because we see this as an important area of future research not yet
reviewed.
We restrict our review to JDM experiments, although we draw
on ?eld studies in considering future research, and focus on four
main areas:
The hierarchical review process: the process is a key control
mechanism within audit ?rms which affects both reviewer
behavior and preparer behavior. Reviewer behavior includes
comparisons of outputs with other group formats (e.g., inter-
acting and nominal groups); different structures of the review
process; factors affecting the output of the review process; and
cognitive effects of taking a reviewer role. Preparer behavior
includes both the effect of the review process as a form of
accountability, as well as the motives/actions of preparers
including persuasion.
Brainstorming: as part of the fraud detection planning process,
auditing standards require a discussion (brainstorming
meeting) between audit team members, including how and
when the client's ?nancial statements may be susceptible to
material misstatement due to fraud. We consider nominal
versus interacting group brainstorming, face-to-face versus
electronic brainstorming, and the relationship between various
judgments made within the brainstorming session.
Consultation within ?rms: in addition to the review process,
consultation with other auditors (i.e., advice seeking and giving)
is common and can reduce uncertainty about dif?cult decisions
and help auditors defend a position to various stakeholders. We
consider the justi?ability of consulting advice, howauditors use
or follow consulting advice, and their willingness to consult, in
both formal and informal settings.
New directions and future research: in addition to future
research discussed within the context of the above three areas,
we consider other areas of future research involving within-?rm
group interactions related to audit teams in context, shared
mental models, and audit team diversity (including sustain-
ability assurance), as well as interactions with groups outside
the audit ?rm, particularly audit committees.
Previous research on multi-auditor judgments (Rich,
Solomon, & Trotman, 1997b; Solomon, 1987) distinguishes be-
tween the terms ‘audit teams’ and ‘audit groups’, where ‘audit
teams’ refers to the hierarchical review process and ‘audit
groups’ refers to situations where auditors solve a particular
problem or perform a task as a group. However, more recent
research in psychology has suggested the distinction is “a rather
arti?cial one”, with Kerr and Tindale (2004, p. 624) explaining
the difference as follows:
“Team research tends to focus on relatively longer-term groups
with multiple task responsibilities, often functioning within an
organization. …Small group performance research, on the other
hand, tends to be basic research …, and is usually studied
experimentally in ad-hoc laboratory groups. However, there are
many exceptions to these general rules. …”
The related research in the audit literature has elements of both
group and team research as described above. For example, in
practice, groups within audit ?rms are normally part of a team
which exists over time, has multiple responsibilities, and generally
operates within the organization. On the other hand, the actual
research conducted on groups in auditing invariably is not con-
ducted within actual teams and generally studies ad-hoc groups
where the participants may or may not have previously worked
together (for exceptions see Tan & Jamal, 2001; Trotman, 1985).
Participants may have never met other group members, may have
met them socially or at training, or may have worked together on a
number of audits. Thus we use the terms ‘team’ and ‘groups’
interchangeably.
1.2. Group audit JDM literature over time
Table 1 divides 57 group audit JDM papers across four leading
international journals e Accounting, Organizations and Society
(AOS), Contemporary Accounting Research (CAR), Journal of Ac-
counting Research (JAR), and The Accounting Review (TAR) (Panel A)
e and across time (Panel B).
2
A number of important observations
can be made. First, under the heading ‘Pre-review interacting
groups’ only three papers are included, which were all published in
JAR in the 1980s. However, we note some of the other papers in the
review process category include an interacting group as a com-
parison (Owhoso, Messier, & Lynch, 2002; Trotman, 1985) and all
the brainstorming papers include an interacting group (except
Hammersley, Bamber, & Carpenter, 2010). However, the nature of
the task in the brainstorming studies involves idea generation,
whereas in the reviewprocess studies the task is usually evaluation
or estimation.
3
Second, there are two quite different streams of research on the
review process, with one looking at the effects of the review pro-
cess on the reviewer (and comparing the quality of the audit post-
versus pre-review) and the other examining the effects on the
preparer via accountability and justi?cation. Both categories have a
substantial number of papers (23 and 18 respectively). However,
we note that there have been only three papers on the effects of the
review process on the reviewer in the last decade (Frank &
Hoffman, 2015; Peecher, Piercey, Rich, & Tubbs, 2010; Tan & Tan,
2008). We do not believe this is due to a decrease in the impor-
tance of this topic, but is more related to the dif?culties of obtaining
participants with suf?cient seniority to carry out the review. On the
other hand, papers on the effects of the review process on the
preparer, while ?rst appearing almost a decade after those papers
on the effects on the reviewer, have continued into the current
decade. Third, the titles of the review process papers on ‘effects on
the reviewer’ and ‘effects on the preparer’ are quite different, with
19 of the 23 ‘effects on the reviewer’ papers including the word
‘review’ in their title, indicating the papers are clearly about the
review process. In contrast, only ?ve of the 18 ‘effects on the pre-
parer’ papers include the word ‘review’, suggesting that these pa-
pers are more about accountability and justi?cation, with the
review process as the means of inducing these behaviors.
Fourth, the brainstorming papers have all been published
2
While we note that Table 1 is limited to these four journals, throughout the
paper we do refer to some important papers in other journals, particularly Auditing:
A Journal of Practice & Theory.
3
This is an important distinction and the results are often not transferable across
tasks (see Kerr & Tindale, 2004).
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 57
between 2007 and 2015, with ?ve of the six papers published in
TAR. This research focus re?ects the requirements of US (AU 316)
and international (ISA 240) auditing standards for a brainstorming
session to be held, which includes an audit team discussion as to
howand when a client's ?nancial statements may be susceptible to
material misstatement from fraud (AICPA, 2002; IAASB, 2004).
Fifth, while the consultation research theme started in the 1990s
with three papers, there has been renewed interest in the
2010e2015 period with a further three papers published. This
renewed interest is likely motivated, in part, by revisions to rules
about audit team consultations (Gold, Knechel, & Wallage, 2012)
and by regulator concerns about audit team reliance on specialists'
work (Bauer & Estep, 2015; PCAOB, 2015), which involves consul-
ting and coordinating with these specialists.
2. The hierarchical review process
The audit review process refers to members of the audit team
evaluating the work of other audit teammembers; it is a key control
mechanism within audit ?rms aimed at increasing audit quality
(Libby & Trotman, 1993; Solomon, 1987). The review process is
generally hierarchical and sequential in that additional review may
occur as the work moves up the hierarchy to partners. The two
main objectives of the review process relate to ensuring the
appropriateness of both the opinion given and the documentation
provided (Rich et al., 1997a). The reviewer identi?es alternative
explanations, omissions, and inconsistencies that were missed by
the preparer, and evaluates whether the documentation is suf?-
cient to support the conclusion reached (Libby & Trotman, 1993;
Tan & Trotman, 2003). Also, the preparer anticipates conse-
quences from the review, and the need to justify and document
their conclusions to the reviewer (Emby &Gibbins, 1988; Rich et al.,
1997a). This in?uences audit quality through the preparer's effort
from feeling accountable to the reviewer and his/her strategic
behavior to stylize workpapers in order to persuade the reviewer
that adequate evidence has been collected and that documentation
is appropriate (Rich et al., 1997a).
In examining the review process from the perspective of the
reviewer, the fundamental issues to address are under what cir-
cumstances does the review process improve the quality of the
audit and why does this occur. Accordingly, we divide the research
on the effects of the review process on the reviewer into the
following categories: comparisons of the review process against
other baselines, such as individual judgments, nominal groups, and
non-hierarchical interacting groups; what performance gains result
from the review process; comparisons of the performance of
alternative forms of the review process (e.g., face-to-face versus
electronic); the effect of environmental factors (e.g., task being
reviewed); and the effect of preparer attributes on the review
process (e.g., stylization). A summary of the papers addressing the
reviewer perspective is provided in Fig. 1 Panel A.
Potential reasons for why reviewed judgments may be of higher
quality than pre-review judgments are the inclusion of an addi-
tional person (a diversi?cation effect), the discussion between the
two group members leading to some form of cognitive stimulation
(an interaction effect), or the addition of a person of a higher rank
who decides on the relative weights for different inputs to the
judgments (a hierarchical effect). To isolate these differences, the
ideal research design is to have individual judgments (e.g., both
seniors and managers), interaction between two auditors (either
two seniors or one senior and one manager), and also the
combining of individual judgments for a nominal group (either two
seniors or one senior and one manager). Differences due to diver-
si?cation, interactions and the hierarchy can then be isolated.
In making the above comparisons of individuals, interacting
groups and a review process, it is necessary for the researcher to
have a benchmark and this is often done by forming nominal teams.
If the task is an estimate (e.g., inventory obsolescence or an
assessment of the strength of an internal control system), this will
be a statistical average of individual estimates. If it is a prediction of
an event (e.g., a company will fail/not fail) it will be a majority vote.
If it is a brainstorming task it will be the combining of all unique
ideas generated. Note that nominal teams can only be formed if
there is a within-subject treatment with auditors completing the
task as individuals ?rst, or a between-subjects design where par-
ticipants in one of the treatments complete the task as individuals.
4
In examining the review process from the perspective of the
preparer, the fundamental issues to be addressed are how
accountability to the reviewer or stylization attempts in?uence
Table 1
Publications of group audit JDM experiments.
The review process Brainstorming Consultation
Pre-review interacting groups Effects on the reviewer
b
Effects on the preparer
b
Panel A: By topic and journal in Accounting, Organizations and Society (AOS), Contemporary Accounting Research (CAR), Journal of Accounting Research (JAR) and
The Accounting Review (TAR) from 1970 to 2015
a
Journal
AOS 0 4 2 0 2
CAR 0 6 1 1 1
JAR 3 6 10 0 1
TAR 0 7 5 5 3
Total 3 23 18 6 7
Panel B: By topic and decade in Accounting, Organizations and Society (AOS), Contemporary Accounting Research (CAR), Journal of Accounting Research (JAR) and
The Accounting Review (TAR) from 1970 to 2015
a
Decade
1970s 0 0 0 0 0
1980s 3 5 0 0 0
1990s 0 9 8 0 3
2000s 0 7 7 4 1
2010e2015 0 2 3 2 3
Total 3 23 18 6 7
a
2015 papers are those published at June 2015.
b
Several papers on the review process consider both the effects on the reviewer and the preparer. These have been categorized based on their primary focus here, but in
later tables the effects on both groups are presented.
4
Such benchmarks are an important example of how an experiment can include
treatments that do not presently exist in practice to tease out the explanations
between different treatments (Libby & Luft, 1993).
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 58
evidence collection and documentation, and ultimately, audit
quality. Accordingly, we divide the research on the effects of the
review process on the preparer into reviewer preferences that are
unknown versus known, as well as alternative review formats. A
summary of the papers addressing the preparer perspective is
given in Fig. 1, Panel B. Bonner and Sprinkle (2002) distinguish
among effort duration, intensity, and direction; one observation
that becomes apparent from distinguishing between unknown
versus known reviewer preferences is that unknown preferences
tend to increase effort duration or intensity, while known prefer-
ences tend to increase effort direction. This observation is helpful in
reconciling the positive versus negative effects on audit quality of
accountability to reviewers that have been documented in the audit
literature.
2.1. Early history of group decision making and review process
research
The early audit JDM literature was dominated by studies of in-
dividual judgments extending the results of Ashton (1974). How-
ever, the importance of group decision making including the
review process was identi?ed in Joyce (1976), Libby and Lewis
(1977), Libby and Blash?eld (1978), and Mock and Turner (1979).
Three early empirical papers in auditing (Schultz & Reckers, 1981;
Solomon, 1982; Trotman, Yetton, & Zimmer, 1983) directly
address the bene?ts of group decision making, including diversi-
?cation and interaction effects.
A natural extension of the above studies was to move to an
examination of the effects of the review process as that was, and
still is, the predominant form of group decision making in audit
?rms. Early research examines the need for reviewers to evaluate
the reliability of preparers (Bamber, 1983) and how the review
process differs from other group structures due to its hierarchical
nature (Trotman, 1985; Trotman & Yetton, 1985).
2.2. The review process versus alternative group structures
Trotman and Yetton (1985) ?nd that the review process (oper-
ationalized by having managers review seniors' work in their
absence) signi?cantly reduces judgment variance relative to indi-
vidual judgments, but there are no differences in performance
between the review process and a two person nominal or inter-
acting group (two seniors). This suggests that the advantage of the
review process may simply be due to a diversi?cation effect. In
contrast, Trotman (1985) examines a more complex task and
operationalizes the review process by having managers review
seniors' work with the senior present, which increases the oppor-
tunity to differentiate the expertise between their own judgments
and those of the senior. Trotman ?nds the review process out-
performs nominal groups (judgments are over 50 percent more
accurate post-versus pre-review), but does not outperform inter-
acting groups of seniors; while the review process is more effective
in reducing systematic bias, the interacting group is more effective
in reducing random error. Finally, Owhoso et al. (2002) extend the
research on nominal versus interacting review teams by including
different types of errors and cases from different industries to
specify the conditions under which process gains from the review
process occur, including the unique contributions of both managers
and seniors to the team's overall effectiveness.
In summary, across the three studies there is evidence of a
diversi?cation effect, an interaction effect and a hierarchical effect,
but the results vary with a range of task and environmental con-
ditions. The following section discusses some of the gains from the
review process and the factors affecting them.
2.3. What performance gains result from the review process?
In the 1990s, many of the studies on the review process
extended the earlier documentation of bene?ts by incorporating
Fig. 1. Summary of review process experiments.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 59
cognitive factors. Libby and Trotman (1993) identify one source of
gain from the review process as the different points at which pre-
parers and reviewers enter the judgment process, which affects the
likelihood that the implications of inconsistent evidence are
considered. Preparers enter the process prior to an initial judgment
being made and therefore have incentives to justify their positions,
whereas reviewers enter the process after an initial judgment has
been made and therefore are expected to be more concerned with
determining whether the judgment is consistent with the facts
they have available. Libby and Trotman ?nd preparers' (reviewers')
relative recall is in the direction of information consistent (incon-
sistent) with their (the preparer) judgment. Ricchiute (1999) ex-
tends these results to a situation where seniors' biased memory for
consistent information may affect their documentation in the audit
workpapers such that the partner, as reviewer, evaluates a subset of
evidence that the preparer evaluates. He ?nds partners' decisions
are biased in the direction of the seniors' decisions.
Ramsay (1994) observes that in many audit ?rms, seniors
perform detailed mechanical reviews and managers perform more
general, conceptual reviews. He ?nds that managers outperform
seniors in detecting conceptual errors, but seniors outperform
managers in detecting mechanical errors; further, Harding and
Trotman (1999) ?nd that seniors outperform staff auditors in
identifying conceptual errors, while staff auditors outperform se-
niors in identifying mechanical errors. The combined results of
Ramsay and Harding and Trotman suggest a more conceptual re-
viewtemplate as auditors move up the ranks of their ?rmfromstaff
to senior to manager.
Owhoso et al. (2002) have auditors complete cases as a nominal
or reviewteam, both within and out of their industry specialization.
Within specialization, managers and seniors both incrementally
contribute to team effectiveness and nominal teams outperform
the interacting reviewteams in detecting mechanical errors but not
conceptual errors; however, out of specialization, nominal teams
outperform review teams in detecting both mechanical and con-
ceptual errors (Owhoso et al., 2002).
While the above studies suggest that the review process im-
proves audit effectiveness, there has been some research showing it
is not always effective. Tan and Tan (2008) ?nd auditors do not
suf?ciently discount the implications of evidence received that is
subsequently invalidated despite being aware of the erroneous
nature of this evidence and that reviewers are also susceptible to
this effect, unless accountability to a superior (reviewer) is
increased. Messier and Tubbs (1994) ?nd the recency effect bias,
which had been found in psychology and individual audit judgment
research, is not mitigated by the reviewprocess. Frank and Hoffman
(2015) show that reviewers do not always mitigate biases when
they cannot readily determine how much the bias in?uences a
preparer's judgment.
2.4. Alternative forms of the review process
There are many different ways the review process can be
operationalized. Traditionally, the workpapers are examined by the
reviewer independently and written review notes are developed
for the preparer to follow up. The reviewer then meets with the
preparer to discuss these review notes, or sends them to the pre-
parer (e.g., electronically) for a response (Payne, Ramsay, &Bamber,
2010). Face-to-face review is also common where the preparer and
reviewer meet in a real-time interactive review (Gibbins &
Trotman, 2002).
Comparisons have been made between reviews with and
without discussion (Ismail &Trotman, 1995), specialized versus all-
encompassing reviews (Bamber & Ramsay, 1997, 2000), and elec-
tronic versus face-to-face reviews (Agoglia, Brazel, Hat?eld, &
Jackson, 2010; Agoglia, Hat?eld, & Brazel, 2009; Brazel, Agoglia, &
Hat?eld, 2004). Ismail and Trotman (1995) examine auditor per-
formance on an analytical procedures task prior to and after review,
where the form of review is manipulated via the presence/absence
of discussion between the reviewer (a manager or a senior) and a
reviewee (a senior). They ?nd that review increases the number of
plausible hypotheses generated for both groups, but to a greater
extent for review groups with discussion. Bamber and Ramsay
(1997) note that in some ?rms, managers are instructed to focus
on conceptual errors while seniors are instructed to focus on me-
chanical errors. In other cases, both seniors and managers carry out
all-encompassing reviews without limit on the focus. They ?nd
bene?ts of hierarchical all-encompassing reviews as these reviews
are signi?cantly more accurate than specialized reviews. The type
of review affects con?dence levels, and both seniors and managers
require more time to perform specialized reviews compared to
those performing all-encompassing reviews (Bamber & Ramsay,
2000).
With technological advances, audit ?rms have increased alter-
natives for how to conduct the review process including the review
of workpapers online and then providing the preparer with the
review notes via email (Brazel et al., 2004). They ?nd that the
method of review affects preparer effectiveness and ef?ciency. In
particular, the preparers anticipating a face-to-face revieware more
concerned with audit effectiveness and produce higher quality
judgments. Gibbins and Trotman (2002) ?nd both preparers and
reviewers generally prefer face-to-face reviews and a survey
research conducted by Agoglia et al. (2010) ?nds that reviewers
perceive face-to-face to be more effective than electronic review,
but that electronic review is considered more convenient. Agoglia
et al. (2009) ?nd the quality of reviewers' going concern judg-
ments are signi?cantly lower for electronic compared to face-to-
face review. They conclude that this results from the effect of re-
view mode on the preparer (see Brazel et al., 2004) persisting
through the review process with the consequence that reviewer
judgments are ultimately affected by the formof reviewanticipated
by the preparer. Agoglia et al. (2010) ?nd that workload pressure
can increase the likelihood of choosing electronic review but only
when misstatement risk is low. When misstatement risk is high,
face-to-face review is chosen irrespective of workload pressures.
2.5. Effects of environmental factors on the review process
Several papers have addressed the effect of environmental fac-
tors on the review process including: information presentation
structures (Boritz, 1985), the planning memorandum (Bamber &
Bylinski, 1987), time pressure (Bamber & Bylinski, 1987), aggres-
sive ?nancial reporting (Phillips, 1999), and audit risk and infor-
mation importance (Sprinkle & Tubbs, 1998). Boritz (1985) is an
early example of reviewresults being highly dependent on the task
and the complexity of the setting. Bamber and Bylinski (1987) ?nd
that while an initial audit plan directs managers' subsequent re-
view, their estimated review times to complete tasks are not
affected by deadline pressure. Higher levels of audit risk increase
reviewers' accuracy of evidence recall (Sprinkle & Tubbs, 1998) and
amount of recall of (i.e., attention to) aggressive ?nancial statement
reporting (Phillips, 1999). Both papers attribute differences in
judgment to different cognitive processes adopted as a result of the
environmental factors in the review process. These different
cognitive processes contribute to both effectiveness and ef?ciency,
which we address further under future research below.
2.6. Effect of the preparer on the review process
Another feature of the audit environment that affects the
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 60
review process is the attributes of the preparer. Bamber (1983)
recognizes the importance of reviewers evaluating the reliability
of preparers as information sources. He ?nds that while reviewers'
assessments of the informativeness of audit evidence collected by
a subordinate are sensitive to the manipulation of source reli-
ability, most managers excessively discount the information's
diagnosticity for reductions in source reliability relative to the
model's predictions, contrary to earlier psychology evidence. Asare
and McDaniel (1996) ?nd that reviewers reperform more of the
work completed by unfamiliar preparers than by familiar pre-
parers, and that preparer familiarity and task complexity interact
to determine the effectiveness of reviewers in detecting conclu-
sion errors.
One of the few studies in the audit literature using actual audit
teams is Tan and Jamal (2001), who investigate whether managers'
(reviewers) assessments of the quality of seniors' (preparers) work
is in?uenced by their prior impression of the preparer and whether
this effect varies depending on whether these preparers are clas-
si?ed as outstanding or average by the audit ?rm. Their key result is
that average managers evaluate memos written by outstanding
seniors more favorably than memos written by average seniors,
only when they know the identity of the author of the memo.
Outstanding managers are not susceptible to the same effect (Tan &
Jamal, 2001).
Frank and Hoffman (2015) examine the responses of experi-
enced reviewers when they review a preparer judgment that ap-
pears to be biased by the preparer's affect (i.e., feeling of personally
liking or disliking) towards a client's controller. They ?nd that re-
viewers rely more on the preparer's judgments when they suspect
it is biased than do reviewers who are not informed about the
preparer's affect. This result is consistent with psychology ?ndings
on the ironic rebound effect which predicts that when individuals
are trying not to rely on information, they sometimes rely on it
more.
The importance of the preparer to the outcome of the review
process is a central theme in Rich et al. (1997a), who characterize
the audit review process from a persuasion perspective where
preparers may stylize evidence collected and documented in the
workpapers, depending on the reviewer and his/her preferences. As
discussed in a later section, preparers do indeed stylize their
workpapers (e.g., Agoglia, Kida, & Hanno, 2003; Shankar & Tan,
2006).
Six papers since 2000 directly address effects on reviewers of
preparer stylization. One method of stylization is via the pre-
parer's use of a justi?cation memo. Yip-Ow and Tan (2000) ?nd
the presence (versus absence) of a preparer's justi?cation memo
for a non-error cause reduces the number of alternative error
hypotheses generated by reviewers, and increases reviewers' as-
sessments of the likelihood of the non-error cause suggested by
the preparer. Agoglia et al. (2003) examine how reviewers respond
to three different justi?cation memos by preparers and ?nd that
the more the memo justi?es the preparers' preferred reporting
outcome, the more the reviewer provides a judgment supporting
that outcome. However, Tan and Yip-Ow (2001) ?nd that re-
viewers place less reliance on a preparer's consistencies when the
preparer's justi?cation memo is stylized to downplay inconsistent
evidence (and emphasize consistent evidence) compared to when
it is neutral, suggesting reviewers are sensitive to preparer styli-
zation attempts. This supports the Rich et al. (1997a) model, which
suggests that reviewers learn to cope with stylization attempts
over time.
Effects of stylization may also depend on interactions be-
tween preparer attributes and either attributes of the reviewer
(Tan & Trotman, 2003) or expectations about the client (Rich,
2004). Tan and Trotman (2003) show that reviewers' responses
to preparer stylization attempts depend on the nature of the
stylization attempt, auditor rank, and reviewer's sensitivity to
stylization. Rich (2004) ?nds that reviewers' expectations about
the client and preparer interact to affect reviewers' evaluation of
the preparer's work and the amount of follow-up work to be
carried out. Peecher et al. (2010) investigate whether reviewers
are in?uenced not only by stylization from preparers, but also by
their own guidance provided to preparers, which motivates
stylization in the ?rst place. Their results show that preparers
are in?uenced by the reviewers' preferences for less versus more
skepticism and that preparer judgments in?uence reviewers'
?nal judgments, with this occurring to an even greater extent
when reviewers provide guidance to preparers on how to make a
judgment.
2.7. Effect of the review process on the preparer
2.7.1. Reviewers with unknown preferences
Studies emerged in the early 1990s that emphasized the
importance of the review process in in?uencing audit quality, by
inducing accountability in preparers and changing how they
perform the work to be reviewed (e.g., Johnson & Kaplan, 1991;
Kennedy, 1993; Koonce, Anderson, & Marchant, 1995; Tan, 1995).
Speci?cally, being accountable to a reviewer with unknown pref-
erences is generally expected to induce vigilance e that is, effort
intensity and duration e to reduce errors or biases. Extant research
has generally manipulated such accountability by either informing
or not informing participants that their responses may be reviewed
and, in some cases, they may be asked to justify their decisions at a
later time. Accordingly, increased accountability is helpful in
reducing the biased use of audit evidence by preparers stemming
from insuf?cient effort or attention, such as a recency bias
(Kennedy, 1993) and a consistency or con?rmation bias (Tan, 1995).
Koonce et al. (1995) also show it increases the amount of justi?-
cation documented by preparers.
However, Kennedy (1995) suggests that accountability to a
reviewer with unknown preferences will not help reduce biases
related to data problems (e.g., faulty memory). Research has
shown that increased accountability does not reduce the
continued in?uence of evidence that has been processed but is
irrelevant (Kennedy, 1995) or is subsequently proven invalid (Tan
& Tan, 2008), or the dilution of the use of diagnostic evidence in
the presence of non-diagnostic evidence (Glover, 1997; Hoffman
& Patton, 1997). Koonce et al. (1995) show that increased
accountability does not increase the type of justi?cations docu-
mented by preparers, while Ricchiute (1999) shows that it does
not decrease preparers' tendency to recall and document for re-
viewers relatively more evidence consistent with their initial
decision (Libby & Trotman, 1993). Moreover, Tan and Kao (1999)
?nd that increased accountability has limited positive in?uence
on effort (or performance) in less complex tasks, or in increas-
ingly complex tasks unless preparers have increasingly high
knowledge and skill.
Two areas where greater accountability may increase both effort
duration/intensity and effort direction are with respect to consis-
tent versus inconsistent evidence and error versus non-error ex-
planations. Tan (1995) and Asare, Wright, and Trompeter (2000)
show that the threat of review by a superior with unknown pref-
erences increases the attention paid to evidence items or expla-
nations with relatively more attention paid to inconsistent
evidence (Tan, 1995) or error explanations (Asare et al., 2000).
Hoffman and Patton (1997) and Johnson and Kaplan (1991) show
such accountability also causes auditors to exhibit more con-
servativism, consensus, or self-insight. DeZoort, Harrison, and
Taylor (2006) further show auditors exhibit increasing levels of
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 61
conservatism and consistency as strength of accountability pres-
sure increases.
2.7.2. Reviewers with known preferences
Peecher (1996) illustrates that known preferences of reviewers
in the audit review process in?uence evidence sought and
weighted by preparers. That is, preparers not only increase effort
or attention in collecting evidence and evaluating outcomes, but
they direct their effort or attention more toward evidence they
believe is consistent with reviewer preferences. Therefore, being
accountable to reviewers with known preferences may reduce or
increase errors and biases, and in turn audit quality, depending on
the nature of the preferences (Rich et al., 1997a). Both Peecher
(1996) and Turner (2001) provide reviewer preferences for less
skepticism or more skepticism and ?nd that preparers generate
error hypotheses or search for evidence consistent with reviewer
preferences, respectively. Peecher ?nds the former effect is miti-
gated when client integrity is low, while Turner ?nds the latter
effect is more pronounced when auditors make explicit decisions
versus when they provide their beliefs for the reviewer to take
action. Wilks (2002) ?nds preparers make judgments consistent
with reviewer preferences (for pessimism versus optimism), and
this effect is even more pronounced when preferences are
communicated before (versus after) preparers evaluate evidence,
because these preparers also subconsciously evaluate the evi-
dence consistent with reviewer preferences. Peecher et al. (2010)
?nd preparers evaluate evidence consistent with reviewer pref-
erences for less skepticism, when those preferences are commu-
nicated to them (versus not communicated at all), and as noted
earlier these preparers' inputs are incorporated into reviewers'
?nal judgments. Overall, these ?ndings suggest preparers are
more (less) skeptical when a reviewer prefers an unfavorable
(favorable) client view.
Peecher et al. (2013) suggest a different preference and
examine whether preparers who are accountable for their
judgments, judgment process, or some combination thereof,
likely in?uence innovation and quality of the audit process and
procedures. Consistent with this notion, Messier, Quick, and
Vandervelde (2014) ?nd that preparers are less likely to simply
follow what was done in the prior year when they are required
(versus not) to justify their decision to a partner who prefers to
focus on the process behind their decision. Lambert and Agoglia
(2011) examine how preparers close out review notes by having
the preparers interact with a pre-programmed client, as well as
receive and respond to reviewer comments. They ?nd that pre-
parers provide greater effort and ?nd more seeded errors when
the review notes are framed to encourage appropriate conclu-
sions versus defensible documentation, but only when the re-
view is timely. Gold, Gronewold, and Salterio (2014) examine
how preparers respond to an audit ?rm that has a more versus
less open climate to learning from errors, when errors are
discovered after the normal review process is complete. They
?nd a more open climate increases reporting of self-made me-
chanical (but not conceptual) errors and both types of peer-made
errors.
Related research motivated by Rich et al. (1997a) on styliza-
tion/persuasion attempts examines the nature of justi?cations
documented by preparers within the audit review process.
Shankar and Tan (2006) ?nd that preparers try to bolster their
position, as needed, when they are required to determine their
own preference prior to receiving the reviewer preference. Spe-
ci?cally, upon learning of their reviewers' preference after form-
ing their own preference, preparers document more (less)
evidence favoring their preference and a greater (lesser) spectrum
of issues when their reviewer's preference is dissimilar (similar),
but only for highly knowledgeable preparers (Shankar & Tan,
2006).
5
Agoglia et al. (2003) examine how preparers stylize
when reviewer preferences are unknown but a preference for
how to structure their written justi?cation is expressed. They ?nd
that preparers, asked to assess a client's overall control environ-
ment, have more positive judgments, and document more total
evidence and judgment-consistent evidence, when they must
summarize balanced (i.e., supporting and contradictory) evidence
and assess components of the control environment than when
they do not assess components but either summarize balanced
evidence, or summarize primarily supporting evidence. Thus, in a
typical setting where there is more supporting than contradictory
evidence, the more detail required, the more supporting (and
total) evidence will be documented (Agoglia et al., 2003).
2.7.3. Effect of alternative formats of the review process on the
preparer
Despite suggestions from Tan and Kao (1999) that variations in
the review process format likely in?uence preparers, few papers
have examined this issue. Variation in the type of review (or
accountability) likely in?uences not only effort duration and in-
tensity but also effort direction and the goals preparers focus on
(e.g., ef?ciency versus effectiveness). To date, only three studies
examine review format. Brazel et al. (2004) and Agoglia et al.
(2009) compare preparer judgments when they anticipate a face-
to-face interacting review versus when they anticipate an elec-
tronic review. They ?nd that preparers who anticipate the more
interactive review spend more time preparing documentation,
anchor less on prior year judgments, are more concerned with
effectiveness, and make more effective judgments (Brazel et al.,
2004), and also have higher quality documentation (Agoglia et al.,
2009). Payne et al. (2010) compare face-to-face interacting review
with written review and ?nd that the former leads to a focus on
more cognitively effortful audit procedures resulting in better
identi?cation of fraud red ?ags. This combination of results in-
dicates that review format in?uences each of effort duration, in-
tensity, and direction.
2.8. Future research on the review process
The need for future research on the review process is still
important given the criticisms by regulators (ASIC, 2014; PCAOB,
2013) and the resources devoted by audit ?rms to the review
process. At present, the audit literature on the review process ex-
amines situations where the review process has been effective and
this effectiveness appears to depend on the nature of the task and
errors, how the review team is constituted, and a range of envi-
ronmental factors. Given the scarcity of participant resources,
dif?cult choices have to be made on what factors and what com-
bination of factors should be included in future studies of the re-
view process.
Some relevant theoretical insights are provided by Einhorn,
Hogarth, and Klempner (1977) who consider various baseline
models for examining the quality of group performance, de?ned as
how close the group judgment is to the actual value of the item
being predicted. They compare four types of model: random se-
lection of a single individual in the group; equal weighted mean of
group judgments; using the judgments of the best member
(identi?ed in advance); and a weighted average of the members
based on expected expertise. They ?nd that the performance of
5
In addition, high tacit knowledge preparers are more likely to frame the evi-
dence consistent with their preference but only when their reviewer's preference is
similar (Shankar & Tan, 2006).
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 62
these models depends on group size and ‘bias’. Experimentally the
cost of varying group size in auditing research is not likely to be
practical, thus emphasis could be placed on ‘bias’ which Einhorn
et al. divide into systematic bias and random error. Systematic bias
in individual judgments is potentially affected by both the indi-
vidual and the task being performed, and such a consideration
would be valuable in designing studies on the review process.
We also suggest that progress can be made by describing the
underlying cognitive processes in enough detail to allow re-
searchers to predict the speci?c bene?ts of the review process and
under what circumstances such bene?ts will be increased or
decreased (Libby & Trotman, 1993) and the interaction between
cognitive factors and environmental factors (Libby & Luft, 1993).
Subsequently, Gibbins and Trotman (2002) use retrospective recall
from managers of their actual review engagements to develop a
model of ?ve components of the audit ?le review process that will
affect reviewer behavior: manager's expectations about the client,
the preparer, the partner, the manager's own approach to review,
and the circumstances faced by the reviewer (e.g., deadline pres-
sure). Factors included in this model could be incorporated in
future research to progress the research not only on when potential
bene?ts are realized but also potential limitations of the review
process (e.g., Frank & Hoffman, 2015).
Research on face-to-face interaction between auditors (e.g.,
Schultz & Reckers, 1981; Solomon, 1982), particularly between
reviewer and reviewee (e.g., Ismail & Trotman, 1995; Trotman,
1985), has become rare (except in the brainstorming literature
discussed subsequently). This interaction is important as it permits
examining one of the key factors in?uencing group performance.
Such interaction can be a rich source of information about process
gains but it does raise additional complications in the design of the
study fromboth a control perspective and in obtaining participants.
In addition to the hierarchical review process discussed above,
auditing standards (e.g., AS 7; ISA 220) require engagement quality
control reviews (EQR) whereby a second partner carries out a re-
view of the audit (previously called concurring partner reviews)
(IAASB, 2008; PCAOB, 2009). The conduct of EQR has been criticized
by inspectors, including suggestions for greater involvement of the
second partner (PCAOB, 2013). There are important questions to
address on the nature, extent, and timing of the EQR in the audit
process. Future research on EQR would bene?t from a better un-
derstanding of how this process should be structured and how the
type of review interacts with various environmental factors. Epps
and Messier (2007) provide insights on audit ?rm practices in
this area and Messier, Kozloski, and Kotchetova-Kozloski (2010)
provide details of SEC enforcement actions including alleged vio-
lations of due care related to lack of professional skepticism. This
material provides the basis for rich research instruments to inves-
tigate the judgments made by the EQR.
Opportunities for signi?cant incremental contribution on the
effects on the preparer of the review process will likely come from
examining how the type of accountability in?uences auditor
judgments, such as recent research on reviewer preferences for a
focus on process, outcome, documentation (e.g., Lambert &Agoglia,
2011; Messier et al., 2014; Peecher et al., 2013), or varying error
climates (Gold et al., 2014). For example, while audit ?rm efforts to
establish an appropriate tone at the top in response to regulator
criticism about the review process may succeed in in?uencing its
professionals, this success could depend on the message and howit
is conveyed. In addition, research can consider looking beyond
preferences for more versus less skepticism, but rather at the type
of skepticism including being outwardly skeptical of evidence
versus inwardly skeptical of one's own judgments (Bell, Peecher, &
Solomon, 2005), and whether the bene?ts of inward skepticism
over and above outward skepticism depend on the judgments
being made (Grenier, 2014; Harding & Trotman, 2014).
A major thrust of Rich et al. (1997a) was to motivate research
on how preparers may engage in strategic, stylized behavior
depending upon different environmental factors or formats in the
review process. However, research on review format has been very
limited and opportunities exist in the intersection of stylization
and review format. For example, increased use of electronic
workpapers, email communication, and electronic review may
increase/reduce preparers' stylization. Since Rich et al. there have
been signi?cant changes in the production of audit working pa-
pers such as greater automation and more structured computer
assisted selection of audit tests which may reduce opportunities
for stylization of documentation and work performed, respec-
tively. Still, the use of written memos to document and defend
auditors' positions remains a critical part of the audit and styli-
zation therein is likely to persist. Future research can examine
how the continued practice of written justi?cations and the evo-
lution of electronic workpapers and communication in?uence the
opportunities for, and results of, preparer stylization efforts and
the review process.
While the above discussion relates to preparers stylizing the
workpapers for review, it is suggested that the likelihood of an
inspection will affect the review carried out by a partner. However,
such inspections have been suggested to lead to defensive auditing
(Peecher et al., 2013) and likely reduced levels of professional
skepticism due to shifting focus to documentation and checklists
(Westermann, Cohen, & Trompeter, 2014). Research examining the
effect of different forms of the review process and expectations
around the inspection process on the professional skepticism of
both preparers and reviewers (particularly at the partner level) is
warranted.
3. Brainstorming
A newline of group research in auditing evolved with Carpenter
(2007) to examine fraud brainstorming sessions conducted as part
of the audit planning process. The issue became important to the
audit literature with the revision of auditing standards related to
the auditor's responsibility to consider fraud in an audit of ?nancial
statements, which require a ‘discussion’ (ISA 240) or ‘brainstorming
session’ (AU 316) among audit team members including how and
when a client's ?nancial statements may be susceptible to material
misstatement due to fraud. The bene?ts of brainstorming have
been proclaimed since Osborn (1957) and there exists decades of
social psychology and organizational behavior research comparing
alternative brainstorming formats and the reasons for process gains
and losses during brainstorming sessions (see review by Stroebe,
Nijstad, & Rietzschel, 2010).
Hammersley (2011) sets out a model of the fraud planning tasks,
which includes four stages of the brainstorming process: fraud risk
factor identi?cation, fraud hypothesis generation, fraud risk
assessment, and audit program modi?cation.
6
Hammersley de-
scribes a wide range of relevant audit research related to each of the
factors in her model. We restrict our discussion to audit research
where participants are involved in the brainstorming task. We
divide this research into face-to-face brainstorming and electronic
brainstorming. The classi?cation of these relevant studies by task
type is provided in Fig. 2.
6
Fraud risk factor identi?cation includes incentives, opportunities and the ability
to rationalize fraud. Fraud hypothesis generation refers to the process of identifying
plausible explanations of how fraud can occur given the risk factors that exist,
which is followed by preliminary risk assessments of the likelihood of fraud
existing (Hammersley, 2011). Finally, audit program modi?cations refer to changes
made to audit programs based on the above factors.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 63
3.1. Face-to-face interacting versus nominal brainstorming
While auditing standards do not specify a formof brainstorming
group, the most commonly used method in practice is a face-to-
face interacting team (Bellovary & Johnstone, 2007; Brazel,
Carpenter, & Jenkins, 2010). Thus, one important question to
address is how the interacting brainstorming team compares to
other possible alternatives including a baseline measure such as a
nominal group. With interacting brainstorming groups there is
discussion between participants whereas for nominal brain-
storming groups, participants complete the task as individuals
without interaction and their unique ideas are combined by the
researcher after completion of the task.
The early group brainstorming literature in psychology advo-
cates interacting group brainstorming as a means of increasing
creativity in idea generation (Osborn, 1957). The bene?ts of group
brainstorming are based on obtaining a large pool of ideas and,
through cognitive stimulation, generating new ideas that combine
and/or change previously generated ideas. However, numerous
studies in the psychology literature ?nd that nominal groups
outperform interacting groups in both the quantity and quality of
ideas generated because of a range of process losses (see Kerr &
Tindale, 2004, for a review). The key process losses identi?ed are
production blocking, evaluation apprehension and social loa?ng.
7
However, the bulk of the brainstorming studies in psychology
have been conducted using undergraduate college students in ad-
hoc groups on hypothetical problems, and have been heavily crit-
icized for lacking realism (Isaksen, 1998).
8
In comparison, auditing standards require brainstorming to be
carried out by auditors with adequate experience and training, the
groups are generally hierarchical, and identifying fraud risks and
generating potential fraud hypotheses are familiar, but complex,
tasks because while they are conducted on every audit they
require participants to draw on their accumulated but limited
knowledge of previous fraud. Despite these important differences
in the audit context, the only fraud brainstorming studies that
have compared face-to-face interacting brainstorming with nom-
inal brainstorming (Carpenter, 2007; Lynch, Murthy, & Engle,
2009) ?nd results that are generally consistent with the
psychology literature. Carpenter (2007) ?nds that nominal groups
generate more unique fraud hypotheses than similar size inter-
acting groups, but that the interacting groups generate more ‘high
quality’ frauds (i.e., those consistent with frauds seeded in the
case). The nominal and interacting groups are manipulated
within-subjects and Carpenter ?nds that while not all identi?ed
frauds by nominal groups are included in the interacting groups'
fraud hypotheses, the group brainstorming generates additional
fraud ideas. This is an example of cognitive stimulation. For a
simpler task involving fraud risk identi?cation, Lynch et al. (2009)
also ?nd that nominal groups generate more ideas than face-to-
face interacting groups.
3.2. Interacting face-to-face brainstorming compared to alternative
brainstorming formats
Here we consider research that examines four alternatives to
the traditional face-to-face brainstorming groups used in practice:
brainstorming guidelines, premortem instructions, strategic
reasoning, and use of prompts. Trotman, Simnett, and Khalifa
(2009) compare the performance of three interacting group
treatments provided with either standard interacting group in-
structions (similar to Carpenter, 2007), brainstorming guidelines
or pre-mortem instructions. They develop their brainstorming
guidelines in line with Osborn's (1957) four guidelines that aim to
defer judgment until after the idea-generation phase.
9
They
develop their pre-mortem instructions based on Klein (1999) and
simulate backward thinking (i.e., putting an individual into the
future and asking them to think backwards) by instructing audi-
tors to imagine it is months after the audit is completed and
?nancial reporting fraud has been discovered for the client, before
asking them to generate the potential frauds that could have
occurred. Trotman et al. ?nd the brainstorming groups provided
with guidelines or pre-mortem instructions identify a greater
quantity of potential frauds, and more frauds matching the list of
frauds identi?ed as most important by a panel of experts, than the
interacting group receiving standard interacting group
instructions.
Hoffman and Zimbelman (2009) suggest that encouraging
strategic reasoning is one way of generating more creative fraud
detection procedures. They describe strategic reasoning in auditing
Fig. 2. Summary of previous fraud brainstorming experiments addressing nominal or interacting teams.
7
Production blocking occurs when only one group member can communicate at
once, which can result in forgetting ideas, suppressing ideas, and remembering and
listening, rather than creating ideas. Evaluation apprehension refers to individuals
withholding ideas due to fear of negative evaluation of their ideas. Social loa?ng or
free riding refers to individuals relying on others to complete the task for such
reasons as perception that their inputs are unidenti?able or dispensable.
8
Isaksen (1998, p. 21) states that “most studies used untrained students working
in arti?cial groups on relatively meaningless tasks with little or no concern for
measuring anything but sheer quantity.”
9
Osborn's (1957, pp. 300e301) four guidelines to defer judgment are: criticism
and adverse judgment of ideas must be withheld until later; wilder and unusual
ideas are desired; quantity is desired as a greater number of ideas should increase
the likelihood of useful ideas; and combination and improvement are sought to
facilitate participants turning the ideas of others into better ideas or combining two
or more ideas into another idea.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 64
as answering the following three questions: What potential frauds
may have occurred? How could management conceal the potential
frauds from the standard audit plan? How could the plan be
modi?ed to detect the concealed frauds? (Hoffman & Zimbelman,
2009). They ?nd that both interacting brainstorming groups
(without guidelines) and individuals engaged in strategic reasoning
provide more effective modi?cations to standard audit processes
than individuals without interventions, but that the combination of
the two interventions does not outperform either of the two in-
terventions used alone.
Finally, two studies examine different forms of prompting
during brainstorming. Lynch et al. (2009) provide content facili-
tation by way of ?ve fraud prompts related to incentives/pressure,
opportunities, attitudes/rationalizations, revenue recognition, and
management override. For both face-to-face and electronic
brainstorming groups, the automatic content facilitation leads to a
larger number of relevant fraud risks being identi?ed. Unlike the
studies discussed above, participants in Hammersley et al. (2010)
do not complete a brainstorming session, but instead list three
risks they would raise during a brainstorming session and watch a
video of the fraud brainstorming session. They ?nd that, when not
primed about the fraud risks identi?ed, auditors with more spe-
ci?c documentation provide higher fraud risk assessments and
request more evidence. However, importantly they show that
decision aids such as priming are not always bene?cial. Partici-
pants who are primed also increase fraud risk assessments and
evidence requests when they receive summary information, but
decrease such judgments when they receive speci?c documenta-
tion because priming makes the client speci?c risks less typical
(Hammersley et al., 2010).
In sum, while unstructured face-to-face brainstorming is the
most common method used by audit ?rms (Brazel et al., 2010),
this method is outperformed with respect to the quantity of fraud
risks identi?ed and fraud hypotheses generated by any of the
structured methods examined above (Lynch et al., 2009; Trotman
et al., 2009). With respect to modifying audit programs, unstruc-
tured face-to-face brainstorming is equaled by prompting in-
dividuals to use strategic reasoning (Hoffman & Zimbelman,
2009). There is mixed evidence with respect to the effects on
the quality of the fraud hypotheses generated (Carpenter, 2007;
Trotman et al., 2009) which is likely partly due to different mea-
sures of ‘quality’ and differences in research design including
whether the different methods are manipulated within- or
between-subjects. Overall, it is suggested that there is consider-
able potential for improvement in brainstorming by adding alter-
native facilitation methods to the unstructured face-to-face
interaction used in practice.
3.3. Electronic brainstorming
Electronic brainstorming has been a popular brainstorming
technique in a range of business ?elds (Pinsonneault, Barki,
Gallupe, & Hoppen, 1999). Previous studies in psychology and
information systems ?nd that, for idea generation tasks, electronic
brainstorming outperforms face-to-face brainstorming (Dennis &
Valacich, 1993; Guzzo & Dickson, 1996; see Kerr & Tindale,
2004, for a review). This has led to suggestions that the brain-
storming literature should move on from comparing face-to-face
brainstorming to electronic brainstorming, and focus on what
forms of electronic brainstorming are most appropriate (Guzzo &
Dickson, 1996). The difference between face-to-face brain-
storming and electronic brainstorming is that the latter uses
computer software to allow individual group members to indi-
vidually input ideas without interruption, which reduces process
losses because of lower production blocking and evaluation
apprehension (particularly when the ideas are provided anony-
mously) (Kerr & Tindale, 2004).
10
This superiority of electronic brainstorming has been supported
by Lynch et al. (2009), the only audit study to compare face-to-face
and electronic brainstorming; they ?nd that both forms of elec-
tronic brainstorming (nominal and interacting) outperform face-
to-face interacting brainstorming.
11
However, the comparison of
electronic interacting and electronic nominal brainstorming is less
clear. Using undergraduate students as participants, Lynch et al.
(2009) ?nd no difference in fraud risk identi?cation between
electronic nominal and interacting brainstorming groups, but note
that their results could be driven by low power from small sample
sizes. Also, use of undergraduate students is likely to result in a low
range of differential expertise in their teams, and would bias
against ?nding signi?cant results.
As electronic brainstorming minimizes production blocking and
evaluation apprehension (Kerr & Tindale, 2004; Pinsonneault et al.,
1999), it allowed Chen, Trotman, and Zhou (2015) to focus on a third
potential process loss (social loa?ng) to better address the under-
lying mechanisms of brainstorming effectiveness and examine
when and why interacting groups do (or do not) outperform
nominal groups in electronic brainstorming. Chen et al. ?nd nom-
inal teams generate a signi?cantly larger number of fraud risk
factors and fraud hypotheses, and generate higher quality fraud
hypotheses (i.e., hypotheses consistent with those identi?ed by
fraud experts), than interacting teams. They also provide evidence
that social loa?ng of seniors drives the differences between nom-
inal and interacting teams in the fraud hypothesis generation task.
3.4. Future research on brainstorming
A key focus of future brainstorming research should be potential
process gains and losses that exist with alternative methods of
fraud brainstorming, including methods presently used in practice
and those that could be adopted. The bulk of the ?ndings in the
audit literature comparing interacting brainstorming to nominal
group brainstorming show that the latter performs better, and no
?ndings demonstrate that interacting brainstorming, as generally
used in practice, outperforms any of the alternatives examined in
the audit literature based on quantity of frauds generated (Chen
et al., 2015; Hammersley, 2011). Thus an important focus of
future research is how to reduce process losses and/or increase
cognitive stimulation in interacting groups.
At this stage we know very little about the nature of process
losses in the audit environment, under what circumstances they are
likely to exist or how they can be reduced. Opportunities arise to
consider how the various process losses differ between audit and
psychology studies, and how they are affected by the audit task
completed and the form of brainstorming involved. Knowledge of
these process losses can be used to develop improvements in group
brainstorming (Kerr & Tindale, 2004). Facilitator training is one
method that has been used to minimize production blocking and
evaluation apprehension (Oxley, Dzindolet, & Paulus, 1996; Paulus,
Nakui, Paulus, &Brown, 2006). These methods have potential given
the role of audit partners and managers in the facilitation of fraud
10
The degree of anonymity can depend on the research setting. For example,
when data is collected at the of?ces of an audit ?rm during staff training or at
another meeting, participants will know who else is in the room including some
details of their background. If the data is collected in small groups the level of
anonymity decreases.
11
In Lynch et al. (2009), the face-to-face group interacts verbally with fraud risk
factors being typed on the laptop by one of the team members. The electronic
interacting group members could input ideas simultaneously and they could see
the ideas of other group members in real time.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 65
brainstorming sessions (Dennis & Johnstone, 2014). For example,
additional guidelines used by the facilitator have included staying
focused on the task, not criticizing, and returning to previous cat-
egories (Oxley et al., 1996; Paulus et al., 2006). However, it is
important to consider the multi-period effects of these structured
approaches. Trotman et al. (2009) ?nd that brainstorming groups,
given Osborn's (1957) brainstorming guidelines, generate a greater
quantity of potential frauds and higher quality potential frauds.
However, whether similar effects would exist across multi-periods
is unknown, as the effects of the brainstorming guidelines may
decrease over time, as auditors and audit teams become more
familiar with the guidelines.
In addition, it is necessary to better understand the circum-
stances where brainstorming can bene?t from cognitive stimula-
tion. In particular, what are the unique attributes of fraud
brainstorming in auditing that are likely to allow cognitive stimu-
lation and under what circumstances does it occur? As noted
earlier, Carpenter (2007) ?nds process gains via cognitive stimu-
lation when group members completed the task as individuals
prior to group interaction. This also suggests there are potential
bene?ts from auditors in practice performing the task as in-
dividuals as an input to the subsequent brainstorming session
(Brazel et al., 2010; KPMG, 2011).
Another possibility for enhancing cognitive stimulation, and
reducing process losses, is open, round robin brainstorming (see
Beasley & Jenkins, 2003; Kerr & Tindale, 2004) which involves
individual brainstorming and then group members take turns
reading their ideas to the rest of the group one at a time. They
then have an opportunity to add additional ideas which may
have been generated while listening to others. This method is
based on nominal group techniques (NGT) (van de Ven &
Delbecq, 1971) developed in the management literature, which
can include additional steps such as discussing each idea for
clari?cation and then ranking ideas. The potential bene?t of
these methods is that they may reduce process losses by
lowering production blocking (via nominal group processes) and
social loa?ng (as all participants present their ideas). Listening to
others is also likely to assist with cognitive stimulation, partic-
ularly when auditors know they have the opportunity to present
additional ideas. Other methods that add structure to the task
should be investigated. For example, in the management ac-
counting literature, Chen, Williamson, and Zhou (2012), apply a
‘parent-child’ idea structure to brainstorming, where one group
member suggests an initial idea (‘parent-idea’) and then other
group members build on that idea with subsequent ideas (‘child-
ideas’).
While there is considerable support in the psychology literature
for electronic brainstorming, it has only been compared to face-to-
face brainstorming in the accounting literature by Lynch et al.
(2009) using student surrogates in a fraud risk identi?cation task.
This leaves many unanswered questions including under what
circumstances electronic brainstorming can outperform face-to-
face brainstorming and what effects do alternative electronic
brainstorming structures have on cognitive stimulation and process
losses. For example, with electronic brainstorming, the researcher
has access to the order of ideas that are typed in and the timing of if/
when these ideas are read by other participants. Such information
provides a much richer source to examine the level of cognitive
stimulation.
The importance of task differences has long been recognized in
the group literatures in both psychology (Guzzo & Dickson, 1996;
Herold, 1979) and accounting (Ashton, 1990; Bonner, 1994;
Trotman, 2005). In fact, Herold's (1979) observations that the se-
lection of tasks in group performance experiments ‘has been any-
thing but systematic’ and ‘may be a major reason for the
inconsistency’ in the results of group performance research also
apply to the audit JDM research on groups. This is not surprising
given the dif?culty of obtaining participants (decreases the number
of potential treatments) and the limited time available for each
experiment (decreases the opportunities for within subject ma-
nipulations). We suggest that without adequate consideration of
task differences between studies, there is likely to continue to be
unexplained differences in the results of brainstorming studies and
more generally group judgments in auditing.
4. Consultation within ?rms
Consultations involve interactions, or anticipated interactions,
among audit group members whereby one auditor seeks advice or
feedback from another auditor(s) on an appropriate course of ac-
tion. These consultations can be a source of justi?cation for audi-
tors' actions (Emby & Gibbins, 1988) and may help justify
undertaking a contemplated course of action or document the
decision process to help defend a position within the audit team or
to inspectors. Consultations can range from binding (i.e., formal) to
advisory (i.e., informal) and accounting ?rms have emphasized the
importance of both formal and informal consultation (Ranzilla,
Chevalier, Herrmann, Glover, & Prawitt, 2011). In Fig. 3 we divide
this consulting research into the perceived justi?ability of consul-
ting advice, and the willingness to followversus willingness to seek
consulting advice.
4.1. Perceived justi?ability of consulting advice
Kennedy, Kleinmuntz, and Peecher (1997) ?nd auditors perceive
another auditor's decision to be more justi?able when s/he infor-
mally consults with a knowledgeable auditor (versus not), and most
justi?able when the advice con?rms the consulting auditor's initial
position. Moreover, Kennedy et al. show that consulting, yet not
following discon?rming advice, is perceived as less justi?able than
following discon?rming advice, but more justi?able than not
consulting at all.
4.2. Willingness to follow consulting advice
Kadous, Leiby, and Peecher (2013) also consider informal con-
sultations, by examining auditors' willingness to follow contrary
advice from a peer. Consistent with previous literature, they ?nd
that auditors tend to revise in line with advice regardless of
treatment; however, non-specialist and specialist auditors react
differently. Speci?cally, they ?nd non-specialists use a ‘trust heu-
ristic’ and view better justi?ed advice as higher (similar) quality
and (do not) weight it more heavily than worse justi?ed advice
when the informal advice comes from a peer with whom the
auditor shares a weaker (stronger) social bond. Specialists, how-
ever, do distinguish between the quality of better versus worse
justi?ed advice when the advisor social bond is stronger, but
quality fails to in?uence weighting; that is, specialists discount
better advice from close advisors (Kadous et al., 2013).
Salterio (1996) and Salterio and Koonce (1997) examine formal
consultation with audit ?rms' research or technical units. These
units are often consulted about accounting issues and provide ev-
idence of precedents in practice, but may not provide a formal
recommendation. Salterio ?nds that auditors tend to follow the
direction suggested in consultation precedents, and this occurs to a
greater extent when precedents are more similar to the issue at
hand, but importantly, auditors are no less likely to followwhen the
precedents discon?rm the client preference. However, Salterio and
Koonce ?nd that while auditors exhibit such behavior when all
precedents provided by a consultation unit point in the same
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 66
direction, when precedents con?ict, auditors tend to follow an
aggressive client preference if it is made known.
Ng and Shankar (2010) demonstrate that before or after formal
consultation, auditors are more likely to accept the client prefer-
ence when the client's justi?cation is stronger. However, after (but
not before) consulting, this effect is diminished by the presence of a
quality assessment standard (QAS) designed to encourage auditors
to assess appropriateness of the client preference prior to assessing
its acceptability, but only if the consultation unit also explicitly
advises that the client should adopt an alternative method. Similar
to the ?ndings of Salterio and Koonce (1997), even if consultation
informs auditors that the client preference is acceptable but less
appropriate than an alternative method, auditors tend to permit
the client preference if they do not receive a de?nitive answer as to
which alternative to follow.
4.3. Willingness to seek consulting advice
Research on auditors' willingness to seek formal consulting
advice has been examined froma fraud perspective, where ?ndings
indicate that auditors are more willing to consult with fraud ex-
perts when fraud risk is higher (Asare & Wright, 2004; Gold et al.,
2012). However, Gold et al. (2012) ?nd this effect when the
requirement to consult is strict (i.e., mandatory and binding) but
not when the requirement is lenient (i.e., advisory and non-
binding), presumably because failing to consult despite high
fraud risk is less justi?able when requirements are strict. This is
consistent with the ?ndings in Kennedy et al. (1997) that consul-
tation increases perceptions of justi?ability. Gold et al. also ?nd that
an auditor's willingness to consult increases when deadline pres-
sure is higher, and to a greater extent under stricter consultation
requirements.
Some current working papers illustrate evolving trends in this
research, as they focus on more informal settings and judgments
and choices about the extent of consultation. For example, Nelson,
Proell, and Randel (2014) show that auditors are more willing to
consult with audit team superiors about audit issues, in particular
issues that increase effectiveness but also audit costs, when these
superiors have a more team-oriented leadership style. Schaefer
(2014) shows that social costs (e.g., reputation concerns) can
decrease auditors' willingness to informally consult, induce themto
consult with peers rather than superiors, or provide a less balanced
set of facts to an advisor. Further, Bauer, Hillison, Peecher, and
Pomeroy (2015) demonstrate that, without a prompt to consider
fraud, auditors make better changes to an audit plan for responding
to a high fraud risk when informally advising a peer than when
deciding for oneself, due to different focal concerns in their
respective mental perspectives.
4.4. Future research on consultation
Since the emergence of audit research on consultations (and
advice), this area has received limited empirical attention and of-
fers many opportunities for future research. Two questions of
interest are what stopping rules auditors use after consulting; and
are they more likely to stop consulting and reach a decision after
receiving con?rming or discon?rming advice? These are potentially
important questions given that con?rming advice is seen as more
justi?able (Kennedy et al., 1997) and auditors have been shown to
seek con?rming evidence (e.g., Kaplan & Reckers, 1989; Turner,
2001). Further, prior research utilizes experimental designs that
often force auditors to make judgments after consultation evidence
is provided, but whether or not auditors would choose to conclude
at that point is unknown. In addition, audit research has only begun
to explore when auditors seek consultation and from whom they
seek it (Gold et al., 2012; Schaefer, 2014), with many task, person,
and environmental variables of potential interest to these
questions.
Within the limited extant consultation research, the advisor role
has received even less attention and it is unclear what factors affect
the quality of the advice given. Further, given the frequency of
informal consultation, both the advisor role and the formality of
consultation offer much potential for future research, particularly
in light of the dynamic and endogenous nature of the consultation
process. For example, advisors often depend on the information or
evidence provided by the advice seeker and likely receive or have
access to more evidence when the process is formal. How the
extent and nature of evidence available in?uences advisors' will-
ingness to indicate a de?nitive conclusion, their likelihood of
requesting or seeking more evidence, and their ultimate decision
can be explored in future research, as well as whether such de-
cisions differ for formal versus informal consultation. In addition,
those consulted may form views about the advisee, based on pre-
vious interactions, including whether they provide a balanced view
of client facts and how they act on the advice given. These issues
can be addressed by manipulating the nature of prior interactions
or by using multi-period experiments.
While formal advice is likely to be handled through written
communication (e.g., emails or completion of standard forms),
informal advice could be face-to-face, via telephone or via email. As
discussed in our earlier sections on the review process and brain-
storming, the mode of communication is likely to affect the content
of the consulting advice given and interact with other environ-
mental factors, and research in this area is warranted.
5. New directions and future research
In our above coverage we have included future research on the
key areas we have addressed: the review process, brainstorming,
and consultation within ?rms. Here we suggest potential changes
to the research agenda by looking at other group related topics that
we believe will progress research in auditing. We consider three
areas of within-?rm group interaction (audit teams in context;
shared mental models; and audit team diversity, including sus-
tainability assurance). We also recognize that auditors frequently
interact outside the audit ?rm, and consider audit committee in-
teractions with auditors as an important example of opportunities
in this area. Research in some areas has already begun and other
Fig. 3. Summary of consultation experiments.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 67
areas are yet to be investigated.
5.1. Audit teams in context
Reviews of the group psychology literature by Guzzo and
Dickson (1996) and Ilgen, Hollenbeck, Johnson, and Jundt (2005)
focus on team performance in organizational contexts. Guzzo and
Dickson emphasize teameffectiveness, which they de?ne as group-
produced outputs (e.g., quality, quantity, speed, etc.), the conse-
quences a group has for its members (e.g., job satisfaction), and the
enhancement of a team's effectiveness in the future (e.g., future
performance). They also consider studies involving interventions
aimed at improving team performance, including whether these
interventions are targeted at individual team members (e.g., the
skills of team members) or at teams as performing units (e.g., team
development). The examination of teams within their organiza-
tional context has generated some signi?cant ?ndings in psychol-
ogy research (see Guzzo & Dickson, 1996; Klein, 1999). Yet, while
audit researchers are interested in audit teams, our experimental
designs generally do not include auditors who are members of the
same team and the experiments rarely consider the effect on the
team or the individuals in the future. However, in a tax setting,
Andiola and Bedard (2015) examine the effects of the review pro-
cess on job satisfaction and turnover intentions.
One context, examined in Guzzo and Dickson (1996), that con-
tains multiple team features with interesting similarities to audit-
ing is the ‘crew resource management’ (CRM) literature which
investigates ?ight crew behavior.
12
One element of CRM research
that is pertinent for audit teams is the use of simulations. This
aspect of CRM research has expanded beyond aviation to other
team-based contexts, including management, education, crisis
management, and medicine. Simulations enable team training for
events that are high-risk, but low frequency, within virtual envi-
ronments that are essentially risk-free (Haerkens, Jenkins, & van
der Hoeven, 2012). These simulated environments allow teams to
experiment and make mistakes, so as to learn valuable lessons
(Wood, Beckmann, & Birney, 2009). In an audit team setting, sim-
ulations can be used to examine the effects of potential frauds on
the ?nancial statements where the resulting insights may lead to
improvements in fraud hypothesis generation and testing.
5.2. Shared mental models in audit teams
Mental models comprise structured knowledge, data on pat-
terns of cues, and rules that causally and/or associatively link this
knowledge with these cues (Bell, Peecher, & Solomon, 2002, p. 14).
One way in which mental models are used is to run mental simu-
lations, whereby a set of initial conditions are provided and the
future trajectory of a systemis then imagined (Markman &Gentner,
2001). Klein (1999) suggests three criteria for evaluating a mental
simulation: coherence (whether it makes sense); applicability
(whether it provides what the decision maker needs); and
completeness (whether it includes too much or too little informa-
tion). In the audit JDM literature, a small body of research has
examined individual mental models developed by auditors (e.g.,
Brewster, 2011; Choo & Trotman, 1991; Hammersley, 2006;
Knechel, Salterio, & Kotchetova-Kozloski, 2010), as well as their
use within audit teams for running mental simulations (Chen et al.,
2015; Trotman et al., 2009).
Integral to effective audit team functioning is the development
of a shared mental model of the audit task, the auditee, and the
audit team itself (including team members' knowledge, attitudes,
and beliefs). A shared mental model is de?ned as “organized un-
derstanding of relevant knowledge that is shared by team mem-
bers” (Mohammed & Dumville, 2001, p. 89). This group level
cognitive construct has been investigated in the psychology and
organizational behavior literatures for more than two decades (see
Cannon-Bowers, Salas, & Converse, 1993), and its importance in an
audit context is discussed by Bell et al. (2002).
Three speci?c areas of research related to the development of
shared mental models are considered. First, examining the extent
to which auditors' shared mental models change over time, in
response to post-task feedback (Cannon-Bowers et al., 1993), is
potentially of interest, given the importance of mental models in
the brainstorming literature (Hammersley, 2011). Second, the
cross-training of teams, where each team member is trained in the
duties of teammates with the aimof better equipping individuals to
anticipate the needs of other group members, has been shown to
lead to greater development of shared mental models and
enhanced team performance (Ilgen et al., 2005; Marks, Sabella,
Burke, & Zaccaro, 2002; Mathieu, Heffner, Goodwin, Salas, &
Cannon-Bowers, 2000). In auditing, cross-training could be used
to better informjunior members of their supervisors' roles and may
help the audit team work with, and integrate the work of, spe-
cialists (e.g., tax, valuation, IT) (Bauer & Estep, 2014). Third, lead-
ership training can be used to foster shared mental models (Marks,
Zaccaro, & Mathieu, 2000). In an audit setting, tone at the top is a
key aspect of quality control with important implications for team
performance (Dennis & Johnstone, 2014). How leadership training
on fostering shared mental models is best operationalized (and the
potential bene?ts fromdoing so) in an audit setting warrants future
experimental research.
In addition, shared mental models can lead to excessive con-
formity in the way group members think (Cannon-Bowers et al.,
1993). Exploring factors which foster positive versus negative ef-
fects of mental models within the audit setting is likely to be fruitful
given the potentially moderating roles of task complexity, knowl-
edge and expertise diversity, and group norms of issue resolution in
mental model research (Bantel & Jackson, 1989; Kellermanns,
Floyd, Pearson, & Spencer, 2008).
5.3. Audit team background diversity including sustainability
assurance
The traditional hierarchical review process in auditing consists
of team members with similar educational backgrounds (e.g., ac-
counting), similar training (e.g., CPA programs, in-house training),
and a degree of similarity in experience which increases as one
moves up the hierarchy. However, as new forms of assurance
emerge, including sustainability assurance, these relationships
change. For example, in the assurance of greenhouse gas (GHG)
emissions, multi-disciplinary audit teams are common, which
include team members with an accounting/auditing background
and those from other disciplines including the environmental sci-
ences (Trotman & Trotman, 2015). Such multi-disciplinary teams
have implications for the review process given the different types
of knowledge held by different team members. This diversity can
improve team performance through the addition of increased
knowledge to a particular task (Guzzo & Dickson, 1996; Williams &
O'Reilly, 1998). However, individuals from different backgrounds
may have different frames of reference, use different terminology
and have problem solving styles that may impede the sharing of
12
Similar features include: crews are hierarchical; each crew member contributes
to the decision making with different levels of knowledge; the duration of a ?ight
crew's existence as a unit is limited; training for crews is done at the individual,
rather the team, level; ?ight crews use standard ?ight brie?ngs where the captain
establishes expectations for the crew; ?ight crews face increasing levels of auto-
mation and regulation; CRM research includes external parties, such as air traf?c
controllers (Guzzo & Dickson, 1996).
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 68
information (van Knippenberg &Schippers, 2007). Such differences
have the potential to affect the performance of the review process
compared to other group formats and the potential effectiveness of
the review process may vary depending on whether the auditor
with the non-accounting background is the reviewee or the
reviewer. Ekasingh (2014) notes the importance of such variables as
cross-understanding (Huber & Lewis, 2010) and information elab-
oration (van Knippenberg, De Dreu, & Homan, 2004) and ?nds that
on a GHG assurance brainstorming task, auditors generate more
risks comparing subject matter and suitable criteria, whereas non-
accountants generate more risks related to subject matter
measurement.
In addition to sustainability assurance, audits of ?nancial
statements (including integrated audits of ?nancial statements and
internal controls over ?nancial reporting) involve a number of
specialists with varying backgrounds in education, training, and
experience including tax, valuation, and IT specialists (Bauer &
Estep, 2015). As suggested by Danos, Eichenseher, and Holt
(1989), not only will the audit team and their non-audit special-
ists differ in knowledge but they will also differ in the language or
jargon they use, which may impede performance of the engage-
ment teamas a whole. van der Vegt and Bunderson (2005) ?nd that
success for teams with a diversity of expertise in their membership
depends on the extent to which they share a team identity.
Consistent with this notion, Bauer and Estep (2015) provide evi-
dence that auditors and IT specialists who have a better working
relationship share a stronger team identity and integrate more
productively within the audit process. Team identity (or other so-
cial identities) may provide an interesting avenue for investigation
of these team effects.
5.4. Audit committee interactions with auditors
Audit committee members (ACMs) are accountable for over-
seeing the ?nancial reporting and audit process (CAQ, 2014; FEE,
2014; FRC, 2013), which often includes the asking of probing
questions by ACMs and responses by management and auditors
(Beasley, Carcello, Hermanson, & Neal, 2009; Gendron & B edard,
2006; Kang, 2014). New research has been emerging that exam-
ines these interactions. One line of research has considered the
probing questions asked by ACMs to their auditors (Kang, 2014;
Kang, Trotman, & Trotman, 2015; Pomeroy, 2010). Yet, auditors
also initiate communication with ACMs, and characteristics of
ACMs likely in?uence auditor accountability and how auditors
choose to interact with these ACMs. Despite their design chal-
lenges, experiments that examine the dynamic process among
ACMs, auditors, and managers are likely to provide important
?ndings. As emerging research has shown, these questions can
likely be explored by manipulating ACM behavior and observing
reactions of auditors or management (e.g., Bhattacharjee, Moreno,
& Pyzoha, 2014; Fiolleau, Hoang, & Pomeroy, 2013), and/or
manipulating auditor behavior and observing ACM questioning
behavior (Kang et al., 2015).
Finally, much recent emphasis has been placed on audit quality
indicators (AQIs) (Francis, 2011) with suggestions that AQIs should
be communicated between auditors and ACMs (CAQ, 2014). How-
ever, given the variation in ACMs' level of involvement in the audit,
background, experience, and ?nancial expertise (e.g., Beasley et al.,
2009; Cohen, Krishnamoorthy, & Wright, 2010; Kang et al., 2015) it
important to understand how ACMs will respond to and use these
AQIs.
6. Conclusion
Research on audit groups began over 30 years ago and
impressively it has incorporated the unique aspects of the audit
environment, and evolved with changes in audit regulation and
practice. We have outlined many of the insights from this research
and the potential directions for both existing research topics and
new lines of research on audit groups. To address these questions
audit researchers can play their part with innovative and rigorous
research designs. We believe it is in the interests of society, via
improved capital markets, and the audit ?rms themselves, that
audit practitioners be involved in this research. Given the criticism
that audit ?rms have faced from regulators and inspectors in the
last decade, a step forward is independent research on how best to
design group interactions between auditors within the ?rm and
with other professionals outside the audit ?rm including man-
agement, audit committees and inspectors.
Acknowledgments
We thank Chris Agoglia, Lindsay Andiola, Mike Bamber, Tina
Carpenter, Chris Chapman, Clara Chen, Rick Hat?eld, Vicky Hoff-
man, Bill Messier, Mark Nelson, Flora Zhou, and the participants at
the AOS 40th Anniversary Event on 1e2 May 2015. We acknowl-
edge the ?nancial support of an ARC Australian Professorial
Fellowship (DP110103382) to Ken Trotman.
References
Agoglia, C. P., Brazel, J. F., Hat?eld, R. C., & Jackson, S. B. (2010). How do audit
workpaper reviewers cope with the con?icting pressures of detecting mis-
statements and balancing client workloads? Auditing: A Journal of Practice &
Theory, 29(2), 27e43.
Agoglia, C. P., Hat?eld, R. C., & Brazel, J. F. (2009). The effects of audit review format
on review team judgments. Auditing: A Journal of Practice & Theory, 28(1),
95e111.
Agoglia, C. P., Kida, T., & Hanno, D. M. (2003). The effects of alternative justi?cation
memos on the judgments of audit reviewees and reviewers. Journal of Ac-
counting Research, 41(1), 33e46.
American Institute of Certi?ed Public Accountants (AICPA). (2002). Consideration of
fraud in a ?nancial statement audit. Interim Auditing Standard (AU) Section 316.
New York, NY: AICPA.
Andiola, L., & Bedard, J. C. (2015). What distinguishes effective audit reviews? Evidence
from the experiences of staff auditors. Bentley University. Working paper.
Asare, S. K., & McDaniel, L. S. (1996). The effects of familiarity with the preparer and
task complexity on the effectiveness of the audit reviewprocess. The Accounting
Review, 71(2), 139e159.
Asare, S. K., & Wright, A. M. (2004). The effectiveness of alternative risk assessment
and program planning tools in a fraud setting. Contemporary Accounting
Research, 21(2), 325e352.
Asare, S. K., Wright, A. M., & Trompeter, G. M. (2000). The effect of accountability
and time budgets on auditors' testing strategies. Contemporary Accounting
Research, 17(4), 539e560.
Ashton, R. H. (1974). An experimental study of internal control judgments. Journal of
Accounting Research, 12(1), 143e157.
Ashton, R. H. (1990). Pressure and performance in accounting decision settings:
paradoxical effects of incentives, feedback, and justi?cation. Journal of Ac-
counting Research, 28(3), 148e180.
Australian Securities & Investments Commission (ASIC). (2014). Report 397: Audit
inspection program report for 2012e13. Available athttp://download.asic.gov.au/
media/1344614/rep397-published-27-June-2014.pdf Accessed January 2015.
Bamber, E. M. (1983). Expert judgment in the audit team: a source reliability
approach. Journal of Accounting Research, 21(2), 396e412.
Bamber, E. M., & Bylinski, J. H. (1987). The effects of the planning memorandum,
time pressure and individual auditor characteristics on audit managers' review
time judgments. Contemporary Accounting Research, 4(1), 127e143.
Bamber, E. M., & Ramsay, R. J. (1997). An investigation of the effects of specialization
in audit workpaper review. Contemporary Accounting Research, 14(3), 501e513.
Bamber, E. M., & Ramsay, R. J. (2000). The effects of specialization in audit work-
paper reviewon reviewef?ciency and reviewers' con?dence. Auditing: A Journal
of Practice & Theory, 19(2), 147e157.
Bantel, K. A., & Jackson, S. E. (1989). Top management and innovations in banking:
does the composition of the top team make a difference? Strategic Management
Journal, 10, 107e124.
Bauer, T., & Estep, C. (2014). The IT auditor function on ?nancial statement and inte-
grated audits: Description of practice and avenues for future research. University
of Illinois. Working paper.
Bauer, T., & Estep, C. (2015). One team or two teams? Exploring the relationship be-
tween auditors and IT specialists and its implications for a collective audit team
identity and audit quality. University of Illinois. Working paper.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 69
Bauer, T., Hillison, S., Peecher, M. E., & Pomeroy, B. (2015). Do auditors make better
fraud planning decisions when advising colleagues versus deciding for themselv-
ess? University of Illinois and University of Waterloo. Working paper.
Beasley, M. S., Carcello, J. V., Hermanson, D. R., & Neal, T. L. (2009). The audit
committee oversight process. Contemporary Accounting Research, 26(1), 65e122.
Beasley, M. S., & Jenkins, J. G. (2003). A primer for brainstorming fraud risks. Journal
of Accountancy, 196(6), 32e38.
Bell, T. B., Peecher, M. E., & Solomon, I. (2002). A guide to selecting SSA cases. In
T. B. Bell, & I. Solomon (Eds.), Cases in strategic-systems auditing. Montvale, NJ:
KPMG and University of Illinois at Urbana-Champaign.
Bell, T. B., Peecher, M. E., & Solomon, I. (2005). The 21st century public company audit:
Conceptual elements of KPMG's global audit methodology. Montvale, New Jersey:
KPMG.
Bellovary, J. L., & Johnstone, K. M. (2007). Descriptive evidence from audit practice
on SAS No. 99 brainstorming activities. Current Issues in Auditing, 1, A1eA11.
Bhattacharjee, S., Moreno, K. K., & Pyzoha, J. S. (2014). The in?uence of audit com-
mittee intervention approach on auditor and client judgments during accounting
disputes. Virginia Tech, Northeastern University, and Case Western Reserve
University. Working paper.
Bonner, S. E. (1994). A model of the effects of audit task complexity. Accounting,
Organizations and Society, 19(3), 213e234.
Bonner, S. E., & Sprinkle, G. B. (2002). The effects of monetary incentives on effort
and task performance: theories, evidence, and a framework for research. Ac-
counting, Organizations and Society, 27(4e5), 303e345.
Boritz, J. E. (1985). The effect of information presentation structures on audit
planning and review judgments. Contemporary Accounting Research, 1(2),
193e218.
Brazel, J. F., Agoglia, C. P., & Hat?eld, R. C. (2004). Electronic versus face-to-face
review: the effects of alternative forms of review on auditors' performance.
The Accounting Review, 79(4), 949e966.
Brazel, J. F., Carpenter, T. D., & Jenkins, J. G. (2010). Auditors' use of brainstorming in
the consideration of fraud: evidence from the ?eld. The Accounting Review,
85(4), 1273e1301.
Brewster, B. E. (2011). How a systems perspective improves knowledge acquisition
and performance in analytical procedures. The Accounting Review, 86(3),
915e943.
Brown, H. L., & Wright, A. M. (2008). Negotiation research in auditing. Accounting
Horizons, 22(1), 91e110.
Cannon-Bowers, J. A., Salas, E., & Converse, S. (1993). Shared mental models in
expert team decision making. In N. J. Castellan, Jr. (Ed.), Individual and group
decision making: Current issues (pp. 221e246). Hillsdale, NJ: Lawrence Erlbaum
Associates, Inc.
Carpenter, T. (2007). Audit team brainstorming, fraud risk identi?cation, and fraud
risk assessment: implications of SAS No. 99. The Accounting Review, 82(5),
1119e1140.
Centre for Audit Quality (CAQ). (2014). CAQ approach to audit quality indicators.
Available athttp://www.thecaq.org/docs/reports-and-publications/caq-
approach-to-audit-quality-indicators-april-2014.pdf?sfvrsn¼2/caq-approach-
to-audit-quality-indicators Accessed May 2014.
Chen, C. X., Trotman, K. T., & Zhou, F. H. (2015). Nominal versus interacting elec-
tronic fraud brainstorming in hierarchical audit teams? The Accounting Review,
90(1), 175e198.
Chen, C. X., Williamson, M. G., & Zhou, F. H. (2012). Reward system design and group
creativity: an experimental investigation. The Accounting Review, 87(6),
1885e1911.
Choo, F., & Trotman, K. T. (1991). The relationship between knowledge structure and
judgments for experienced and inexperienced auditors. The Accounting Review,
66(3), 464e485.
Cohen, J., Krishnamoorthy, G., & Wright, A. (2010). Corporate governance in the
post-Sarbanes-Oxley era: auditors' experiences. Contemporary Accounting
Research, 27(3), 751e786.
Danos, P., Eichenseher, J. W., & Holt, D. L. (1989). Specialized knowledge and its
communication in auditing. Contemporary Accounting Research, 6(1), 91e109.
Dennis, A. R., & Valacich, J. S. (1993). Computer brainstorms: more heads are better
than one. Journal of Applied Psychology, 78(4), 531e537.
Dennis, S., & Johnstone, K. M. (2014). Audit partner leadership tone and professional
skepticism in fraud brainstorming. University of Wisconsin. Working paper.
DeZoort, T., Harrison, P., & Taylor, M. (2006). Accountability and auditors' materi-
ality judgments: the effects of differential pressure strength on conservatism,
variability, and effort. Accounting, Organizations and Society, 31(4), 373e390.
Einhorn, H. J., Hogarth, R. M., & Klempner, E. (1977). Quality of group judgment.
Psychological Bulletin, 84(1), 158e172.
Ekasingh, E. (2014). Examining the performance of multidisciplinary greenhouse gas
assurance engagement teams (PhD dissertation). Australia: UNSW.
Emby, C., & Gibbins, M. (1988). Good judgment in public accounting: quality and
justi?cation. Contemporary Accounting Research, 4(1), 287e313.
Epps, K. K., & Messier, W. F., Jr. (2007). Engagement quality reviews: a comparison of
audit ?rm practices. Auditing: A Journal of Practice & Theory, 26(2), 167e181.
Federation of European Accountants (FEE). (2014). Opening a discussion: The future
of audit and assurance. Available athttp://www.fee.be/images/Future_of_Audit_
and_Assurance_Discussion_Paper_1402.pdf Accessed May 2014.
Financial Reporting Council (FRC). (2013). Audit quality thematic review: Materiality.
Available athttps://www.frc.org.uk/Our-Work/Publications/Audit-Quality-
Review/Audit-Quality-Thematic-Review-Materiality.pdf Accessed May 2014.
Fiolleau, K. J., Hoang, K. J., & Pomeroy, B. (2013). Auditors' communications with audit
committees: The in?uence of the audit committee's oversight approach. Tulane
University and University of Waterloo. Working paper.
Francis, J. (2011). A framework for understanding and researching audit quality.
Auditing: A Journal of Practice & Theory, 30(2), 125e152.
Frank, M. L., & Hoffman, V. B. (2015). How audit reviewers respond to an audit
preparer's affective bias: the ironic rebound effect. The Accounting Review, 90(2),
559e577.
Gendron, Y., & B edard, J. (2006). On the constitution of audit committee effec-
tiveness. Accounting, Organizations and Society, 31(3), 211e239.
Gibbins, M., & Trotman, K. T. (2002). Audit review: managers' interpersonal ex-
pectations and conduct of the review. Contemporary Accounting Research, 19(3),
411e444.
Glover, S. M. (1997). The in?uence of time pressure and accountability on auditors'
processing of nondiagnostic information. Journal of Accounting Research, 35(2),
213e226.
Gold, A., Gronewold, U., & Salterio, S. E. (2014). Error management in audit ?rms:
error climate, type, and originator. The Accounting Review, 89(1), 303e330.
Gold, A., Knechel, W. R., & Wallage, P. (2012). The effect of the strictness of
consultation requirements on fraud consultation. The Accounting Review, 87(3),
925e949.
Grenier, J. H. (2014). Encouraging professional skepticism in the industry specialization
era. Miami University. Working paper.
Guzzo, R. A., & Dickson, M. W. (1996). Teams in organizations: recent research on
performance and effectiveness. Annual Review of Psychology, 47, 307e338.
Haerkens, M. H. T. M., Jenkins, D. H., & van der Hoeven, J. G. (2012). Crew resource
management in the ICU: the need for culture change. Annals of Intensive Care,
2(39), 1e5.
Hammersley, J. S. (2006). Pattern identi?cation and industry-specialist auditors. The
Accounting Review, 81(2), 309e336.
Hammersley, J. S. (2011). A review and model of auditor judgments in fraud-related
planning tasks. Auditing: A Journal of Practice & Theory, 30(40), 101e128.
Hammersley, J. S., Bamber, E. M., & Carpenter, T. D. (2010). The in?uence of docu-
mentation speci?city and priming on auditors' fraud risk assessments and ev-
idence evaluation decisions. The Accounting Review, 85(2), 547e571.
Harding, N., & Trotman, K. T. (1999). Hierarchical differences in audit workpaper
review performance. Contemporary Accounting Research, 16(4), 671e685.
Harding, N., & Trotman, K. T. (2014). Enhancing professional skepticism via partner
communication of fraud brainstorming outcomes. Australia: UNSW. Working
paper.
Herold, D. (1979). The effectiveness of work groups. In S. Kerr (Ed.), Organizational
behavior. Columbus, OH: Grid.
Hoffman, V. B., & Patton, J. M. (1997). Accountability, the dilution effect, and
conservatism in auditors' fraud judgments. Journal of Accounting Research,
35(2), 227e237.
Hoffman, V., & Zimbelman, M. (2009). Do strategic reasoning and brainstorming
help auditors change their standard audit procedures in response to fraud risk?
The Accounting Review, 84(3), 811e837.
Huber, G. P., & Lewis, K. (2010). Cross-understanding: implications for group
cognition and performance. Academy of Management Review, 35(1), 6e26.
Ilgen, D. R., Hollenbeck, J. R., Johnson, M., & Jundt, D. (2005). Teams in organiza-
tions: from input-process-output models to IMOI models. Annual Review of
Psychology, 56, 517e543.
International Auditing and Assurance Standards Board (IAASB). (2004). Interna-
tional standard on auditing (ISA) 240. The auditor's responsibility to consider fraud
in an audit of ?nancial statements. New York, NY 10017: International Federation
of Accountants (IFAC).
International Auditing and Assurance Standards Board (IAASB). (2008). Interna-
tional standard on auditing (ISA) 220. Quality control for an audit of ?nancial
statements. New York, NY 10017: IFAC.
Isaksen, S. G. (1998). A review of brainstorming research: Six critical issues for enquiry
(Monograph #302). Buffalo, NY: Creative Problem Solving Group-Buffalo.
Ismail, Z., & Trotman, K. T. (1995). The impact of the review process in hypothesis
generation tasks. Accounting, Organizations and Society, 20(5), 345e357.
Johnson, V. E., & Kaplan, S. E. (1991). Experimental evidence on the effects of
accountability on auditor judgments. Auditing: A Journal of Practice & Theory,
10(Suppl.), 96e107.
Joyce, E. J. (1976). Expert judgment in audit program planning. Journal of Accounting
Research, 14(3), 29e60.
Kadous, K., Leiby, J., & Peecher, M. E. (2013). How do auditors weight informal
contrary advice? The joint in?uence of advisor social bond and advice justi?-
ability. The Accounting Review, 88(6), 2061e2087.
Kang, Y. J. (2014). Are audit committees more challenging given a sophisticated investor
base? Does the answer change given anticipation of additional mandatory audit
report disclosure? University of Massachusetts-Amherst. Working paper.
Kang, Y. J., Trotman, A. J., & Trotman, K. T. (2015). The effect of an audit judgment
rule on audit committees' questioning on accounting estimates. Accounting,
Organizations and Society (forthcoming).
Kaplan, S. E., & Reckers, P. M. J. (1989). An examination of information search during
initial audit planning. Accounting, Organizations and Society, 14(5/6), 539e550.
Kellermanns, F. W., Floyd, S. W., Pearson, A. W., & Spencer, B. (2008). The contingent
effect of constructive confrontation on the relationship between shared mental
models and decision quality. Journal of Organizational Behavior, 29, 119e137.
Kennedy, J. (1993). Debiasing audit judgment with accountability: a framework and
experimental results. Journal of Accounting Research, 31(2), 231e245.
Kennedy, J. (1995). Debiasing the curse of knowledge in audit judgment. The
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 70
Accounting Review, 70(2), 249e273.
Kennedy, J., Kleinmuntz, D. N., & Peecher, M. E. (1997). Determinants of the justi-
?ability of performance in ill-structured audit tasks. Journal of Accounting
Research, 35(3), 105e123.
Kerr, N. L., & Tindale, R. S. (2004). Group performance and decision making. Annual
Review of Psychology, 55, 623e655.
Klein, G. (1999). Sources of power: How people make decisions. Cambridge, MA: MIT
Press.
Knechel, W. R., Salterio, S. E., & Kotchetova-Kozloski, N. (2010). The effect of
benchmarked performance measures and strategic analysis on auditors' risk
assessments and mental models. Accounting, Organizations and Society, 35,
316e333.
Koonce, L., Anderson, U., & Marchant, G. (1995). Justi?cation of decisions in audit-
ing. Journal of Accounting Research, 33(2), 369e384.
KPMG. (2011). Evaluating professional judgment in auditing and accounting: The
KPMG profession judgment framework. KPMG.
Lambert, T. A., & Agoglia, C. P. (2011). Closing the loop: review process factors
affecting audit staff follow-through. Journal of Accounting Research, 49(5),
1275e1306.
Libby, R., & Blash?eld, R. K. (1978). Performance of a composition as a function of
the number of judges. Organizational Behavior and Human Performance, 21,
121e129.
Libby, R., & Lewis, B. L. (1977). Human information processing research in ac-
counting: the state of the art. Accounting, Organizations and Society, 2(3),
245e268.
Libby, R., & Lewis, B. L. (1982). Human information processing research in ac-
counting: the state of the art in 1982. Accounting, Organizations and Society, 7(3),
231e285.
Libby, R., & Luft, J. (1993). Determinants of judgment performance in accounting
settings: ability, knowledge, motivation, and environment. Accounting, Organi-
zations and Society, 18(5), 425e450.
Libby, R., & Trotman, K. T. (1993). The review process as a control for differential
recall of evidence in auditor judgments. Accounting, Organizations and Society,
18(6), 559e574.
Lorge, I., Fox, D., Davitz, J., & Brenner, M. (1958). A survey of studies contrasting the
quality of group performance and individual performance, 1920e1957. Psy-
chological Bulletin, 55(6), 337e372.
Lynch, A., Murthy, U., & Engle, T. (2009). Fraud brainstorming using computer-
mediated communication: the effects of brainstorming technique and facilita-
tion. The Accounting Review, 84(4), 1209e1232.
Markman, A. B., & Gentner, D. (2001). Thinking. Annual Review of Psychology, 52,
223e247.
Marks, M. A., Sabella, M. J., Burke, C. S., & Zaccaro, S. J. (2002). The impact of cross-
training on team effectiveness. Journal of Applied Psychology, 87(1), 3e13.
Marks, M. A., Zaccaro, S. J., & Mathieu, J. E. (2000). Performance implications of
leader brie?ngs and team-interaction training for team adaptation to novel
environments. Journal of Applied Psychology, 85(6), 971e986.
Mathieu, J. E., Heffner, T. S., Goodwin, G. F., Salas, E., & Cannon-Bowers, J. A. (2000).
The in?uence of shared mental models on team process and performance.
Journal of Applied Psychology, 85(2), 273e283.
McGrath, J. E., & Kravitz, D. A. (1982). Group research. Annual Review of Psychology,
33, 195e230.
Messier, W. F., Jr., Kozloski, T. M., & Kotchetova-Kozloski, N. (2010). An analysis of
SEC and PCAOB enforcement actions against engagement quality reviewers.
Auditing: A Journal of Practice & Theory, 29(2), 233e252.
Messier, W. F., Jr., Quick, L. A., & Vandervelde, S. D. (2014). The in?uence of process
accountability and accounting standard type on auditor usage of a status quo
heuristic. Accounting, Organizations and Society, 39(1), 59e74.
Messier, W. F., Jr., & Tubbs, R. M. (1994). Recency effects in belief revision: the
impact of audit experience and the review process. Auditing: A Journal of
Practice & Theory, 13(1), 57e72.
Mock, T. J., & Turner, J. L. (1979). A ?eld test of the effects of changes in internal
controls on audit programs. In T. J. Burns (Ed.), Behavioral experiments in ac-
counting II. Columbus: College of Administrative Science, Ohio State University.
Mohammed, S., & Dumville, B. C. (2001). Team mental models in a team knowledge
framework: expanding theory and measurement across disciplinary bound-
aries. Journal of Organizational Behavior, 22, 89e106.
Nelson, M. W., Proell, C. A., & Randel, A. E. (2014). Team-oriented leadership, audit
risk and auditors' willingness to raise audit issues. Cornell University, Texas
Christian University and San Diego State University. Working paper.
Nelson, M., & Tan, H.-T. (2005). Judgment and decision making research in auditing:
a task, person, and interpersonal interaction perspective. Auditing: A Journal of
Practice & Theory, 24(Suppl.), 41e71.
Ng, T., & Shankar, P. (2010). Effects of technical department's advice, quality
assessment standards, and client justi?cations on auditors' propensity to accept
client-preferred accounting methods. The Accounting Review, 85(5), 1743e1761.
Osborn, A. F. (1957). Applied imagination: Principles and procedures of creative
thinking (2nd ed.). New York, NY: Charles Scribner's Sons.
Owhoso, V. E., Messier, W. F., Jr., & Lynch, J. J. G. (2002). Error detection by industry-
specialized teams during sequential audit review. Journal of Accounting
Research, 40(3), 883e900.
Oxley, N. L., Dzindolet, M. T., & Paulus, P. B. (1996). The effects of facilitators on the
performance of brainstorming groups. Journal of Social Behavior and Personality,
11, 633e646.
Paulus, P. B., Nakui, T., Paulus, V. L., & Brown, V. R. (2006). Effects of task instructions
and brief breaks on brainstorming. Group Dynamics: Theory, Research and
Practice, 10(3), 206e219.
Payne, E. A., Ramsay, R. J., & Bamber, M. E. (2010). The effect of alternative types of
review on auditors' procedures and performance. Auditing: A Journal of Practice
& Theory, 29(1), 207e220.
Peecher, M. E. (1996). The in?uence of auditors' justi?cation processes on their
decisions: a cognitive model and experimental evidence. Journal of Accounting
Research, 34(1), 125e140.
Peecher, M. E., Piercey, M. D., Rich, J. S., & Tubbs, R. M. (2010). The effects of a su-
pervisor's active intervention in subordinates' judgments, directional goals, and
perceived technical knowledge advantage on audit team judgments. The Ac-
counting Review, 85(5), 1763e1786.
Peecher, M. E., Solomon, I., & Trotman, K. T. (2013). An accountability framework for
?nancial statement auditors and related research questions. Accounting, Orga-
nizations and Society, 38(8), 596e620.
Phillips, F. (1999). Auditor attention to and judgments of aggressive ?nancial
reporting. Journal of Accounting Research, 37(1), 167e189.
Pinsonneault, A., Barki, H., Gallupe, R. B., & Hoppen, N. (1999). Electronic brain-
storming: the illusion of productivity. Information Systems Research, 10(2),
110e133.
Pomeroy, B. (2010). Audit committee members investigation of signi?cant ac-
counting decisions. Auditing: A Journal of Practice & Theory, 29(1), 173e205.
Public Company Accounting Oversight Board (PCAOB). (2009). Engagement quality
review. Auditing standard (AS) No. 7. Washington, D.C.: PCAOB.
Public Company Accounting Oversight Board (PCAOB). (2013). Observations related
to the implementation of the auditing standard on engagement quality review.
Release No. 2013-011. Washington, D.C.: PCAOB.
Public Company Accounting Oversight Board (PCAOB). (2015). Staff consultation
paper no. 2015-01: The auditor's use of the work of specialists. Washington, D.C.:
PCAOB. Available athttp://pcaobus.org/Standards/Pages/SCP_Specialists.aspx
Accessed July 2015.
Ramsay, R. J. (1994). Senior/manager differences in audit workpaper review per-
formance. Journal of Accounting Research, 32(1), 127e135.
Ranzilla, S., Chevalier, R. E., Herrmann, G., Glover, S. M., & Prawitt, D. F. (2011).
Elevating professional judgment in auditing: The KPMG professional judgment
framework. New York, NY: KPMG LLP.
Ricchiute, D. N. (1999). The effect of audit seniors' decisions on working paper
documentation and on partners' decisions. Accounting, Organizations and Soci-
ety, 24(2), 155e171.
Rich, J. S. (2004). Reviewers' responses to expectations about the client and the
preparer. The Accounting Review, 79(2), 497e517.
Rich, J. S., Solomon, I., & Trotman, K. T. (1997a). Multi-auditor judgment/decision
making research: a decade later. Journal of Accounting Literature, 16, 86e126.
Rich, J. S., Solomon, I., & Trotman, K. T. (1997b). The audit review process: a char-
acterization from the persuasion perspective. Accounting, Organizations and
Society, 22(5), 481e505.
Salterio, S. (1996). The effects of precedents and client position on auditors'
?nancial accounting policy judgment. Accounting, Organizations and Society,
21(5), 467e486.
Salterio, S., & Koonce, L. (1997). The persuasiveness of audit evidence: the case of
accounting policy decisions. Accounting, Organizations and Society, 22(6),
573e587.
Schaefer, T. J. (2014). The effects of social costs and internal quality reviews on auditor
consultation strategies (Dissertation). University of South Carolina.
Schultz, J. J., Jr., & Reckers, P. M. J. (1981). The impact of group processing on selected
audit disclosure decisions. Journal of Accounting Research, 19(2), 482e501.
Shankar, P. G., & Tan, H.-T. (2006). Determinants of audit preparers' workpaper
justi?cations. The Accounting Review, 81(2), 473e495.
Solomon, I. (1982). Probability of assessment by individual auditors and audit
teams: an empirical investigation. Journal of Accounting Research, 20(2),
689e710.
Solomon, I. (1987). Multi-auditor judgment decision making research. Journal of
Accounting Literature, 6, 1e25.
Sprinkle, G. B., & Tubbs, R. M. (1998). The effects of audit risk and information
importance on auditor memory during working paper review. The Accounting
Review, 73(4), 475e502.
Stroebe, W., Nijstad, B. A., & Rietzschel, E. F. (2010). Beyond productivity loss in
brainstorming groups: the evolution of a question. Advances in Experimental
Social Psychology, 43, 157e203.
Tan, H.-T. (1995). Effects of expectations, prior involvement, and review awareness
on memory for audit evidence and judgment. Journal of Accounting Research,
33(1), 113e135.
Tan, H.-T., & Jamal, K. (2001). Do auditors objectively evaluate their subordinates'
work? The Accounting Review, 76(1), 99e110.
Tan, H.-T., & Kao, A. (1999). Accountability effects on auditors' performance: the
in?uence of knowledge, problem-solving ability, and task complexity. Journal of
Accounting Research, 37(1), 209e223.
Tan, H.-T., & Trotman, K. T. (2003). Reviewers' responses to anticipated stylization
attempts by preparers of audit workpapers. The Accounting Review, 78(2),
581e604.
Tan, H.-T., & Yip-Ow, J. (2001). Are reviewers' judgements in?uenced by memo
structure and conclusions documented in audit workpapers? Contemporary
Accounting Research, 18(4), 663e678.
Tan, S.-K., & Tan, H.-T. (2008). Effects of exposure to subsequently invalidated evi-
dence on judgements of audit workpaper preparers and reviewers.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 71
Contemporary Accounting Research, 25(3), 921e946.
Trotman, A. J., & Trotman, K. T. (2015). Internal audit's role in GHG emissions and
energy reporting: evidence from audit committees, senior accountants and
internal auditors. Auditing: A Journal of Practice & Theory, 34(1), 199e230.
Trotman, K. T. (1985). The review process and the accuracy of auditor judgments.
Journal of Accounting Research, 23(2), 740e752.
Trotman, K. T. (2005). Discussion of judgment and decision making researching
auditing: a task, person, and interpersonal interaction perspective. Auditing: A
Journal of Practice & Theory, 24, 73e87.
Trotman, K. T., Simnett, R., & Khalifa, A. (2009). Impact of the type of audit team
discussions on auditors' generation of material frauds. Contemporary Accounting
Research, 26(4), 1115e1142.
Trotman, K. T., & Yetton, P. W. (1985). The effect of the review process on auditor
judgments. Journal of Accounting Research, 23(1), 256e267.
Trotman, K. T., Yetton, P. W., & Zimmer, I. R. (1983). Individual and group judgments
of internal control systems. Journal of Accounting Research, 21(1), 286e292.
Turner, C. W. (2001). Accountability demands and the auditor's evidence search
strategy: the in?uence of reviewer preferences and the nature of the response
(belief vs. action). Journal of Accounting Research, 39(3), 683e706.
van der Vegt, G., & Bunderson, J. S. (2005). Learning and performance in multi-
disciplinary teams: the importance of collective team identi?cation. Academy of
Management Journal, 48(3), 532e547.
van de Ven, A., & Delbecq, A. L. (1971). Nominal versus interacting group processes
for committee decision-making effectiveness. Academy of Management Journal,
14(2), 203e212.
van Knippenberg, D., De Dreu, C. K. W., & Homan, A. C. (2004). Work group diversity
and group performance: an integrative model and research agenda. Journal of
Applied Psychology, 89, 1008e1022.
van Knippenberg, D., & Schippers, M. C. (2007). Work group diversity. Annual Re-
view of Psychology, 58, 515e541.
Westermann, K. D., Cohen, J., & Trompeter, G. (2014). A qualitative examination of the
perceived in?uence of accountability on professional skepticism. Florida Interna-
tional University, Boston College and University of Central Florida. Working
paper.
Wilks, T. J. (2002). Predecisional distortion of evidence as a consequence of real-
time audit review. The Accounting Review, 77(1), 51e72.
Williams, K. Y., & O'Reilly, C. A. (1998). Demography and diversity in organizations:
a review of 40 years of research. In Research in organizational behavior (Vol. 20,
pp. 77e140). Stamford: Jai Press Inc.
Wood, R. E., Beckmann, J. F., & Birney, D. P. (2009). Simulations, learning and real
world capabilities. Education þ Training, 51(5/6), 491e510.
Yip-Ow, J., & Tan, H.-T. (2000). Effects of the preparer's justi?cation on the re-
viewer's hypothesis generation and judgment in analytical procedures. Ac-
counting, Organizations & Society, 25(2), 203e215.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 72
doc_253802788.pdf
We examine the research literature on audit groups/teams focusing on three main areas: the hierarchical
review process, brainstorming as part of the fraud detection planning process, and consultation within
firms. We restrict our discussion of these three literatures to judgment and decision making (JDM) experiments.
We consider research where two or more individuals within the audit firm interact with one
another face-to-face, electronically, or where one person prepares/reviews working papers for another.
Group judgment and decision making in auditing: Past and future
research
Ken T. Trotman
a, *
, Tim D. Bauer
b
, Kerry A. Humphreys
a
a
School of Accounting, UNSW Australia, Australia
b
University of Illinois at UrbanaeChampaign, USA
a r t i c l e i n f o
Article history:
Received 25 August 2015
Accepted 11 September 2015
Available online 9 October 2015
Keywords:
Review process
Brainstorming
Consultation
a b s t r a c t
We examine the research literature on audit groups/teams focusing on three main areas: the hierarchical
review process, brainstorming as part of the fraud detection planning process, and consultation within
?rms. We restrict our discussion of these three literatures to judgment and decision making (JDM) ex-
periments. We consider research where two or more individuals within the audit ?rm interact with one
another face-to-face, electronically, or where one person prepares/reviews working papers for another.
We outline future research within each of the above areas, as well as considering other areas of future
research involving within-?rm group interactions related to audit teams in context, shared mental
models, and audit team diversity (including sustainability assurance), as well as interactions with groups
outside the audit ?rm, particularly audit committees.
© 2015 Elsevier Ltd. All rights reserved.
1. Introduction
The judgment and decision making (JDM) literature in auditing
emerged in the early 1970s, with Accounting, Organizations and
Society (AOS) playing a signi?cant role in the early development of
this ?eld (Libby & Lewis, 1977, 1982). While most JDM audit liter-
ature concentrates on individual auditor judgments (Nelson &Tan,
2005), audits are generally conducted by groups. They often involve
a hierarchical team where preparers may stylize audit tests and
working papers depending on the reviewer, and the reviewprocess
will be affected by a range of preparer, task and environmental
factors (Rich, Solomon, &Trotman, 1997a, 1997b). Concentrating on
single-person studies is problematic, as psychology research shows
the results of these studies often do not generalize to multi-person
settings (Solomon, 1987).
Audit group research draws not only on the wider audit JDM
experimental literature for both theory and methods, but also on
the social psychology/organizational behavior literature. The latter
literature has a long history (see Lorge, Fox, Davitz, &Brenner, 1958,
for an early review). Some evidence of the extent of group research
is its inclusion as one of three key parts of the Annual Review of
Psychology in 1961 (McGrath & Kravitz, 1982) leading to regular
review papers of new research being published in the Annual Re-
view of Psychology from 1962 to 1991, approximately every three
years. With the increased impetus on group research, the reviews
in the Annual Review of Psychology from 1996 then began to be split
into more specialized subsections of group/team research.
1
1.1. Scope of this review
Our working de?nition of research on audit groups/teams in-
cludes research where two or more individuals within the audit
?rm interact with one another. This interaction can be face-to-face,
electronic, or one person preparing/reviewing working papers for
another. It implies that the persons are mutually aware of one
another and take into account the actions of other group members
(McGrath & Kravitz, 1982). Our de?nition is broad enough to
include group interactions between an auditor and other internal
* Corresponding author. School of Accounting, UNSW Business School, UNSW,
Sydney, NSW 2052, Australia.
E-mail address: [email protected] (K.T. Trotman).
1
Guzzo and Dickson (1996) and Ilgen et al. (2005) distinguish their reviews from
previous research based on their focus on team performance in organizational
contexts, where work groups and teams are normally embedded in organizations
and exist over a period in time. In comparison, Kerr and Tindale (2004) restrict
themselves to cover group performance and decision making with a focus on the
groups' outputs (e.g., solutions, decisions) and the processes employed by groups.
Contents lists available at ScienceDirect
Accounting, Organizations and Society
j ournal homepage: www. el sevi er. com/ l ocat e/ aoshttp://dx.doi.org/10.1016/j.aos.2015.09.004
0361-3682/© 2015 Elsevier Ltd. All rights reserved.
Accounting, Organizations and Society 47 (2015) 56e72
parties, for example studies where the auditor consults with others
in the ?rm (both auditors and other specialists).
Auditors also interact with other professions outside the audit
?rm including others involved in corporate governance such as
client management and the audit committee, and external spe-
cialists especially for complex estimates. We exclude the literature
on auditor-client negotiations given its coverage in recent review
papers (e.g., Brown & Wright, 2008). We include interactions be-
tween auditors and audit committees in future research, given the
enhanced role of audit committee members (ACMs) for effectively
overseeing the ?nancial reporting and auditing process (CAQ, 2014;
FEE, 2014; FRC, 2013; Peecher, Solomon, & Trotman, 2013) and
because we see this as an important area of future research not yet
reviewed.
We restrict our review to JDM experiments, although we draw
on ?eld studies in considering future research, and focus on four
main areas:
The hierarchical review process: the process is a key control
mechanism within audit ?rms which affects both reviewer
behavior and preparer behavior. Reviewer behavior includes
comparisons of outputs with other group formats (e.g., inter-
acting and nominal groups); different structures of the review
process; factors affecting the output of the review process; and
cognitive effects of taking a reviewer role. Preparer behavior
includes both the effect of the review process as a form of
accountability, as well as the motives/actions of preparers
including persuasion.
Brainstorming: as part of the fraud detection planning process,
auditing standards require a discussion (brainstorming
meeting) between audit team members, including how and
when the client's ?nancial statements may be susceptible to
material misstatement due to fraud. We consider nominal
versus interacting group brainstorming, face-to-face versus
electronic brainstorming, and the relationship between various
judgments made within the brainstorming session.
Consultation within ?rms: in addition to the review process,
consultation with other auditors (i.e., advice seeking and giving)
is common and can reduce uncertainty about dif?cult decisions
and help auditors defend a position to various stakeholders. We
consider the justi?ability of consulting advice, howauditors use
or follow consulting advice, and their willingness to consult, in
both formal and informal settings.
New directions and future research: in addition to future
research discussed within the context of the above three areas,
we consider other areas of future research involving within-?rm
group interactions related to audit teams in context, shared
mental models, and audit team diversity (including sustain-
ability assurance), as well as interactions with groups outside
the audit ?rm, particularly audit committees.
Previous research on multi-auditor judgments (Rich,
Solomon, & Trotman, 1997b; Solomon, 1987) distinguishes be-
tween the terms ‘audit teams’ and ‘audit groups’, where ‘audit
teams’ refers to the hierarchical review process and ‘audit
groups’ refers to situations where auditors solve a particular
problem or perform a task as a group. However, more recent
research in psychology has suggested the distinction is “a rather
arti?cial one”, with Kerr and Tindale (2004, p. 624) explaining
the difference as follows:
“Team research tends to focus on relatively longer-term groups
with multiple task responsibilities, often functioning within an
organization. …Small group performance research, on the other
hand, tends to be basic research …, and is usually studied
experimentally in ad-hoc laboratory groups. However, there are
many exceptions to these general rules. …”
The related research in the audit literature has elements of both
group and team research as described above. For example, in
practice, groups within audit ?rms are normally part of a team
which exists over time, has multiple responsibilities, and generally
operates within the organization. On the other hand, the actual
research conducted on groups in auditing invariably is not con-
ducted within actual teams and generally studies ad-hoc groups
where the participants may or may not have previously worked
together (for exceptions see Tan & Jamal, 2001; Trotman, 1985).
Participants may have never met other group members, may have
met them socially or at training, or may have worked together on a
number of audits. Thus we use the terms ‘team’ and ‘groups’
interchangeably.
1.2. Group audit JDM literature over time
Table 1 divides 57 group audit JDM papers across four leading
international journals e Accounting, Organizations and Society
(AOS), Contemporary Accounting Research (CAR), Journal of Ac-
counting Research (JAR), and The Accounting Review (TAR) (Panel A)
e and across time (Panel B).
2
A number of important observations
can be made. First, under the heading ‘Pre-review interacting
groups’ only three papers are included, which were all published in
JAR in the 1980s. However, we note some of the other papers in the
review process category include an interacting group as a com-
parison (Owhoso, Messier, & Lynch, 2002; Trotman, 1985) and all
the brainstorming papers include an interacting group (except
Hammersley, Bamber, & Carpenter, 2010). However, the nature of
the task in the brainstorming studies involves idea generation,
whereas in the reviewprocess studies the task is usually evaluation
or estimation.
3
Second, there are two quite different streams of research on the
review process, with one looking at the effects of the review pro-
cess on the reviewer (and comparing the quality of the audit post-
versus pre-review) and the other examining the effects on the
preparer via accountability and justi?cation. Both categories have a
substantial number of papers (23 and 18 respectively). However,
we note that there have been only three papers on the effects of the
review process on the reviewer in the last decade (Frank &
Hoffman, 2015; Peecher, Piercey, Rich, & Tubbs, 2010; Tan & Tan,
2008). We do not believe this is due to a decrease in the impor-
tance of this topic, but is more related to the dif?culties of obtaining
participants with suf?cient seniority to carry out the review. On the
other hand, papers on the effects of the review process on the
preparer, while ?rst appearing almost a decade after those papers
on the effects on the reviewer, have continued into the current
decade. Third, the titles of the review process papers on ‘effects on
the reviewer’ and ‘effects on the preparer’ are quite different, with
19 of the 23 ‘effects on the reviewer’ papers including the word
‘review’ in their title, indicating the papers are clearly about the
review process. In contrast, only ?ve of the 18 ‘effects on the pre-
parer’ papers include the word ‘review’, suggesting that these pa-
pers are more about accountability and justi?cation, with the
review process as the means of inducing these behaviors.
Fourth, the brainstorming papers have all been published
2
While we note that Table 1 is limited to these four journals, throughout the
paper we do refer to some important papers in other journals, particularly Auditing:
A Journal of Practice & Theory.
3
This is an important distinction and the results are often not transferable across
tasks (see Kerr & Tindale, 2004).
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 57
between 2007 and 2015, with ?ve of the six papers published in
TAR. This research focus re?ects the requirements of US (AU 316)
and international (ISA 240) auditing standards for a brainstorming
session to be held, which includes an audit team discussion as to
howand when a client's ?nancial statements may be susceptible to
material misstatement from fraud (AICPA, 2002; IAASB, 2004).
Fifth, while the consultation research theme started in the 1990s
with three papers, there has been renewed interest in the
2010e2015 period with a further three papers published. This
renewed interest is likely motivated, in part, by revisions to rules
about audit team consultations (Gold, Knechel, & Wallage, 2012)
and by regulator concerns about audit team reliance on specialists'
work (Bauer & Estep, 2015; PCAOB, 2015), which involves consul-
ting and coordinating with these specialists.
2. The hierarchical review process
The audit review process refers to members of the audit team
evaluating the work of other audit teammembers; it is a key control
mechanism within audit ?rms aimed at increasing audit quality
(Libby & Trotman, 1993; Solomon, 1987). The review process is
generally hierarchical and sequential in that additional review may
occur as the work moves up the hierarchy to partners. The two
main objectives of the review process relate to ensuring the
appropriateness of both the opinion given and the documentation
provided (Rich et al., 1997a). The reviewer identi?es alternative
explanations, omissions, and inconsistencies that were missed by
the preparer, and evaluates whether the documentation is suf?-
cient to support the conclusion reached (Libby & Trotman, 1993;
Tan & Trotman, 2003). Also, the preparer anticipates conse-
quences from the review, and the need to justify and document
their conclusions to the reviewer (Emby &Gibbins, 1988; Rich et al.,
1997a). This in?uences audit quality through the preparer's effort
from feeling accountable to the reviewer and his/her strategic
behavior to stylize workpapers in order to persuade the reviewer
that adequate evidence has been collected and that documentation
is appropriate (Rich et al., 1997a).
In examining the review process from the perspective of the
reviewer, the fundamental issues to address are under what cir-
cumstances does the review process improve the quality of the
audit and why does this occur. Accordingly, we divide the research
on the effects of the review process on the reviewer into the
following categories: comparisons of the review process against
other baselines, such as individual judgments, nominal groups, and
non-hierarchical interacting groups; what performance gains result
from the review process; comparisons of the performance of
alternative forms of the review process (e.g., face-to-face versus
electronic); the effect of environmental factors (e.g., task being
reviewed); and the effect of preparer attributes on the review
process (e.g., stylization). A summary of the papers addressing the
reviewer perspective is provided in Fig. 1 Panel A.
Potential reasons for why reviewed judgments may be of higher
quality than pre-review judgments are the inclusion of an addi-
tional person (a diversi?cation effect), the discussion between the
two group members leading to some form of cognitive stimulation
(an interaction effect), or the addition of a person of a higher rank
who decides on the relative weights for different inputs to the
judgments (a hierarchical effect). To isolate these differences, the
ideal research design is to have individual judgments (e.g., both
seniors and managers), interaction between two auditors (either
two seniors or one senior and one manager), and also the
combining of individual judgments for a nominal group (either two
seniors or one senior and one manager). Differences due to diver-
si?cation, interactions and the hierarchy can then be isolated.
In making the above comparisons of individuals, interacting
groups and a review process, it is necessary for the researcher to
have a benchmark and this is often done by forming nominal teams.
If the task is an estimate (e.g., inventory obsolescence or an
assessment of the strength of an internal control system), this will
be a statistical average of individual estimates. If it is a prediction of
an event (e.g., a company will fail/not fail) it will be a majority vote.
If it is a brainstorming task it will be the combining of all unique
ideas generated. Note that nominal teams can only be formed if
there is a within-subject treatment with auditors completing the
task as individuals ?rst, or a between-subjects design where par-
ticipants in one of the treatments complete the task as individuals.
4
In examining the review process from the perspective of the
preparer, the fundamental issues to be addressed are how
accountability to the reviewer or stylization attempts in?uence
Table 1
Publications of group audit JDM experiments.
The review process Brainstorming Consultation
Pre-review interacting groups Effects on the reviewer
b
Effects on the preparer
b
Panel A: By topic and journal in Accounting, Organizations and Society (AOS), Contemporary Accounting Research (CAR), Journal of Accounting Research (JAR) and
The Accounting Review (TAR) from 1970 to 2015
a
Journal
AOS 0 4 2 0 2
CAR 0 6 1 1 1
JAR 3 6 10 0 1
TAR 0 7 5 5 3
Total 3 23 18 6 7
Panel B: By topic and decade in Accounting, Organizations and Society (AOS), Contemporary Accounting Research (CAR), Journal of Accounting Research (JAR) and
The Accounting Review (TAR) from 1970 to 2015
a
Decade
1970s 0 0 0 0 0
1980s 3 5 0 0 0
1990s 0 9 8 0 3
2000s 0 7 7 4 1
2010e2015 0 2 3 2 3
Total 3 23 18 6 7
a
2015 papers are those published at June 2015.
b
Several papers on the review process consider both the effects on the reviewer and the preparer. These have been categorized based on their primary focus here, but in
later tables the effects on both groups are presented.
4
Such benchmarks are an important example of how an experiment can include
treatments that do not presently exist in practice to tease out the explanations
between different treatments (Libby & Luft, 1993).
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 58
evidence collection and documentation, and ultimately, audit
quality. Accordingly, we divide the research on the effects of the
review process on the preparer into reviewer preferences that are
unknown versus known, as well as alternative review formats. A
summary of the papers addressing the preparer perspective is
given in Fig. 1, Panel B. Bonner and Sprinkle (2002) distinguish
among effort duration, intensity, and direction; one observation
that becomes apparent from distinguishing between unknown
versus known reviewer preferences is that unknown preferences
tend to increase effort duration or intensity, while known prefer-
ences tend to increase effort direction. This observation is helpful in
reconciling the positive versus negative effects on audit quality of
accountability to reviewers that have been documented in the audit
literature.
2.1. Early history of group decision making and review process
research
The early audit JDM literature was dominated by studies of in-
dividual judgments extending the results of Ashton (1974). How-
ever, the importance of group decision making including the
review process was identi?ed in Joyce (1976), Libby and Lewis
(1977), Libby and Blash?eld (1978), and Mock and Turner (1979).
Three early empirical papers in auditing (Schultz & Reckers, 1981;
Solomon, 1982; Trotman, Yetton, & Zimmer, 1983) directly
address the bene?ts of group decision making, including diversi-
?cation and interaction effects.
A natural extension of the above studies was to move to an
examination of the effects of the review process as that was, and
still is, the predominant form of group decision making in audit
?rms. Early research examines the need for reviewers to evaluate
the reliability of preparers (Bamber, 1983) and how the review
process differs from other group structures due to its hierarchical
nature (Trotman, 1985; Trotman & Yetton, 1985).
2.2. The review process versus alternative group structures
Trotman and Yetton (1985) ?nd that the review process (oper-
ationalized by having managers review seniors' work in their
absence) signi?cantly reduces judgment variance relative to indi-
vidual judgments, but there are no differences in performance
between the review process and a two person nominal or inter-
acting group (two seniors). This suggests that the advantage of the
review process may simply be due to a diversi?cation effect. In
contrast, Trotman (1985) examines a more complex task and
operationalizes the review process by having managers review
seniors' work with the senior present, which increases the oppor-
tunity to differentiate the expertise between their own judgments
and those of the senior. Trotman ?nds the review process out-
performs nominal groups (judgments are over 50 percent more
accurate post-versus pre-review), but does not outperform inter-
acting groups of seniors; while the review process is more effective
in reducing systematic bias, the interacting group is more effective
in reducing random error. Finally, Owhoso et al. (2002) extend the
research on nominal versus interacting review teams by including
different types of errors and cases from different industries to
specify the conditions under which process gains from the review
process occur, including the unique contributions of both managers
and seniors to the team's overall effectiveness.
In summary, across the three studies there is evidence of a
diversi?cation effect, an interaction effect and a hierarchical effect,
but the results vary with a range of task and environmental con-
ditions. The following section discusses some of the gains from the
review process and the factors affecting them.
2.3. What performance gains result from the review process?
In the 1990s, many of the studies on the review process
extended the earlier documentation of bene?ts by incorporating
Fig. 1. Summary of review process experiments.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 59
cognitive factors. Libby and Trotman (1993) identify one source of
gain from the review process as the different points at which pre-
parers and reviewers enter the judgment process, which affects the
likelihood that the implications of inconsistent evidence are
considered. Preparers enter the process prior to an initial judgment
being made and therefore have incentives to justify their positions,
whereas reviewers enter the process after an initial judgment has
been made and therefore are expected to be more concerned with
determining whether the judgment is consistent with the facts
they have available. Libby and Trotman ?nd preparers' (reviewers')
relative recall is in the direction of information consistent (incon-
sistent) with their (the preparer) judgment. Ricchiute (1999) ex-
tends these results to a situation where seniors' biased memory for
consistent information may affect their documentation in the audit
workpapers such that the partner, as reviewer, evaluates a subset of
evidence that the preparer evaluates. He ?nds partners' decisions
are biased in the direction of the seniors' decisions.
Ramsay (1994) observes that in many audit ?rms, seniors
perform detailed mechanical reviews and managers perform more
general, conceptual reviews. He ?nds that managers outperform
seniors in detecting conceptual errors, but seniors outperform
managers in detecting mechanical errors; further, Harding and
Trotman (1999) ?nd that seniors outperform staff auditors in
identifying conceptual errors, while staff auditors outperform se-
niors in identifying mechanical errors. The combined results of
Ramsay and Harding and Trotman suggest a more conceptual re-
viewtemplate as auditors move up the ranks of their ?rmfromstaff
to senior to manager.
Owhoso et al. (2002) have auditors complete cases as a nominal
or reviewteam, both within and out of their industry specialization.
Within specialization, managers and seniors both incrementally
contribute to team effectiveness and nominal teams outperform
the interacting reviewteams in detecting mechanical errors but not
conceptual errors; however, out of specialization, nominal teams
outperform review teams in detecting both mechanical and con-
ceptual errors (Owhoso et al., 2002).
While the above studies suggest that the review process im-
proves audit effectiveness, there has been some research showing it
is not always effective. Tan and Tan (2008) ?nd auditors do not
suf?ciently discount the implications of evidence received that is
subsequently invalidated despite being aware of the erroneous
nature of this evidence and that reviewers are also susceptible to
this effect, unless accountability to a superior (reviewer) is
increased. Messier and Tubbs (1994) ?nd the recency effect bias,
which had been found in psychology and individual audit judgment
research, is not mitigated by the reviewprocess. Frank and Hoffman
(2015) show that reviewers do not always mitigate biases when
they cannot readily determine how much the bias in?uences a
preparer's judgment.
2.4. Alternative forms of the review process
There are many different ways the review process can be
operationalized. Traditionally, the workpapers are examined by the
reviewer independently and written review notes are developed
for the preparer to follow up. The reviewer then meets with the
preparer to discuss these review notes, or sends them to the pre-
parer (e.g., electronically) for a response (Payne, Ramsay, &Bamber,
2010). Face-to-face review is also common where the preparer and
reviewer meet in a real-time interactive review (Gibbins &
Trotman, 2002).
Comparisons have been made between reviews with and
without discussion (Ismail &Trotman, 1995), specialized versus all-
encompassing reviews (Bamber & Ramsay, 1997, 2000), and elec-
tronic versus face-to-face reviews (Agoglia, Brazel, Hat?eld, &
Jackson, 2010; Agoglia, Hat?eld, & Brazel, 2009; Brazel, Agoglia, &
Hat?eld, 2004). Ismail and Trotman (1995) examine auditor per-
formance on an analytical procedures task prior to and after review,
where the form of review is manipulated via the presence/absence
of discussion between the reviewer (a manager or a senior) and a
reviewee (a senior). They ?nd that review increases the number of
plausible hypotheses generated for both groups, but to a greater
extent for review groups with discussion. Bamber and Ramsay
(1997) note that in some ?rms, managers are instructed to focus
on conceptual errors while seniors are instructed to focus on me-
chanical errors. In other cases, both seniors and managers carry out
all-encompassing reviews without limit on the focus. They ?nd
bene?ts of hierarchical all-encompassing reviews as these reviews
are signi?cantly more accurate than specialized reviews. The type
of review affects con?dence levels, and both seniors and managers
require more time to perform specialized reviews compared to
those performing all-encompassing reviews (Bamber & Ramsay,
2000).
With technological advances, audit ?rms have increased alter-
natives for how to conduct the review process including the review
of workpapers online and then providing the preparer with the
review notes via email (Brazel et al., 2004). They ?nd that the
method of review affects preparer effectiveness and ef?ciency. In
particular, the preparers anticipating a face-to-face revieware more
concerned with audit effectiveness and produce higher quality
judgments. Gibbins and Trotman (2002) ?nd both preparers and
reviewers generally prefer face-to-face reviews and a survey
research conducted by Agoglia et al. (2010) ?nds that reviewers
perceive face-to-face to be more effective than electronic review,
but that electronic review is considered more convenient. Agoglia
et al. (2009) ?nd the quality of reviewers' going concern judg-
ments are signi?cantly lower for electronic compared to face-to-
face review. They conclude that this results from the effect of re-
view mode on the preparer (see Brazel et al., 2004) persisting
through the review process with the consequence that reviewer
judgments are ultimately affected by the formof reviewanticipated
by the preparer. Agoglia et al. (2010) ?nd that workload pressure
can increase the likelihood of choosing electronic review but only
when misstatement risk is low. When misstatement risk is high,
face-to-face review is chosen irrespective of workload pressures.
2.5. Effects of environmental factors on the review process
Several papers have addressed the effect of environmental fac-
tors on the review process including: information presentation
structures (Boritz, 1985), the planning memorandum (Bamber &
Bylinski, 1987), time pressure (Bamber & Bylinski, 1987), aggres-
sive ?nancial reporting (Phillips, 1999), and audit risk and infor-
mation importance (Sprinkle & Tubbs, 1998). Boritz (1985) is an
early example of reviewresults being highly dependent on the task
and the complexity of the setting. Bamber and Bylinski (1987) ?nd
that while an initial audit plan directs managers' subsequent re-
view, their estimated review times to complete tasks are not
affected by deadline pressure. Higher levels of audit risk increase
reviewers' accuracy of evidence recall (Sprinkle & Tubbs, 1998) and
amount of recall of (i.e., attention to) aggressive ?nancial statement
reporting (Phillips, 1999). Both papers attribute differences in
judgment to different cognitive processes adopted as a result of the
environmental factors in the review process. These different
cognitive processes contribute to both effectiveness and ef?ciency,
which we address further under future research below.
2.6. Effect of the preparer on the review process
Another feature of the audit environment that affects the
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 60
review process is the attributes of the preparer. Bamber (1983)
recognizes the importance of reviewers evaluating the reliability
of preparers as information sources. He ?nds that while reviewers'
assessments of the informativeness of audit evidence collected by
a subordinate are sensitive to the manipulation of source reli-
ability, most managers excessively discount the information's
diagnosticity for reductions in source reliability relative to the
model's predictions, contrary to earlier psychology evidence. Asare
and McDaniel (1996) ?nd that reviewers reperform more of the
work completed by unfamiliar preparers than by familiar pre-
parers, and that preparer familiarity and task complexity interact
to determine the effectiveness of reviewers in detecting conclu-
sion errors.
One of the few studies in the audit literature using actual audit
teams is Tan and Jamal (2001), who investigate whether managers'
(reviewers) assessments of the quality of seniors' (preparers) work
is in?uenced by their prior impression of the preparer and whether
this effect varies depending on whether these preparers are clas-
si?ed as outstanding or average by the audit ?rm. Their key result is
that average managers evaluate memos written by outstanding
seniors more favorably than memos written by average seniors,
only when they know the identity of the author of the memo.
Outstanding managers are not susceptible to the same effect (Tan &
Jamal, 2001).
Frank and Hoffman (2015) examine the responses of experi-
enced reviewers when they review a preparer judgment that ap-
pears to be biased by the preparer's affect (i.e., feeling of personally
liking or disliking) towards a client's controller. They ?nd that re-
viewers rely more on the preparer's judgments when they suspect
it is biased than do reviewers who are not informed about the
preparer's affect. This result is consistent with psychology ?ndings
on the ironic rebound effect which predicts that when individuals
are trying not to rely on information, they sometimes rely on it
more.
The importance of the preparer to the outcome of the review
process is a central theme in Rich et al. (1997a), who characterize
the audit review process from a persuasion perspective where
preparers may stylize evidence collected and documented in the
workpapers, depending on the reviewer and his/her preferences. As
discussed in a later section, preparers do indeed stylize their
workpapers (e.g., Agoglia, Kida, & Hanno, 2003; Shankar & Tan,
2006).
Six papers since 2000 directly address effects on reviewers of
preparer stylization. One method of stylization is via the pre-
parer's use of a justi?cation memo. Yip-Ow and Tan (2000) ?nd
the presence (versus absence) of a preparer's justi?cation memo
for a non-error cause reduces the number of alternative error
hypotheses generated by reviewers, and increases reviewers' as-
sessments of the likelihood of the non-error cause suggested by
the preparer. Agoglia et al. (2003) examine how reviewers respond
to three different justi?cation memos by preparers and ?nd that
the more the memo justi?es the preparers' preferred reporting
outcome, the more the reviewer provides a judgment supporting
that outcome. However, Tan and Yip-Ow (2001) ?nd that re-
viewers place less reliance on a preparer's consistencies when the
preparer's justi?cation memo is stylized to downplay inconsistent
evidence (and emphasize consistent evidence) compared to when
it is neutral, suggesting reviewers are sensitive to preparer styli-
zation attempts. This supports the Rich et al. (1997a) model, which
suggests that reviewers learn to cope with stylization attempts
over time.
Effects of stylization may also depend on interactions be-
tween preparer attributes and either attributes of the reviewer
(Tan & Trotman, 2003) or expectations about the client (Rich,
2004). Tan and Trotman (2003) show that reviewers' responses
to preparer stylization attempts depend on the nature of the
stylization attempt, auditor rank, and reviewer's sensitivity to
stylization. Rich (2004) ?nds that reviewers' expectations about
the client and preparer interact to affect reviewers' evaluation of
the preparer's work and the amount of follow-up work to be
carried out. Peecher et al. (2010) investigate whether reviewers
are in?uenced not only by stylization from preparers, but also by
their own guidance provided to preparers, which motivates
stylization in the ?rst place. Their results show that preparers
are in?uenced by the reviewers' preferences for less versus more
skepticism and that preparer judgments in?uence reviewers'
?nal judgments, with this occurring to an even greater extent
when reviewers provide guidance to preparers on how to make a
judgment.
2.7. Effect of the review process on the preparer
2.7.1. Reviewers with unknown preferences
Studies emerged in the early 1990s that emphasized the
importance of the review process in in?uencing audit quality, by
inducing accountability in preparers and changing how they
perform the work to be reviewed (e.g., Johnson & Kaplan, 1991;
Kennedy, 1993; Koonce, Anderson, & Marchant, 1995; Tan, 1995).
Speci?cally, being accountable to a reviewer with unknown pref-
erences is generally expected to induce vigilance e that is, effort
intensity and duration e to reduce errors or biases. Extant research
has generally manipulated such accountability by either informing
or not informing participants that their responses may be reviewed
and, in some cases, they may be asked to justify their decisions at a
later time. Accordingly, increased accountability is helpful in
reducing the biased use of audit evidence by preparers stemming
from insuf?cient effort or attention, such as a recency bias
(Kennedy, 1993) and a consistency or con?rmation bias (Tan, 1995).
Koonce et al. (1995) also show it increases the amount of justi?-
cation documented by preparers.
However, Kennedy (1995) suggests that accountability to a
reviewer with unknown preferences will not help reduce biases
related to data problems (e.g., faulty memory). Research has
shown that increased accountability does not reduce the
continued in?uence of evidence that has been processed but is
irrelevant (Kennedy, 1995) or is subsequently proven invalid (Tan
& Tan, 2008), or the dilution of the use of diagnostic evidence in
the presence of non-diagnostic evidence (Glover, 1997; Hoffman
& Patton, 1997). Koonce et al. (1995) show that increased
accountability does not increase the type of justi?cations docu-
mented by preparers, while Ricchiute (1999) shows that it does
not decrease preparers' tendency to recall and document for re-
viewers relatively more evidence consistent with their initial
decision (Libby & Trotman, 1993). Moreover, Tan and Kao (1999)
?nd that increased accountability has limited positive in?uence
on effort (or performance) in less complex tasks, or in increas-
ingly complex tasks unless preparers have increasingly high
knowledge and skill.
Two areas where greater accountability may increase both effort
duration/intensity and effort direction are with respect to consis-
tent versus inconsistent evidence and error versus non-error ex-
planations. Tan (1995) and Asare, Wright, and Trompeter (2000)
show that the threat of review by a superior with unknown pref-
erences increases the attention paid to evidence items or expla-
nations with relatively more attention paid to inconsistent
evidence (Tan, 1995) or error explanations (Asare et al., 2000).
Hoffman and Patton (1997) and Johnson and Kaplan (1991) show
such accountability also causes auditors to exhibit more con-
servativism, consensus, or self-insight. DeZoort, Harrison, and
Taylor (2006) further show auditors exhibit increasing levels of
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 61
conservatism and consistency as strength of accountability pres-
sure increases.
2.7.2. Reviewers with known preferences
Peecher (1996) illustrates that known preferences of reviewers
in the audit review process in?uence evidence sought and
weighted by preparers. That is, preparers not only increase effort
or attention in collecting evidence and evaluating outcomes, but
they direct their effort or attention more toward evidence they
believe is consistent with reviewer preferences. Therefore, being
accountable to reviewers with known preferences may reduce or
increase errors and biases, and in turn audit quality, depending on
the nature of the preferences (Rich et al., 1997a). Both Peecher
(1996) and Turner (2001) provide reviewer preferences for less
skepticism or more skepticism and ?nd that preparers generate
error hypotheses or search for evidence consistent with reviewer
preferences, respectively. Peecher ?nds the former effect is miti-
gated when client integrity is low, while Turner ?nds the latter
effect is more pronounced when auditors make explicit decisions
versus when they provide their beliefs for the reviewer to take
action. Wilks (2002) ?nds preparers make judgments consistent
with reviewer preferences (for pessimism versus optimism), and
this effect is even more pronounced when preferences are
communicated before (versus after) preparers evaluate evidence,
because these preparers also subconsciously evaluate the evi-
dence consistent with reviewer preferences. Peecher et al. (2010)
?nd preparers evaluate evidence consistent with reviewer pref-
erences for less skepticism, when those preferences are commu-
nicated to them (versus not communicated at all), and as noted
earlier these preparers' inputs are incorporated into reviewers'
?nal judgments. Overall, these ?ndings suggest preparers are
more (less) skeptical when a reviewer prefers an unfavorable
(favorable) client view.
Peecher et al. (2013) suggest a different preference and
examine whether preparers who are accountable for their
judgments, judgment process, or some combination thereof,
likely in?uence innovation and quality of the audit process and
procedures. Consistent with this notion, Messier, Quick, and
Vandervelde (2014) ?nd that preparers are less likely to simply
follow what was done in the prior year when they are required
(versus not) to justify their decision to a partner who prefers to
focus on the process behind their decision. Lambert and Agoglia
(2011) examine how preparers close out review notes by having
the preparers interact with a pre-programmed client, as well as
receive and respond to reviewer comments. They ?nd that pre-
parers provide greater effort and ?nd more seeded errors when
the review notes are framed to encourage appropriate conclu-
sions versus defensible documentation, but only when the re-
view is timely. Gold, Gronewold, and Salterio (2014) examine
how preparers respond to an audit ?rm that has a more versus
less open climate to learning from errors, when errors are
discovered after the normal review process is complete. They
?nd a more open climate increases reporting of self-made me-
chanical (but not conceptual) errors and both types of peer-made
errors.
Related research motivated by Rich et al. (1997a) on styliza-
tion/persuasion attempts examines the nature of justi?cations
documented by preparers within the audit review process.
Shankar and Tan (2006) ?nd that preparers try to bolster their
position, as needed, when they are required to determine their
own preference prior to receiving the reviewer preference. Spe-
ci?cally, upon learning of their reviewers' preference after form-
ing their own preference, preparers document more (less)
evidence favoring their preference and a greater (lesser) spectrum
of issues when their reviewer's preference is dissimilar (similar),
but only for highly knowledgeable preparers (Shankar & Tan,
2006).
5
Agoglia et al. (2003) examine how preparers stylize
when reviewer preferences are unknown but a preference for
how to structure their written justi?cation is expressed. They ?nd
that preparers, asked to assess a client's overall control environ-
ment, have more positive judgments, and document more total
evidence and judgment-consistent evidence, when they must
summarize balanced (i.e., supporting and contradictory) evidence
and assess components of the control environment than when
they do not assess components but either summarize balanced
evidence, or summarize primarily supporting evidence. Thus, in a
typical setting where there is more supporting than contradictory
evidence, the more detail required, the more supporting (and
total) evidence will be documented (Agoglia et al., 2003).
2.7.3. Effect of alternative formats of the review process on the
preparer
Despite suggestions from Tan and Kao (1999) that variations in
the review process format likely in?uence preparers, few papers
have examined this issue. Variation in the type of review (or
accountability) likely in?uences not only effort duration and in-
tensity but also effort direction and the goals preparers focus on
(e.g., ef?ciency versus effectiveness). To date, only three studies
examine review format. Brazel et al. (2004) and Agoglia et al.
(2009) compare preparer judgments when they anticipate a face-
to-face interacting review versus when they anticipate an elec-
tronic review. They ?nd that preparers who anticipate the more
interactive review spend more time preparing documentation,
anchor less on prior year judgments, are more concerned with
effectiveness, and make more effective judgments (Brazel et al.,
2004), and also have higher quality documentation (Agoglia et al.,
2009). Payne et al. (2010) compare face-to-face interacting review
with written review and ?nd that the former leads to a focus on
more cognitively effortful audit procedures resulting in better
identi?cation of fraud red ?ags. This combination of results in-
dicates that review format in?uences each of effort duration, in-
tensity, and direction.
2.8. Future research on the review process
The need for future research on the review process is still
important given the criticisms by regulators (ASIC, 2014; PCAOB,
2013) and the resources devoted by audit ?rms to the review
process. At present, the audit literature on the review process ex-
amines situations where the review process has been effective and
this effectiveness appears to depend on the nature of the task and
errors, how the review team is constituted, and a range of envi-
ronmental factors. Given the scarcity of participant resources,
dif?cult choices have to be made on what factors and what com-
bination of factors should be included in future studies of the re-
view process.
Some relevant theoretical insights are provided by Einhorn,
Hogarth, and Klempner (1977) who consider various baseline
models for examining the quality of group performance, de?ned as
how close the group judgment is to the actual value of the item
being predicted. They compare four types of model: random se-
lection of a single individual in the group; equal weighted mean of
group judgments; using the judgments of the best member
(identi?ed in advance); and a weighted average of the members
based on expected expertise. They ?nd that the performance of
5
In addition, high tacit knowledge preparers are more likely to frame the evi-
dence consistent with their preference but only when their reviewer's preference is
similar (Shankar & Tan, 2006).
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 62
these models depends on group size and ‘bias’. Experimentally the
cost of varying group size in auditing research is not likely to be
practical, thus emphasis could be placed on ‘bias’ which Einhorn
et al. divide into systematic bias and random error. Systematic bias
in individual judgments is potentially affected by both the indi-
vidual and the task being performed, and such a consideration
would be valuable in designing studies on the review process.
We also suggest that progress can be made by describing the
underlying cognitive processes in enough detail to allow re-
searchers to predict the speci?c bene?ts of the review process and
under what circumstances such bene?ts will be increased or
decreased (Libby & Trotman, 1993) and the interaction between
cognitive factors and environmental factors (Libby & Luft, 1993).
Subsequently, Gibbins and Trotman (2002) use retrospective recall
from managers of their actual review engagements to develop a
model of ?ve components of the audit ?le review process that will
affect reviewer behavior: manager's expectations about the client,
the preparer, the partner, the manager's own approach to review,
and the circumstances faced by the reviewer (e.g., deadline pres-
sure). Factors included in this model could be incorporated in
future research to progress the research not only on when potential
bene?ts are realized but also potential limitations of the review
process (e.g., Frank & Hoffman, 2015).
Research on face-to-face interaction between auditors (e.g.,
Schultz & Reckers, 1981; Solomon, 1982), particularly between
reviewer and reviewee (e.g., Ismail & Trotman, 1995; Trotman,
1985), has become rare (except in the brainstorming literature
discussed subsequently). This interaction is important as it permits
examining one of the key factors in?uencing group performance.
Such interaction can be a rich source of information about process
gains but it does raise additional complications in the design of the
study fromboth a control perspective and in obtaining participants.
In addition to the hierarchical review process discussed above,
auditing standards (e.g., AS 7; ISA 220) require engagement quality
control reviews (EQR) whereby a second partner carries out a re-
view of the audit (previously called concurring partner reviews)
(IAASB, 2008; PCAOB, 2009). The conduct of EQR has been criticized
by inspectors, including suggestions for greater involvement of the
second partner (PCAOB, 2013). There are important questions to
address on the nature, extent, and timing of the EQR in the audit
process. Future research on EQR would bene?t from a better un-
derstanding of how this process should be structured and how the
type of review interacts with various environmental factors. Epps
and Messier (2007) provide insights on audit ?rm practices in
this area and Messier, Kozloski, and Kotchetova-Kozloski (2010)
provide details of SEC enforcement actions including alleged vio-
lations of due care related to lack of professional skepticism. This
material provides the basis for rich research instruments to inves-
tigate the judgments made by the EQR.
Opportunities for signi?cant incremental contribution on the
effects on the preparer of the review process will likely come from
examining how the type of accountability in?uences auditor
judgments, such as recent research on reviewer preferences for a
focus on process, outcome, documentation (e.g., Lambert &Agoglia,
2011; Messier et al., 2014; Peecher et al., 2013), or varying error
climates (Gold et al., 2014). For example, while audit ?rm efforts to
establish an appropriate tone at the top in response to regulator
criticism about the review process may succeed in in?uencing its
professionals, this success could depend on the message and howit
is conveyed. In addition, research can consider looking beyond
preferences for more versus less skepticism, but rather at the type
of skepticism including being outwardly skeptical of evidence
versus inwardly skeptical of one's own judgments (Bell, Peecher, &
Solomon, 2005), and whether the bene?ts of inward skepticism
over and above outward skepticism depend on the judgments
being made (Grenier, 2014; Harding & Trotman, 2014).
A major thrust of Rich et al. (1997a) was to motivate research
on how preparers may engage in strategic, stylized behavior
depending upon different environmental factors or formats in the
review process. However, research on review format has been very
limited and opportunities exist in the intersection of stylization
and review format. For example, increased use of electronic
workpapers, email communication, and electronic review may
increase/reduce preparers' stylization. Since Rich et al. there have
been signi?cant changes in the production of audit working pa-
pers such as greater automation and more structured computer
assisted selection of audit tests which may reduce opportunities
for stylization of documentation and work performed, respec-
tively. Still, the use of written memos to document and defend
auditors' positions remains a critical part of the audit and styli-
zation therein is likely to persist. Future research can examine
how the continued practice of written justi?cations and the evo-
lution of electronic workpapers and communication in?uence the
opportunities for, and results of, preparer stylization efforts and
the review process.
While the above discussion relates to preparers stylizing the
workpapers for review, it is suggested that the likelihood of an
inspection will affect the review carried out by a partner. However,
such inspections have been suggested to lead to defensive auditing
(Peecher et al., 2013) and likely reduced levels of professional
skepticism due to shifting focus to documentation and checklists
(Westermann, Cohen, & Trompeter, 2014). Research examining the
effect of different forms of the review process and expectations
around the inspection process on the professional skepticism of
both preparers and reviewers (particularly at the partner level) is
warranted.
3. Brainstorming
A newline of group research in auditing evolved with Carpenter
(2007) to examine fraud brainstorming sessions conducted as part
of the audit planning process. The issue became important to the
audit literature with the revision of auditing standards related to
the auditor's responsibility to consider fraud in an audit of ?nancial
statements, which require a ‘discussion’ (ISA 240) or ‘brainstorming
session’ (AU 316) among audit team members including how and
when a client's ?nancial statements may be susceptible to material
misstatement due to fraud. The bene?ts of brainstorming have
been proclaimed since Osborn (1957) and there exists decades of
social psychology and organizational behavior research comparing
alternative brainstorming formats and the reasons for process gains
and losses during brainstorming sessions (see review by Stroebe,
Nijstad, & Rietzschel, 2010).
Hammersley (2011) sets out a model of the fraud planning tasks,
which includes four stages of the brainstorming process: fraud risk
factor identi?cation, fraud hypothesis generation, fraud risk
assessment, and audit program modi?cation.
6
Hammersley de-
scribes a wide range of relevant audit research related to each of the
factors in her model. We restrict our discussion to audit research
where participants are involved in the brainstorming task. We
divide this research into face-to-face brainstorming and electronic
brainstorming. The classi?cation of these relevant studies by task
type is provided in Fig. 2.
6
Fraud risk factor identi?cation includes incentives, opportunities and the ability
to rationalize fraud. Fraud hypothesis generation refers to the process of identifying
plausible explanations of how fraud can occur given the risk factors that exist,
which is followed by preliminary risk assessments of the likelihood of fraud
existing (Hammersley, 2011). Finally, audit program modi?cations refer to changes
made to audit programs based on the above factors.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 63
3.1. Face-to-face interacting versus nominal brainstorming
While auditing standards do not specify a formof brainstorming
group, the most commonly used method in practice is a face-to-
face interacting team (Bellovary & Johnstone, 2007; Brazel,
Carpenter, & Jenkins, 2010). Thus, one important question to
address is how the interacting brainstorming team compares to
other possible alternatives including a baseline measure such as a
nominal group. With interacting brainstorming groups there is
discussion between participants whereas for nominal brain-
storming groups, participants complete the task as individuals
without interaction and their unique ideas are combined by the
researcher after completion of the task.
The early group brainstorming literature in psychology advo-
cates interacting group brainstorming as a means of increasing
creativity in idea generation (Osborn, 1957). The bene?ts of group
brainstorming are based on obtaining a large pool of ideas and,
through cognitive stimulation, generating new ideas that combine
and/or change previously generated ideas. However, numerous
studies in the psychology literature ?nd that nominal groups
outperform interacting groups in both the quantity and quality of
ideas generated because of a range of process losses (see Kerr &
Tindale, 2004, for a review). The key process losses identi?ed are
production blocking, evaluation apprehension and social loa?ng.
7
However, the bulk of the brainstorming studies in psychology
have been conducted using undergraduate college students in ad-
hoc groups on hypothetical problems, and have been heavily crit-
icized for lacking realism (Isaksen, 1998).
8
In comparison, auditing standards require brainstorming to be
carried out by auditors with adequate experience and training, the
groups are generally hierarchical, and identifying fraud risks and
generating potential fraud hypotheses are familiar, but complex,
tasks because while they are conducted on every audit they
require participants to draw on their accumulated but limited
knowledge of previous fraud. Despite these important differences
in the audit context, the only fraud brainstorming studies that
have compared face-to-face interacting brainstorming with nom-
inal brainstorming (Carpenter, 2007; Lynch, Murthy, & Engle,
2009) ?nd results that are generally consistent with the
psychology literature. Carpenter (2007) ?nds that nominal groups
generate more unique fraud hypotheses than similar size inter-
acting groups, but that the interacting groups generate more ‘high
quality’ frauds (i.e., those consistent with frauds seeded in the
case). The nominal and interacting groups are manipulated
within-subjects and Carpenter ?nds that while not all identi?ed
frauds by nominal groups are included in the interacting groups'
fraud hypotheses, the group brainstorming generates additional
fraud ideas. This is an example of cognitive stimulation. For a
simpler task involving fraud risk identi?cation, Lynch et al. (2009)
also ?nd that nominal groups generate more ideas than face-to-
face interacting groups.
3.2. Interacting face-to-face brainstorming compared to alternative
brainstorming formats
Here we consider research that examines four alternatives to
the traditional face-to-face brainstorming groups used in practice:
brainstorming guidelines, premortem instructions, strategic
reasoning, and use of prompts. Trotman, Simnett, and Khalifa
(2009) compare the performance of three interacting group
treatments provided with either standard interacting group in-
structions (similar to Carpenter, 2007), brainstorming guidelines
or pre-mortem instructions. They develop their brainstorming
guidelines in line with Osborn's (1957) four guidelines that aim to
defer judgment until after the idea-generation phase.
9
They
develop their pre-mortem instructions based on Klein (1999) and
simulate backward thinking (i.e., putting an individual into the
future and asking them to think backwards) by instructing audi-
tors to imagine it is months after the audit is completed and
?nancial reporting fraud has been discovered for the client, before
asking them to generate the potential frauds that could have
occurred. Trotman et al. ?nd the brainstorming groups provided
with guidelines or pre-mortem instructions identify a greater
quantity of potential frauds, and more frauds matching the list of
frauds identi?ed as most important by a panel of experts, than the
interacting group receiving standard interacting group
instructions.
Hoffman and Zimbelman (2009) suggest that encouraging
strategic reasoning is one way of generating more creative fraud
detection procedures. They describe strategic reasoning in auditing
Fig. 2. Summary of previous fraud brainstorming experiments addressing nominal or interacting teams.
7
Production blocking occurs when only one group member can communicate at
once, which can result in forgetting ideas, suppressing ideas, and remembering and
listening, rather than creating ideas. Evaluation apprehension refers to individuals
withholding ideas due to fear of negative evaluation of their ideas. Social loa?ng or
free riding refers to individuals relying on others to complete the task for such
reasons as perception that their inputs are unidenti?able or dispensable.
8
Isaksen (1998, p. 21) states that “most studies used untrained students working
in arti?cial groups on relatively meaningless tasks with little or no concern for
measuring anything but sheer quantity.”
9
Osborn's (1957, pp. 300e301) four guidelines to defer judgment are: criticism
and adverse judgment of ideas must be withheld until later; wilder and unusual
ideas are desired; quantity is desired as a greater number of ideas should increase
the likelihood of useful ideas; and combination and improvement are sought to
facilitate participants turning the ideas of others into better ideas or combining two
or more ideas into another idea.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 64
as answering the following three questions: What potential frauds
may have occurred? How could management conceal the potential
frauds from the standard audit plan? How could the plan be
modi?ed to detect the concealed frauds? (Hoffman & Zimbelman,
2009). They ?nd that both interacting brainstorming groups
(without guidelines) and individuals engaged in strategic reasoning
provide more effective modi?cations to standard audit processes
than individuals without interventions, but that the combination of
the two interventions does not outperform either of the two in-
terventions used alone.
Finally, two studies examine different forms of prompting
during brainstorming. Lynch et al. (2009) provide content facili-
tation by way of ?ve fraud prompts related to incentives/pressure,
opportunities, attitudes/rationalizations, revenue recognition, and
management override. For both face-to-face and electronic
brainstorming groups, the automatic content facilitation leads to a
larger number of relevant fraud risks being identi?ed. Unlike the
studies discussed above, participants in Hammersley et al. (2010)
do not complete a brainstorming session, but instead list three
risks they would raise during a brainstorming session and watch a
video of the fraud brainstorming session. They ?nd that, when not
primed about the fraud risks identi?ed, auditors with more spe-
ci?c documentation provide higher fraud risk assessments and
request more evidence. However, importantly they show that
decision aids such as priming are not always bene?cial. Partici-
pants who are primed also increase fraud risk assessments and
evidence requests when they receive summary information, but
decrease such judgments when they receive speci?c documenta-
tion because priming makes the client speci?c risks less typical
(Hammersley et al., 2010).
In sum, while unstructured face-to-face brainstorming is the
most common method used by audit ?rms (Brazel et al., 2010),
this method is outperformed with respect to the quantity of fraud
risks identi?ed and fraud hypotheses generated by any of the
structured methods examined above (Lynch et al., 2009; Trotman
et al., 2009). With respect to modifying audit programs, unstruc-
tured face-to-face brainstorming is equaled by prompting in-
dividuals to use strategic reasoning (Hoffman & Zimbelman,
2009). There is mixed evidence with respect to the effects on
the quality of the fraud hypotheses generated (Carpenter, 2007;
Trotman et al., 2009) which is likely partly due to different mea-
sures of ‘quality’ and differences in research design including
whether the different methods are manipulated within- or
between-subjects. Overall, it is suggested that there is consider-
able potential for improvement in brainstorming by adding alter-
native facilitation methods to the unstructured face-to-face
interaction used in practice.
3.3. Electronic brainstorming
Electronic brainstorming has been a popular brainstorming
technique in a range of business ?elds (Pinsonneault, Barki,
Gallupe, & Hoppen, 1999). Previous studies in psychology and
information systems ?nd that, for idea generation tasks, electronic
brainstorming outperforms face-to-face brainstorming (Dennis &
Valacich, 1993; Guzzo & Dickson, 1996; see Kerr & Tindale,
2004, for a review). This has led to suggestions that the brain-
storming literature should move on from comparing face-to-face
brainstorming to electronic brainstorming, and focus on what
forms of electronic brainstorming are most appropriate (Guzzo &
Dickson, 1996). The difference between face-to-face brain-
storming and electronic brainstorming is that the latter uses
computer software to allow individual group members to indi-
vidually input ideas without interruption, which reduces process
losses because of lower production blocking and evaluation
apprehension (particularly when the ideas are provided anony-
mously) (Kerr & Tindale, 2004).
10
This superiority of electronic brainstorming has been supported
by Lynch et al. (2009), the only audit study to compare face-to-face
and electronic brainstorming; they ?nd that both forms of elec-
tronic brainstorming (nominal and interacting) outperform face-
to-face interacting brainstorming.
11
However, the comparison of
electronic interacting and electronic nominal brainstorming is less
clear. Using undergraduate students as participants, Lynch et al.
(2009) ?nd no difference in fraud risk identi?cation between
electronic nominal and interacting brainstorming groups, but note
that their results could be driven by low power from small sample
sizes. Also, use of undergraduate students is likely to result in a low
range of differential expertise in their teams, and would bias
against ?nding signi?cant results.
As electronic brainstorming minimizes production blocking and
evaluation apprehension (Kerr & Tindale, 2004; Pinsonneault et al.,
1999), it allowed Chen, Trotman, and Zhou (2015) to focus on a third
potential process loss (social loa?ng) to better address the under-
lying mechanisms of brainstorming effectiveness and examine
when and why interacting groups do (or do not) outperform
nominal groups in electronic brainstorming. Chen et al. ?nd nom-
inal teams generate a signi?cantly larger number of fraud risk
factors and fraud hypotheses, and generate higher quality fraud
hypotheses (i.e., hypotheses consistent with those identi?ed by
fraud experts), than interacting teams. They also provide evidence
that social loa?ng of seniors drives the differences between nom-
inal and interacting teams in the fraud hypothesis generation task.
3.4. Future research on brainstorming
A key focus of future brainstorming research should be potential
process gains and losses that exist with alternative methods of
fraud brainstorming, including methods presently used in practice
and those that could be adopted. The bulk of the ?ndings in the
audit literature comparing interacting brainstorming to nominal
group brainstorming show that the latter performs better, and no
?ndings demonstrate that interacting brainstorming, as generally
used in practice, outperforms any of the alternatives examined in
the audit literature based on quantity of frauds generated (Chen
et al., 2015; Hammersley, 2011). Thus an important focus of
future research is how to reduce process losses and/or increase
cognitive stimulation in interacting groups.
At this stage we know very little about the nature of process
losses in the audit environment, under what circumstances they are
likely to exist or how they can be reduced. Opportunities arise to
consider how the various process losses differ between audit and
psychology studies, and how they are affected by the audit task
completed and the form of brainstorming involved. Knowledge of
these process losses can be used to develop improvements in group
brainstorming (Kerr & Tindale, 2004). Facilitator training is one
method that has been used to minimize production blocking and
evaluation apprehension (Oxley, Dzindolet, & Paulus, 1996; Paulus,
Nakui, Paulus, &Brown, 2006). These methods have potential given
the role of audit partners and managers in the facilitation of fraud
10
The degree of anonymity can depend on the research setting. For example,
when data is collected at the of?ces of an audit ?rm during staff training or at
another meeting, participants will know who else is in the room including some
details of their background. If the data is collected in small groups the level of
anonymity decreases.
11
In Lynch et al. (2009), the face-to-face group interacts verbally with fraud risk
factors being typed on the laptop by one of the team members. The electronic
interacting group members could input ideas simultaneously and they could see
the ideas of other group members in real time.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 65
brainstorming sessions (Dennis & Johnstone, 2014). For example,
additional guidelines used by the facilitator have included staying
focused on the task, not criticizing, and returning to previous cat-
egories (Oxley et al., 1996; Paulus et al., 2006). However, it is
important to consider the multi-period effects of these structured
approaches. Trotman et al. (2009) ?nd that brainstorming groups,
given Osborn's (1957) brainstorming guidelines, generate a greater
quantity of potential frauds and higher quality potential frauds.
However, whether similar effects would exist across multi-periods
is unknown, as the effects of the brainstorming guidelines may
decrease over time, as auditors and audit teams become more
familiar with the guidelines.
In addition, it is necessary to better understand the circum-
stances where brainstorming can bene?t from cognitive stimula-
tion. In particular, what are the unique attributes of fraud
brainstorming in auditing that are likely to allow cognitive stimu-
lation and under what circumstances does it occur? As noted
earlier, Carpenter (2007) ?nds process gains via cognitive stimu-
lation when group members completed the task as individuals
prior to group interaction. This also suggests there are potential
bene?ts from auditors in practice performing the task as in-
dividuals as an input to the subsequent brainstorming session
(Brazel et al., 2010; KPMG, 2011).
Another possibility for enhancing cognitive stimulation, and
reducing process losses, is open, round robin brainstorming (see
Beasley & Jenkins, 2003; Kerr & Tindale, 2004) which involves
individual brainstorming and then group members take turns
reading their ideas to the rest of the group one at a time. They
then have an opportunity to add additional ideas which may
have been generated while listening to others. This method is
based on nominal group techniques (NGT) (van de Ven &
Delbecq, 1971) developed in the management literature, which
can include additional steps such as discussing each idea for
clari?cation and then ranking ideas. The potential bene?t of
these methods is that they may reduce process losses by
lowering production blocking (via nominal group processes) and
social loa?ng (as all participants present their ideas). Listening to
others is also likely to assist with cognitive stimulation, partic-
ularly when auditors know they have the opportunity to present
additional ideas. Other methods that add structure to the task
should be investigated. For example, in the management ac-
counting literature, Chen, Williamson, and Zhou (2012), apply a
‘parent-child’ idea structure to brainstorming, where one group
member suggests an initial idea (‘parent-idea’) and then other
group members build on that idea with subsequent ideas (‘child-
ideas’).
While there is considerable support in the psychology literature
for electronic brainstorming, it has only been compared to face-to-
face brainstorming in the accounting literature by Lynch et al.
(2009) using student surrogates in a fraud risk identi?cation task.
This leaves many unanswered questions including under what
circumstances electronic brainstorming can outperform face-to-
face brainstorming and what effects do alternative electronic
brainstorming structures have on cognitive stimulation and process
losses. For example, with electronic brainstorming, the researcher
has access to the order of ideas that are typed in and the timing of if/
when these ideas are read by other participants. Such information
provides a much richer source to examine the level of cognitive
stimulation.
The importance of task differences has long been recognized in
the group literatures in both psychology (Guzzo & Dickson, 1996;
Herold, 1979) and accounting (Ashton, 1990; Bonner, 1994;
Trotman, 2005). In fact, Herold's (1979) observations that the se-
lection of tasks in group performance experiments ‘has been any-
thing but systematic’ and ‘may be a major reason for the
inconsistency’ in the results of group performance research also
apply to the audit JDM research on groups. This is not surprising
given the dif?culty of obtaining participants (decreases the number
of potential treatments) and the limited time available for each
experiment (decreases the opportunities for within subject ma-
nipulations). We suggest that without adequate consideration of
task differences between studies, there is likely to continue to be
unexplained differences in the results of brainstorming studies and
more generally group judgments in auditing.
4. Consultation within ?rms
Consultations involve interactions, or anticipated interactions,
among audit group members whereby one auditor seeks advice or
feedback from another auditor(s) on an appropriate course of ac-
tion. These consultations can be a source of justi?cation for audi-
tors' actions (Emby & Gibbins, 1988) and may help justify
undertaking a contemplated course of action or document the
decision process to help defend a position within the audit team or
to inspectors. Consultations can range from binding (i.e., formal) to
advisory (i.e., informal) and accounting ?rms have emphasized the
importance of both formal and informal consultation (Ranzilla,
Chevalier, Herrmann, Glover, & Prawitt, 2011). In Fig. 3 we divide
this consulting research into the perceived justi?ability of consul-
ting advice, and the willingness to followversus willingness to seek
consulting advice.
4.1. Perceived justi?ability of consulting advice
Kennedy, Kleinmuntz, and Peecher (1997) ?nd auditors perceive
another auditor's decision to be more justi?able when s/he infor-
mally consults with a knowledgeable auditor (versus not), and most
justi?able when the advice con?rms the consulting auditor's initial
position. Moreover, Kennedy et al. show that consulting, yet not
following discon?rming advice, is perceived as less justi?able than
following discon?rming advice, but more justi?able than not
consulting at all.
4.2. Willingness to follow consulting advice
Kadous, Leiby, and Peecher (2013) also consider informal con-
sultations, by examining auditors' willingness to follow contrary
advice from a peer. Consistent with previous literature, they ?nd
that auditors tend to revise in line with advice regardless of
treatment; however, non-specialist and specialist auditors react
differently. Speci?cally, they ?nd non-specialists use a ‘trust heu-
ristic’ and view better justi?ed advice as higher (similar) quality
and (do not) weight it more heavily than worse justi?ed advice
when the informal advice comes from a peer with whom the
auditor shares a weaker (stronger) social bond. Specialists, how-
ever, do distinguish between the quality of better versus worse
justi?ed advice when the advisor social bond is stronger, but
quality fails to in?uence weighting; that is, specialists discount
better advice from close advisors (Kadous et al., 2013).
Salterio (1996) and Salterio and Koonce (1997) examine formal
consultation with audit ?rms' research or technical units. These
units are often consulted about accounting issues and provide ev-
idence of precedents in practice, but may not provide a formal
recommendation. Salterio ?nds that auditors tend to follow the
direction suggested in consultation precedents, and this occurs to a
greater extent when precedents are more similar to the issue at
hand, but importantly, auditors are no less likely to followwhen the
precedents discon?rm the client preference. However, Salterio and
Koonce ?nd that while auditors exhibit such behavior when all
precedents provided by a consultation unit point in the same
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 66
direction, when precedents con?ict, auditors tend to follow an
aggressive client preference if it is made known.
Ng and Shankar (2010) demonstrate that before or after formal
consultation, auditors are more likely to accept the client prefer-
ence when the client's justi?cation is stronger. However, after (but
not before) consulting, this effect is diminished by the presence of a
quality assessment standard (QAS) designed to encourage auditors
to assess appropriateness of the client preference prior to assessing
its acceptability, but only if the consultation unit also explicitly
advises that the client should adopt an alternative method. Similar
to the ?ndings of Salterio and Koonce (1997), even if consultation
informs auditors that the client preference is acceptable but less
appropriate than an alternative method, auditors tend to permit
the client preference if they do not receive a de?nitive answer as to
which alternative to follow.
4.3. Willingness to seek consulting advice
Research on auditors' willingness to seek formal consulting
advice has been examined froma fraud perspective, where ?ndings
indicate that auditors are more willing to consult with fraud ex-
perts when fraud risk is higher (Asare & Wright, 2004; Gold et al.,
2012). However, Gold et al. (2012) ?nd this effect when the
requirement to consult is strict (i.e., mandatory and binding) but
not when the requirement is lenient (i.e., advisory and non-
binding), presumably because failing to consult despite high
fraud risk is less justi?able when requirements are strict. This is
consistent with the ?ndings in Kennedy et al. (1997) that consul-
tation increases perceptions of justi?ability. Gold et al. also ?nd that
an auditor's willingness to consult increases when deadline pres-
sure is higher, and to a greater extent under stricter consultation
requirements.
Some current working papers illustrate evolving trends in this
research, as they focus on more informal settings and judgments
and choices about the extent of consultation. For example, Nelson,
Proell, and Randel (2014) show that auditors are more willing to
consult with audit team superiors about audit issues, in particular
issues that increase effectiveness but also audit costs, when these
superiors have a more team-oriented leadership style. Schaefer
(2014) shows that social costs (e.g., reputation concerns) can
decrease auditors' willingness to informally consult, induce themto
consult with peers rather than superiors, or provide a less balanced
set of facts to an advisor. Further, Bauer, Hillison, Peecher, and
Pomeroy (2015) demonstrate that, without a prompt to consider
fraud, auditors make better changes to an audit plan for responding
to a high fraud risk when informally advising a peer than when
deciding for oneself, due to different focal concerns in their
respective mental perspectives.
4.4. Future research on consultation
Since the emergence of audit research on consultations (and
advice), this area has received limited empirical attention and of-
fers many opportunities for future research. Two questions of
interest are what stopping rules auditors use after consulting; and
are they more likely to stop consulting and reach a decision after
receiving con?rming or discon?rming advice? These are potentially
important questions given that con?rming advice is seen as more
justi?able (Kennedy et al., 1997) and auditors have been shown to
seek con?rming evidence (e.g., Kaplan & Reckers, 1989; Turner,
2001). Further, prior research utilizes experimental designs that
often force auditors to make judgments after consultation evidence
is provided, but whether or not auditors would choose to conclude
at that point is unknown. In addition, audit research has only begun
to explore when auditors seek consultation and from whom they
seek it (Gold et al., 2012; Schaefer, 2014), with many task, person,
and environmental variables of potential interest to these
questions.
Within the limited extant consultation research, the advisor role
has received even less attention and it is unclear what factors affect
the quality of the advice given. Further, given the frequency of
informal consultation, both the advisor role and the formality of
consultation offer much potential for future research, particularly
in light of the dynamic and endogenous nature of the consultation
process. For example, advisors often depend on the information or
evidence provided by the advice seeker and likely receive or have
access to more evidence when the process is formal. How the
extent and nature of evidence available in?uences advisors' will-
ingness to indicate a de?nitive conclusion, their likelihood of
requesting or seeking more evidence, and their ultimate decision
can be explored in future research, as well as whether such de-
cisions differ for formal versus informal consultation. In addition,
those consulted may form views about the advisee, based on pre-
vious interactions, including whether they provide a balanced view
of client facts and how they act on the advice given. These issues
can be addressed by manipulating the nature of prior interactions
or by using multi-period experiments.
While formal advice is likely to be handled through written
communication (e.g., emails or completion of standard forms),
informal advice could be face-to-face, via telephone or via email. As
discussed in our earlier sections on the review process and brain-
storming, the mode of communication is likely to affect the content
of the consulting advice given and interact with other environ-
mental factors, and research in this area is warranted.
5. New directions and future research
In our above coverage we have included future research on the
key areas we have addressed: the review process, brainstorming,
and consultation within ?rms. Here we suggest potential changes
to the research agenda by looking at other group related topics that
we believe will progress research in auditing. We consider three
areas of within-?rm group interaction (audit teams in context;
shared mental models; and audit team diversity, including sus-
tainability assurance). We also recognize that auditors frequently
interact outside the audit ?rm, and consider audit committee in-
teractions with auditors as an important example of opportunities
in this area. Research in some areas has already begun and other
Fig. 3. Summary of consultation experiments.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 67
areas are yet to be investigated.
5.1. Audit teams in context
Reviews of the group psychology literature by Guzzo and
Dickson (1996) and Ilgen, Hollenbeck, Johnson, and Jundt (2005)
focus on team performance in organizational contexts. Guzzo and
Dickson emphasize teameffectiveness, which they de?ne as group-
produced outputs (e.g., quality, quantity, speed, etc.), the conse-
quences a group has for its members (e.g., job satisfaction), and the
enhancement of a team's effectiveness in the future (e.g., future
performance). They also consider studies involving interventions
aimed at improving team performance, including whether these
interventions are targeted at individual team members (e.g., the
skills of team members) or at teams as performing units (e.g., team
development). The examination of teams within their organiza-
tional context has generated some signi?cant ?ndings in psychol-
ogy research (see Guzzo & Dickson, 1996; Klein, 1999). Yet, while
audit researchers are interested in audit teams, our experimental
designs generally do not include auditors who are members of the
same team and the experiments rarely consider the effect on the
team or the individuals in the future. However, in a tax setting,
Andiola and Bedard (2015) examine the effects of the review pro-
cess on job satisfaction and turnover intentions.
One context, examined in Guzzo and Dickson (1996), that con-
tains multiple team features with interesting similarities to audit-
ing is the ‘crew resource management’ (CRM) literature which
investigates ?ight crew behavior.
12
One element of CRM research
that is pertinent for audit teams is the use of simulations. This
aspect of CRM research has expanded beyond aviation to other
team-based contexts, including management, education, crisis
management, and medicine. Simulations enable team training for
events that are high-risk, but low frequency, within virtual envi-
ronments that are essentially risk-free (Haerkens, Jenkins, & van
der Hoeven, 2012). These simulated environments allow teams to
experiment and make mistakes, so as to learn valuable lessons
(Wood, Beckmann, & Birney, 2009). In an audit team setting, sim-
ulations can be used to examine the effects of potential frauds on
the ?nancial statements where the resulting insights may lead to
improvements in fraud hypothesis generation and testing.
5.2. Shared mental models in audit teams
Mental models comprise structured knowledge, data on pat-
terns of cues, and rules that causally and/or associatively link this
knowledge with these cues (Bell, Peecher, & Solomon, 2002, p. 14).
One way in which mental models are used is to run mental simu-
lations, whereby a set of initial conditions are provided and the
future trajectory of a systemis then imagined (Markman &Gentner,
2001). Klein (1999) suggests three criteria for evaluating a mental
simulation: coherence (whether it makes sense); applicability
(whether it provides what the decision maker needs); and
completeness (whether it includes too much or too little informa-
tion). In the audit JDM literature, a small body of research has
examined individual mental models developed by auditors (e.g.,
Brewster, 2011; Choo & Trotman, 1991; Hammersley, 2006;
Knechel, Salterio, & Kotchetova-Kozloski, 2010), as well as their
use within audit teams for running mental simulations (Chen et al.,
2015; Trotman et al., 2009).
Integral to effective audit team functioning is the development
of a shared mental model of the audit task, the auditee, and the
audit team itself (including team members' knowledge, attitudes,
and beliefs). A shared mental model is de?ned as “organized un-
derstanding of relevant knowledge that is shared by team mem-
bers” (Mohammed & Dumville, 2001, p. 89). This group level
cognitive construct has been investigated in the psychology and
organizational behavior literatures for more than two decades (see
Cannon-Bowers, Salas, & Converse, 1993), and its importance in an
audit context is discussed by Bell et al. (2002).
Three speci?c areas of research related to the development of
shared mental models are considered. First, examining the extent
to which auditors' shared mental models change over time, in
response to post-task feedback (Cannon-Bowers et al., 1993), is
potentially of interest, given the importance of mental models in
the brainstorming literature (Hammersley, 2011). Second, the
cross-training of teams, where each team member is trained in the
duties of teammates with the aimof better equipping individuals to
anticipate the needs of other group members, has been shown to
lead to greater development of shared mental models and
enhanced team performance (Ilgen et al., 2005; Marks, Sabella,
Burke, & Zaccaro, 2002; Mathieu, Heffner, Goodwin, Salas, &
Cannon-Bowers, 2000). In auditing, cross-training could be used
to better informjunior members of their supervisors' roles and may
help the audit team work with, and integrate the work of, spe-
cialists (e.g., tax, valuation, IT) (Bauer & Estep, 2014). Third, lead-
ership training can be used to foster shared mental models (Marks,
Zaccaro, & Mathieu, 2000). In an audit setting, tone at the top is a
key aspect of quality control with important implications for team
performance (Dennis & Johnstone, 2014). How leadership training
on fostering shared mental models is best operationalized (and the
potential bene?ts fromdoing so) in an audit setting warrants future
experimental research.
In addition, shared mental models can lead to excessive con-
formity in the way group members think (Cannon-Bowers et al.,
1993). Exploring factors which foster positive versus negative ef-
fects of mental models within the audit setting is likely to be fruitful
given the potentially moderating roles of task complexity, knowl-
edge and expertise diversity, and group norms of issue resolution in
mental model research (Bantel & Jackson, 1989; Kellermanns,
Floyd, Pearson, & Spencer, 2008).
5.3. Audit team background diversity including sustainability
assurance
The traditional hierarchical review process in auditing consists
of team members with similar educational backgrounds (e.g., ac-
counting), similar training (e.g., CPA programs, in-house training),
and a degree of similarity in experience which increases as one
moves up the hierarchy. However, as new forms of assurance
emerge, including sustainability assurance, these relationships
change. For example, in the assurance of greenhouse gas (GHG)
emissions, multi-disciplinary audit teams are common, which
include team members with an accounting/auditing background
and those from other disciplines including the environmental sci-
ences (Trotman & Trotman, 2015). Such multi-disciplinary teams
have implications for the review process given the different types
of knowledge held by different team members. This diversity can
improve team performance through the addition of increased
knowledge to a particular task (Guzzo & Dickson, 1996; Williams &
O'Reilly, 1998). However, individuals from different backgrounds
may have different frames of reference, use different terminology
and have problem solving styles that may impede the sharing of
12
Similar features include: crews are hierarchical; each crew member contributes
to the decision making with different levels of knowledge; the duration of a ?ight
crew's existence as a unit is limited; training for crews is done at the individual,
rather the team, level; ?ight crews use standard ?ight brie?ngs where the captain
establishes expectations for the crew; ?ight crews face increasing levels of auto-
mation and regulation; CRM research includes external parties, such as air traf?c
controllers (Guzzo & Dickson, 1996).
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 68
information (van Knippenberg &Schippers, 2007). Such differences
have the potential to affect the performance of the review process
compared to other group formats and the potential effectiveness of
the review process may vary depending on whether the auditor
with the non-accounting background is the reviewee or the
reviewer. Ekasingh (2014) notes the importance of such variables as
cross-understanding (Huber & Lewis, 2010) and information elab-
oration (van Knippenberg, De Dreu, & Homan, 2004) and ?nds that
on a GHG assurance brainstorming task, auditors generate more
risks comparing subject matter and suitable criteria, whereas non-
accountants generate more risks related to subject matter
measurement.
In addition to sustainability assurance, audits of ?nancial
statements (including integrated audits of ?nancial statements and
internal controls over ?nancial reporting) involve a number of
specialists with varying backgrounds in education, training, and
experience including tax, valuation, and IT specialists (Bauer &
Estep, 2015). As suggested by Danos, Eichenseher, and Holt
(1989), not only will the audit team and their non-audit special-
ists differ in knowledge but they will also differ in the language or
jargon they use, which may impede performance of the engage-
ment teamas a whole. van der Vegt and Bunderson (2005) ?nd that
success for teams with a diversity of expertise in their membership
depends on the extent to which they share a team identity.
Consistent with this notion, Bauer and Estep (2015) provide evi-
dence that auditors and IT specialists who have a better working
relationship share a stronger team identity and integrate more
productively within the audit process. Team identity (or other so-
cial identities) may provide an interesting avenue for investigation
of these team effects.
5.4. Audit committee interactions with auditors
Audit committee members (ACMs) are accountable for over-
seeing the ?nancial reporting and audit process (CAQ, 2014; FEE,
2014; FRC, 2013), which often includes the asking of probing
questions by ACMs and responses by management and auditors
(Beasley, Carcello, Hermanson, & Neal, 2009; Gendron & B edard,
2006; Kang, 2014). New research has been emerging that exam-
ines these interactions. One line of research has considered the
probing questions asked by ACMs to their auditors (Kang, 2014;
Kang, Trotman, & Trotman, 2015; Pomeroy, 2010). Yet, auditors
also initiate communication with ACMs, and characteristics of
ACMs likely in?uence auditor accountability and how auditors
choose to interact with these ACMs. Despite their design chal-
lenges, experiments that examine the dynamic process among
ACMs, auditors, and managers are likely to provide important
?ndings. As emerging research has shown, these questions can
likely be explored by manipulating ACM behavior and observing
reactions of auditors or management (e.g., Bhattacharjee, Moreno,
& Pyzoha, 2014; Fiolleau, Hoang, & Pomeroy, 2013), and/or
manipulating auditor behavior and observing ACM questioning
behavior (Kang et al., 2015).
Finally, much recent emphasis has been placed on audit quality
indicators (AQIs) (Francis, 2011) with suggestions that AQIs should
be communicated between auditors and ACMs (CAQ, 2014). How-
ever, given the variation in ACMs' level of involvement in the audit,
background, experience, and ?nancial expertise (e.g., Beasley et al.,
2009; Cohen, Krishnamoorthy, & Wright, 2010; Kang et al., 2015) it
important to understand how ACMs will respond to and use these
AQIs.
6. Conclusion
Research on audit groups began over 30 years ago and
impressively it has incorporated the unique aspects of the audit
environment, and evolved with changes in audit regulation and
practice. We have outlined many of the insights from this research
and the potential directions for both existing research topics and
new lines of research on audit groups. To address these questions
audit researchers can play their part with innovative and rigorous
research designs. We believe it is in the interests of society, via
improved capital markets, and the audit ?rms themselves, that
audit practitioners be involved in this research. Given the criticism
that audit ?rms have faced from regulators and inspectors in the
last decade, a step forward is independent research on how best to
design group interactions between auditors within the ?rm and
with other professionals outside the audit ?rm including man-
agement, audit committees and inspectors.
Acknowledgments
We thank Chris Agoglia, Lindsay Andiola, Mike Bamber, Tina
Carpenter, Chris Chapman, Clara Chen, Rick Hat?eld, Vicky Hoff-
man, Bill Messier, Mark Nelson, Flora Zhou, and the participants at
the AOS 40th Anniversary Event on 1e2 May 2015. We acknowl-
edge the ?nancial support of an ARC Australian Professorial
Fellowship (DP110103382) to Ken Trotman.
References
Agoglia, C. P., Brazel, J. F., Hat?eld, R. C., & Jackson, S. B. (2010). How do audit
workpaper reviewers cope with the con?icting pressures of detecting mis-
statements and balancing client workloads? Auditing: A Journal of Practice &
Theory, 29(2), 27e43.
Agoglia, C. P., Hat?eld, R. C., & Brazel, J. F. (2009). The effects of audit review format
on review team judgments. Auditing: A Journal of Practice & Theory, 28(1),
95e111.
Agoglia, C. P., Kida, T., & Hanno, D. M. (2003). The effects of alternative justi?cation
memos on the judgments of audit reviewees and reviewers. Journal of Ac-
counting Research, 41(1), 33e46.
American Institute of Certi?ed Public Accountants (AICPA). (2002). Consideration of
fraud in a ?nancial statement audit. Interim Auditing Standard (AU) Section 316.
New York, NY: AICPA.
Andiola, L., & Bedard, J. C. (2015). What distinguishes effective audit reviews? Evidence
from the experiences of staff auditors. Bentley University. Working paper.
Asare, S. K., & McDaniel, L. S. (1996). The effects of familiarity with the preparer and
task complexity on the effectiveness of the audit reviewprocess. The Accounting
Review, 71(2), 139e159.
Asare, S. K., & Wright, A. M. (2004). The effectiveness of alternative risk assessment
and program planning tools in a fraud setting. Contemporary Accounting
Research, 21(2), 325e352.
Asare, S. K., Wright, A. M., & Trompeter, G. M. (2000). The effect of accountability
and time budgets on auditors' testing strategies. Contemporary Accounting
Research, 17(4), 539e560.
Ashton, R. H. (1974). An experimental study of internal control judgments. Journal of
Accounting Research, 12(1), 143e157.
Ashton, R. H. (1990). Pressure and performance in accounting decision settings:
paradoxical effects of incentives, feedback, and justi?cation. Journal of Ac-
counting Research, 28(3), 148e180.
Australian Securities & Investments Commission (ASIC). (2014). Report 397: Audit
inspection program report for 2012e13. Available athttp://download.asic.gov.au/
media/1344614/rep397-published-27-June-2014.pdf Accessed January 2015.
Bamber, E. M. (1983). Expert judgment in the audit team: a source reliability
approach. Journal of Accounting Research, 21(2), 396e412.
Bamber, E. M., & Bylinski, J. H. (1987). The effects of the planning memorandum,
time pressure and individual auditor characteristics on audit managers' review
time judgments. Contemporary Accounting Research, 4(1), 127e143.
Bamber, E. M., & Ramsay, R. J. (1997). An investigation of the effects of specialization
in audit workpaper review. Contemporary Accounting Research, 14(3), 501e513.
Bamber, E. M., & Ramsay, R. J. (2000). The effects of specialization in audit work-
paper reviewon reviewef?ciency and reviewers' con?dence. Auditing: A Journal
of Practice & Theory, 19(2), 147e157.
Bantel, K. A., & Jackson, S. E. (1989). Top management and innovations in banking:
does the composition of the top team make a difference? Strategic Management
Journal, 10, 107e124.
Bauer, T., & Estep, C. (2014). The IT auditor function on ?nancial statement and inte-
grated audits: Description of practice and avenues for future research. University
of Illinois. Working paper.
Bauer, T., & Estep, C. (2015). One team or two teams? Exploring the relationship be-
tween auditors and IT specialists and its implications for a collective audit team
identity and audit quality. University of Illinois. Working paper.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 69
Bauer, T., Hillison, S., Peecher, M. E., & Pomeroy, B. (2015). Do auditors make better
fraud planning decisions when advising colleagues versus deciding for themselv-
ess? University of Illinois and University of Waterloo. Working paper.
Beasley, M. S., Carcello, J. V., Hermanson, D. R., & Neal, T. L. (2009). The audit
committee oversight process. Contemporary Accounting Research, 26(1), 65e122.
Beasley, M. S., & Jenkins, J. G. (2003). A primer for brainstorming fraud risks. Journal
of Accountancy, 196(6), 32e38.
Bell, T. B., Peecher, M. E., & Solomon, I. (2002). A guide to selecting SSA cases. In
T. B. Bell, & I. Solomon (Eds.), Cases in strategic-systems auditing. Montvale, NJ:
KPMG and University of Illinois at Urbana-Champaign.
Bell, T. B., Peecher, M. E., & Solomon, I. (2005). The 21st century public company audit:
Conceptual elements of KPMG's global audit methodology. Montvale, New Jersey:
KPMG.
Bellovary, J. L., & Johnstone, K. M. (2007). Descriptive evidence from audit practice
on SAS No. 99 brainstorming activities. Current Issues in Auditing, 1, A1eA11.
Bhattacharjee, S., Moreno, K. K., & Pyzoha, J. S. (2014). The in?uence of audit com-
mittee intervention approach on auditor and client judgments during accounting
disputes. Virginia Tech, Northeastern University, and Case Western Reserve
University. Working paper.
Bonner, S. E. (1994). A model of the effects of audit task complexity. Accounting,
Organizations and Society, 19(3), 213e234.
Bonner, S. E., & Sprinkle, G. B. (2002). The effects of monetary incentives on effort
and task performance: theories, evidence, and a framework for research. Ac-
counting, Organizations and Society, 27(4e5), 303e345.
Boritz, J. E. (1985). The effect of information presentation structures on audit
planning and review judgments. Contemporary Accounting Research, 1(2),
193e218.
Brazel, J. F., Agoglia, C. P., & Hat?eld, R. C. (2004). Electronic versus face-to-face
review: the effects of alternative forms of review on auditors' performance.
The Accounting Review, 79(4), 949e966.
Brazel, J. F., Carpenter, T. D., & Jenkins, J. G. (2010). Auditors' use of brainstorming in
the consideration of fraud: evidence from the ?eld. The Accounting Review,
85(4), 1273e1301.
Brewster, B. E. (2011). How a systems perspective improves knowledge acquisition
and performance in analytical procedures. The Accounting Review, 86(3),
915e943.
Brown, H. L., & Wright, A. M. (2008). Negotiation research in auditing. Accounting
Horizons, 22(1), 91e110.
Cannon-Bowers, J. A., Salas, E., & Converse, S. (1993). Shared mental models in
expert team decision making. In N. J. Castellan, Jr. (Ed.), Individual and group
decision making: Current issues (pp. 221e246). Hillsdale, NJ: Lawrence Erlbaum
Associates, Inc.
Carpenter, T. (2007). Audit team brainstorming, fraud risk identi?cation, and fraud
risk assessment: implications of SAS No. 99. The Accounting Review, 82(5),
1119e1140.
Centre for Audit Quality (CAQ). (2014). CAQ approach to audit quality indicators.
Available athttp://www.thecaq.org/docs/reports-and-publications/caq-
approach-to-audit-quality-indicators-april-2014.pdf?sfvrsn¼2/caq-approach-
to-audit-quality-indicators Accessed May 2014.
Chen, C. X., Trotman, K. T., & Zhou, F. H. (2015). Nominal versus interacting elec-
tronic fraud brainstorming in hierarchical audit teams? The Accounting Review,
90(1), 175e198.
Chen, C. X., Williamson, M. G., & Zhou, F. H. (2012). Reward system design and group
creativity: an experimental investigation. The Accounting Review, 87(6),
1885e1911.
Choo, F., & Trotman, K. T. (1991). The relationship between knowledge structure and
judgments for experienced and inexperienced auditors. The Accounting Review,
66(3), 464e485.
Cohen, J., Krishnamoorthy, G., & Wright, A. (2010). Corporate governance in the
post-Sarbanes-Oxley era: auditors' experiences. Contemporary Accounting
Research, 27(3), 751e786.
Danos, P., Eichenseher, J. W., & Holt, D. L. (1989). Specialized knowledge and its
communication in auditing. Contemporary Accounting Research, 6(1), 91e109.
Dennis, A. R., & Valacich, J. S. (1993). Computer brainstorms: more heads are better
than one. Journal of Applied Psychology, 78(4), 531e537.
Dennis, S., & Johnstone, K. M. (2014). Audit partner leadership tone and professional
skepticism in fraud brainstorming. University of Wisconsin. Working paper.
DeZoort, T., Harrison, P., & Taylor, M. (2006). Accountability and auditors' materi-
ality judgments: the effects of differential pressure strength on conservatism,
variability, and effort. Accounting, Organizations and Society, 31(4), 373e390.
Einhorn, H. J., Hogarth, R. M., & Klempner, E. (1977). Quality of group judgment.
Psychological Bulletin, 84(1), 158e172.
Ekasingh, E. (2014). Examining the performance of multidisciplinary greenhouse gas
assurance engagement teams (PhD dissertation). Australia: UNSW.
Emby, C., & Gibbins, M. (1988). Good judgment in public accounting: quality and
justi?cation. Contemporary Accounting Research, 4(1), 287e313.
Epps, K. K., & Messier, W. F., Jr. (2007). Engagement quality reviews: a comparison of
audit ?rm practices. Auditing: A Journal of Practice & Theory, 26(2), 167e181.
Federation of European Accountants (FEE). (2014). Opening a discussion: The future
of audit and assurance. Available athttp://www.fee.be/images/Future_of_Audit_
and_Assurance_Discussion_Paper_1402.pdf Accessed May 2014.
Financial Reporting Council (FRC). (2013). Audit quality thematic review: Materiality.
Available athttps://www.frc.org.uk/Our-Work/Publications/Audit-Quality-
Review/Audit-Quality-Thematic-Review-Materiality.pdf Accessed May 2014.
Fiolleau, K. J., Hoang, K. J., & Pomeroy, B. (2013). Auditors' communications with audit
committees: The in?uence of the audit committee's oversight approach. Tulane
University and University of Waterloo. Working paper.
Francis, J. (2011). A framework for understanding and researching audit quality.
Auditing: A Journal of Practice & Theory, 30(2), 125e152.
Frank, M. L., & Hoffman, V. B. (2015). How audit reviewers respond to an audit
preparer's affective bias: the ironic rebound effect. The Accounting Review, 90(2),
559e577.
Gendron, Y., & B edard, J. (2006). On the constitution of audit committee effec-
tiveness. Accounting, Organizations and Society, 31(3), 211e239.
Gibbins, M., & Trotman, K. T. (2002). Audit review: managers' interpersonal ex-
pectations and conduct of the review. Contemporary Accounting Research, 19(3),
411e444.
Glover, S. M. (1997). The in?uence of time pressure and accountability on auditors'
processing of nondiagnostic information. Journal of Accounting Research, 35(2),
213e226.
Gold, A., Gronewold, U., & Salterio, S. E. (2014). Error management in audit ?rms:
error climate, type, and originator. The Accounting Review, 89(1), 303e330.
Gold, A., Knechel, W. R., & Wallage, P. (2012). The effect of the strictness of
consultation requirements on fraud consultation. The Accounting Review, 87(3),
925e949.
Grenier, J. H. (2014). Encouraging professional skepticism in the industry specialization
era. Miami University. Working paper.
Guzzo, R. A., & Dickson, M. W. (1996). Teams in organizations: recent research on
performance and effectiveness. Annual Review of Psychology, 47, 307e338.
Haerkens, M. H. T. M., Jenkins, D. H., & van der Hoeven, J. G. (2012). Crew resource
management in the ICU: the need for culture change. Annals of Intensive Care,
2(39), 1e5.
Hammersley, J. S. (2006). Pattern identi?cation and industry-specialist auditors. The
Accounting Review, 81(2), 309e336.
Hammersley, J. S. (2011). A review and model of auditor judgments in fraud-related
planning tasks. Auditing: A Journal of Practice & Theory, 30(40), 101e128.
Hammersley, J. S., Bamber, E. M., & Carpenter, T. D. (2010). The in?uence of docu-
mentation speci?city and priming on auditors' fraud risk assessments and ev-
idence evaluation decisions. The Accounting Review, 85(2), 547e571.
Harding, N., & Trotman, K. T. (1999). Hierarchical differences in audit workpaper
review performance. Contemporary Accounting Research, 16(4), 671e685.
Harding, N., & Trotman, K. T. (2014). Enhancing professional skepticism via partner
communication of fraud brainstorming outcomes. Australia: UNSW. Working
paper.
Herold, D. (1979). The effectiveness of work groups. In S. Kerr (Ed.), Organizational
behavior. Columbus, OH: Grid.
Hoffman, V. B., & Patton, J. M. (1997). Accountability, the dilution effect, and
conservatism in auditors' fraud judgments. Journal of Accounting Research,
35(2), 227e237.
Hoffman, V., & Zimbelman, M. (2009). Do strategic reasoning and brainstorming
help auditors change their standard audit procedures in response to fraud risk?
The Accounting Review, 84(3), 811e837.
Huber, G. P., & Lewis, K. (2010). Cross-understanding: implications for group
cognition and performance. Academy of Management Review, 35(1), 6e26.
Ilgen, D. R., Hollenbeck, J. R., Johnson, M., & Jundt, D. (2005). Teams in organiza-
tions: from input-process-output models to IMOI models. Annual Review of
Psychology, 56, 517e543.
International Auditing and Assurance Standards Board (IAASB). (2004). Interna-
tional standard on auditing (ISA) 240. The auditor's responsibility to consider fraud
in an audit of ?nancial statements. New York, NY 10017: International Federation
of Accountants (IFAC).
International Auditing and Assurance Standards Board (IAASB). (2008). Interna-
tional standard on auditing (ISA) 220. Quality control for an audit of ?nancial
statements. New York, NY 10017: IFAC.
Isaksen, S. G. (1998). A review of brainstorming research: Six critical issues for enquiry
(Monograph #302). Buffalo, NY: Creative Problem Solving Group-Buffalo.
Ismail, Z., & Trotman, K. T. (1995). The impact of the review process in hypothesis
generation tasks. Accounting, Organizations and Society, 20(5), 345e357.
Johnson, V. E., & Kaplan, S. E. (1991). Experimental evidence on the effects of
accountability on auditor judgments. Auditing: A Journal of Practice & Theory,
10(Suppl.), 96e107.
Joyce, E. J. (1976). Expert judgment in audit program planning. Journal of Accounting
Research, 14(3), 29e60.
Kadous, K., Leiby, J., & Peecher, M. E. (2013). How do auditors weight informal
contrary advice? The joint in?uence of advisor social bond and advice justi?-
ability. The Accounting Review, 88(6), 2061e2087.
Kang, Y. J. (2014). Are audit committees more challenging given a sophisticated investor
base? Does the answer change given anticipation of additional mandatory audit
report disclosure? University of Massachusetts-Amherst. Working paper.
Kang, Y. J., Trotman, A. J., & Trotman, K. T. (2015). The effect of an audit judgment
rule on audit committees' questioning on accounting estimates. Accounting,
Organizations and Society (forthcoming).
Kaplan, S. E., & Reckers, P. M. J. (1989). An examination of information search during
initial audit planning. Accounting, Organizations and Society, 14(5/6), 539e550.
Kellermanns, F. W., Floyd, S. W., Pearson, A. W., & Spencer, B. (2008). The contingent
effect of constructive confrontation on the relationship between shared mental
models and decision quality. Journal of Organizational Behavior, 29, 119e137.
Kennedy, J. (1993). Debiasing audit judgment with accountability: a framework and
experimental results. Journal of Accounting Research, 31(2), 231e245.
Kennedy, J. (1995). Debiasing the curse of knowledge in audit judgment. The
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 70
Accounting Review, 70(2), 249e273.
Kennedy, J., Kleinmuntz, D. N., & Peecher, M. E. (1997). Determinants of the justi-
?ability of performance in ill-structured audit tasks. Journal of Accounting
Research, 35(3), 105e123.
Kerr, N. L., & Tindale, R. S. (2004). Group performance and decision making. Annual
Review of Psychology, 55, 623e655.
Klein, G. (1999). Sources of power: How people make decisions. Cambridge, MA: MIT
Press.
Knechel, W. R., Salterio, S. E., & Kotchetova-Kozloski, N. (2010). The effect of
benchmarked performance measures and strategic analysis on auditors' risk
assessments and mental models. Accounting, Organizations and Society, 35,
316e333.
Koonce, L., Anderson, U., & Marchant, G. (1995). Justi?cation of decisions in audit-
ing. Journal of Accounting Research, 33(2), 369e384.
KPMG. (2011). Evaluating professional judgment in auditing and accounting: The
KPMG profession judgment framework. KPMG.
Lambert, T. A., & Agoglia, C. P. (2011). Closing the loop: review process factors
affecting audit staff follow-through. Journal of Accounting Research, 49(5),
1275e1306.
Libby, R., & Blash?eld, R. K. (1978). Performance of a composition as a function of
the number of judges. Organizational Behavior and Human Performance, 21,
121e129.
Libby, R., & Lewis, B. L. (1977). Human information processing research in ac-
counting: the state of the art. Accounting, Organizations and Society, 2(3),
245e268.
Libby, R., & Lewis, B. L. (1982). Human information processing research in ac-
counting: the state of the art in 1982. Accounting, Organizations and Society, 7(3),
231e285.
Libby, R., & Luft, J. (1993). Determinants of judgment performance in accounting
settings: ability, knowledge, motivation, and environment. Accounting, Organi-
zations and Society, 18(5), 425e450.
Libby, R., & Trotman, K. T. (1993). The review process as a control for differential
recall of evidence in auditor judgments. Accounting, Organizations and Society,
18(6), 559e574.
Lorge, I., Fox, D., Davitz, J., & Brenner, M. (1958). A survey of studies contrasting the
quality of group performance and individual performance, 1920e1957. Psy-
chological Bulletin, 55(6), 337e372.
Lynch, A., Murthy, U., & Engle, T. (2009). Fraud brainstorming using computer-
mediated communication: the effects of brainstorming technique and facilita-
tion. The Accounting Review, 84(4), 1209e1232.
Markman, A. B., & Gentner, D. (2001). Thinking. Annual Review of Psychology, 52,
223e247.
Marks, M. A., Sabella, M. J., Burke, C. S., & Zaccaro, S. J. (2002). The impact of cross-
training on team effectiveness. Journal of Applied Psychology, 87(1), 3e13.
Marks, M. A., Zaccaro, S. J., & Mathieu, J. E. (2000). Performance implications of
leader brie?ngs and team-interaction training for team adaptation to novel
environments. Journal of Applied Psychology, 85(6), 971e986.
Mathieu, J. E., Heffner, T. S., Goodwin, G. F., Salas, E., & Cannon-Bowers, J. A. (2000).
The in?uence of shared mental models on team process and performance.
Journal of Applied Psychology, 85(2), 273e283.
McGrath, J. E., & Kravitz, D. A. (1982). Group research. Annual Review of Psychology,
33, 195e230.
Messier, W. F., Jr., Kozloski, T. M., & Kotchetova-Kozloski, N. (2010). An analysis of
SEC and PCAOB enforcement actions against engagement quality reviewers.
Auditing: A Journal of Practice & Theory, 29(2), 233e252.
Messier, W. F., Jr., Quick, L. A., & Vandervelde, S. D. (2014). The in?uence of process
accountability and accounting standard type on auditor usage of a status quo
heuristic. Accounting, Organizations and Society, 39(1), 59e74.
Messier, W. F., Jr., & Tubbs, R. M. (1994). Recency effects in belief revision: the
impact of audit experience and the review process. Auditing: A Journal of
Practice & Theory, 13(1), 57e72.
Mock, T. J., & Turner, J. L. (1979). A ?eld test of the effects of changes in internal
controls on audit programs. In T. J. Burns (Ed.), Behavioral experiments in ac-
counting II. Columbus: College of Administrative Science, Ohio State University.
Mohammed, S., & Dumville, B. C. (2001). Team mental models in a team knowledge
framework: expanding theory and measurement across disciplinary bound-
aries. Journal of Organizational Behavior, 22, 89e106.
Nelson, M. W., Proell, C. A., & Randel, A. E. (2014). Team-oriented leadership, audit
risk and auditors' willingness to raise audit issues. Cornell University, Texas
Christian University and San Diego State University. Working paper.
Nelson, M., & Tan, H.-T. (2005). Judgment and decision making research in auditing:
a task, person, and interpersonal interaction perspective. Auditing: A Journal of
Practice & Theory, 24(Suppl.), 41e71.
Ng, T., & Shankar, P. (2010). Effects of technical department's advice, quality
assessment standards, and client justi?cations on auditors' propensity to accept
client-preferred accounting methods. The Accounting Review, 85(5), 1743e1761.
Osborn, A. F. (1957). Applied imagination: Principles and procedures of creative
thinking (2nd ed.). New York, NY: Charles Scribner's Sons.
Owhoso, V. E., Messier, W. F., Jr., & Lynch, J. J. G. (2002). Error detection by industry-
specialized teams during sequential audit review. Journal of Accounting
Research, 40(3), 883e900.
Oxley, N. L., Dzindolet, M. T., & Paulus, P. B. (1996). The effects of facilitators on the
performance of brainstorming groups. Journal of Social Behavior and Personality,
11, 633e646.
Paulus, P. B., Nakui, T., Paulus, V. L., & Brown, V. R. (2006). Effects of task instructions
and brief breaks on brainstorming. Group Dynamics: Theory, Research and
Practice, 10(3), 206e219.
Payne, E. A., Ramsay, R. J., & Bamber, M. E. (2010). The effect of alternative types of
review on auditors' procedures and performance. Auditing: A Journal of Practice
& Theory, 29(1), 207e220.
Peecher, M. E. (1996). The in?uence of auditors' justi?cation processes on their
decisions: a cognitive model and experimental evidence. Journal of Accounting
Research, 34(1), 125e140.
Peecher, M. E., Piercey, M. D., Rich, J. S., & Tubbs, R. M. (2010). The effects of a su-
pervisor's active intervention in subordinates' judgments, directional goals, and
perceived technical knowledge advantage on audit team judgments. The Ac-
counting Review, 85(5), 1763e1786.
Peecher, M. E., Solomon, I., & Trotman, K. T. (2013). An accountability framework for
?nancial statement auditors and related research questions. Accounting, Orga-
nizations and Society, 38(8), 596e620.
Phillips, F. (1999). Auditor attention to and judgments of aggressive ?nancial
reporting. Journal of Accounting Research, 37(1), 167e189.
Pinsonneault, A., Barki, H., Gallupe, R. B., & Hoppen, N. (1999). Electronic brain-
storming: the illusion of productivity. Information Systems Research, 10(2),
110e133.
Pomeroy, B. (2010). Audit committee members investigation of signi?cant ac-
counting decisions. Auditing: A Journal of Practice & Theory, 29(1), 173e205.
Public Company Accounting Oversight Board (PCAOB). (2009). Engagement quality
review. Auditing standard (AS) No. 7. Washington, D.C.: PCAOB.
Public Company Accounting Oversight Board (PCAOB). (2013). Observations related
to the implementation of the auditing standard on engagement quality review.
Release No. 2013-011. Washington, D.C.: PCAOB.
Public Company Accounting Oversight Board (PCAOB). (2015). Staff consultation
paper no. 2015-01: The auditor's use of the work of specialists. Washington, D.C.:
PCAOB. Available athttp://pcaobus.org/Standards/Pages/SCP_Specialists.aspx
Accessed July 2015.
Ramsay, R. J. (1994). Senior/manager differences in audit workpaper review per-
formance. Journal of Accounting Research, 32(1), 127e135.
Ranzilla, S., Chevalier, R. E., Herrmann, G., Glover, S. M., & Prawitt, D. F. (2011).
Elevating professional judgment in auditing: The KPMG professional judgment
framework. New York, NY: KPMG LLP.
Ricchiute, D. N. (1999). The effect of audit seniors' decisions on working paper
documentation and on partners' decisions. Accounting, Organizations and Soci-
ety, 24(2), 155e171.
Rich, J. S. (2004). Reviewers' responses to expectations about the client and the
preparer. The Accounting Review, 79(2), 497e517.
Rich, J. S., Solomon, I., & Trotman, K. T. (1997a). Multi-auditor judgment/decision
making research: a decade later. Journal of Accounting Literature, 16, 86e126.
Rich, J. S., Solomon, I., & Trotman, K. T. (1997b). The audit review process: a char-
acterization from the persuasion perspective. Accounting, Organizations and
Society, 22(5), 481e505.
Salterio, S. (1996). The effects of precedents and client position on auditors'
?nancial accounting policy judgment. Accounting, Organizations and Society,
21(5), 467e486.
Salterio, S., & Koonce, L. (1997). The persuasiveness of audit evidence: the case of
accounting policy decisions. Accounting, Organizations and Society, 22(6),
573e587.
Schaefer, T. J. (2014). The effects of social costs and internal quality reviews on auditor
consultation strategies (Dissertation). University of South Carolina.
Schultz, J. J., Jr., & Reckers, P. M. J. (1981). The impact of group processing on selected
audit disclosure decisions. Journal of Accounting Research, 19(2), 482e501.
Shankar, P. G., & Tan, H.-T. (2006). Determinants of audit preparers' workpaper
justi?cations. The Accounting Review, 81(2), 473e495.
Solomon, I. (1982). Probability of assessment by individual auditors and audit
teams: an empirical investigation. Journal of Accounting Research, 20(2),
689e710.
Solomon, I. (1987). Multi-auditor judgment decision making research. Journal of
Accounting Literature, 6, 1e25.
Sprinkle, G. B., & Tubbs, R. M. (1998). The effects of audit risk and information
importance on auditor memory during working paper review. The Accounting
Review, 73(4), 475e502.
Stroebe, W., Nijstad, B. A., & Rietzschel, E. F. (2010). Beyond productivity loss in
brainstorming groups: the evolution of a question. Advances in Experimental
Social Psychology, 43, 157e203.
Tan, H.-T. (1995). Effects of expectations, prior involvement, and review awareness
on memory for audit evidence and judgment. Journal of Accounting Research,
33(1), 113e135.
Tan, H.-T., & Jamal, K. (2001). Do auditors objectively evaluate their subordinates'
work? The Accounting Review, 76(1), 99e110.
Tan, H.-T., & Kao, A. (1999). Accountability effects on auditors' performance: the
in?uence of knowledge, problem-solving ability, and task complexity. Journal of
Accounting Research, 37(1), 209e223.
Tan, H.-T., & Trotman, K. T. (2003). Reviewers' responses to anticipated stylization
attempts by preparers of audit workpapers. The Accounting Review, 78(2),
581e604.
Tan, H.-T., & Yip-Ow, J. (2001). Are reviewers' judgements in?uenced by memo
structure and conclusions documented in audit workpapers? Contemporary
Accounting Research, 18(4), 663e678.
Tan, S.-K., & Tan, H.-T. (2008). Effects of exposure to subsequently invalidated evi-
dence on judgements of audit workpaper preparers and reviewers.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 71
Contemporary Accounting Research, 25(3), 921e946.
Trotman, A. J., & Trotman, K. T. (2015). Internal audit's role in GHG emissions and
energy reporting: evidence from audit committees, senior accountants and
internal auditors. Auditing: A Journal of Practice & Theory, 34(1), 199e230.
Trotman, K. T. (1985). The review process and the accuracy of auditor judgments.
Journal of Accounting Research, 23(2), 740e752.
Trotman, K. T. (2005). Discussion of judgment and decision making researching
auditing: a task, person, and interpersonal interaction perspective. Auditing: A
Journal of Practice & Theory, 24, 73e87.
Trotman, K. T., Simnett, R., & Khalifa, A. (2009). Impact of the type of audit team
discussions on auditors' generation of material frauds. Contemporary Accounting
Research, 26(4), 1115e1142.
Trotman, K. T., & Yetton, P. W. (1985). The effect of the review process on auditor
judgments. Journal of Accounting Research, 23(1), 256e267.
Trotman, K. T., Yetton, P. W., & Zimmer, I. R. (1983). Individual and group judgments
of internal control systems. Journal of Accounting Research, 21(1), 286e292.
Turner, C. W. (2001). Accountability demands and the auditor's evidence search
strategy: the in?uence of reviewer preferences and the nature of the response
(belief vs. action). Journal of Accounting Research, 39(3), 683e706.
van der Vegt, G., & Bunderson, J. S. (2005). Learning and performance in multi-
disciplinary teams: the importance of collective team identi?cation. Academy of
Management Journal, 48(3), 532e547.
van de Ven, A., & Delbecq, A. L. (1971). Nominal versus interacting group processes
for committee decision-making effectiveness. Academy of Management Journal,
14(2), 203e212.
van Knippenberg, D., De Dreu, C. K. W., & Homan, A. C. (2004). Work group diversity
and group performance: an integrative model and research agenda. Journal of
Applied Psychology, 89, 1008e1022.
van Knippenberg, D., & Schippers, M. C. (2007). Work group diversity. Annual Re-
view of Psychology, 58, 515e541.
Westermann, K. D., Cohen, J., & Trompeter, G. (2014). A qualitative examination of the
perceived in?uence of accountability on professional skepticism. Florida Interna-
tional University, Boston College and University of Central Florida. Working
paper.
Wilks, T. J. (2002). Predecisional distortion of evidence as a consequence of real-
time audit review. The Accounting Review, 77(1), 51e72.
Williams, K. Y., & O'Reilly, C. A. (1998). Demography and diversity in organizations:
a review of 40 years of research. In Research in organizational behavior (Vol. 20,
pp. 77e140). Stamford: Jai Press Inc.
Wood, R. E., Beckmann, J. F., & Birney, D. P. (2009). Simulations, learning and real
world capabilities. Education þ Training, 51(5/6), 491e510.
Yip-Ow, J., & Tan, H.-T. (2000). Effects of the preparer's justi?cation on the re-
viewer's hypothesis generation and judgment in analytical procedures. Ac-
counting, Organizations & Society, 25(2), 203e215.
K.T. Trotman et al. / Accounting, Organizations and Society 47 (2015) 56e72 72
doc_253802788.pdf