You are on page 1of 16

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/49278972

A Research Note on the Effects of Gender and


Task Complexity on an Audit Judgment

Article in Behavioral Research in Accounting · February 2001


DOI: 10.2308/bria.2001.13.1.111 · Source: OAI

CITATIONS READS

56 350

2 authors, including:

Gary S. Monroe
UNSW Sydney
25 PUBLICATIONS 392 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Audit risk & quality View project

All content following this page was uploaded by Gary S. Monroe on 10 March 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Behavioral Research In Accounting
Volume 13, 2001
Printed in USA

A Research Note on the Effects of


Gender and Task Complexity on an
Audit Judgment
Janne Chung
York University

Gary S. Monroe
Edith Cowan University

ABSTRACT
This study examines the effects of gender and task complexity on the
accuracy of audit judgments. Because research in cognitive psychol-
ogy and marketing suggests that females may be more accurate deci-
sion makers in complex decision tasks, we hypothesize that there will
be a significant interaction between gender and task complexity on the
accuracy of an audit judgment. A 2 × 2 full factorial experiment (males/
females by high-/low-task complexity) was carried out. The number and
consistency of cues was manipulated to create the high- and low-
complexity conditions. Participants were required to judge whether an
inventory balance was fairly presented based on case material that con-
tained a material misstatement in the inventory account balance. The
results support the hypothesis.

INTRODUCTION
A significant body of audit research has demonstrated that a vari-
ety of individual-level factors influences auditor judgment (see Solomon

This study was supported by grants from the School of Accounting, Edith Cowan
University and the Small and Medium Enterprise Research Centre, Edith Cowan
University. The authors thank John Anderson, Vincent Chong, Don Pallais, Roger
Simnett, Ken Trotman, and participants at the 1998 AAA/ABO Conference and the
seminar series at Edith Cowan University and the University of New South Wales for
their helpful comments. We also thank Karen Pincus for the use of her case materials,
and Roger Simnett, Peter Schelluch, and Grant Gay for their assistance with data
collection. Finally, we thank the editor and two anonymous reviewers for their very
helpful comments and suggestions. Data are available upon request from the first
author.
112 Chung and Monroe

and Shields [1995] for a review) and that the influence of these indi-
vidual-level factors changes as task complexity increases (e.g., Tan
and Kao 1999; Libby 1995). The examination of the effects of these
factors on task complexity is important given the complexity of many
audit tasks. For example, these tasks frequently contain numerous
and inconsistent information cues and have a high degree of diffi-
culty. Bonner (1994) provides three reasons for the examination of
task complexity in an audit situation. First, the complexity of a task
may have a significant effect on auditor performance. Second, cur-
rent decision aids and training techniques may be improved when
researchers understand the complexity of the different audit tasks.
Third, understanding the complexity of a task helps audit firm man-
agement find the best match between audit staff and audit task.
Research findings in the cognitive psychology and marketing lit-
erature suggest that gender may be an individual-level factor that
influences performance in judgment tasks and that the impact of
gender changes as task complexity changes. This literature suggests
that females may be more efficient and effective information proces-
sors in complex decision tasks than are males because of females’
superior ability to differentiate between and integrate decision cues.
Males have been shown to analyze information in less depth relative
to females by using fewer decision cues. However, apart from Chung
and Monroe (1998) and O’Donnell and Johnson (1999), the influence
of gender on information processing and judgment has not been ex-
amined in an auditing context.
One of the tasks that auditors have to perform is to form an opin-
ion on whether an account balance is fairly stated. The complexity of
this task varies between the type of account balance, the size of the
balance, and the number of cues contained in the data set including
the consistency of these cues. This study examines whether auditor
gender and task complexity affect decision accuracy with respect to
judging whether financial statements contain a material misstate-
ment in the inventory account balance. We find a significant inter-
action effect between gender and task complexity on judgment
accuracy. If supported by additional research, these results may have
important implications for audit firms in terms of training, decision
aids, assignment of personnel to audit engagements, and the audit
review process, all of which typically ignore gender differences in
information processing and decision-making abilities.
The remainder of this paper is organized as follows. First, the
relevant literature is reviewed to illustrate prior research into gen-
der-based differences in information processing to provide a basis
for developing a testable hypothesis. This is followed by a descrip-
tion of the experimental task, administration procedures and
A Research Note on the Effects of Gender and Task Complexity on an Audit Judgment 113

participants. Next, the results of the statistical tests of the hypoth-


esis are discussed. The final section presents a discussion of the
research findings, their implications for audit research and practice,
suggestions for future research, and limitations of the study.

BACKGROUND AND THEORY DEVELOPMENT


Gender Differences in Information Processing
Based on research into socialized gender differences in information
processing, Meyers-Levy (1986) developed a unifying theoretical frame-
work to explain the observed differences between male and female infor-
mation processing styles. This is called the “selectivity hypothesis.”
Gender-based differences in information processing and decision making
are based on different approaches that males and females adopt to ob-
tain and process information cues to solve problems or make decisions.
Males generally do not use all the information available in problem solv-
ing, nor do they process information comprehensively. Instead, males tend
to be limited information processors. Females are viewed as detailed pro-
cessors who process most of the available information cues (Meyers-
Levy 1986). They tend to conduct a detailed and “piecemeal” evaluation
of more of the available information and rely less on heuristics.
Females become more efficient relative to males as information
loads increase. In a complex task with multiple cues, males take more
time than do females to complete the task (Fairweather and Hutt 1972).
In addition, females have higher recall of new information (Erngrund
et al. 1996) and experience less information overload compared with
males (Allen and Griffeth 1997). Two auditing studies provide support
for the prediction of differences in information processing between
males and females. Chung and Monroe (1998) found that male ac-
counting students adopted a hypothesis-confirming strategy in a going-
concern evaluation task, whereas female accounting students did not.
In addition, they found that male students were selective information
processors and that female students were comprehensive information
processors. O’Donnell and Johnson (1999) investigated whether audi-
tor gender is related to differences in information-processing effort
during planning analytical procedures. They found that, when finan-
cial statement balances were consistent with information about client
business activity, females exerted more processing effort by analyzing
decision cues in more depth than male auditors did. However, when
unexpected fluctuations were seeded in their case materials (thus in-
creasing task complexity), the female auditors exerted less processing
effort than did the male auditors did in completing the more complex
task. These studies suggest that in a multi-cue task, females may
have an advantage over males as males’ processing capacity is ex-
ceeded earlier than are the females’ processing capacity.
114 Chung and Monroe

Research Hypothesis
Task complexity is a result of a decision-maker’s limited process-
ing capabilities (Hogarth and Einhorn 1992), memory (Bonner 1994),
and ability to integrate cues (Driver and Mock 1975). Low-complexity
tasks are characterized by fewer information cues, information cues
that are consistent with the event to be predicted, relatively high con-
sistency between expected and actual decision cues, relatively low levels
of ambiguity in the decision context, and few potential decision out-
comes (Bonner 1994). By contrast, high-complexity tasks contain more
information cues and a large number of these cues are inconsistent
with the event being predicted. At the same time, these tasks may
contain relatively high levels of ambiguity and many potential deci-
sion outcomes. As task complexity increases, females should perform
better than males due to their better ability to process multiple and/
or inconsistent cues, make finer distinctions among similar cues, re-
call cues, and integrate cues. Because of their limited information-
processing capacity, males should be able to cope relatively better when
task complexity is low compared to when task complexity is high.
Decision cues from multiple information sources are typically
used to form judgments about whether an account balance is mate-
rially misstated. Auditors identify misstatements in financial state-
ments by acquiring and processing information cues about inherent
risk, control risk, results of analytical procedures, tests of controls,
and results of substantive tests. One factor that affects the com-
plexity of the experimental task used in the present study is the
number of information cues. Bonner (1994) provides the following
review of task complexity. Task difficulty and task structure are two
aspects of task complexity. Task difficulty is related to the amount
of information, whereas task structure is related to information clarity
(see Bonner [1994] for a review). Bonner (1994) divides an information-
processing task into three stages: input, processing, and output. At
the input and processing stages, task complexity is increased by in-
creasing the number of cues. There is a difference between number of
cues available and number of cues processed. If the former is increased
and the latter does not similarly increase, then task complexity is not
increased.
With a large number of cues, the decision maker must divide his/
her attention between these cues (including selecting and weighting
them), and integrate them into a judgment. Judgment suffers when
the number of cues being considered exceeds the decision-maker’s
ability to attend to them. Another factor affecting the complexity of
this task is the consistency of cues. A task also becomes more com-
plex if there are inconsistent cues, and the performance of a decision
maker who is not adept at integrating inconsistent cues will not be as
good in comparison to a decision maker who is better able to integrate
incongruent cues.
A Research Note on the Effects of Gender and Task Complexity on an Audit Judgment 115

The selectivity hypothesis suggests that males will be as good as


or better than females at judging whether there is a material mis-
statement in an account balance when task complexity is relatively
low. In a less complex audit task characterized by relatively consis-
tent cues and fewer cues, a limited information-processing capacity
should be adequate for identifying and processing the cues that facili-
tate an accurate judgment. However, the comprehensive information-
processing capacity of females might not necessarily be the best
approach to a less complex task.
When task complexity is relatively high, a comprehensive informa-
tion processor better matches the task than does a limited information
processor. Males may be disadvantaged relative to females because
their limited information-processing capacity causes them to miss
those cues that determine accurate judgments. As females process a
higher proportion of cues (relative to males), they are in a better
position to identify these cues. Females are also better able to pro-
cess multiple and/or inconsistent cues, make finer distinctions among
similar cues, recall cues, and integrate cues. As a consequence, fe-
males should make more accurate judgments than would males when
task complexity is high because females’ comprehensive information
processing is needed to accomplish the task. This suggests that there
should be an interaction effect between gender and task complexity
on the accuracy of audit judgments.1 Accordingly, the following hy-
pothesis is proposed:

H1: There will be a significant interaction effect between gen-


der and task complexity on the accuracy of an audit judg-
ment about whether an inventory account balance
contains a material misstatement.

THE EXPERIMENT
Task
The experimental task required the participants to evaluate a modi-
fied version of the case material developed by Pincus (1991) and form

1 If a task is made more complex by increasing the number of cues available, then the
nature of the additional cues may impact on whether judgments are more accurate, as
accurate, or less accurate compared with the less complex task. For example, if the
additional cues in the more complex task are cues that are relevant to making the cor-
rect decision, then the accuracy of females’ judgments should increase as they incorpo-
rate these additional relevant cues into their decision. Males’ judgments would not nec-
essarily improve as they will not incorporate these additional relevant cues into their
decision or their decision accuracy might deteriorate due to information overload. A dif-
ferent effect is expected if the additional cues in the more complex task contain irrel-
evant cues or inconsistent cues. In this case, the accuracy of the judgments made by
males would most likely decrease due to information overload. However, the accuracy of
the judgments made by females would either not be affected by the additional cues or
would decrease, but not by as much as the accuracy of the judgments made by males.
116 Chung and Monroe

a judgment on whether inventory was materially misstated.2 The case


material contained a material misstatement of inventory by manage-
ment. The experiment was a 2 × 2 full factorial design (males/females
by high-/low-task complexity). This task was chosen because, first, a
sufficiently complex task was required to successfully test the variables
manipulated in this study. Second, it had an externally verifiable judg-
ment, which was necessary to test for judgment accuracy.3 Material
provided included information on the background of the company as
well as the current year’s and two prior years’ inventory and other fi-
nancial data. The information included both positive and negative indi-
cators of the fair presentation of the inventory balance and several items
that were not relevant to the judgment. Therefore, the judgment was
not clear-cut. The results of the current year’s audit procedures for
inventory were also available together with selected client records. The
management fraud was committed by overstating ending inventory quan-
tities and values. This could be detected via the various risk indicators
such as excessive ending inventory, inventory count procedures, man-
agement control, override of internal control procedures by manage-
ment, and inconsistent closing inventory amounts for the year being
audited and the opening balances for the following year.
Manipulating the number and consistency (inconsistency) of cues
in the case material operationalized task complexity. In the less com-
plex condition, the case material included 38 out of the 70 information
items originally used by Pincus (1991). In the more complex condition,
48 out of the original 70 items were used. This represented an in-
crease of ten information items (26 percent). The increase in the total
number of words was 23 percent.4 The risk indicators identified in the
2
It was necessary to shorten the case material so that the case could be completed in
the allotted time. The modifications to the original case developed by Pincus (1991)
included the conversion of weights and measurements to metric, the use of current
dates, and food items and proper names to reflect an Australian environment. The
dollar values were not changed. Such changes were not considered necessary be-
cause the type of restaurant had been changed from a Mexican restaurant to a fast-
food restaurant. Some of the information items from the original case materials were
omitted and only the information for the latest three years was given. The items that
were dropped were generally either redundant or irrelevant to the “fair presentation”
judgment. For example, items such as the names of the directors, officer/director
indemnification agreement and liability insurance, comments of board members re-
garding CEO and CFO, number of employees, and types of employee training and
compensation were dropped from the case material. Two experts in the hospitality
and restaurant industry examined the final versions of the cases for realism. The
experts did not examine these cases for complexity. Both experts were of the opinion
that the cases were realistic and reflective of current Australian restaurant prac-
tices. While the cases were simplified, they still contained sufficient information to
successfully test the variables manipulated in this study.
3
Based on the results of Pincus (1991), where low accuracy was observed (44.7 per-
cent of her participants made the correct judgment), low accuracy was similarly ex-
pected for the participants in this study.
4
The low-complexity condition comprised 4,170 words whereas the high-complexity
condition comprised 5,140 words.
A Research Note on the Effects of Gender and Task Complexity on an Audit Judgment 117

previous paragraph together with the inconsistent closing and opening


balances were found in both versions. Three of the ten cues that com-
prised the difference between the high- and low-task-complexity condi-
tions were inconsistent information in that the cues suggested that the
inventory balance was fairly stated.5 For example, item 37 indicated
that the internal audit reports revealed no exceptions. Item 38 stated
that the company was not involved in litigation and that their legal
adviser had acted in that position since inception of the company. Item
39 provided information on rental of additional warehouse space that
was consistent with their expansion plans. The other seven cues that
comprised the difference between the high- and low-task-complexity
conditions did not provide evidence of the misstatement in inventory.
They were either not relevant to the decision or, at best, suggested the
need to exercise professional skepticism. For example, items 2 and 3
provided information detailing the members of the board of directors,
board structure, and number of meetings. Three cues showed that the
company was expanding (items 7, 8, and 9). Item 27 showed a slow-
down in growth of earnings per share and item 28 showed a decrease in
the company’s price-earning ratio. None of the additional ten cues al-
lowed a more accurate judgment to be made in that none of the items
were related to fraud and several cues provided consistent information.
Therefore, we did not expect judgment accuracy to increase in the high-
complexity treatment for either males or females.

Procedures
All participants received the case materials in the form of a book-
let. The front of the booklet provided some introductory information
that explained the purpose of the experiment (which was to examine
auditors’ inventory judgments) and contained an assurance of confi-
dentiality. The participants were told to spend 15 minutes familiariz-
ing themselves with the case materials before answering the task
questions. They read the initial information of the case that contained
a general description of the operations of a chain of fast-food restau-
rants. The instructions to all participants read:
Instructions
In order for you to perform an audit of Tucker House’s inventory
balance, the following information items are available. You should
only evaluate those information items you consider are necessary for
forming an opinion on whether the inventory balance is fairly stated.

The participants were then presented with the content pages show-
ing the types of information available. This was followed by the actual
information items. After acquainting themselves with the information,

5
We did not undertake manipulation checks about the number and consistency of
the cues.
118 Chung and Monroe

the participants performed the task of evaluating whether the inven-


tory balance was fairly presented. The participants recorded their
answer on a nine-point Likert-type scale anchored by “fairly pre-
sented” (1) and “not fairly presented” (9). They also made a dichoto-
mous judgment as to whether the inventory balance was fairly
presented (i.e., fairly presented/not fairly presented). In the post-
test questionnaire, the participants provided various demographic
and diagnostic information.

Participants
The participants were 159 (101 males and 58 females) accoun-
tants attending a national audit-training program. All participants had
completed their first university degree and were employed by char-
tered accounting firms in auditing, business services, or taxation. The
experiment was carried out after the participants had completed four
days of audit training spread over three weeks, which included topics
such as analytical procedures, risk analysis, and inventory audit pro-
cedures. Their average age was 24.3 years (std. dev. 3.35) and their
average working experience was 35.1 months (std. dev. 28.41). Fifty-
three percent of the participants had audit experience, ranging from 1
to 108 months (mean 19.6 months [std. dev. 16.6]). There were no
significant gender differences for age, working experience, and audit
experience (t = .40, p = .69; t = 1.27, p = .21; and t = .69, p = .49,
respectively). They were each paid Australian $20 for their efforts.

RESULTS
Diagnostic Tests
To test the success of the task complexity manipulation, three vari-
ables were measured on nine-point scales. They were: perceived com-
plexity of the task (1 = not at all complex; 9 = extremely complex),
amount of information processed (1 = very little information; 9 = a
great deal of information), and task difficulty (1 = not at all difficult;
9 = extremely difficult). Two-way ANOVAs were carried out with task
complexity (high and low) and gender as the independent variables
and these results are reported in Table 1. There was no gender main
effect across all three diagnostic variables. There was also no task
complexity main effect for the participants’ perception of the complex-
ity of the task and the amount of information processed. For the per-
ception of task difficulty, there was a highly significant main effect for
task complexity (F = 14.62, p = .000), although there was no signifi-
cant main effect for gender and no significant interaction effect. Par-
ticipants in the high-complexity condition experienced significantly
higher task difficulty compared to participants in the low-complexity
condition. This is consistent with Bonner (1994) who states that in-
creasing the number of cues and the inconsistency of cues increases
the task difficulty aspect of task complexity.
A Research Note on the Effects of Gender and Task Complexity on an Audit Judgment 119

TABLE 1
The Effects of Gender and Task Complexity on the Diagnostic Measures

Panel A: Summary Results of ANCOVA Testing

Perceived Amount of
Complexity Information
Source of Variation of Taska Processeda Task Difficultya
p-value p-value p-value
Gender (G) .383 .628 .610
Task Complexity (T) .414 .978 .000
G×T .544 .721 .890

Panel B: Means
Perceived Amount of
Complexity Information
of Task Processed Task Difficulty
Task Complexity Male Female Male Female Male Female
High 7.00 6.00 5.88 6.00 5.72 5.88
n = 57 n = 30 n = 57 n = 30 n = 57 n = 30

Low 6.30 5.82 5.50 6.32 4.70 4.88


n = 44 n = 28 n = 44 n = 28 n = 44 n = 28

a
These variables were measured on nine-point scales.

Two-sample t-tests were performed by collapsing perceived com-


plexity of task and amount of information processed across the task
complexity dimension and testing for gender differences. The results
were t = 1.l0 (p = .29) and t = –1.84 (p = .07), respectively. The negative
t-statistic for the latter indicated that the female participants pro-
cessed marginally more information relative to the males.

Analyses of H1
Hypothesis 1 predicted an interaction effect between gender and
task complexity on the accuracy of the participants’ judgments about
whether inventory was materially misstated. Control variables that may
potentially have an effect on the participants’ judgment were included
in the ANCOVA model that was used to test the hypothesis. These were
general work experience, domain work experience, motivation, effort
expended, and amount of information processed. The dependent vari-
able was the participants’ assessments of whether the inventory bal-
ance was fairly presented using their judgments on the nine-point
120 Chung and Monroe

scale anchored by “fairly presented” (1) and “not fairly presented” (9).
As the inventory balance in the case material was not fairly presented,
a higher score represents a more accurate judgment. The results are
shown in Panel A, Table 2. The means (std. dev.) are shown in Panel
B, Table 2. Apart from effort expended (which had a marginally sig-
nificant effect [F = 2.94, p = .09]), the rest of the covariates did not
have significant effects on the participants’ judgment.6 There was a
significant main effect for task complexity (F = 4.32, p = .04) and no
significant main effect for gender (F < 1, p > .10) (Panel A, Table 2). A
significant interaction effect between gender and task complexity was
observed (F = 4.91, p = .03) (Table 2, Panel A) supporting H1.7
T-tests were carried out to investigate the direction of the effects.
In the low-complexity condition, females were less accurate compared
to the males and the difference in judgments was significant (t = 1.93,
p = .029, one-tailed). Females were marginally more accurate than
were males in the high-complexity condition (t = 1.61, p = .055, one-
tailed). The differences across the high- and low-complexity conditions
could have been caused by males making less accurate judgments in
the high-complexity condition relative to the judgments of males in the
low-complexity condition. The differences could also have been caused
by better performance by females in the high-complexity condition com-
pared to females in the low-complexity condition. A third explanation is
that both occurred. To test this, we compared the accuracy of the males
across both the low- and high-complexity conditions and the accuracy
of the females across both complexity conditions. Males in the high-
complexity condition made significantly less accurate judgments (mean
= 4.13) relative to males in the low-complexity condition (mean = 5.66)
(t = 3.76, p = .00, one-tailed). Therefore, males’ performance deterio-
rated as task complexity increased. Females’ mean judgments were
not significantly different between the two complexity conditions (high-
task complexity mean = 4.86 and low-task complexity mean = 4.64,
t < 1, p > .10 based on one-tailed test). The females’ mean judgments
were not significantly affected by the additional information cues, many
of which were inconsistent cues. The performance of females in the
high-complexity condition was not expected to be better than the per-
formance of females in the low-complexity situation as the cues in the
less complex task already included the cues necessary to make the

6
General and domain experience did not have any effects on the participants’ judg-
ment. As the experiment was carried out after the participants had completed four
days of intensive audit training, this may have contributed to the elimination of
differences in performance between those with audit experience and those without.
7
While ANCOVA can overcome unequal cell sizes, the results may be affected by the
overrepresentation of males in the participant pool. Ten randomly selected samples
with 58 males in the two complexity conditions were also tested. In all 10 samples,
the interaction effect was significant at p < .05 and the post hoc comparisons were all
significant and in the same direction as reported in the main text.
A Research Note on the Effects of Gender and Task Complexity on an Audit Judgment 121

TABLE 2
The Effects of Gender and Task Complexity on the Fair Presentation of
the Inventory Balance

Panel A: Source of Variation


SS df F-value p-valuea
Covariates
Work experience 3.63 1 .78 ns
Audit experience 1.05 1 .24 ns
Effort expended 12.73 1 2.94 .09
Motivation 1.62 1 .37 ns
Amount of information processed 6.45 1 1.49 ns
Main effects
Gender (G) .13 1 .03 ns
Task Complexity (T) 18.71 1 4.32 .04
Interaction effect
G×T 21.23 1 4.91 .03
Residual 618.96 150

Panel B: Means of Fairly Presented Judgmentb (Std. Dev.)


Task Complexity
Gender Low High
Females 4.64 4.86
(2.18) (1.98)
n = 28 (n = 30)
Males 5.66 4.13
(2.18) (1.98)
n = 44 n = 57

a ns = not significant.
b 1 = fairly presented; 9 = not fairly presented.

correct judgment. The additional cues in the more complex task con-
tained irrelevant cues, inconsistent cues, and cues that did not relate
directly to the misstatement in inventory, although they may have
suggested a higher risk of misstatement in the financial statements.
Analyses were also conducted using the responses on the dichoto-
mous scale (fairly presented or not fairly presented). These results are
reported in Table 3. In the low-complexity condition, 59 percent (26
out of 44) of the males and 50 percent (14 out of 28) of the females
made accurate judgments. The difference was not significant (χ 2 < 1,
p > .1, one-tailed. In the high-complexity condition, 37 percent (11
out of 30) of the females made accurate judgments compared to 23
percent (13 out of 57) of the males and this difference was marginally
significant (χ2 = 2.13, p = .07, one-tailed. As expected, males in the
low-complexity condition were significantly more accurate compared
to males in the high-complexity condition (59 percent and 23 percent
122 Chung and Monroe

TABLE 3
Accuracy Rates Using the Dichotomous Measure of Accuracy

Task Complexity
Gender Lowa Higha χa
Females 50% 37% χ2 < 1, p > .10
14 11
(n = 28) (n = 30)
Males 59% 23% χ2 = 13,79, p = .00
26 13
(n = 44) (n = 57)
χ2 < 1, p > .10 χ2 = 2.13, p = .07

a
% represent the percent of participants with a correct judgment.

accurate, respectively) (χ2 = 13.79, p = .00, one-tailed). Females in the


high-complexity condition were not significantly more accurate com-
pared to females in the low-complexity condition (percent accurate
were 38 and 50, respectively) (χ2 < 1, p > .10, one-tailed).

Additional Analyses
While many prior studies examined the developmental aspects of gen-
der differences in information processing, no study has examined the ef-
fects of professional training and experience on these differences. To shed
light on these potential effects, additional ANCOVA analyses were carried
out on the genders separately. The results showed that neither males’
nor females’ judgment were affected by general work experience (p >
.10 for both males and females) or by domain experience (p > .10 for
both genders). Next, general and domain work experience were dichoto-
mized at the means8 to produce two dichotomous variables with high
and low categories for each (general work experience: n = 110 [low]
and 49 [high] and domain work experience: n = 97 [low] and 62 [high]).
Additional ANCOVA analyses were carried out to test if there was any
interaction effect between general and domain experience and gender.
The results showed no significant interaction effects between gender
and general work experience (p > .10) and between gender and domain
work experience (p > .10) on accuracy judgments.

CONCLUSION
This study compared the accuracy of audit judgments made by
males and females at two levels of task complexity. One hundred fifty-
nine participants (comprising 101 males and 58 females) were asked

8
The mean (std. dev.) for general work experience was 35.08 (28.41) months and
10.35 (15.52) months for domain experience.
A Research Note on the Effects of Gender and Task Complexity on an Audit Judgment 123

to evaluate inventory-related case materials containing a material mis-


statement. The number and consistency of cues were manipulated to
create two levels of task complexity. In the less complex task, males
were more accurate than females. In the more complex task, females
were more accurate than males. These differences across the com-
plexity conditions were caused by a decrease in the accuracy of males’
judgments in the more complex task.
Presently, gender differences in information-processing ability are
routinely ignored in training, personnel assignment to engagements,
the audit decision-making process and the audit review process. The
results of this study suggest that this practice should not change un-
til more research have been carried out as gender-typing of audit tasks
is a complex issue. More research is also necessary before implica-
tions can be made to training of audit staff and designing of decision
aids. In addition, effectiveness must be balanced with efficiency prior
to drawing any such conclusions from the results of this study.
Although accurate judgments are ideal, sometimes accuracy is traded
off against efficiency because inefficiency is costly, especially in an
auditing context. This study was not designed to measure efficiency,
therefore, it is unclear whether males’ limited information processing
in a high-complexity task resulted in more efficient information pro-
cessing. Similarly, it is unclear whether females’ more accurate judg-
ments in the more complex task were achieved as a trade-off for
efficiency. However, the results do suggest that training should be
provided to both genders to try to improve judgment. Gender-based
comparisons made after training in information processing provides a
clearer picture of whether audit tasks should gender-typed based on
complexity.
As this is one of a very limited number of studies to suggest the
presence of gender differences in information processing for an audit
task, further research is needed before firm conclusions can be drawn.
In this study, only two levels of task complexity were manipulated. It is
unclear whether males’ judgments will continue to be less accurate
compared to females’ judgments when the number of cues is increased
from (say) 48 to 58 or whether females will be less accurate than are
males in much simpler tasks. Future studies using multiple levels of com-
plexity would be able to answer this question. In addition, future studies
could examine the interaction between gender and the other elements of
task complexity on judgment outcome. For example, the effects of cue
characteristics could be investigated. The manipulation of high complex-
ity could include additional relevant cues to examine whether females’
judgments become more accurate under these conditions.
Finally, there are limitations to the generalizability of the results
of this study that should be considered. First, a general observation of
the accuracy levels of the participants can be made, that is, the accu-
racy levels were generally low. Apart from males in the low-complexity
124 Chung and Monroe

condition (whose mean judgment was greater than the midpoint [5]),
accuracy scores of the rest of the cells were below 5. There are two
explanations for this. First, the participants were relatively inexperi-
enced, and this inexperience may have resulted in the lack of accu-
racy. Second, and perhaps more probable, was the inherent complexity
of the case materials. These were adapted from an actual audit where
the original auditors failed to identify the misstatement (Pincus 1991).
In spite of the fact that the case had been modified in the present
study, it remained comparatively complex. It is unclear what effects,
if any, a simpler case would have on the observed results. Second, as
relatively inexperienced participants took part in this study, it is un-
clear whether the generalizability of the results is affected by their
lack of experience. Covariate analysis, however, suggested that audit
experience did not have an effect on the results. The third limitation
of the study concerns the method of remunerating the participants.
This was based on participation and was not related to judgment ac-
curacy. What effect this method of remunerating participants would
have on the results remains untested.

REFERENCES
Allen, D. G., and R. W. Griffeth. 1997. Vertical and lateral information
processing: The effects of gender, employee classification level, and
media richness on communication and work outcomes. Human Rela-
tions 50 (10): 6201–6223.
Bonner, S. E. 1994. A model of the effects of audit task complexity.
Accounting, Organizations and Society 19 (3): 213–234.
Chung, J., and G. S. Monroe. 1998. Gender differences in information
processing: An empirical test of the hypothesis-confirming strategy
in an audit context. Accounting and Finance 38 (2): 265–279.
Driver, M. J., and T. J. Mock. 1975. Human information processing,
decision style theory, and accounting information systems. The Ac-
counting Review (July): 490–508.
Erngrund, K., T. Mantyle, and L. Nilsson. 1996. Adult age differences in
source recall: A population-based study. The Journal of Gerontology
51(B): P335–P340.
Fairweather, H., and S. J. Hutt. 1972. Gender differences in perceptual
motor skills. In Gender Differences: Their Ontogeny and Significance,
edited by C. Ounsted, and D. C. Taylor. Edinburgh, U.K.: Churchill
Livingstone.
Hogarth, R. M., and H. J. Einhorn. 1992. Order effects in belief updat-
ing: The belief-adjustment mode. Cognitive Psychology 24: 1–55.
Libby, R. 1995. The role of knowledge and memory in audit judgment.
In Judgment and Decision-Making Research in Accounting and Audit-
ing, edited by R. Ashton, and A. Ashton. New York, NY: Cambridge
University Press.
Meyers-Levy, J. 1986. Gender differences in information processing: A
selectivity interpretation. In Cognitive and Affective Responses to Ad-
vertising, edited by P. Cafferata, and A. M. Tybout, 219–260. Lexington,
MA: Lexington Books.
A Research Note on the Effects of Gender and Task Complexity on an Audit Judgment 125

O’Donnell, E., and E. N. Johnson. 1999. The effects of gender and


task complexity on information processing effort during planning
analytical procedures. Working paper, Arizona State University,
Tempe, Arizona.
Pincus, K. V. 1991. Audit judgment confidence. Behavioral Research In
Accounting 3: 39–64.
Solomon, I., and M. Shields. 1995. Judgment and decision-making re-
search in auditing. In Judgment and Decision-Making Research in
Accounting and Auditing, edited by R. Ashton, and A. Ashton. New
York, NY: Cambridge University Press.
Tan, H., and A. Kao. 1999. Accountability effects on auditors’ perfor-
mance: Influence of knowledge, problem-solving ability, and task com-
plexity. Journal of Accounting Research 37: 209–223.

View publication stats

You might also like