Why would South African auditors
avoid the use of statistical sampling on
the audit?
Pieter von Wielligh
Professor:
Department of
Accounting,
Stellenbosch
University
Elmarie Swanepoel
Lecturer: Department
of Accounting,
Stellenbosch University
The recent
INTRODUCTION
global
economic crisis and the numerous
international audit scandals over the
past decade have raised concerns about
the quality of external audits across the
world. Regulators in the USA and the
UK have made specific references to
the place of sampling in these concerns
(McCollum 2009:16).
It is a well-known fact that auditors
cannot test all transactions underlying
the financial statements in the process
of obtaining sufficient appropriate
audit evidence on which to base the
audit opinion. One of the ways in
which auditors can address this issue is
to make use of audit sampling. Audit
sampling is defined in ISA530
(SAICA 2013:ISA530-2 para 5(a)) as
selecting “less than 100% of items
within a population… such that all
sampling units have a chance of
selection … to provide the auditor with
a reasonable basis on which to
draw conclusions about the entire
population.”
However, the use of audit sampling
introduces sampling risk to the wellknown audit risk formula. Sampling
risk is the risk that the auditor’s
conclusion, on the basis of the sample
results, differs from the conclusion
he would have reached had the test
been applied in the same way to the
entire population (SAICA 2013:
GLOSSARY-15).
According
to
ISA500 (SAICA 2013: ISA500-12
para A52), when planning the extent of
the audit procedures required to reduce
sampling risk to an acceptable level,
the auditor has the following selection
alternatives:
• testing the entire population that
makes up a specific balance or class
of transactions (this eliminates
sampling risk completely);
• selecting specific items on the basis
of certain criteria, for example: high
value, key items, items over a certain
value threshold, or items containing
specific information; and
• using audit sampling.
Audit sampling can be either statistical
or non-statistical. Statistical sampling
involves the use of scientific probability theory to evaluate sample results,
whereas non-statistical sampling
Summer 2013/14
Auditing SA
15
Why would SA auditors avoid the use of statistical sampling on the audit?
Auditing SA
Why would SA auditors avoid the use of statistical sampling on the audit?
Auditing SA
tation of all such considerations
easier (Cosserat & Rodda 2009:289).
It therefore provides a direct link
between risk assessment, precision
level, and the resulting sample size
and evaluation (Wilburn 1984:16).
involves the use of professional judgement instead of probability theory.
A recent South African study indicated
that only 38% of respondents use a full
statistical sampling approach for audit
sampling (Swanepoel 2013:8-10). The
authors conducted the study discussed
in this article to determine the reasons
auditors would choose not to use
statistical sampling as an audit tool.
The study also considered the types of
assistance and resources available to
audit teams that do consider using
statistical sampling as an audit tool.
The assistance and resources available
to audit teams were investigated
inter alia to determine whether a lack
of these could be a contributing factor
to auditors choosing not to use
statistical sampling.
The research method included
a questionnaire distributed for
completion to South African
Registered Auditors accredited by the
JSE. A response rate of 64% was
achieved, representing the auditors of
91% by number and 99% by total
market capitalisation of JSE listed
companies at the time of the study.
THE ADVANTAGES AND
DISADVANTAGES OF
STATISTICAL SAMPLING
The advantages of using statistical
sampling as an audit tool have been
promoted in textbooks for many years.
They include the following:
•
The statistical sampling approach
is a scientific approach based on
the well-proven principles of probability theory (Cosserat & Rodda
2009:289). This increases the objectivity of the sample size, the items
selected, and the evaluation of the
results (Wilburn 1984:16). This
greater objectivity makes the
sampling process followed by the
auditor more defensible in a court of
law (Arkin 1974:9).
•
Since statistical sampling is based
on specific scientific principles,
different auditors at different locations can use the same basis for
testing, and the results of these tests
can be combined and evaluated for
the population as a whole. This also
ensures continuity, as one auditor can
start the process and another can take
it over from any point, and arrive at
the same answer (Arkin 1974:11).
•
Statistical sampling is able to
quantify the auditor’s level of
assurance derived from applying a
particular audit procedure to a sample
of items.
In recent years the following disadvantages have caused auditors to opt for
non-statistical sampling techniques:
•
The training costs related to being
able to use statistical sampling are
higher than those for non-statistical
sampling (Cosserat & Rodda
2009:289).
•
Statistical sampling is more
complex than non-statistical
sampling,
and
non-statistical
sampling is therefore easier and takes
less time to apply (Cosserat & Rodda
2009:289).
•
•
•
The consideration of all the significant factors is an explicit
prerequisite when using statistical
sampling, thus making the documen-
16
Auditing SA
Since statistical sampling is
perceived to take more time to
apply, it is also perceived to be more
costly, as auditing firms apply hourly
rates to time spent on audits.
Summer 2013/14
One of the limitations of statistical
sampling is that it ignores differences in the nature of the items in the
population being tested. Therefore
unusual transactions do not have a
higher probability than normal transactions to be included in the sample.
PROBLEMATIC AREAS IN THE
USE OF
STATISTICAL SAMPLING
AS AN AUDIT TOOL
Sampling typically comprises the
following three steps:
•
•
Determining the number of items
to test (sample size);
Selecting the items to be tested
(inclusion of items in the sample);
and
•
Evaluating the results of the test
performed.
76% of the respondents in the study
indicated that evaluating the sample
results is the most problematic of the
three steps when using statistical
sampling, while only 16% indicated
that determining the sample size, and
8% indicated the selection of sample
items was most problematic to them.
These findings are similar to those
from international studies. From this it
is clear that the perceived complexity
of the statistical evaluation of sample
results is a challenge for auditors both
locally and internationally. This challenge may be a deterrent to the use of
statistical sampling plans.
THE MAIN REASONS
AUDITORS THINK THEIR PEERS
AVOID THE USE OF
STATISTICAL SAMPLING
The respondents were asked in an
open-ended question why, in their
opinion and experience, auditors
would prefer not to use statistical
sampling. 62% indicated that a lack of
confidence in using statistical
sampling plans in the auditing
environment could be a reason for
auditors not to use it. Furthermore,
76% indicated that it was perceived to
be inefficient and therefore too expensive. The aforementioned two reasons
are similar to those identified in
international studies.
These findings raise the question
whether the perceived inefficiency of a
full statistical sampling approach
TRAINEES’ KNOWLEDGE OF
STATISTICAL SAMPLING
The respondents were asked to
consider the competence in using
statistical sampling of first-year
trainees at entry-level into their
firms. Respondents were then
asked to indicate to what extent
the coverage of statistical
sampling
in
university
programmes followed by these
trainees met their expectations.
The findings are presented in
Table 1 below.
a
88% of the respondents indicated that
they have in place a firm guide on the
application of statistical sampling
(written instructions on the steps to
follow when using statistical
sampling). 69% of the respondents
indicated that they offer a training
course to their staff that covers the
application of statistical sampling
within
their
respective
audit
methodologies. These findings
indicate that the availability of
88% of the respondents
these resources has improved
indicated that they have in place
significantly since the previous
South African study in this
firm guide on the application of
regard in 1981.
statistical sampling
for auditors not to use statistical
sampling.
Table 1:
Satisfaction with coverage of statistical sampling at tertiary education level
Extent to which
expectations are met
Number of
respondents
Proportion
Meets
4
25%
Does not meet
Exceeds
The finding that the majority of
respondents were not satisfied with
their first-year trainees’ knowledge of
statistical sampling is in line with
international findings in this regard.
It is interesting to note that the proportion of respondents whose expectations
are not being met by tertiary education
(75% in this study) has increased since
a study done in 1981, in which this
proportion was only 58%.
The abovementioned finding created
the expectation on the part of the
authors that substantial training would
be provided for trainees by the firms of
these respondents, in order to address
this perceived insufficiency. Unexpectedly, it was found that 56% of
respondents provide only minor
coverage of statistical sampling in their
firms’ training programmes, while
44% of respondents claimed to provide
substantial coverage of this topic.
Auditing SA
a contributing factor to auditors
choosing not to use statistical
sampling.
12
75%
0
-
THE ASSISTANCE AND
RESOURCES AVAILABLE FOR
USING
STATISTICAL SAMPLING
IN PRACTICE
The assistance and resources available
to audit teams (in respect of using
statistical sampling as an audit tool),
were investigated to determine
whether a lack of these could be
Computers could be another
useful resource in the application of
statistical sampling. The respondents
were asked to indicate whether
the three main statistical sampling
steps were computerised within their
respective firms. In the case of a step
being computerised, they were asked
to further indicate whether the
software used had been specifically
self-developed
(in-house),
or
purchased from external suppliers as
an off-the-shelf application. The results
are presented in Table 2 below.
It is clear from Table 2 that the
majority of the auditing firms use
computer software to assist in each of
the three sampling steps. Bought-in
software is used more extensively than
in-house developed software for determining sample sizes and for selecting
sample items, whereas in-house developed software is used more extensively
for evaluating sample results.
Table 2: Computerisation of statistical sampling steps
Sampling step
computerised
Determining size
Proportion of
respondents
63%
Selecting items
63%
Evaluating
results
56%
Software
development
Proportion of
respondents
Self-developed
40%
Bought in
60%
Bought in
80%
Bought in
33%
Self-developed
Self-developed
Summer 2013/14
Auditing SA
20%
67%
17
Why would SA auditors avoid the use of statistical sampling on the audit?
One of the reasons for this unexpected
finding could be that, since the use of
statistical sampling is not a requirement of the auditing standards,
auditing firms are deciding to use a
non-statistical approach and to accept
higher sampling risk, rather than
investing in the costly training of
trainees in statistical sampling. If this
is the case, it could contribute to the
perceived lack of knowledge and experience. This could in turn exacerbate
the perceived lack of confidence in this
regard, which was found to be a reason
might result from a lack of knowledge
and experience in the use of statistical
sampling by auditors. In fact, if it is
applied by a knowledgeable and
experienced person, a full statistical
sampling approach could add
more value than that perceived by
respondents.
Why would SA auditors avoid the use of statistical sampling on the audit?
Auditing SA
Given the perceived lack of knowledge, experience, and
confidence in the application of statistical sampling as an
audit tool, the finding that the majority of firms that use
computerisation in the evaluation of sample results make use
of in-house developed software was unexpected. It was
expected that these firms would rather have sought the
expertise of an external software developer with the
necessary knowledge to assist in developing software to be
used for this sampling step.
SAIGA
ADVANCING AUDITING AND ACCOUNTABILITY
CONCLUSION
Although statistical sampling as an audit tool holds many
benefits, auditors in South Africa and internationally
sometimes avoid its use. Reasons for this include the
complexity of the underlying statistical models, a lack of
confidence in the ability of auditors to use it, and its
perceived inefficiency and the associated costliness.
Although the majority of respondents in the study are of the
opinion that their first-year audit trainees have insufficient
knowledge of statistical sampling, auditing firms nevertheless do not provide extensive training in this regard to
trainees during their training contracts. This may exacerbate
the abovementioned lack of confidence in the use of
statistical sampling as an audit tool.
Auditing firms’ own guides or manuals and training courses
on the application of statistical sampling on the audit, as well
as computer software, are widely available to auditors
considering the use of statistical sampling as an audit tool. It
can therefore be concluded that the choice auditors make not
to use statistical sampling as an audit tool is not significantly
influenced by a lack of resources for and assistance with the
use of statistical sampling as an audit tool.
Elmarie Swanepoel & Pieter von Wielligh ~ Auditing SA
Bibliography
Arkin, H. 1974. Handbook of sampling for auditing and accounting.
New York: McGraw-Hill, Inc.
Cosserat, W.G. & Rodda, N. 2009. Modern Auditing. 3rd Edition.
West Sussex: John Wiley & Sons.
McCollum, T. 2009. Regulators Report on Audit Performance.
Internal Auditor, 66(1):16.
SAICA (South African Institute of Chartered Accountants). 2013.
SAICA Handbook 2012/2013: Volume 2 Auditing. Durban:
LexisNexis.
Swanepoel, E. 2011. An empirical survey of certain key aspects of
the use of statistical sampling by South African Registered Auditors
accredited by the Johannesburg Securities Exchange.
Stellenbosch: Universiteit van Stellenbosch. (MRek).
Wilburn, A.J. 1984. Practical Statistical Sampling for Auditors.
New York: Marcel Dekker.
18
Auditing SA
Summer 2013/14
building
a brighter
future for
government
auditing in
South Africa