The Journal of Applied Business Research – September/October 2009 Volume 25, Number 5

Design And Application Of Individuals

And Moving Range Control Charts
Neil B. Marks, Miami University, USA
Timothy C. Krehbiel, Miami University, USA

ABSTRACT

We evaluate the design of individuals and moving range charts through extensive simulations. Via
a SAS program using 8000 replications, average run length is assessed for several shifts in
process mean and variation. From these computations recommendations are made concerning
when to use the individuals chart only, when to use a combined individuals and moving range
chart, and the optimal design parameters when the combined approach is used. The paper
includes discussion, recommendations, and examples concerning the monitoring of accounting
data.

Keywords: statistical process control, individuals control charts, moving range

INTRODUCTION

S ome processes are such that forming subgroups is impossible or undesirable. For example, ongoing
accounting data such as weekly, monthly, or quarterly sales totals are impossible to subgroup. In these
cases, individual measurements are typically plotted in an individuals chart, sometimes called an X chart.
Sometimes, the moving range (i.e., the difference between the current period and the previous period) is plotted in a
moving range chart, known as an MR chart, and presented in combination with the individuals chart (which situation
we shall call a combination chart or an XMR). In general, the X and MR portions are used to monitor the process
mean and variability, respectively. On other occasions, the individuals chart is used alone. There is much
disagreement as to whether the individuals chart alone or an XMR is best. Moreover, if the XMR is employed, there
is the practical matter of the design of the chart, i.e., what combination of control limits is optimal.

Leading textbooks (e.g., Montgomery, 2009) often suggest an XMR but raise doubt about the usefulness of
the MR portion. Practitioner-oriented publications often use XMR (e.g., Conklin, 2002) but fail to mention the
ongoing debate over the design of the XMR, or if the MR portion should be included at all.

MONITORING WEEKLY SALES

Weekly sales (in thousands of dollars) for Company A are found in Table 1. The mean and standard
deviation of weekly sales are 25 and 3, respectively. Figure 1 illustrates the use of an XMR with 3-sigma control
limits. The process is deemed to be in control since both the X portion and the MR portion of the chart are in
control.

As a second example, consider the weekly sales (in thousands of dollars) for Company B, also located in
Table 1. Again, the mean and standard deviation of weekly sales are 25 and 3, respectively. The XMR chart in
Figure 2 indicates an out-of-control process. In week 18, the sales total of 35 is above the upper control limit, and
the moving range of 10, the difference between 35 and 25, is also above its upper control limit. This illustration of
both the X and MR signaling an out-of-control situation at the same time is not unusual; in fact, it is quite common.
When an individual value is large enough, or small enough, to be outside the limits, it usually produces a moving
range value above the upper control limit. Thus, the value of including the MR portion with the X portion is highly
debatable.

31
The Journal of Applied Business Research – September/October 2009 Volume 25, Number 5

Table 1: Weekly Sales (in thousands of dollars) used in XMR Charts

Week Company A Company B
1 25 25
2 22 22
3 24 24
4 29 29
5 27 27
6 22 22
7 24 24
8 27 27
9 27 27
10 24 24
11 25 25
12 24 24
13 25 25
14 22 22
15 25 25
16 24 24
17 25 25
18 25 35
19 29 29
20 25 25

Figure 1: In Control Process Monitored with 3-Sigma Control Limits

XMR Chart for Company A

U C L=31
30.0
Individual V alue

27.5
_
25.0 X=25

22.5

20.0
LC L=19
1 3 5 7 9 11 13 15 17 19
Week

8
U C L=7.371

6
M oving Range

4
__
2 M R=2.256

0 LC L=0
1 3 5 7 9 11 13 15 17 19
Week

32
The Journal of Applied Business Research – September/October 2009 Volume 25, Number 5

Figure 2: Out of Control Process Monitored with 3-Sigma Control Limits

XMR Chart for Company B

36 1

32
Individual V alue

U C L=31

28
_
X=25
24

20
LC L=19
1 3 5 7 9 11 13 15 17 19
Week

1
10.0

7.5
M oving Range

U C L=7.37

5.0

__
2.5 M R=2.26

0.0 LC L=0
1 3 5 7 9 11 13 15 17 19
Week

LITERATURE REVIEW

Although originally designed for manufacturing data, the application of control charts for accounting data
has become relatively common in the last 20 years. Krehbiel, Havelka, and Scharfenort (2007) list five major areas
where transactional processes lend themselves to control charts: financial reporting, internal auditing, external
auditing, tax accounting and business operations. Other applications of control charts in accounting are found in
Reeve and Philpot (1988), Roth (1990), Bruch (1994), Long, Castellano, and Roehm (2002), Davies (2004) and
Grabski (2004). Dull and Tegarden (2004) argue that use of control charts on accounting data will become more
common as companies increase the frequency of reporting.

Snee (2004) and Krehbiel, Havelka, and Scharfenort (2007) put forth the argument that control charts in
financial and accounting functions will become even more common as more companies adopt Six Sigma quality
initiatives. Many authors, including Neuschler-Fritsch and Norris (2001), Friedman and Gitlow (2002), and Rudisill
and Clary (2004 & 2005) discuss accountants’ roles and responsibilities in successful Six Sigma projects. Case
studies involving successful accounting applications include the elimination of inefficiencies in an accounts payable
process (Brewer and Bagranoff, 2004) and improvements to the quarterly financial reporting process (Brewer and
Eighme, 2005; Krehbiel, Eighme & Cottell, 2009). Jones and Hain (2005) set up control charts to monitor key
performance indicators (KPIs) in the control stage of a Six Sigma project aimed at reducing obsolete inventory in
the materials management branch of a public works facility, and Falton and Falton (2002) discuss the integration of
Six Sigma metrics with management dashboards. Six Sigma applications concerning Sarbanes-Oxley are discussed
by Hofmann (2005), Liebesman (2005), LaComb and Senturk (2006), Senturk, LaComb, Neagu and Doganaksoy
(2006), and Nanda (2008). Six Sigma applications involving balanced scorecards are illustrated by Brewer (2004)
and Nilakantasrinivasan and Nair (2005).

Virtually all the applications mentioned above involve the use of data that are impossible to subgroup, and
thus lend themselves to individuals and moving range charts. In discussing control charts for individual

33
The Journal of Applied Business Research – September/October 2009 Volume 25, Number 5

measurements, Nelson (1982, 1990), argues strongly against the use of moving range charts, given that the moving
range statistic is correlated, thus making interpretation of the chart quite difficult. Acosta-Mejia and Pignatiello
(2000) point out that the correlation increases false alarms over the typical range chart for n=2, but conclude that an
XMR can be a somewhat effective method to identify increases in dispersion. Roes, Does, and Schurink (1993)
conclude that adding an MR chart can increase sensitivity to changes in variability, but this is insufficient cause for
support of greater use. Similarly, Rigdon, Cruthis, and Champ (1994) conclude that the individuals chart alone is
nearly as efficient as the combination chart for detecting changes in variability, and therefore also argue against
using the MR chart. Trip and Wieringa (2006) conclude that many of the problems in XMR applications result from
the design of the chart, but that even the best-designed XMR is only slightly better than that of an individuals chart
alone and therefore advise against using an MR chart.

On the other side of the argument, Crowder (1987a, 1987b) presents significant work concerning design
possibilities for XMR. Crowder’s work implies that XMR charts, as opposed to individuals charts alone, should be
used when there is concern about changes in variability. Amin and Ethridge (1998) as well as Adke and Hong
(1997) conclude that the XMR chart does provide some additional information useful in detecting variation shifts
and conclude that in some situations it should be used.

The average run length (ARL) of a control chart is the average number of time periods before an out-of-
control signal is observed. If the process is in control and deemed “All-OK,” then the ARL should be as long as
possible. For a 3-sigma control chart based on normally distributed data (e.g., the individuals chart) the All-OK ARL
is approximately 370, and a combination XMR chart has an All-OK ARL of approximately 200 (see Montgomery,
2009). When the process is out of control, a short ARL is desired, i.e., the user wants to be notified as soon as
possible of this condition. In general, increasing the All-OK ARL (a good thing) usually increases the out-of-control
ARLs (a bad thing); therefore, optimal design setting for the chart is of great interest.

Crowder presents a computer program that allows for the design of various XMR combinations which all
provide All-OK ARLs of desired length. The control limits of the XMR are defined as:

UCLX = μ + Mσ (upper bound on X chart)

LCLX = μ – Mσ (lower bound on X chart)
UCLR = Rσ (upper bound on MR chart)
LCLR = 0 (lower bound on MR chart)

Table 2: Design of XMR Charts

All OK X-chart only Very tight X Tight X Wide X Very wide X
ARL Very wide MR Wide MR Balanced Tight MR Very tight MR
50 M = 2.326 M = 2.33 M = 2.40 M = 2.55 M = 2.80 M = 3. 30
R = 4.50 R = 3.66 R = 3.41 R = 3.28 R = 3.24
100 M = 2.576 M = 2.58 M = 2.65 M = 2.80 M = 3.00 M = 3.50
R = 5.00 R = 4.04 R = 3.77 R = 3.67 R = 3.60
250 M = 2.878 M = 2.88 M = 2.95 M = 3.10 M = 3.30 M = 3.80
R = 5.00 R = 4.47 R = 4.22 R = 4.11 R = 4.05
370 M = 3.000 M = 3.00 M = 3.10 M = 3.20 M = 3.40 M = 3.80
R = 6.00 R = 4.57 R = 4.40 R = 4.29 R = 4.23
500 M = 3.090 M = 3.09 M = 3.20 M = 3.30 M = 3.50 M = 4. 00
R = 6.00 R = 4.67 R = 4.53 R = 4.42 R = 4.36
750 M = 3.209 M = 3.21 M = 3.30 M = 3.45 M = 3.60 M = 4.00
R = 6.00 R = 4.88 R = 4.66 R = 4.59 R = 4.55
1000 M = 3.291 M = 3.29 M = 3.40 M = 3.50 M = 3.65 M = 4.00
R = 6.50 R = 4.96 R = 4.82 R = 4.72 R = 4.65
X-chart: (LCL, UCL) = μ ± Mσ, MR-chart: LCL = 0, UCL = Rσ

34
The Journal of Applied Business Research – September/October 2009 Volume 25, Number 5

For each ARL considered, various combinations of M and R can be used. Crowder (1987a, p. 101) concludes that in
the selection of M and R (step one in his design procedure), “If possible changes in the process mean are of primary
concern, choosing the combination from step one with smallest M will yield a control scheme with smallest ARL
shifts from nominal in μ. Similarly, if possible changes in the process standard deviation are of primary concern,
choosing the combination from step one with smallest R will provide the greatest sensitivity to changes in σ.” Many
of the design combinations discussed in Crowder are reproduced in Vardeman and Jobe (1999) and included here in
Table 2. (We add a column to Table 2, the X-chart only case, to give added perspective and to help decide if the MR
portion is needed at all.) Vardeman and Jobe concur with Crowder that the columns on the left-hand side are best for
when the major concern is detecting a change in the mean, and the columns on the right-hand side are best for when
the major concern is detecting an increase in variability.

STUDY DESIGN

Each of the ARLs produced in this study was generated by 8000 replications of a simulation model
executed by a SAS program, whose logic is as follows. For given M and R and arbitrarily specified (without loss of
generality) μ and σ (50 and 10, respectively), control limits for the associated X and MR charts are computed:

UCLX = 50 + 10M (upper bound on X chart)

LCLX = 50 – 10M (lower bound on X chart)
UCLR = 10R (upper bound on MR chart)
LCLR = 0 (lower bound on MR chart)

Then the generation of 8000 ARLs is initiated. For shifts in σ, the standard deviation (10) is multiplied by either1.5,
2, or 2.5, known as λ. Shifts in μ are achieved by adding 10k (k = 0.5, 1, 1.5, or 2) to the mean (50). In other words,
k is the shift of the mean in standard deviation units, and λ = σ'/σ, where σ' is the new standard deviation, represents
the relative increase in process variation. A succession of random observations from a normal distribution is
produced according to the formula X = μ* + z(σ*), where μ*, and σ* refer to the shifted values just described, and z
comes from the SAS function RANNOR. Examples of X are 55 + 10z (k = 0.5) and 50 + 20z (λ = 2.0). The
moving range (MR), the absolute difference between two adjacent observations, is computed, and for a given
replication, run length (RL) is established if either MR > UCLR or X < LCLX or X > UCLX. Through PROC
MEANS the mean (ARL) and standard deviation of the 8000 simulated run lengths are found.

The appropriate number of replications for a study may be set as that value which achieves a satisfactorily
small relative error in the estimation of the critical output, that is, half-width of the estimating confidence interval
divided by the mean. In this case the standard deviation of RL is always equal to or just below the mean, so a
sample size of 8000 could always produce a relative error in the vicinity of 0.02 in every case, judged to be an
indicator of excellent precision in every case.1

RESULTS

Table 3 shows ARLs using the format of Vardeman and Jobe’s chart for various settings of k and λ, plus
the X-chart only case. As expected, the ARL relative to k rises from left to right since detection of the shift becomes
increasingly difficult. There is little distinction to be observed between the X-chart only and Very Tight X/Very
Wide MR cases. On the far right (Very Wide X/Very Tight MR), the simulated ARLs are approximately the All-
OK ARLs.

By contrast, ARL relative to shifts in variability follows a parabolic path from left to right, ending with
values approximating those of an MR chart only since at that point the control limits on the X chart are of little
consequence. The increase in ARLs in the far right columns compared to the middle columns of Table 3 is both
counter intuitive and contradictory to Crowder’s assertion. Our results indicate that the X chart has become so wide
at this point that it loses almost all ability to detect increases in process variation while the tighter MR chart is only
slightly more sensitive. The result is that these combination charts are actually worse at detecting an increase in
process variation. For example, when using an All-OK ARL of 50, the Very Wide X/Very Tight MR design detects
a doubling of the process variation in 4.05 subgroups on average, while the balanced design takes only an average of

35
The Journal of Applied Business Research – September/October 2009 Volume 25, Number 5

3.49 subgroups. Moreover, the Very Wide X/Very Tight MR design is extremely slow at detecting changes in the
mean. Of great interest is that the designs in the last three columns (Balanced, Wide X/Tight MR, and Very Wide X/
Very Tight MR) are statistically equal to or worse than the first three columns for all 21 combinations of ARLs and
increased variability the study investigated.2

Table 3: ARLs for k = (0.5, 1, 2) and λ = (1.5, 2, 2.5)

All X-chart only Very tight X Tight X Wide X Very wide X
OK Very wide MR Wide MR Balanced Tight MR Very tight MR
ARL
50 27.46 8.14 27.90 7.93 29.44 7.54 34.76 7.51 41.21 7.79 48.63 8.58
10.69 4.07 10.90 3.74 12.01 3.47 15.14 3.49 21.89 3.65 39.17 4.05
2.72 2.86 2.73 2.49 2.88 2.33 3.44 2.34 4.65 2.42 9.65 2.58
100 50.03 11.50 50.32 11.32 54.60 10.51 65.23 10.60 79.72 11.13 96.13 12.04
17.34 5.01 7.49 4.71 19.61 4.38 25.51 4.37 36.39 4.54 71.64 5.01
3.59 3.34 3.61 2.99 3.91 2.72 4.72 2.71 6.28 2.82 14.45 3.05
250 111.09 18.11 109.11 17.91 123.78 16.30 153.84 16.56 189.46 17.26 250.90 19.19
32.96 6.62 33.15 6.31 38.71 5.83 52.21 5.80 77.79 6.10 174.18 6.73
5.27 3.99 5.29 3.64 5.87 3.31 7.33 3.34 10.28 3.46 26.75 3.83
370 155.48 21.85 155.55 21.75 185.55 19.63 212.13 19.66 271.89 20.75 376.90 22.67
43.99 7.40 43.96 7.18 54.80 6.52 67.35 6.50 103.86 6.77 217.24 7.48
6.35 4.33 6.35 4.04 7.36 3.60 8.70 3.62 12.41 3.77 27.32 4.06
500 200.37 25.10 200.24 24.96 243.80 22.57 284.49 22.85 369.48 23.94 504.10 26.86
53.89 8.05 55.00 7.79 70.71 7.12 86.59 7.19 139.78 7.52 470.48 8.31
7.26 4.59 7.25 4.27 8.72 3.85 10.35 3.87 14.91 4.03 43.33 4.47
750 288.78 30.34 286.87 30.08 342.90 27.38 433.79 27.42 528.72 28.70 749.70 32.20
73.04 9.07 72.92 8.75 89.93 8.00 128.79 8.09 189.63 8.38 334.37 9.31
8.92 4.98 8.88 4.60 10.36 4.16 13.67 4.20 18.27 4.37 42.19 4.79
1000 372.56 34.72 365.86 34.60 458.75 31.36 549.77 31.73 660.89 32.53 997.90 35.33
89.21 9.85 90.96 9.61 116.72 8.70 149.91 8.78 215.54 9.06 426.29 9.80
10.14 5.27 10.19 4.97 12.49 4.41 14.97 4.45 20.34 4.57 43.19 4.97
The first column of numbers in a cell represent k = 0.5, 1, and 2, respectively, when λ = σ'/σ = 1, and the second column of
numbers in a cell represent λ = σ'/σ = 1.5, 2, and 2.5, respectively, when k = 0.

RECOMMENDATIONS

The moving range chart alone is seen to be of little value since its benefit, a marginal ability to detect shifts
in process variation, is overwhelmed by a substantial insensitivity to shifts in the mean. Further, in contrast to
suggestions made by other researchers, we find that pertaining to detection of increased process variation, the
combined charts having narrow moving range control limits are actually inferior to those with wider limits. Based
on the results of our simulations, we offer the following recommendations:

1. An X-chart only is the most sensitive to changes in the mean. The chart is also relatively sensitive to
increases in variability.
2. Adding a MR-chart decreases the chance of detecting a change in the mean.
3. Adding a MR-chart slightly increases the chance of detecting a change in the process variation.
4. As a direct result of 1-3, there is very little value from using the MR-chart, since the substantial loss of
sensitivity to mean changes overshadows the slight gain in sensitivity to variation changes.
5. In specific instances where the analyst is more concerned with increases in variation than mean shifts, the
Tight X/Wide MR combinations of an XMR chart are appropriate.
6. Combinations on the right-hand-side of Table 2 or Table 3 (e.g., Balanced, Wide X/Tight MR, Very Wide
X/Very Tight MR) are actually equal to or worse at detecting changes in the process variation that those
charts in the middle. Therefore, we recommend that these three columns never be used.
7. Since accounting data are typically collected weekly, monthly, or quarterly, the All-OK ARLs should be
much shorter than those typically used in the data-rich manufacturing environment. If collection occurs

36
The Journal of Applied Business Research – September/October 2009 Volume 25, Number 5

once a month, a false alarm every 370 time periods equates to one about every 31 years, whereas, a
frequency of four collections each hour produces one about every 4 days.
8. In the common case that only an X chart is employed, changes in the mean and increases in variation will
be detected. The analyst, however, needs to investigate the nature of the out-of-control signal to determine
if the process is experiencing a change in the mean or an increase in variation. The two examples that
follow illustrate the concept.

Table 4: Weekly Sales ( in thousands of dollars) used in X Charts

Week Company C Company D
1 27 27
2 23 23
3 24 24
4 27 27
5 23 23
6 23 23
7 25 25
8 24 24
9 27 27
10 27 27
11 22 22
12 24 26
13 35 31
14 23 22
15 28 26
16 27 30
17 28 23
18 25 19
19 25 28
20 23 23

Figure 3. Out of Control Signal Due to a Change in the Mean

X Chart for Company C

36 1

34

32
Individual Value

30 UCL=30.152

28

26 _
X=25
24

22

20 LCL=19.848

1 3 5 7 9 11 13 15 17 19
Week

37
The Journal of Applied Business Research – September/October 2009 Volume 25, Number 5

Figure 4: Out of Control Signal Due to an Increase in Variation

X Chart for Company D

32

30 UCL=30.152

28
Individual Value

26
_
X=25
24

22

20 LCL=19.848

1 3 5 7 9 11 13 15 17 19
Week

EXAMPLES

Company C designs an X chart with an All-OK ARL of 100. The mean and standard deviation of weekly
sales are 25 and 3, respectively. Thus the upper control limit is placed 2.576 sigma above the mean and the lower
control limit 2.576 sigma below the mean. The sales for 20 weeks are located in Table 4 and the control chart
appears in Figure 3. The out-of-control signal is produced by a one-time increase in central tendency, and the
variation around the center line appears not to increase before or after the spike.

Company D designs an X chart with an All-OK ARL of 100. Once again, the mean and standard deviation
of weekly sales are 25 and 3, respectively, and the control limits are the same as above. The data are located in Table
4 and the control chart in Figure 4 indicates an out-of-control signal. In this example, the variability around the
center line approximately doubles after week 10. Although the X chart picked up the out-of-control condition, the
mean of the last 10 days was still 25. The increased variability was detected because of the tight control limits on the
X chart made possible by omitting the MR chart.

SUGGESTIONS FOR FUTURE RESEARCH

Although individuals and moving range charts are easily designed and implemented and relatively
effective, recent advancements in process monitoring in the manufacturing sector using highly sophisticated
techniques may be adaptable to accounting data. In particular, exponentially weighted moving average charts with a
fast-initial-response (FIR) enhancement and the similarly constructed cumulative sum chart with an FIR
enhancement warrant investigation.

38
The Journal of Applied Business Research – September/October 2009 Volume 25, Number 5

NOTES

1. The relative error is calculated as follows for a 95% confidence interval:

(1 / mean) ∙ ((1.96 ∙ stdev) / (√8000)).

As examples, in column 5 of Table 3 with in-control ARL of 1000 and lambda of 2.5, the mean 4.572 and
standard deviation 4.513 give a relative error of 0.0216. In column1 (in-control ARL equal to 250 and
lambda of 1.5), the mean 18.110 and standard deviation 17.752 produced a relative error 0.0215. In
column 3 (in-control ARL equal to 50 and k = 1), the mean 12.013 and standard deviation 11.545 give a
relative error 0.0211.
2. Hypothesis tests comparing ARLs at the 5% level of significance were conducted. Column 3 (Tight X,
Wide MR) was found indistinguishable from column 4 (Balanced), i.e., the null hypothesis of equal ARLs
was not rejected in all 21 cases. Column 5 (Wide X, Tight M) was found to be inferior to Column 3, i.e., in
all 21 cases the null hypothesis was rejected. Likewise, Column 6 (Very Wide X, Very Tight M) was found
to be inferior to Column 3.

AUTHOR INFORMATION

Neil B. Marks earned a B. S. (Applied Mathematics) and M. B. A. at Washington University in St. Louis and a Ph.
D. (Management Science) at The Ohio State University. He has taught at several American universities, the last
being Miami University since 1985, holding now the rank of Associate Professor of Decision Sciences. His research
interests lie in applied statistics and simulation of random processes. His work has appeared in International
Journal of Production Research, Journal of the Operational Research Society, Communications in Statistics,
Production and Inventory Management Journal, Decision Sciences Journal of Innovative Education, and Journal of
Applied Statistics.

Timothy C. Krehbiel earned a B. A. (History) at McPherson College and a M.S. (Statistics) and Ph. D. (Statistics)
at The University of Wyoming. He has taught at Miami University since 1990, holding now the rank of Professor
and Endres Faculty Fellow. His research interests include statistical process control and six sigma quality. His
work has appeared in Quality Management Journal, International Journal of Production Research, Communications
in Statistics, and Journal of Purchasing and Supply Management. He is also the co-author of three textbooks: Basic
Business Statistics, Business Statistics a First Course, and Statistics for Managers Using Microsoft Excel.

REFERENCES

1. 1.Acosta-Mejia, C. A., Pignatiello, Jr., J. J. (2000). Monitoring dispersion without subgrouping. Journal of
Quality Technology 32(2):89-102.
2. Adke, S. R., Hong, X. (1997). A supplementary test based on the control chart for individuals. Journal of
Quality Technology 29(1):16-20.
3. Amin, R. W., Ethridge, R. A. (1998). A note on individual and moving range control charts. Journal of
Quality Technology 30(1):70-74.
4. Brewer, P. C. 2004. Six Sigma helps a company create a culture of accountability. Journal of
Organizational Excellence (Summer) 45-59.
5. Brewer, P. C. and N. A. Bagranoff. 2004. Near zero-defect accounting with Six Sigma. Journal of
Corporate Accounting and Finance (January/February) 67-72.
6. Brewer, P. C. and J. E. Eighme. 2005. Using Six Sigma to improve the finance function. Strategic Finance
(May) 27-33.
7. Bruch, N. M. 1994. Using control charts to help manage accounts receivable. Healthcare Financial
Management 48(7):44-48.
8. Conklin, J. D. (2002). Control charts and administration. Quality Progress 35(3):99-102.
9. Crowder, S. V. (1987a). Computation of ARL for combined individual measurement and moving range
charts. Journal of Quality Technology 19(2):98-102.
10. Crowder, S. V. (1987b). A program for the computation of ARL for combined individual measurement and
moving range charts. Journal of Quality Technology 19(2):103-106.
39
The Journal of Applied Business Research – September/October 2009 Volume 25, Number 5

11. Davies, M. C. 2004. Discussion of supporting continuous monitoring using control charts. International
Journal of Accounting Information System 5(2):131-134.
12. Dull, R. B. and D. P. Tegarden. 2004. Using control charts to monitor financial reporting of public
companies. International Journal of Accounting Information Systems 5(2):109-127.
13. Falton, D. M., and F. W. Falton. 2003. Toe the line: No more WorldComs. Quality Progress (January) 29-
35.
14. Friedman M. and H. Gitlow. 2002. The CPA in industry: Six Sigma primer for CPAs. The CPA Journal
(November) 56-59.
15. Grabski, S. 2004. Discussion of supporting continuous monitoring using control charts. International
Journal of Accounting Information Systems 5(2):129-130.
16. Hofmann, A. 2005. Two controls, one result. Quality Progress (July) 68-71.
17. Jones, E. C., and J. A. Hain. 2005. Use what you have. Six Sigma Forum Magazine 4(3):23-28.
18. Krehbiel, T. C., Eighme, J. E., and P. G. Cottell. 2009. Morgan Systems, Inc.: Application of Six Sigma to
the finance function. Forthcoming in Journal of Accounting Education.
19. Krehbiel, T. C., Havelka, D., and M. Scharfenort. 2007. Process monitoring in accounting: Implementing
pre-control charts. Journal of Applied Business Research.23(4):93-104.
20. 20. LaComb, C., and D. Senturk. 2006. The house that fraud built. Quality Progress (January) 52-60.
21. Liebesman, S. 2005. Mitigate SOX risk with ISO 9001 and 14001. Quality Progress (September) 91-93.
22. Long, J. A., J. F. Castellano and A. Roehm. 2002. A user friendly financial reporting system. Quality
Progress (January) 60-66.
23. Montgomery, D. C. 2009. Introduction to Statistical Quality Control. New Jersey: John Wiley and Sons,
Inc.
24. Nanda, V. 2008. Two are better than one. Quality Progress (June) 42-47.
25. Nelson, L. S. (1982). Control charts for individual measurements. Journal of Quality Technology
14(2):172-173.
26. Nelson, L. S. (1990). Monitoring reduction in variation with a range chart. Journal of Quality Technology
22(2):163-165.
27. Nilakantasrinivasan, N., and A. Nair. 2005. DMAIC failure modes. Six Sigma Forum Magazine 4(3):30-34.
28. Neuschler-Fritsch, D. and R. Norris. 2001. Capturing financial benefits from Six Sigma. Quality Progress
(May) 39-44.
29. Reeve, J. M., and J. W. Philpot. 1988. Applications of statistical process control for financial management.
Journal of Cost Management (Fall) 33-40.
30. Rigdon, S. E., Cruthis, E. N., Champ, C. W. (1994). Design strategies for individuals and moving range
control charts. Journal of Quality Technology 26(4):274-287.
31. Roes, K.C.B., Does, R. J. M. M., Schurink, Y. (1993). Shewhart-type control charts for individual
observations. Journal of Quality Technology 25:188-198.
32. Roth, H. P. 1990. Applications of control charts. The CPA Journal (April) 90-95.
33. Rudisill, F. and D. Clary. 2004. The management accountant’s role in Six Sigma. Strategic Finance
(November) 35-39.
34. Rudisill, F. and D. Clary. 2005. Six Sigma: Saving and making money. Chartered Accountant’s Journal
(April) 21-25.
35. Senturk, D., LaComb, C., Neagu, R. and M. Doganaksoy. 2006. Detect financial problems with Six Sigma.
Quality Progress (April) 41-47.
36. Snee, R. D. 2004. Weave Six Sigma into the fabric of an organization. Quality Progress (September) 69-
72.
37. Trip, A., Wieringa, J. E. (2006). Individuals charts and additional tests for changes in spread. Quality and
Reliability Engineering International 22:239-249.
38. Vardeman, S. B., Jobe, J. M. (1999). Statistical Quality Assurance Methods for Engineers. New York:
Wiley.

40