Download as pdf or txt
Download as pdf or txt
You are on page 1of 27

Communications in Statistics - Simulation and

Computation

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/lssp20

Performance of Max-EWMA control chart for


joint monitoring of mean and variance with
measurement error

Amjad Javaid , Muhammad Noor-ul-Amin & Muhammad Hanif

To cite this article: Amjad Javaid , Muhammad Noor-ul-Amin & Muhammad Hanif (2020):
Performance of Max-EWMA control chart for joint monitoring of mean and variance with
measurement error, Communications in Statistics - Simulation and Computation, DOI:
10.1080/03610918.2020.1842886

To link to this article: https://doi.org/10.1080/03610918.2020.1842886

Published online: 05 Nov 2020.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at


https://www.tandfonline.com/action/journalInformation?journalCode=lssp20
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R

https://doi.org/10.1080/03610918.2020.1842886

Performance of Max-EWMA control chart for joint monitoring


of mean and variance with measurement error
Amjad Javaida, Muhammad Noor-ul-Aminb, and Muhammad Hanif c
a
Pakistan Bureau of Statistics, Islamabad, Pakistan; bCOMSATS Institute of Information Technology, Lahore,
Pakistan; cNational College of Business Administration and Economics, Lahore, Pakistan

ABSTRACT ARTICLE HISTORY


Measurement error has negative effects on the monitoring of mean and Received 12 February 2020
variance shifts during industrial production. It is important to consider the Accepted 22 October 2020
measurement error for the development of quality control techniques to
KEYWORDS
monitor the process in the right direction. This article unfolds that how the
Control charts; Joint
measurement error is affecting the performance of control charts used for monitoring; Max-EWMA
joint monitoring of the process mean and variance shifts under the control chart;
scheme of Max-EMWA control charts. The covariate model has been used Measurement error
to study the effect of measurement error and multiple measurements have
been tried to reduce this effect. The calculations of the ARLs and SDRLs
are obtained by using Monte Carlo simulations to study the effects of
measurement error on the Max-EWMA control chart. In addition to com-
parison with other charts, an illustrative example is also included compris-
ing real life data to support the simulation results.

1. Introduction
To maintain the quality of a product, Walter A. Shewhart introduced the concept of control
charts in the 1920s who is known as the pioneer of the control charts. He developed X,  R, S, S2
type of control charts which are not based on past data but the information of the last sample or
observation collected from the production process. Later on, researchers developed control charts
like the exponentially weighted moving average (EMWA) chart by Roberts (1959) and the cumu-
lative sum (CUSUM) chart by Page (1954) which are more sensitive as compared to the previous
charts to monitor the small and moderate shifts. The latest charts used past information from the
production process which makes these charts efficient in detecting small shifts in the process
mean and / or variance. Afterward, many authors like Xie (1999), Chen, Cheng, and Xie (2001,
2004), Zhang and Chen (2005), Khoo, Teh, and Wu (2010), Lee and Lin (2012), Haq, Brown, and
Moltchanova (2015) and Lu, Tsai, and Huang (2017), Raza et al. (2019), Noor-ul-Amin, Arif, and
Hanif (2019), Javaid, Noor-ul-Amin, and Hanif (2020) developed control charts for joint monitor-
ing of mean and variance during the production process but they could not address the issue of
measurement error. Their control charts have been proved efficient assuming that the measure-
ments taken in the samples selected are correct, as they could not mention the effect of measure-
ment error in their proposed process control schemes.
Montgomery (2009) said that measurements are the most important part of quality control to
develop effective statistical process control. In real life situations, it is rarely possible that meas-
urements taken or observations collected are hundred percent precise but may vary to some
extent from the actual values. The measurement error refers to the value of a variable that we

CONTACT Amjad Javaid amjadsandhu@hotmail.com Pakistan Bureau of Statistics, Islamabad, Pakistan.


ß 2020 Taylor & Francis Group, LLC
2 A. JAVAID ET AL.

observe or collect but is randomly different from the actual or original value of our interest.
There is always a chance of error in the measurements taken for the development of control
charts to monitor the quality of the product. It has been noticed by the researchers that a signifi-
cant effect of measurement error exists in the application of statistical process control techniques
as these techniques are based on the measurement of observations or sample values.
Measurement error causes reduced efficiency of the control charts in terms of average run lengths
(ARLs). Montgomery and Runger (1993) said that the quality control process depends upon the
capability of gauges and tools used for measurement of observations or data collection regarding
desired variables. Sanders (1995) studied the effect of measurement error in detecting disturban-
ces in the continuous production process of chemicals. Mittag and Stemann (1998) highlighted
that imprecise gauge caused measurement error which contaminated the data used for control
charts, ultimately affecting the control chart efficiency. Linna and Woodall (2001) also studied
the negative effect of measurement error on the efficiency of control charts. Maravelakis,
Panaretos, and Psarakis (2004) highlighted that EWMA charts are affected by the measurement
error. Li and Huang (2009), Wu (2011), Maravelakis (2012), Hu et al. (2015), Ghashghaei et al.
(2016), Sabahno and Amiri (2017), Cheng and Wang (2018), Tang et al. (2018), and Salmasnia,
Maleki, and Niaki (2018) also studied the effects of errors occurred during measurement of qual-
ity characteristic, in one or the other way.
In addition to above mentioned research work on control charts with measurement error,
authors also tried in recent years to explore the effects of measurement error on the efficiency of
process monitoring techniques. Riaz et al. (2019) developed a control chart by mixing EWMA
and CUSUM charts using a regression estimator and used it for process monitoring of the mean.
They observed that their proposed chart is performing better even in the presence of measure-
ment error than other existing counterparts. Noor-ul-Amin, Riaz, and Safeer (2019) studied and
concluded that the efficiency of their proposed EWMA chart using auxiliary information for pro-
cess monitoring of the mean was affected by the measurement error. They used the covariate
model to study the error effect and tried multiple measurements to reduce the error effect, in
addition to using the linearly increasing variance method. Asif, Khan, and Noor-ul-Amin (2020)
proposed a hybrid exponentially weighted moving average control chart and studied the effect of
measurement error on the chart. They concluded that their proposed chart is efficient than the
simple EWMA chart based on average run lengths, to deal with measurement error for process
monitoring. Noor-ul-Amin et al. (2020) analyzed that the performance of an efficient Max-
EWMA chart using auxiliary information is affected if measurement error exists when sampling
units are considered for the development of control charts.
All these researchers analyzed the effects of measurement error on mean shift and / or vari-
ance shift in the process monitoring and concluded that measurement error has a negative effect
on detecting the shifts during the production process. In this research paper, we tried to analyze
how much the measurement error affects the joint monitoring of mean as well as variance. For
joint monitoring of mean and variance shifts, the Max-EWMA chart has been used under the
covariate model, and the effects of measurement error on the chart have been analyzed. Multiple
measurements of the same sample, using the covariate model technique, have been studied to
reduce the effect of measurement error.
After the introduction, the Max-EWMA control chart for joint monitoring has been explained
in Sec. 2. Then the covariate model for the Max-EWMA chart has been discussed in Sec. 3 and
the first two tables have been generated. Afterward, in Sec. 4, the covariate model with multiple
measurements has been explained and Tables 3–5 have been produced. Then comparison with
other charts has been elucidated in Sec. 5 showing Table 6. An illustrative example using real life
data is given in Sec. 6. The main findings are highlighted in Sec. 7 and the conclusion is drawn
in Sec. 8, while references are given at the end of this article.
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
3

2. Max-EWMA chart for joint monitoring


After Shewhart’s control charts, various authors proposed memory based control charts using the
past as well as current information in the statistics used for monitoring mean and variance shifts
separately in the production process. Roberts (1959) proposed the EWMA statistic for jth sample
having sample size n, as:
 j þ ð1  kÞZj1 ,
Zj ¼ kY (1)
where Y  j is the sample mean, k is the smoothing constant having values 0 < k  1, and Z0 is the
target mean for the first sample i.e. j ¼ 1: It is important to note that k is a weight assigned to
the current information and rate assigned to the past information of the production process. If
k ¼ 1, the EWMA chart will become a Shewhart chart depending upon current information
only. The smaller values of k are also used to detect smaller process shifts while large values of k
help to detect larger shifts in the process. The EWMA statistic (1) was used for monitoring a sin-
gle parameter shift in the production process. Then Xie (1999) and Chen, Cheng, and Xie (2001)
introduced the concept of joint monitoring of two parameters shift in mean and/or variance with
a single control chart and name it maximum exponentially weighted moving average
(Max-EWMA) control chart. Although notations and abbreviations have been explained in the
text, however, these are elaborated for the easement and understanding of the readers.

Notation and abbreviation Explanation


Zj Statistic for the jth sample
Zj1 Statistic for previous to the jth sample
Y j Sample mean of variable Y
k Smoothing constant
n Sample size
l0 Known population parameter for mean
r20 Known population parameter for variance
a Shift in process mean
b Shift in process variance
Uj , Ut Transformed estimator for mean
Vj , Vt Transformed estimator for variance
Pj , Pt EWMA statistic for mean
Qj , Qt EWMA statistic for variance
ARL Average run length
SDRL Standard deviation of run lengths
EWMA Exponentially weighted moving average
Max-EWMA Maximum exponentially weighted moving average
Max-EWMAME Maximum exponentially weighted moving average with measurement error
UCL Upper control limit

If Y is a variable of interest in a production process which is normally distributed with mean


l ¼ l0 þ ar0 and variance r2 ¼ b2 r20 , where l0 and r20 are the known parameters for mean and
variance whereas, a and b are the shifts in process mean and variance, having zero and unit val-
ues for an in control process with no shift, respectively. If a shift in the process mean and/or
variance takes place, the Max-EWMA statistic can detect, even a small shift for corrective
measures well in time. For an in control process the transformed statistics for mean and variance,
which are independent and both follow the normal distribution with mean zero and unit
variance, for the jth sample, are:
 j  l0
Y
Uj ¼ qffiffiffiffiffiffiffiffi
ffi , (2)
r2
0
n
4 A. JAVAID ET AL.

" ( )#
ðn  1ÞS2j
Vj ¼ /1 H , ðn  1Þ , (3)
r20
Pn
where j ¼ 1, 2, 3, :::::, n is the size of jth sample, Y  j ¼ i¼1 Yij is the mean of jth sample, and S2j ¼
Pn n
ðY Y j Þ2
is the variance of jth sample, while Hðn, tÞ follow the chi-square distribution with
i¼1 ij
n1
degrees of freedom t, and /1 is the inverse of the standard normal distribution function. We
can get two EWMA statistics for mean and variance by using above expressions (2) and (3) as:
Pj ¼ kUj þ ð1  kÞPj1 , (4)
Qj ¼ kVj þ ð1  kÞQj1 , (5)
where P0 ¼ Q0 ¼ 0 for the first sample i.e. j ¼ 1, while Pj and Qj are independent of each other,
as the expressions (2) and (3) used for these EWMA statistics are mutually independent and
follow the normal distribution. Therefore, EWMA statistics (4) and (5) also follow the normal
distribution with mean zero and variance k=ð2  kÞ for largely repetitive samples, when the
production process is in control. We can use only one statistic for joint monitoring of mean
and variance rather than use separate statistics (4) and (5) for mean and variance monitoring
respectively, which can be the maximum out of absolute values of both the (4) and (5) statistics.
To plot the maximum EWMA statistics out of two EWMA statistics expressed in (4) and (5), on
the control chart, the Max-EWMA statistic can be written as:
    
Max  EWMA ¼ Max Pj , Qj  , (6)

where j ¼ 1, 2, 3, :::::, the Max-EWMA is the maximum out of two absolute values of mean and
variance EWMA statistics. The absolute values guide us that only the upper control limit (UCL)
is sufficient to plot the Max-EWMA statistic to control the production process, which can be
written as derived by Xie (1999):
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2j

UCL ¼ ð1:128379 þ 0:602810  LÞ k½1ð1kÞ 
ð2kÞ , which can be written for largely repetitive sam-
ples as:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k
UCL ¼ ð1:128379 þ 0:602810  LÞ , (7)
ð2  kÞ

where L is the control constant and is obtained to approach the desired average run length
k
denoted by ARL0 for in control process, whereas, VarðPj Þ ¼ VarðQj Þ ¼ 2k for largely repetitive
samples when Pj and Qj follow the standard normal distribution. Let us study how the measure-
ment error using the covariate model, affects the efficiency of the Max-EWMA chart for joint
monitoring of mean and/or variance in a production process.

3. Covariate model and Max-EWMA chart


If we monitor any quality characteristic, say X, during an experiment or production process, the
measurement system may have a covariate with that characteristic. For example, if we measure
the length of a particular product, there might be several ways to measure the length e.g., tape,
ruler, centimeters, inches, coordinate measurement machines, etc. called covariates. The covariates
are the characteristics of participants, other than the actual treatment or measurement, in an
experiment. If we study the tolerance of a plant against drought, the level of drought is the actual
treatment, but the other factor affecting the tolerance level e.g., plant size is considered as a
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
5

covariate. Hence the other variables correlating with the quality characteristic might also affect
controlling the characteristic X:
It is assumed that for an in control production process, the quality characteristic X is normally
distributed with mean l and variance r2p : We are unable to get the true value of X, but another
related value Y, which has covariance with the X and can be expressed in terms of linear covari-
ate model as:
Y ¼ A þ BX þ e, (8)
where A, B are constants and e is the error term which is random and is distributed normally
and independently of X, with mean zero and variance r2m : It is also assumed that all the parame-
ters of this covariate model are known. It can be easily seen that if A ¼ 0 and B ¼ 1, then the
measurement error model Y ¼ X þ e developed by Bennet (1954) becomes a special case of this
covariate model (8). As X has a normal distribution, so the relationship between Y and X indi-
cates that Y is also normally distributed with mean and variance
ly ¼ A þ Blx and r2y ¼ B2 r2p þ r2m : (9)
To develop a Max-EWMA chart using the covariate model, as X cannot be measured correctly, we
select a sample of n values for a quality characteristic Y from the production process and calculate
the transformed statistics for mean and variance using expression (9) in equations (2) and (3) as:
  ðA þ Blx Þ
Y
Ut ¼ rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
. , (10)
ðB rp þrm Þ
2 2 2

and
" ( )#
1 ðn  1ÞS2
Vt ¼ / H 2 2 , ðn  1Þ (11)
B rp þ r2m
Pn Pn
Y  Þ2
ðY Y
where Y ¼ j¼1 j , and S2 ¼ j¼1 j , while Hðn, tÞ follow the chi-square distribution with
n n1
1
degrees of freedom t, and / is the inverse of the standard normal distribution function. Then
the EWMA statistics for mean and variance are calculated as:
Pt ¼ kUt þ ð1  kÞPt1 , (12)
Qt ¼ kVt þ ð1  kÞQt1 , (13)
where t ¼ 1, 2, 3, … .., P0 ¼ Q0 ¼ 0 for t ¼ 1, while Pt and Qt are independent of each
other and follow the normal distribution with mean zero and variance k=ð2  kÞ for largely
repetitive samples, when the production process is in control. To plot one maximum
EWMA statistic out of two (12) and (13) for joint monitoring, the Max-EWMA statistic can
be written as:
Max  EWMA ¼ MaxðjPt j, jQt jÞ, (14)
As the Max-EWMA statistic is the absolute value, so we need only the upper control limit (UCL) to
plot the statistic for joint monitoring. In order to plot Max-EWMA, the UCL can be written as:
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
k
UCL ¼ ð1:128379 þ 0:602810  LÞ , (15)
ð2  kÞ

where L is the control constant and 0 < k  1: For a particular sample, if Max-EWMA falls
under the UCL, the process is in control which shows that no shift occurs and another sample
can be taken to proceed further. However, when Max-EWMA lies beyond the UCL, the process
6 A. JAVAID ET AL.

shows some shift which indicates to stop the process for corrective measures, and it is a run
length (RL). After correction, the production process is again started and another sample is
selected and so on. A large number of samples are selected for replications to reach the average
and standard deviation of run length distribution i.e. ARL and SDRL:
The algorithm to calculate the average run length (ARL) and standard deviation of run length
(SDRL) for Max-EWMA with measurement error chart is given as:

1. First of all, generate samples of size n ¼ 5:


2. Determine the ARL0 :
3. Fix values of parameters and k ¼ 0:05, and arrive the value of L, on the basis of ARL0 :
4. Calculate transformed statistics of mean and variance to arrive at Max-EWMA.
5. Calculate the UCL.
6. Compare Max-EWMA and UCL. If Max-EWMA remains below the UCL, the process is in
control for this sample. After that go to generate another sample.
7. However, if the Max-EWMA statistic is beyond the UCL, it indicates that the process has
been shifted to an out of control position and it is the run length sufficient to stop the pro-
duction process and to solve the problem causing the shift.

We have used 50,000 replicates to calculate each ARL and SDRL: In order to get ARL0 ¼370,
we arrived at L¼ 2.710. Different combinations of shift values in mean i.e. a ¼ 0.00, 0.25, 0.50,
0.75, 1.00, 1.25, 1.50, 1.75, 2.00, 2.25, 2.50, 2.75, 3.00 and shift values in variance i.e. b ¼ 0.25,
0.50, 0.75, 0.90, 1, 1.10, 1.25, 1.50, 2.00, 2.50, 3.00 are used in this study. To observe the impact
of measurement error, these shifts are studied under different combinations of r2m  r2 with B ¼ 1:
Calculations are shown in Table 1.
Table 1 reveals that the Max-EWMA chart efficiently detects process shifts in mean and vari-
ance jointly, as every increase in mean shift reduces the values of ARLs: Similarly, every shift in
variance also reduces ARLs which indicates that this chart is capable of detecting process shifts at
the earliest for joint monitoring.
However, the efficiency of this chart is badly affected by the measurement error during data
collection for quality characteristic. Table 1 uncovers that under no error column, the ARL values
are equal to that of the Max-EWMA chart. However, as the ratio of error variance r2m  r2
increased from 0.1 to 1, the ARL and SDRL values also increased. It shows that the Max-EWMA
chart detects shift in mean or variance or both, in the presence of measurement error, later than
before when there was no measurement error. This error effect can be reduced by different meth-
ods. First of all, we can study the impact of the slope B to reduce the error effect having error
variance equal to process variance i.e. r2m ¼ r2 , and considering remaining all combinations of
Table 1. The calculated values of ARLs and SDRLs are shown in Table 2 to see the impact of B to
reduce the error effect.
Table 2 is generated to study the impact of coefficient of covariate X on the Max-EWMA con-
trol chart. Table 2 uncovers that every increase in B reduces the error effect for fixed error i.e.
r2m  r2 ¼ 1: Although increasing values of B from 1 to 5 resulted to decrease the ARLs and
SDRLs but even at B¼ 5, the values of ARLs are more than the values of no error column due to
severe effect of the error. The no error column has been taken the same as was in Table 1 just to
show the same pattern for comparison between no error and measurement error. It can be con-
strued from these results that moderate and small shifts in mean and variance are detected
quickly in this technique, as shown in Tables 1 and 2. We can also observe that the A in covari-
ate model has no impact on the ARLs in this research work.
To reduce the impact of measurement error, the number of samples can be increased or more
than one measurement for a selected sample/observation can be taken. We adopted multiple
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
7

Table 1. ARLs and SDRLs of Max-EWMA chart using covariate model for B ¼ 1:
r2m  r2
No error 0.1 0.2 0.3 0.5 1
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
0.25 0.00 4.16 0.49 4.33 0.55 4.75 0.64 5.42 0.81 7.34 1.46 17.29 6.50
0.25 4.16 0.48 4.31 0.53 4.75 0.64 5.40 0.81 7.34 1.44 17.01 6.32
0.50 4.16 0.48 4.31 0.55 4.74 0.64 5.40 0.81 7.33 1.40 13.72 3.81
0.75 4.16 0.49 4.31 0.54 4.74 0.63 5.38 0.75 6.66 0.97 9.58 2.23
1.00 4.14 0.47 4.26 0.50 4.57 0.52 4.86 0.49 5.36 0.67 6.99 1.40
1.25 3.91 0.25 3.92 0.22 3.97 0.20 4.03 0.29 4.31 0.52 5.53 1.00
1.50 3.04 0.19 3.05 0.23 3.15 0.36 3.31 0.46 3.48 0.50 4.58 0.76
1.75 2.89 0.02 2.90 0.03 2.94 0.03 2.99 0.05 3.07 0.27 3.93 0.61
2.00 2.54 0.47 2.55 0.48 2.60 0.43 2.66 0.37 2.83 0.25 3.45 0.53
2.25 2.00 0.05 2.01 0.08 2.04 0.19 2.14 0.35 2.53 0.50 3.11 0.39
2.50 2.00 0.00 2.00 0.00 2.00 0.00 2.00 0.02 2.09 0.29 2.86 0.37
2.75 2.00 0.00 2.00 0.00 2.00 0.00 2.00 0.00 2.00 0.05 2.62 0.48
3.00 1.99 0.00 1.99 0.00 1.99 0.00 2.00 0.00 2.00 0.01 2.34 0.47
0.50 0.00 7.36 1.45 7.46 1.48 7.97 1.68 8.70 1.99 11.17 3.09 25.58 12.39
0.25 7.35 1.45 7.47 1.50 7.93 1.67 8.68 1.96 11.11 3.03 23.49 10.15
0.50 7.26 1.31 7.37 1.35 7.76 1.46 8.35 1.63 9.95 1.15 15.08 4.72
0.75 6.20 0.83 6.28 0.86 6.42 0.91 6.67 1.00 7.29 1.28 9.73 2.51
1.00 4.82 0.59 4.83 0.61 4.92 0.64 5.04 0.69 5.44 0.85 7.07 1.56
1.25 3.91 0.43 3.92 0.43 3.97 0.45 4.05 0.49 4.33 0.61 5.53 1.08
1.50 3.19 0.40 3.21 0.41 3.27 0.45 3.37 0.49 3.62 0.53 4.59 0.80
1.75 2.89 0.20 2.90 0.19 2.94 0.20 2.99 0.21 3.12 0.36 3.94 0.65
2.00 2.54 0.49 2.55 0.49 2.60 0.48 2.66 0.44 2.83 0.33 3.46 0.56
2.25 2.08 0.27 2.10 0.30 2.14 0.35 2.23 0.42 2.53 0.50 3.12 0.43
2.50 2.00 0.04 2.00 0.05 2.00 0.08 2.02 0.14 2.15 0.36 2.86 0.41
2.75 2.00 0.00 2.00 0.00 2.00 0.00 2.00 0.01 2.02 0.13 2.62 0.49
3.00 1.99 0.00 1.99 0.00 1.99 0.01 2.00 0.03 2.00 0.03 2.36 0.48
0.75 0.00 19.57 7.98 19.94 8.11 20.88 8.91 22.80 10.26 29.25 15.14 69.15 49.89
0.25 17.41 6.10 17.54 6.13 18.40 6.61 19.40 7.16 22.65 9.25 37.77 20.79
0.50 10.22 2.52 10.33 2.57 10.54 2.70 10.87 2.86 11.86 3.40 16.24 6.11
0.75 6.58 1.31 6.62 1.34 6.73 1.39 6.92 1.46 7.46 1.71 9.83 2.85
1.00 4.86 0.85 4.88 0.84 4.99 0.88 5.11 0.92 5.48 1.09 7.09 1.73
1.25 3.92 0.62 3.93 0.63 3.98 0.65 4.07 0.67 4.36 0.77 5.55 1.18
1.50 3.27 0.48 3.28 0.50 3.33 0.51 3.41 0.54 3.65 0.61 4.61 0.90
1.75 2.89 0.37 2.90 0.37 2.94 0.37 2.99 0.38 3.16 0.46 3.94 0.71
2.00 2.54 0.50 2.55 0.50 2.60 0.49 2.66 0.47 2.83 0.42 3.47 0.60
2.25 2.18 0.38 2.19 0.39 2.23 0.42 2.30 0.46 2.53 0.50 3.13 0.49
2.50 2.03 0.16 2.03 0.16 2.04 0.19 2.07 0.26 2.21 0.41 2.86 0.46
2.75 2.00 0.04 2.00 0.03 2.00 0.06 2.00 0.07 2.05 0.21 2.62 0.50
3.00 1.99 0.04 1.99 0.05 1.99 0.04 2.00 0.03 2.00 0.07 2.37 0.48
0.90 0.00 90.70 71.07 92.68 72.27 96.55 76.11 105.1 84.45 131.3 111.1 238.6 220.3
0.25 25.73 12.25 25.86 12.55 26.65 13.00 27.47 13.68 30.76 16.43 44.28 28.06
0.50 10.61 3.31 10.63 3.27 10.84 3.41 11.15 3.51 12.16 4.09 16.38 6.66
0.75 6.63 1.58 6.67 1.61 6.75 1.64 6.96 1.74 7.51 1.98 9.87 3.11
1.00 4.89 1.00 4.94 1.01 4.99 1.04 5.12 1.07 5.49 1.23 7.08 1.84
1.25 3.92 0.72 3.95 0.73 3.99 0.75 4.09 0.77 4.38 0.88 5.55 1.28
1.50 3.30 0.56 3.31 0.56 3.35 0.57 3.43 0.60 3.66 0.67 4.61 0.96
1.75 2.89 0.46 2.90 0.46 2.94 0.47 3.00 0.47 3.17 0.52 3.95 0.78
2.00 2.54 0.51 2.55 0.50 2.60 0.50 2.66 0.50 2.83 0.47 3.48 0.64
2.25 2.21 0.41 2.22 0.41 2.27 0.44 2.34 0.48 2.53 0.51 3.13 0.53
2.50 2.05 0.21 2.06 0.23 2.07 0.25 2.10 0.31 2.24 0.43 2.86 0.49
2.75 2.00 0.08 2.00 0.08 2.00 0.10 2.00 0.13 2.06 0.25 2.62 0.51
3.00 1.99 0.10 1.99 0.08 1.99 0.09 2.00 0.07 2.01 0.12 2.38 0.49
1.00 0.00 370.1 349.2 369.9 349.4 369.3 356.8 372.5 356.6 372.2 356.4 370.9 356.0
0.25 26.54 14.16 26.67 14.26 27.35 14.71 28.40 15.68 31.37 18.23 45.14 30.33
0.50 10.62 3.67 10.75 3.73 10.84 3.72 11.22 3.98 12.17 4.43 16.47 7.09
0.75 6.66 1.76 6.70 1.78 6.82 1.84 6.99 1.91 7.56 2.14 9.91 3.26
1.00 4.89 1.09 4.94 1.12 5.03 1.14 5.12 1.20 5.53 1.33 7.11 1.95
1.25 3.93 0.79 3.96 0.80 4.02 0.81 4.11 0.85 4.39 0.94 5.59 1.36
(continued)
8 A. JAVAID ET AL.

Table 1. Continued.
r2m  r2
No error 0.1 0.2 0.3 0.5 1
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
1.50 3.30 0.60 3.33 0.61 3.36 0.62 3.43 0.64 3.69 0.72 4.65 1.03
1.75 2.90 0.51 2.91 0.50 2.95 0.52 3.01 0.51 3.18 0.57 3.98 0.81
2.00 2.55 0.52 2.55 0.52 2.60 0.52 2.66 0.51 2.83 0.51 3.48 0.65
2.25 2.24 0.43 2.26 0.44 2.29 0.46 2.34 0.78 2.53 0.51 3.14 0.56
2.50 2.07 0.26 2.07 0.26 2.09 0.29 2.12 0.33 2.26 0.44 2.86 0.51
2.75 2.01 0.12 2.01 0.13 2.01 0.13 2.03 0.17 2.09 0.28 2.62 0.52
3.00 1.99 0.13 1.99 0.13 1.99 0.12 2.00 0.10 2.02 0.14 2.38 0.49
1.10 0.00 73.26 58.26 73.51 58.60 76.34 60.85 81.26 65.21 96.57 78.87 166.1 148.5
0.25 24.20 13.12 24.17 12.84 24.71 13.37 25.55 14.30 28.51 16.25 41.35 26.96
0.50 10.55 3.90 10.65 3.89 10.84 4.01 11.15 4.25 12.09 4.68 16.34 7.20
0.75 6.66 1.92 6.70 1.95 6.81 1.98 6.99 2.07 7.54 2.31 9.83 3.37
1.00 4.89 1.21 4.94 1.20 5.03 1.25 5.13 1.26 5.53 1.44 7.08 2.03
1.25 3.93 0.85 3.96 0.86 4.02 0.88 4.10 0.91 4.39 1.01 5.59 1.42
1.50 3.30 0.66 3.33 0.66 3.36 0.67 3.43 0.70 3.68 0.77 4.64 1.07
1.75 2.89 0.56 2.90 0.56 2.94 0.57 3.00 0.57 3.18 0.61 3.97 0.84
2.00 2.54 0.53 2.55 0.53 2.60 0.53 2.65 0.53 2.83 0.54 3.48 0.69
2.25 2.24 0.44 2.26 0.45 2.29 0.46 2.34 0.48 2.53 0.52 3.14 0.58
2.50 2.07 0.27 2.07 0.28 2.10 0.31 2.12 0.36 2.26 0.45 2.86 0.53
2.75 2.01 0.17 2.01 0.16 2.01 0.18 2.03 0.21 2.10 0.30 2.61 0.53
3.00 1.98 0.15 1.98 0.16 1.98 0.16 1.99 0.14 2.02 0.18 2.39 0.49
1.25 0.00 20.52 10.70 21.01 11.52 21.79 11.71 22.89 12.60 26.91 15.64 48.90 35.03
0.25 16.20 7.54 16.28 7.54 16.68 7.84 17.38 8.21 19.69 9.88 30.29 17.61
0.50 9.93 3.76 10.00 3.77 10.18 3.84 10.51 4.02 11.44 4.53 15.65 6.91
0.75 6.65 2.10 6.70 2.14 6.80 2.16 6.97 2.26 7.54 2.49 9.81 3.57
1.00 4.90 1.34 4.94 1.36 5.03 1.82 5.13 1.41 5.53 1.56 7.08 2.19
1.25 3.93 0.95 3.96 0.97 4.02 0.98 4.11 1.02 4.40 1.11 5.59 1.51
1.50 3.30 0.74 3.33 0.75 3.36 0.75 3.43 0.76 3.68 0.84 4.63 1.15
1.75 2.89 0.62 2.90 0.61 2.94 0.63 3.00 0.63 3.19 0.67 3.97 0.89
2.00 2.54 0.55 2.55 0.55 2.60 0.56 2.65 0.57 2.83 0.59 3.48 0.73
2.25 2.24 0.46 2.26 0.46 2.30 0.48 2.34 0.50 2.53 0.54 3.14 0.63
2.50 2.07 0.32 2.08 0.34 2.10 0.35 2.13 0.38 2.26 0.46 2.86 0.56
2.75 2.01 0.21 2.01 0.22 2.01 0.23 2.03 0.25 2.10 0.34 2.61 0.54
3.00 1.97 0.22 1.97 0.21 1.97 0.21 1.99 0.19 2.03 0.22 2.39 0.50
1.50 0.00 9.23 3.75 9.32 3.78 9.55 3.92 9.95 4.11 11.32 4.91 18.23 9.45
0.25 8.83 3.38 8.93 3.49 9.05 3.54 9.45 3.77 10.72 4.40 16.55 7.99
0.50 7.57 2.67 7.62 2.64 7.80 2.76 8.04 2.90 8.91 3.29 12.56 5.23
0.75 6.05 1.96 6.12 1.97 6.20 2.04 6.38 2.11 6.92 2.30 9.21 3.41
1.00 4.85 1.46 4.87 1.48 4.94 1.50 5.08 1.54 5.45 1.70 7.08 2.32
1.25 3.93 1.10 3.96 1.12 4.02 1.10 4.11 1.15 4.40 1.23 5.59 1.64
1.50 3.30 0.85 3.33 0.88 3.37 0.89 3.44 0.91 3.68 0.96 4.63 1.23
1.75 2.89 0.71 2.90 0.72 2.94 0.73 3.01 0.73 3.20 0.77 3.98 0.99
2.00 2.54 0.61 2.55 0.61 2.60 0.62 2.66 0.63 2.83 0.66 3.52 0.83
2.25 2.25 0.50 2.26 0.51 2.30 0.51 2.34 0.54 2.53 0.57 3.15 0.68
2.50 2.07 0.40 2.08 0.40 2.11 0.41 2.13 0.44 2.26 0.50 2.86 0.62
2.75 2.01 0.31 2.01 0.31 2.01 0.32 2.03 0.33 2.10 0.39 6.61 0.57
3.00 1.94 0.31 1.95 0.30 1.97 0.30 1.99 0.29 2.03 0.30 2.39 0.52
2.00 0.00 6.63 1.55 4.69 1.58 4.81 1.62 4.94 1.68 5.46 1.89 7.88 2.97
0.25 4.63 1.52 4.67 1.55 4.72 1.56 4.89 1.65 5.40 1.86 7.76 2.93
0.50 4.50 1.47 4.53 1.46 4.60 1.50 4.75 1.54 5.22 1.74 7.45 2.69
0.75 4.23 1.34 4.23 1.33 4.35 1.36 4.48 1.43 4.89 1.55 6.76 2.34
1.00 3.92 1.19 3.94 1.19 3.99 1.23 4.09 1.25 4.44 1.38 5.96 1.94
1.25 3.52 1.05 3.54 1.05 3.61 1.06 3.68 1.11 4.00 1.19 5.20 1.64
1.50 3.18 0.93 3.19 0.92 3.24 0.94 3.31 0.96 3.56 1.03 4.53 1.36
1.75 2.85 0.80 2.88 0.82 2.89 0.81 2.97 0.84 3.17 0.90 3.97 1.14
2.00 2.55 0.70 2.55 0.70 2.60 0.71 2.66 0.71 2.83 0.77 3.52 0.97
2.25 2.26 0.60 2.26 0.61 2.30 0.62 2.34 0.64 2.53 0.66 3.15 0.83
2.50 2.07 0.52 2.08 0.53 2.11 0.53 2.13 0.55 2.26 0.58 2.86 0.73
2.75 2.01 0.46 2.01 0.47 2.01 0.47 2.03 0.48 2.10 0.49 2.61 0.65
3.00 1.92 0.44 1.93 0.44 1.95 0.44 1.98 0.44 2.03 0.44 2.39 0.57
(continued)
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
9

Table 1. Continued.
r2m  r2
No error 0.1 0.2 0.3 0.5 1
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
2.50 0.00 3.26 1.03 3.30 1.08 3.32 1.08 3.43 1.12 3.72 1.21 5.12 1.75
0.25 3.28 1.04 3.30 1.06 3.31 1.05 3.43 1.11 3.72 1.21 5.10 1.72
0.50 3.23 1.04 3.23 1.03 3.29 1.05 3.39 1.08 3.67 1.15 5.00 1.67
0.75 3.12 0.97 3.15 1.01 3.23 1.01 3.31 1.05 3.58 1.14 4.83 1.58
1.00 3.03 0.94 3.05 0.94 3.10 0.97 3.17 0.98 3.44 1.07 4.63 1.48
1.25 2.90 0.89 2.91 0.88 2.96 0.90 3.04 0.93 3.28 0.99 4.29 1.33
1.50 2.73 0.82 2.75 0.82 2.81 0.85 2.86 0.86 3.07 0.92 3.98 1.20
1.75 2.57 0.75 2.59 0.75 2.63 0.78 2.70 0.80 2.88 0.84 3.67 1.08
2.00 2.41 0.70 2.42 0.71 2.46 0.71 2.52 0.72 2.67 0.77 3.37 0.98
2.25 2.26 0.65 2.27 0.65 2.30 0.66 2.34 0.65 2.50 0.71 3.09 0.87
2.50 2.07 0.59 2.08 0.60 2.11 0.61 2.13 0.61 2.26 0.64 2.86 0.79
2.75 1.99 0.55 2.00 0.56 2.01 0.56 2.03 0.56 2.10 0.58 2.61 0.71
3.00 1.89 0.53 1.89 0.54 1.92 0.53 1.96 0.53 2.03 0.54 2.40 0.64
3.00 0.00 2.57 0.84 2.59 0.84 2.61 0.84 2.70 0.88 2.91 0.95 3.86 1.26
0.25 2.57 0.85 2.59 0.85 2.62 0.84 2.70 0.88 2.91 0.94 3.85 1.25
0.50 2.55 0.84 2.57 0.84 2.61 0.85 2.68 0.85 2.88 0.92 3.83 1.26
0.75 2.51 0.81 2.53 0.82 2.55 0.81 2.64 0.86 2.84 0.91 3.75 1.20
1.00 2.47 0.80 2.48 0.80 2.53 0.81 2.58 0.82 2.78 0.87 3.65 1.15
1.25 2.42 0.77 2.42 0.76 2.47 0.79 2.50 0.79 2.71 0.85 3.53 1.09
1.50 2.33 0.75 2.34 0.74 2.39 0.75 2.44 0.76 2.61 0.81 3.39 1.04
1.75 2.25 0.70 2.28 0.71 2.29 0.71 2.34 0.73 2.53 0.77 3.24 0.97
2.00 2.15 0.67 2.18 0.68 2.21 0.68 2.24 0.69 2.42 0.73 3.06 0.90
2.25 2.06 0.63 2.08 0.63 2.10 0.64 2.14 0.66 2.31 0.69 2.89 0.83
2.50 1.98 0.61 1.98 0.61 2.02 0.62 2.05 0.62 2.20 0.65 2.73 0.79
2.75 1.89 0.59 1.89 0.59 1.91 0.59 1.96 0.59 2.07 0.61 2.56 0.73
3.00 1.80 0.56 1.81 0.57 1.83 0.57 1.86 0.57 1.98 0.59 2.39 0.66

measurements of the same sample for this study which reduce the measurement error. Let us
study this technique of multiple measurements in the next section.

4. Covariate model with multiple measurements


To cope with the measurement error issue, it was proposed by Linna and Woodall (2001) that
more than one measurements of each sample can be taken by the quality control expert. If meas-
urements are taken a number of times and the average of these measurements is used as quality
characteristic, it will lead to the precise measurement. This strategy will reduce the chances of
error in measurement of any sample or observation. If we take multiple measurements of the
same sample or observation, it will reduce the variance of the error component even to zero for
infinite measurements. However, multiple measurements involve cost and time which may be
considered at the time of planning, as if there is no measurement error, then additional measure-
ments will just increase the cost and time without any benefit. Therefore, all factors must be con-
sidered by experts before decision making.
To calculate the Max-EWMA statistic with measurement error, we take more than one meas-
urements of each sample. Our sample size is n ¼ 5 as mentioned in algorithm described in Sec.
3, and if k is the number of measurements of each sample, then k ¼ 1 for first two tables.
However, if we take a number of measurements more than one i.e. k>1, then the total number
of observations in one sample will be n ¼ 5  k: Under these circumstances, the transformed sta-
tistics for mean and variance will become
  ðA þ Blx Þ
Y
Ut ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi , (16)
B2 r2p r2m
n þ ðnkÞ
10 A. JAVAID ET AL.

Table 2. ARLs and SDRLs of Max-EWMA chart using covariate model for r2m  r2 ¼ 1:
B
No error 1 2 3 5
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
0.25 0.00 4.16 0.49 17.26 6.57 7.34 1.45 5.69 0.88 4.75 0.64
0.25 4.16 0.48 17.06 6.17 7.32 1.43 5.68 0.89 4.74 0.64
0.50 4.16 0.48 13.69 3.82 7.30 1.40 5.66 0.88 4.76 0.64
0.75 4.16 0.49 9.57 2.11 6.68 0.96 5.59 0.79 4.75 0.63
1.00 4.14 0.47 7.00 1.41 5.34 0.67 4.94 0.50 4.58 0.52
1.25 3.91 0.25 5.53 1.00 4.30 0.51 4.06 0.32 3.97 0.19
1.50 3.04 0.19 4.58 0.77 3.63 0.49 3.37 0.48 3.14 0.35
1.75 2.89 0.02 3.93 0.61 3.08 0.28 3.00 0.07 2.94 0.03
2.00 2.54 0.47 3.46 0.54 2.83 0.25 2.68 0.36 2.60 0.43
2.25 2.00 0.05 3.10 0.40 2.53 0.50 2.20 0.40 2.04 0.19
2.50 2.00 0.00 2.89 0.37 2.09 0.29 2.00 0.06 2.00 0.00
2.75 2.00 0.00 2.62 0.48 2.00 0.05 2.00 0.00 2.00 0.00
3.00 1.99 0.00 2.34 0.48 2.00 0.00 2.00 0.00 1.99 0.00
0.50 0.00 7.36 1.45 25.71 12.31 11.19 3.12 9.01 2.14 7.93 1.66
0.25 7.35 1.45 23.35 9.86 11.18 3.07 8.99 2.12 7.92 1.67
0.50 7.26 1.31 15.17 4.79 9.98 2.13 8.54 1.69 7.76 1.46
0.75 6.20 0.83 9.77 2.51 7.28 1.28 6.75 1.04 6.43 0.90
1.00 4.82 0.59 7.06 1.55 5.42 0.85 5.10 0.71 4.93 0.63
1.25 3.91 0.43 5.53 1.07 4.33 0.62 4.09 0.51 3.97 0.46
1.50 3.19 0.40 4.59 0.80 3.63 0.53 3.42 0.50 3.27 0.45
1.75 2.89 0.20 3.94 0.66 3.13 0.36 3.01 0.23 2.94 0.20
2.00 2.54 0.49 3.47 0.56 2.83 0.33 2.68 0.42 2.60 0.47
2.25 2.08 0.27 3.13 0.44 2.53 0.50 2.27 0.45 2.14 0.35
2.50 2.00 0.04 2.86 0.41 2.15 0.36 2.03 0.17 2.01 0.08
2.75 2.00 0.00 2.62 0.49 2.01 0.12 2.00 0.02 2.00 0.00
3.00 1.99 0.00 2.35 0.48 2.00 0.03 2.00 0.00 1.99 0.00
0.75 0.00 19.57 7.98 70.31 51.83 29.22 14.98 23.71 10.95 20.89 8.86
0.25 17.41 6.10 37.73 21.04 22.98 9.33 19.77 7.41 18.17 6.44
0.50 10.22 2.52 16.26 6.11 11.91 3.48 11.01 2.97 10.55 2.68
0.75 6.58 1.31 9.82 2.87 7.48 1.74 6.96 1.49 6.73 1.39
1.00 4.86 0.85 7.07 1.74 5.49 1.07 5.15 0.94 4.96 0.88
1.25 3.92 0.62 5.56 1.20 4.35 0.76 4.10 0.68 3.97 0.65
1.50 3.27 0.48 4.62 0.90 3.64 0.61 3.44 0.55 3.33 0.51
1.75 2.89 0.37 3.96 0.73 3.16 0.47 3.02 0.40 2.94 0.37
2.00 2.54 0.50 3.48 0.60 2.83 0.43 2.68 0.47 2.60 0.49
2.25 2.18 0.38 3.13 0.48 2.53 0.50 2.33 0.47 2.23 0.42
2.50 2.03 0.16 2.86 0.45 2.20 0.40 2.08 0.27 2.04 0.19
2.75 2.00 0.04 2.62 0.50 2.05 0.21 2.01 0.10 2.00 0.05
3.00 1.99 0.04 2.36 0.48 2.01 0.08 2.00 0.04 1.99 0.04
0.90 0.00 90.70 71.07 236.3 217.7 130.6 112.0 109.3 89.39 97.16 78.61
0.25 25.73 12.25 44.83 28.75 30.56 16.28 27.98 14.02 26.29 12.92
0.50 10.61 3.31 16.37 6.66 12.10 4.11 11.32 3.65 10.80 3.39
0.75 6.63 1.58 9.82 3.08 7.50 1.95 7.05 1.76 6.79 1.66
1.00 4.89 1.00 7.10 1.87 5.50 1.22 5.18 1.11 5.01 1.04
1.25 3.92 0.72 5.59 1.28 4.38 0.86 4.12 0.78 3.99 0.73
1.50 3.30 0.56 4.62 0.96 3.66 0.67 3.46 0.61 3.36 0.57
1.75 2.89 0.46 3.97 0.77 3.17 0.52 3.02 0.49 2.94 0.48
2.00 2.54 0.51 3.49 0.62 2.83 0.47 2.68 0.50 2.60 0.50
2.25 2.21 0.41 3.14 0.52 2.53 0.51 2.35 0.48 2.26 0.44
2.50 2.05 0.21 2.86 0.48 2.24 0.43 2.12 0.33 2.07 0.25
2.75 2.00 0.08 2.62 0.51 2.07 0.25 2.02 0.15 2.01 0.10
3.00 1.99 0.10 2.38 0.49 2.01 0.11 2.00 0.08 1.99 0.09
1.00 0.00 370.1 349.2 370.3 358.2 370.2 354.2 369.9 354.9 369.8 353.4
0.25 26.54 14.16 44.87 29.93 31.36 18.06 28.31 15.77 27.12 14.67
0.50 10.62 3.67 16.39 6.97 12.15 4.45 11.34 4.00 10.85 3.72
0.75 6.66 1.76 9.87 3.23 7.52 2.11 7.10 1.96 6.81 1.83
1.00 4.89 1.09 7.12 1.97 5.53 1.32 5.18 1.21 5.02 1.14
1.25 3.93 0.79 5.59 1.34 4.39 0.93 4.13 0.84 4.01 0.82
(continued)
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
11

Table 2. Continued.
B
No error 1 2 3 5
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
1.50 3.30 0.60 4.62 0.99 3.67 0.71 3.48 0.66 3.38 0.62
1.75 2.90 0.51 3.97 0.81 3.19 0.57 3.02 0.52 2.94 0.52
2.00 2.55 0.52 3.50 0.66 2.83 0.51 2.68 0.51 2.60 0.52
2.25 2.24 0.43 3.14 0.56 2.53 0.52 2.36 0.49 2.29 0.46
2.50 2.07 0.26 2.87 0.51 2.25 0.44 2.14 0.34 2.09 0.29
2.75 2.01 0.12 2.62 0.51 2.09 0.29 2.03 0.19 2.01 0.15
3.00 1.99 0.13 2.38 0.49 2.02 0.16 2.00 0.11 1.99 0.11
1.10 0.00 73.26 58.26 166.7 151.8 98.05 81.13 83.53 68.17 76.48 62.22
0.25 24.20 13.12 40.97 26.50 28.23 16.11 26.07 14.52 24.66 13.49
0.50 10.55 3.90 16.35 7.25 12.09 4.70 11.20 4.27 10.81 3.98
0.75 6.66 1.92 9.87 3.41 7.52 2.30 7.10 2.07 6.81 2.01
1.00 4.89 1.21 7.12 2.05 5.53 1.44 5.18 1.31 5.02 1.25
1.25 3.93 0.85 5.59 1.42 4.39 0.99 4.13 0.92 4.01 0.87
1.50 3.30 0.66 4.62 1.07 3.67 0.76 3.48 0.70 3.38 0.68
1.75 2.89 0.56 3.97 0.84 3.19 0.61 3.02 0.57 2.94 0.57
2.00 2.54 0.53 3.50 0.68 2.83 0.55 2.68 0.53 2.59 0.54
2.25 2.24 0.44 3.14 0.58 2.53 0.53 2.36 0.49 2.29 0.46
2.50 2.07 0.27 2.86 0.53 2.25 0.45 2.14 0.37 2.10 0.32
2.75 2.01 0.17 2.62 0.52 2.09 0.30 2.03 0.22 2.01 0.19
3.00 1.98 0.15 2.38 0.50 2.02 0.18 2.00 0.15 1.99 0.15
1.25 0.00 20.52 10.70 49.56 35.34 27.03 15.90 23.38 13.04 21.58 11.60
0.25 16.20 7.54 30.30 17.41 19.51 9.70 17.83 8.47 16.59 7.60
0.50 9.93 3.76 15.58 6.97 11.50 4.56 10.66 4.10 10.22 3.86
0.75 6.65 2.10 9.77 3.54 7.51 2.47 7.05 2.26 6.81 2.14
1.00 4.90 1.34 7.12 2.18 5.53 1.56 5.19 1.46 5.03 1.38
1.25 3.93 0.95 5.59 1.51 4.39 1.10 4.13 1.01 4.02 0.99
1.50 3.30 0.74 4.62 1.12 3.68 0.84 3.48 0.79 3.38 0.76
1.75 2.89 0.62 3.97 0.88 3.19 0.68 3.02 0.65 2.94 0.63
2.00 2.54 0.55 3.50 0.75 2.83 0.59 2.68 0.57 2.59 0.56
2.25 2.24 0.46 3.14 0.62 2.53 0.54 2.36 0.51 2.29 0.48
2.50 2.07 0.32 2.86 0.56 2.25 0.47 2.14 0.39 2.10 0.35
2.75 2.01 0.21 2.61 0.54 2.09 0.34 2.03 0.27 2.01 0.23
3.00 1.97 0.22 2.38 0.50 2.02 0.23 2.00 0.20 1.98 0.21
1.50 0.00 9.23 3.75 18.41 9.36 11.39 4.92 10.12 4.18 9.52 3.79
0.25 8.83 3.38 16.45 7.79 10.65 4.40 9.62 3.79 9.10 3.54
0.50 7.57 2.67 12.56 5.14 8.92 3.25 8.14 2.92 7.76 2.76
0.75 6.05 1.96 9.28 3.38 6.93 2.36 6.48 2.11 6.23 2.02
1.00 4.85 1.46 7.05 2.31 5.45 1.67 5.14 1.57 4.96 1.49
1.25 3.93 1.10 5.59 1.67 4.39 1.26 4.13 1.17 4.01 1.12
1.50 3.30 0.85 4.62 1.26 3.68 0.98 3.48 0.91 3.38 0.88
1.75 2.89 0.71 3.97 1.00 3.19 0.78 3.02 0.75 2.94 0.73
2.00 2.54 0.61 3.50 0.82 2.83 0.67 2.68 0.64 2.59 0.62
2.25 2.25 0.50 3.14 0.70 2.53 0.58 2.36 0.54 2.29 0.53
2.50 2.07 0.40 2.86 0.63 2.25 0.50 2.14 0.45 2.10 0.42
2.75 2.01 0.31 2.61 0.57 2.09 0.39 2.03 0.34 2.01 0.32
3.00 1.94 0.31 2.38 0.52 2.02 0.30 1.99 0.29 1.96 0.31
2.00 0.00 6.63 1.55 7.92 3.03 5.47 1.90 5.01 1.70 4.80 1.57
0.25 4.63 1.52 7.72 2.87 5.41 1.86 4.98 1.67 4.73 1.59
0.50 4.50 1.47 7.35 2.64 5.20 1.72 4.81 1.60 4.62 1.49
0.75 4.23 1.34 6.73 2.29 4.88 1.56 4.54 1.45 4.35 1.39
1.00 3.92 1.19 5.92 1.93 4.46 1.37 4.13 1.26 3.98 1.20
1.25 3.52 1.05 5.18 1.62 4.00 1.20 3.76 1.12 3.61 1.08
1.50 3.18 0.93 4.52 1.33 3.56 1.04 3.37 0.99 3.26 0.95
1.75 2.85 0.80 3.97 1.14 3.18 0.90 3.01 0.84 2.91 0.82
2.00 2.55 0.70 3.50 0.97 2.83 0.77 2.68 0.73 2.59 0.72
2.25 2.26 0.60 3.14 0.84 2.53 0.67 2.36 0.63 2.29 0.62
2.50 2.07 0.52 2.86 0.73 2.25 0.59 2.14 0.55 2.10 0.53
2.75 2.01 0.46 2.61 0.64 2.09 0.51 2.03 0.48 2.01 0.48
3.00 1.92 0.44 2.38 0.58 2.02 0.45 1.99 0.44 1.95 0.44
(continued)
12 A. JAVAID ET AL.

Table 2. Continued.
B
No error 1 2 3 5
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
2.50 0.00 3.26 1.03 5.10 1.75 3.75 1.22 3.49 1.13 3.35 1.07
0.25 3.28 1.04 5.08 1.73 3.73 1.20 3.46 1.11 3.33 1.08
0.50 3.23 1.04 4.99 1.67 3.67 1.17 3.42 1.09 3.30 1.06
0.75 3.12 0.97 4.85 1.59 3.57 1.12 3.34 1.05 3.22 1.01
1.00 3.03 0.94 4.60 1.47 3.44 1.06 3.23 1.00 3.11 0.97
1.25 2.90 0.89 4.31 1.34 3.29 0.99 3.06 0.94 2.97 0.90
1.50 2.73 0.82 4.01 1.21 3.09 0.92 2.90 0.87 2.80 0.83
1.75 2.57 0.75 3.69 1.08 2.88 0.85 2.71 0.81 2.62 0.77
2.00 2.41 0.70 3.38 0.97 2.67 0.76 2.53 0.72 2.47 0.71
2.25 2.26 0.65 3.10 0.88 2.50 0.71 2.36 0.67 2.29 0.65
2.50 2.07 0.59 2.86 0.80 2.25 0.64 2.14 0.62 2.10 0.60
2.75 1.99 0.55 2.61 0.70 2.09 0.58 2.03 0.56 2.01 0.55
3.00 1.89 0.53 2.38 0.64 2.02 0.53 1.97 0.53 1.91 0.53
3.00 0.00 2.57 0.84 3.86 1.25 2.93 0.94 2.75 0.90 2.65 0.87
0.25 2.57 0.85 3.84 1.24 2.91 0.94 2.72 0.88 2.64 0.86
0.50 2.55 0.84 3.82 1.24 2.89 0.92 2.70 0.86 2.60 0.84
0.75 2.51 0.81 3.77 1.20 2.84 0.92 2.67 0.86 2.58 0.83
1.00 2.47 0.80 3.64 1.15 2.79 0.87 2.62 0.83 2.53 0.81
1.25 2.42 0.77 3.54 1.11 2.72 0.85 2.55 0.80 2.46 0.77
1.50 2.33 0.75 3.39 1.04 2.61 0.81 2.47 0.78 2.38 0.74
1.75 2.25 0.70 3.24 0.98 2.53 0.77 2.37 0.73 2.28 0.71
2.00 2.15 0.67 3.07 0.91 2.41 0.72 2.28 0.69 2.21 0.69
2.25 2.06 0.63 2.90 0.85 2.29 0.69 2.17 0.67 2.11 0.65
2.50 1.98 0.61 2.73 0.79 2.19 0.64 2.08 0.62 2.01 0.62
2.75 1.89 0.59 2.57 0.73 2.08 0.61 1.97 0.60 1.92 0.59
3.00 1.80 0.56 2.38 0.67 1.98 0.58 1.89 0.57 1.82 0.57

" ( )#
1 ððn  kÞ  1ÞS2
Vt ¼ / H , ððn  kÞ  1Þ (17)
B2 r2p þ r2m

PðnkÞ PðnkÞ
Yj  Þ2
ðYj Y
 ¼
where Y j¼1
, and S2 ¼ j¼1
: The EWMA statistics, Max-EWMA statistic and
ðnkÞ ððnkÞ1Þ
UCL will remain the same as explained in Sec. 3 in expressions (12), (13), (14) and (15). Hence,
the calculations of ARLs and SDRLs for multiple measurements have been shown in Tables 3–5.
Table 3 is prepared taking k ¼ 5 while rest of the parameters are taken just like in Table 1.
Table 3 tells us that using covariate model and taking multiple measurements of the quality
characteristic, the sensitivity of Max-EWMA chart has increased. In this table, five measurements
of each sample are taken and then the EWMA statistics and ARLs are calculated. However, no
error column is taken the same as in previous two tables, just to take the ARLs and SDRLs of
Max-EWMA chart without measurement error, in order to make the comparison on the same
grounds. It is clear from the values of ARLs and SDRLs shown in Table 3 that these values have
been reduced by approximately 50 % as compared to Table 1. It shows positive aspect of multiple
measurements that shift in mean and variance in joint monitoring has been detected far earlier as
compared to taking only one measurement. However, as the effect of error is increased from 0.1
to 1, the detection of shift is affected negatively as was also shown in Table 1. When we compare
Max-EWMA without error, the ARLs and SDRLs under five different values of r2m  r2 , are less
than those under no error column. This situation of multiple measurements is favorable for early
detection of a shift in joint monitoring, as compared to Tables 1 and 2, where the impact of
measurement error caused a delay in the detection of shift during the production process.
Therefore, multiple measurements of samples or observations of quality characteristic, is a good
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
13

Table 3. ARLs and SDRLs of Max-EWMA chart using covariate model for B ¼ 1 and k ¼ 5:
r2m  r2
No error 0.1 0.2 0.3 0.5 1
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
0.25 0.00 4.16 0.49 2.00 0.00 2.00 0.01 2.08 0.27 2.94 0.32 5.56 1.11
0.25 4.16 0.48 2.00 0.00 2.00 0.02 2.08 0.28 2.94 0.32 5.57 1.10
0.50 4.16 0.48 2.00 0.00 2.00 0.03 2.08 0.27 2.94 0.31 5.22 0.87
0.75 4.16 0.49 2.00 0.00 2.00 0.02 2.08 0.27 2.93 0.28 4.05 0.61
1.00 4.14 0.47 2.00 0.00 2.00 0.00 2.01 0.11 2.53 0.50 3.12 0.41
1.25 3.91 0.25 2.00 0.00 2.00 0.00 2.00 0.00 2.00 0.04 2.57 0.49
1.50 3.04 0.19 1.91 0.00 1.95 0.00 1.95 0.00 2.00 0.00 2.07 0.25
1.75 2.89 0.02 1.55 0.46 1.62 0.40 1.68 0.31 1.94 0.13 2.00 0.03
2.00 2.54 0.47 1.00 0.00 1.00 0.03 1.03 0.16 1.42 0.49 1.96 0.09
2.25 2.00 0.05 1.00 0.00 1.00 0.00 1.00 0.00 1.01 0.08 1.82 0.31
2.50 2.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.50 0.50
2.75 2.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.11 0.32
3.00 1.99 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.01 0.08
0.50 0.00 7.36 1.45 2.97 0.31 3.07 0.35 3.26 0.46 3.99 0.64 7.56 1.86
0.25 7.35 1.45 2.97 0.31 3.07 0.35 3.26 0.47 3.99 0.64 7.42 1.75
0.50 7.26 1.31 2.96 0.30 3.06 0.35 3.26 0.46 3.92 0.58 5.88 1.07
0.75 6.20 0.83 2.92 0.26 2.96 0.21 3.00 0.18 3.20 0.41 4.09 0.68
1.00 4.82 0.59 2.11 0.31 2.16 0.37 2.26 0.44 2.55 0.50 3.13 0.44
1.25 3.91 0.43 2.00 0.00 2.00 0.00 2.00 0.00 2.01 0.09 2.57 0.50
1.50 3.19 0.40 1.91 0.06 1.95 0.04 1.95 0.04 2.00 0.02 2.08 0.28
1.75 2.89 0.20 1.55 0.49 1.62 0.46 1.68 0.40 1.94 0.23 2.00 0.04
2.00 2.54 0.49 1.01 0.12 1.04 0.19 1.09 0.29 1.44 0.50 1.96 0.11
2.25 2.08 0.27 1.00 0.00 1.00 0.00 1.00 0.00 1.03 0.16 1.82 0.34
2.50 2.00 0.04 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.50 0.50
2.75 2.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.13 0.34
3.00 1.99 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.01 0.11
0.75 0.00 19.57 7.98 6.19 1.34 6.45 1.44 6.86 1.61 8.25 2.17 15.85 6.24
0.25 17.41 6.10 6.08 1.22 6.31 1.29 6.67 1.40 7.72 1.74 11.95 3.57
0.50 10.22 2.52 4.31 0.69 4.38 0.71 4.49 0.75 4.86 0.88 6.24 1.41
0.75 6.58 1.31 3.00 0.38 3.04 0.38 3.09 0.41 3.29 0.51 4.12 0.76
1.00 4.86 0.85 2.21 0.41 2.25 0.43 2.33 0.47 2.54 0.50 3.15 0.50
1.25 3.92 0.62 2.00 0.03 2.00 0.05 2.00 0.06 2.03 0.18 2.57 0.50
1.50 3.27 0.48 1.91 0.18 1.95 0.16 1.95 0.13 1.98 0.06 2.10 0.31
1.75 2.89 0.37 1.55 0.49 1.62 0.48 1.68 0.45 1.84 0.31 2.00 0.08
2.00 2.54 0.50 1.07 0.26 1.11 0.31 1.17 0.38 1.46 0.50 1.96 0.14
2.25 2.18 0.38 1.00 0.04 1.00 0.04 1.01 0.08 1.06 0.24 1.82 0.37
2.50 2.03 0.16 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.04 1.50 0.50
2.75 2.00 0.04 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.16 0.37
3.00 1.99 0.04 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.02 0.14
0.90 0.00 90.70 71.07 19.93 8.92 20.68 9.43 22.33 10.74 27.60 14.39 59.97 42.10
0.25 25.73 12.25 9.11 2.53 9.31 2.60 9.55 2.75 10.46 3.23 14.02 5.32
0.50 10.61 3.31 4.39 0.86 4.45 0.87 4.57 0.92 4.89 1.03 6.28 1.54
0.75 6.63 1.58 3.01 0.46 3.05 0.49 3.11 0.50 3.30 0.56 4.12 0.81
1.00 4.89 1.00 2.25 0.43 2.28 0.45 2.35 0.48 2.54 0.51 3.16 0.53
1.25 3.92 0.72 2.00 0.08 2.00 0.09 2.01 0.11 2.05 0.23 2.57 0.51
1.50 3.30 0.56 1.91 0.25 1.95 0.22 1.95 0.18 1.98 0.10 2.13 0.34
1.75 2.89 0.46 1.55 0.50 1.62 0.49 1.68 0.46 1.84 0.35 2.00 0.09
2.00 2.54 0.51 1.11 0.31 1.14 0.35 1.21 0.41 1.46 0.50 1.96 0.18
2.25 2.21 0.41 1.00 0.07 1.01 0.09 1.02 0.13 1.09 0.29 1.82 0.38
2.50 2.05 0.21 1.00 0.01 1.00 0.01 1.00 0.02 1.01 0.07 1.50 0.50
2.75 2.00 0.08 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.01 1.18 0.38
3.00 1.99 0.10 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.03 0.18
1.00 0.00 370.1 349.2 372.7 356.6 370.7 358.0 370.0 354.4 371.2 351.5 370.0 350.1
0.25 26.54 14.16 9.39 2.97 9.47 3.07 9.83 3.22 10.63 3.60 14.02 5.58
0.50 10.62 3.67 4.43 0.95 4.49 0.97 4.61 1.01 4.92 1.10 6.30 1.61
0.75 6.66 1.76 3.01 0.54 3.05 0.53 3.12 0.54 3.31 0.61 4.15 0.85
1.00 4.89 1.09 2.29 0.44 2.33 0.46 2.38 0.48 2.55 0.51 3.16 0.55
1.25 3.93 0.79 2.01 0.12 2.01 0.12 2.02 0.15 2.08 0.25 2.57 0.52
(continued)
14 A. JAVAID ET AL.

Table 3. Continued.
r2m  r2
No error 0.1 0.2 0.3 0.5 1
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
1.50 3.30 0.60 1.91 0.29 1.95 0.25 1.95 0.21 1.98 0.13 2.16 0.35
1.75 2.90 0.51 1.55 0.50 1.60 0.49 1.68 0.47 1.84 0.36 2.01 0.12
2.00 2.55 0.52 1.16 0.34 1.19 0.37 1.25 0.42 1.46 0.50 1.96 0.19
2.25 2.24 0.43 1.01 0.09 1.02 0.13 1.04 0.16 1.13 0.32 1.82 0.39
2.50 2.07 0.26 1.00 0.00 1.00 0.02 1.00 0.02 1.02 0.09 1.51 0.50
2.75 2.01 0.12 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.01 1.20 0.40
3.00 1.99 0.13 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.04 0.19
1.10 0.00 73.26 58.26 20.14 10.07 21.13 10.60 22.33 11.46 26.89 14.74 52.53 37.25
0.25 24.20 13.12 9.03 2.94 9.24 3.11 9.50 3.12 10.39 3.66 13.85 5.49
0.50 10.55 3.90 4.42 1.02 4.49 1.04 4.61 1.09 4.92 1.20 6.29 1.69
0.75 6.66 1.92 3.01 0.57 3.05 0.58 3.12 0.60 3.31 0.65 4.15 0.90
1.00 4.89 1.21 2.29 0.46 2.33 0.47 2.38 0.49 2.55 0.53 3.16 0.59
1.25 3.93 0.85 2.00 0.15 2.01 0.17 2.02 0.19 2.08 0.28 2.57 0.53
1.50 3.30 0.66 1.89 0.31 1.92 0.28 1.94 0.23 1.98 0.16 2.16 0.37
1.75 2.89 0.56 1.55 0.50 1.60 0.49 1.67 0.47 1.84 0.37 2.01 0.14
2.00 2.54 0.53 1.16 0.37 1.19 0.39 1.25 0.44 1.46 0.50 1.96 0.20
2.25 2.24 0.44 1.01 0.13 1.02 0.15 1.04 0.19 1.13 0.33 1.79 0.40
2.50 2.07 0.27 1.00 0.03 1.00 0.02 1.00 0.05 1.02 0.12 1.51 0.50
2.75 2.01 0.17 1.00 0.01 1.00 0.00 1.00 0.01 1.00 0.02 1.20 0.40
3.00 1.98 0.15 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.00 1.04 0.21
1.25 0.00 20.52 10.70 6.88 2.13 7.03 2.20 7.37 2.31 8.46 2.83 14.05 5.92
0.25 16.20 7.54 6.26 1.72 6.46 1.79 6.71 1.87 7.47 2.20 10.84 3.73
0.50 9.93 3.76 4.31 1.04 4.39 1.05 4.51 1.08 4.86 1.22 6.24 1.73
0.75 6.65 2.10 3.01 0.65 3.05 0.65 3.12 0.66 3.31 0.71 4.15 0.94
1.00 4.90 1.34 2.29 0.48 2.33 0.49 2.38 0.50 2.55 0.54 3.16 0.63
1.25 3.93 0.95 2.00 0.22 2.01 0.22 2.02 0.24 2.08 0.32 2.57 0.53
1.50 3.30 0.74 1.86 0.35 1.88 0.32 1.91 0.29 1.97 0.20 2.16 0.38
1.75 2.89 0.62 1.55 0.50 1.59 0.49 1.64 0.48 1.80 0.40 2.01 0.17
2.00 2.54 0.55 1.16 0.38 1.19 0.41 1.25 0.45 1.46 0.50 1.95 0.23
2.25 2.24 0.46 1.01 0.17 1.02 0.19 1.04 0.23 1.13 0.36 1.78 0.41
2.50 2.07 0.32 1.00 0.05 1.00 0.05 1.00 0.07 1.02 0.16 1.51 0.50
2.75 2.01 0.21 1.00 0.00 1.00 0.01 1.00 0.01 1.00 0.04 1.20 0.42
3.00 1.97 0.22 1.00 0.00 1.00 0.00 1.00 0.00 1.00 0.01 1.04 0.24
1.50 0.00 9.23 3.75 3.51 0.86 3.59 0.87 3.70 0.91 4.14 1.04 6.20 1.85
0.25 8.83 3.38 3.50 0.84 3.56 0.85 3.69 0.89 4.08 1.00 6.07 1.72
0.50 7.57 2.67 3.29 0.73 3.36 0.75 3.47 0.76 3.80 0.86 5.20 1.27
0.75 6.05 1.96 2.83 0.61 2.87 0.62 2.94 0.64 3.16 0.69 4.04 0.95
1.00 4.85 1.46 2.29 0.49 2.33 0.50 2.38 0.52 2.55 0.57 3.16 0.69
1.25 3.93 1.10 2.00 0.30 2.01 0.31 2.02 0.31 2.08 0.37 2.57 0.56
1.50 3.30 0.85 1.82 0.39 1.85 0.37 1.88 0.34 1.95 0.28 2.16 0.41
1.75 2.89 0.71 1.55 0.50 1.57 0.50 1.63 0.48 1.77 0.43 2.01 0.23
2.00 2.54 0.61 1.16 0.42 1.19 0.44 1.25 0.46 1.46 0.50 1.92 0.28
2.25 2.25 0.50 1.01 0.24 1.02 0.26 1.04 0.29 1.13 0.40 1.76 0.43
2.50 2.07 0.40 1.00 0.09 1.00 0.10 1.00 0.12 1.02 0.21 1.51 0.50
2.75 2.01 0.31 1.00 0.02 1.00 0.02 1.00 0.04 1.00 0.09 1.20 0.43
3.00 1.94 0.31 1.00 0.00 1.00 0.00 1.00 0.01 1.00 0.02 1.04 0.28
2.00 0.00 6.63 1.55 1.99 0.43 2.02 0.44 2.07 0.44 2.24 0.51 3.05 0.73
0.25 4.63 1.52 1.99 0.43 2.02 0.44 2.07 0.44 2.24 0.51 3.05 0.71
0.50 4.50 1.47 1.99 0.42 2.02 0.43 2.05 0.43 2.23 0.49 3.03 0.69
0.75 4.23 1.34 1.96 0.40 1.99 0.40 2.03 0.40 2.18 0.46 2.93 0.65
1.00 3.92 1.19 1.91 0.39 1.95 0.38 1.97 0.37 2.10 0.40 2.73 0.59
1.25 3.52 1.05 1.82 0.41 1.86 0.39 1.89 0.38 1.99 0.34 2.46 0.54
1.50 3.18 0.93 1.67 0.47 1.71 0.46 1.75 0.44 1.87 0.37 2.16 0.42
1.75 2.85 0.80 1.46 0.50 1.50 0.50 1.56 0.50 1.69 0.47 2.00 0.32
2.00 2.55 0.70 1.16 0.44 1.19 0.45 1.25 0.47 1.46 0.50 1.88 0.35
2.25 2.26 0.60 1.01 0.31 1.02 0.33 1.04 0.35 1.13 0.43 1.70 0.46
2.50 2.07 0.52 1.00 0.18 1.00 0.19 1.00 0.22 1.02 0.30 1.51 0.50
2.75 2.01 0.46 1.00 0.09 1.00 0.09 1.00 0.11 1.00 0.17 1.20 0.45
3.00 1.92 0.44 1.00 0.04 1.00 0.04 1.00 0.05 1.00 0.09 1.04 0.34
(continued)
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
15

Table 3. Continued.
r2m  r2
No error 0.1 0.2 0.3 0.5 1
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
2.50 0.00 3.26 1.03 1.38 0.49 1.41 0.49 1.47 0.50 1.63 0.50 2.13 0.47
0.25 3.28 1.04 1.38 0.49 1.41 0.49 1.46 0.50 1.63 0.49 2.13 0.47
0.50 3.23 1.04 1.38 0.49 1.41 0.49 1.46 0.50 1.63 0.49 2.12 0.45
0.75 3.12 0.97 1.38 0.49 1.41 0.49 1.46 0.50 1.63 0.49 2.11 0.45
1.00 3.03 0.94 1.36 0.48 1.40 0.49 1.45 0.50 1.61 0.49 2.08 0.43
1.25 2.90 0.89 1.33 0.47 1.36 0.48 1.41 0.49 1.58 0.50 2.04 0.39
1.50 2.73 0.82 1.28 0.45 1.30 0.46 1.35 0.48 1.51 0.50 1.97 0.36
1.75 2.57 0.75 1.20 0.40 1.23 0.42 1.27 0.44 1.41 0.49 1.89 0.37
2.00 2.41 0.70 1.12 0.33 1.15 0.35 1.18 0.38 1.30 0.46 1.79 0.42
2.25 2.26 0.65 1.01 0.25 1.02 0.26 1.02 0.29 1.13 0.39 1.65 0.48
2.50 2.07 0.59 1.00 0.16 1.00 0.18 1.00 0.21 1.02 0.30 1.49 0.50
2.75 1.99 0.55 1.00 0.11 1.00 0.11 1.00 0.13 1.00 0.19 1.20 0.46
3.00 1.89 0.53 1.00 0.06 1.00 0.06 1.00 0.07 1.00 0.11 1.04 0.37
3.00 0.00 2.57 0.84 1.07 0.25 1.08 0.28 1.11 0.31 1.20 0.40 1.69 0.49
0.25 2.57 0.85 1.07 0.26 1.08 0.28 1.10 0.30 1.19 0.39 1.68 0.49
0.50 2.55 0.84 1.07 0.26 1.08 0.28 1.10 0.31 1.19 0.40 1.68 0.48
0.75 2.51 0.81 1.07 0.26 1.08 0.27 1.10 0.30 1.19 0.39 1.68 0.49
1.00 2.47 0.80 1.06 0.24 1.08 0.27 1.10 0.30 1.17 0.38 1.67 0.48
1.25 2.42 0.77 1.06 0.24 1.07 0.25 1.09 0.28 1.17 0.37 1.65 0.49
1.50 2.33 0.75 1.05 0.22 1.06 0.24 1.08 0.27 1.15 0.36 1.64 0.49
1.75 2.25 0.70 1.04 0.20 1.05 0.21 1.06 0.24 1.12 0.33 1.60 0.49
2.00 2.15 0.67 1.03 0.17 1.03 0.18 1.04 0.20 1.09 0.29 1.53 0.50
2.25 2.06 0.63 1.01 0.12 1.02 0.13 1.02 0.16 1.07 0.25 1.44 0.50
2.50 1.98 0.61 1.00 0.09 1.00 0.10 1.00 0.12 1.02 0.19 1.33 0.47
2.75 1.89 0.59 1.00 0.06 1.00 0.08 1.00 0.08 1.00 0.14 1.20 0.43
3.00 1.80 0.56 1.00 0.04 1.00 0.04 1.00 0.05 1.00 0.09 1.04 0.35

solution to reduce the effect of measurement error in Max-EWMA chart, according to the results
shown in Table 3.
After studying the impact of multiple measurements of quality characteristic on measure-
ment error by taking k ¼ 5, B ¼ 1 and different values of ratio r2m  r2 in Table 3, let us con-
sider another aspect of these combinations by taking r2m  r2 ¼ 1, k ¼ 5 and different values
of B in Table 4. Just like Tables 2 and 3, the no error column has been taken from Table 1
in order to make a comparison with the Max-EWMA without measurement error. The ARLs
and SDRLs have been calculated for different values of B remaining other things same as in
Table 3.
Table 4 tells us that for multiple measurements having k ¼ 5, as the value of B increased from
1 to 5, the values of ARLs are successively decreased. These results are similar to the results of
Table 2 for early detection of process shift with increase in B to reduce the error effect. However,
it is identified that the ARLs in Table 4 are lesser than ARLs in Table 2, which is a clear indica-
tion that multiple measurements are a good solution to reduce the effect of measurement error in
detection of shift in joint monitoring of the production process. Let us see another aspect of mul-
tiple measurements by increasing the values of k from 5 to 50. For the purpose, calculations of
ARLs and SDRLs are shown in Table 5.
Table 5 tells us the story of multiple measurements to reduce the impact of measurement error
that as we increase the values of k from 5 to 10, 20, and 50, the values of ARLs and SDRLs are
reduced with every increase in the value of k: However, it is critical for the practitioner that how
to choose the value of k, as every increase in k is an addition of cost. The quality control expert
has to find out a breakeven point between the degree of measurement error and the value of k in
order to get the minimum value of ARL for early detection of process shift.
16 A. JAVAID ET AL.

Table 4. ARLs and SDRLs of Max-EWMA chart using covariate model for r2m  r2 ¼ 1, k ¼ 5 and different values of B.
B
No error 1 2 3 5
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
0.25 0.00 4.16 0.49 5.56 1.11 2.94 0.32 2.20 0.40 2.00 0.02
0.25 4.16 0.48 5.57 1.10 2.94 0.32 2.20 0.40 2.00 0.02
0.50 4.16 0.48 5.22 0.87 2.94 0.32 2.20 0.40 2.00 0.02
0.75 4.16 0.49 4.05 0.61 2.92 0.28 2.20 0.40 2.00 0.02
1.00 4.14 0.47 3.12 0.41 2.53 0.50 2.05 0.21 2.00 0.00
1.25 3.91 0.25 2.57 0.49 2.00 0.03 2.00 0.00 2.00 0.00
1.50 3.04 0.19 2.07 0.25 2.00 0.00 1.96 0.03 1.93 0.03
1.75 2.89 0.02 2.00 0.03 1.90 0.14 1.70 0.28 1.61 0.40
2.00 2.54 0.47 1.96 0.09 1.43 0.49 1.05 0.22 1.00 0.03
2.25 2.00 0.05 1.82 0.31 1.01 0.09 1.00 0.00 1.00 0.00
2.50 2.00 0.00 1.50 0.50 1.00 0.00 1.00 0.00 1.00 0.00
2.75 2.00 0.00 1.11 0.32 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.99 0.00 1.01 0.08 1.00 0.00 1.00 0.00 1.00 0.00
0.50 0.00 7.36 1.45 7.56 1.86 4.00 0.64 3.35 0.51 3.07 0.35
0.25 7.35 1.45 7.42 1.75 3.99 0.64 3.35 0.50 3.07 0.35
0.50 7.26 1.31 5.88 1.07 3.91 0.57 3.34 0.49 3.07 0.35
0.75 6.20 0.83 4.09 0.68 3.19 0.41 3.02 0.21 2.96 0.20
1.00 4.82 0.59 3.13 0.44 2.55 0.50 2.30 0.46 2.16 0.37
1.25 3.91 0.43 2.57 0.50 2.01 0.10 2.00 0.02 2.00 0.00
1.50 3.19 0.40 2.08 0.28 2.00 0.03 1.96 0.04 1.93 0.05
1.75 2.89 0.20 2.00 0.04 1.90 0.23 1.70 0.37 1.61 0.46
2.00 2.54 0.49 1.96 0.11 1.44 0.50 1.13 0.33 1.03 0.18
2.25 2.08 0.27 1.82 0.34 1.03 0.16 1.00 0.02 1.00 0.00
2.50 2.00 0.04 1.50 0.50 1.00 0.00 1.00 0.00 1.00 0.00
2.75 2.00 0.00 1.13 0.34 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.99 0.00 1.01 0.11 1.00 0.00 1.00 0.00 1.00 0.00
0.75 0.00 19.57 7.98 15.85 6.24 8.30 2.21 7.04 1.67 6.45 1.44
0.25 17.41 6.10 11.95 3.57 7.70 1.73 6.81 1.45 6.30 1.28
0.50 10.22 2.52 6.24 1.41 4.85 0.86 4.55 0.77 4.38 0.71
0.75 6.58 1.31 4.12 0.76 3.29 0.52 3.12 0.42 3.04 0.39
1.00 4.86 0.85 3.15 0.50 2.55 0.50 2.35 0.48 2.25 0.43
1.25 3.92 0.62 2.57 0.50 2.03 0.18 2.01 0.07 2.00 0.05
1.50 3.27 0.48 2.10 0.31 2.00 0.07 1.96 0.12 1.93 0.16
1.75 2.89 0.37 2.00 0.08 1.90 0.31 1.70 0.43 1.61 0.48
2.00 2.54 0.50 1.96 0.14 1.45 0.50 1.20 0.40 1.11 0.31
2.25 2.18 0.38 1.82 0.37 1.06 0.25 1.01 0.09 1.00 0.04
2.50 2.03 0.16 1.50 0.50 1.00 0.04 1.00 0.01 1.00 0.00
2.75 2.00 0.04 1.16 0.37 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.99 0.04 1.02 0.14 1.00 0.00 1.00 0.00 1.00 0.00
0.90 0.00 90.70 71.07 59.97 42.10 27.46 14.53 23.14 11.13 20.82 9.49
0.25 25.73 12.25 14.02 5.32 10.35 3.15 9.69 2.78 9.27 2.59
0.50 10.61 3.31 6.28 1.54 4.89 1.02 4.60 0.91 4.45 0.88
0.75 6.63 1.58 4.12 0.81 3.30 0.56 3.13 0.50 3.05 0.49
1.00 4.89 1.00 3.16 0.53 2.54 0.51 2.37 0.48 2.28 0.45
1.25 3.92 0.72 2.57 0.51 2.05 0.23 2.02 0.13 2.01 0.09
1.50 3.30 0.56 2.13 0.34 1.99 0.10 1.96 0.17 1.93 0.22
1.75 2.89 0.46 2.00 0.09 1.84 0.34 1.70 0.45 1.61 0.48
2.00 2.54 0.51 1.96 0.18 1.46 0.50 1.24 0.43 1.15 0.35
2.25 2.21 0.41 1.82 0.38 1.09 0.29 1.02 0.14 1.01 0.09
2.50 2.05 0.21 1.50 0.50 1.01 0.07 1.00 0.02 1.00 0.01
2.75 2.00 0.08 1.18 0.38 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.99 0.10 1.03 0.18 1.00 0.00 1.00 0.00 1.00 0.00
1.00 0.00 370.1 349.2 370.0 350.1 370.4 354.8 369.5 355.9 368.1 358.6
0.25 26.54 14.16 14.02 5.58 10.56 3.63 9.91 3.20 9.55 3.07
0.50 10.62 3.67 6.30 1.61 4.93 1.11 4.64 1.01 4.49 0.96
0.75 6.66 1.76 4.15 0.85 3.30 0.60 3.14 0.56 3.05 0.54
1.00 4.89 1.09 3.16 0.55 2.55 0.52 2.38 0.49 2.32 0.47
1.25 3.93 0.79 2.57 0.52 2.06 0.25 2.02 0.16 2.01 0.13
(continued)
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
17

Table 4. Continued.
B
No error 1 2 3 5
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
1.50 3.30 0.60 2.16 0.35 1.99 0.13 1.96 0.20 1.93 0.25
1.75 2.90 0.51 2.01 0.12 1.84 0.36 1.70 0.46 1.61 0.49
2.00 2.55 0.52 1.96 0.19 1.46 0.50 1.26 0.44 1.16 0.37
2.25 2.24 0.43 1.82 0.39 1.11 0.31 1.04 0.19 1.01 0.12
2.50 2.07 0.26 1.51 0.50 1.01 0.10 1.00 0.04 1.00 0.01
2.75 2.01 0.12 1.20 0.40 1.00 0.02 1.00 0.00 1.00 0.00
3.00 1.99 0.13 1.04 0.19 1.00 0.00 1.00 0.00 1.00 0.00
1.10 0.00 73.26 58.26 52.53 37.25 27.01 15.40 22.97 12.12 20.97 10.45
0.25 24.20 13.12 13.85 5.49 10.36 3.60 9.58 3.18 9.22 3.04
0.50 10.55 3.90 6.29 1.69 4.93 1.18 4.63 1.09 4.49 1.06
0.75 6.66 1.92 4.15 0.90 3.30 0.65 3.14 0.61 3.05 0.59
1.00 4.89 1.21 3.16 0.59 2.55 0.53 2.38 0.50 2.32 0.47
1.25 3.93 0.85 2.57 0.53 2.06 0.28 2.02 0.19 2.01 0.15
1.50 3.30 0.66 2.16 0.37 1.98 0.16 1.94 0.23 1.91 0.29
1.75 2.89 0.56 2.01 0.14 1.82 0.38 1.70 0.46 1.59 0.49
2.00 2.54 0.53 1.96 0.20 1.46 0.50 1.26 0.45 1.16 0.39
2.25 2.24 0.44 1.79 0.40 1.11 0.33 1.04 0.21 1.01 0.14
2.50 2.07 0.27 1.51 0.50 1.01 0.12 1.00 0.06 1.00 0.04
2.75 2.01 0.17 1.20 0.40 1.00 0.03 1.00 0.00 1.00 0.00
3.00 1.98 0.15 1.04 0.21 1.00 0.00 1.00 0.00 1.00 0.00
1.25 0.00 20.52 10.70 14.05 5.92 8.46 2.83 7.53 2.43 7.09 2.22
0.25 16.20 7.54 10.84 3.73 7.44 2.17 6.79 1.90 6.42 1.79
0.50 9.93 3.76 6.24 1.73 4.83 1.21 4.53 1.10 4.38 1.04
0.75 6.65 2.10 4.15 0.94 3.30 0.71 3.14 0.66 3.05 0.64
1.00 4.90 1.34 3.16 0.63 2.55 0.54 2.38 0.51 2.32 0.49
1.25 3.93 0.95 2.57 0.53 2.06 0.32 2.02 0.25 2.01 0.22
1.50 3.30 0.74 2.16 0.38 1.97 0.20 1.92 0.28 1.89 0.32
1.75 2.89 0.62 2.01 0.17 1.79 0.41 1.67 0.47 1.59 0.49
2.00 2.54 0.55 1.95 0.23 1.46 0.50 1.26 0.46 1.16 0.41
2.25 2.24 0.46 1.78 0.41 1.11 0.36 1.04 0.24 1.01 0.19
2.50 2.07 0.32 1.51 0.50 1.01 0.17 1.00 0.08 1.00 0.05
2.75 2.01 0.21 1.20 0.42 1.00 0.05 1.00 0.01 1.00 0.02
3.00 1.97 0.22 1.04 0.24 1.00 0.00 1.00 0.00 1.00 0.00
1.50 0.00 9.23 3.75 6.20 1.85 4.13 1.05 3.77 0.92 3.60 0.87
0.25 8.83 3.38 6.07 1.72 4.08 1.02 3.74 0.91 3.57 0.84
0.50 7.57 2.67 5.20 1.27 3.78 0.85 3.51 0.78 3.37 0.74
0.75 6.05 1.96 4.04 0.95 3.16 0.68 2.97 0.64 2.87 0.61
1.00 4.85 1.46 3.16 0.69 2.55 0.56 2.38 0.53 2.32 0.51
1.25 3.93 1.10 2.57 0.56 2.06 0.37 2.02 0.32 2.01 0.30
1.50 3.30 0.85 2.16 0.41 1.95 0.28 1.88 0.34 1.84 0.38
1.75 2.89 0.71 2.01 0.23 1.77 0.42 1.64 0.48 1.57 0.50
2.00 2.54 0.61 1.92 0.28 1.46 0.50 1.26 0.47 1.16 0.44
2.25 2.25 0.50 1.76 0.43 1.11 0.40 1.04 0.31 1.01 0.26
2.50 2.07 0.40 1.51 0.50 1.01 0.22 1.00 0.14 1.00 0.10
2.75 2.01 0.31 1.20 0.43 1.00 0.09 1.00 0.05 1.00 0.03
3.00 1.94 0.31 1.04 0.28 1.00 0.03 1.00 0.01 1.00 0.01
2.00 0.00 6.63 1.55 3.05 0.73 2.24 0.50 2.09 0.45 2.02 0.44
0.25 4.63 1.52 3.05 0.71 2.23 0.50 2.09 0.45 2.02 0.44
0.50 4.50 1.47 3.03 0.69 2.23 0.50 2.09 0.44 2.01 0.43
0.75 4.23 1.34 2.93 0.65 2.19 0.46 2.06 0.41 1.99 0.41
1.00 3.92 1.19 2.73 0.59 2.10 0.41 1.99 0.38 1.94 0.38
1.25 3.52 1.05 2.46 0.54 1.99 0.36 1.91 0.36 1.85 0.39
1.50 3.18 0.93 2.16 0.42 1.86 0.38 1.77 0.43 1.71 0.46
1.75 2.85 0.80 2.00 0.32 1.69 0.47 1.58 0.49 1.51 0.50
2.00 2.55 0.70 1.88 0.35 1.45 0.50 1.35 0.48 1.16 0.45
2.25 2.26 0.60 1.70 0.46 1.11 0.43 1.04 0.37 1.01 0.33
2.50 2.07 0.52 1.51 0.50 1.01 0.30 1.00 0.22 1.00 0.20
2.75 2.01 0.46 1.20 0.45 1.00 0.17 1.00 0.12 1.00 0.09
3.00 1.92 0.44 1.04 0.34 1.00 0.08 1.00 0.05 1.00 0.02
(continued)
18 A. JAVAID ET AL.

Table 4. Continued.
B
No error 1 2 3 5
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
2.50 0.00 3.26 1.03 2.13 0.47 1.63 0.50 1.49 0.51 1.42 0.50
0.25 3.28 1.04 2.13 0.47 1.63 0.49 1.49 0.50 1.42 0.50
0.50 3.23 1.04 2.12 0.45 1.63 0.50 1.49 0.50 1.41 0.49
0.75 3.12 0.97 2.11 0.45 1.61 0.50 1.48 0.50 1.41 0.49
1.00 3.03 0.94 2.08 0.43 1.61 0.49 1.46 0.50 1.39 0.49
1.25 2.90 0.89 2.04 0.39 1.58 0.50 1.43 0.50 1.36 0.48
1.50 2.73 0.82 1.97 0.36 1.52 0.50 1.38 0.49 1.31 0.46
1.75 2.57 0.75 1.89 0.37 1.42 0.49 1.28 0.45 1.22 0.42
2.00 2.41 0.70 1.79 0.42 1.30 0.46 1.20 0.40 1.14 0.34
2.25 2.26 0.65 1.65 0.48 1.11 0.38 1.04 0.31 1.01 0.27
2.50 2.07 0.59 1.49 0.50 1.01 0.29 1.00 0.22 1.00 0.18
2.75 1.99 0.55 1.20 0.46 1.00 0.20 1.00 0.14 1.00 0.11
3.00 1.89 0.53 1.04 0.37 1.00 0.11 1.00 0.08 1.00 0.06
3.00 0.00 2.57 0.84 1.69 0.49 1.19 0.39 1.12 0.32 1.09 0.28
0.25 2.57 0.85 1.68 0.49 1.19 0.40 1.12 0.32 1.09 0.28
0.50 2.55 0.84 1.68 0.48 1.19 0.40 1.12 0.32 1.08 0.28
0.75 2.51 0.81 1.68 0.49 1.19 0.39 1.11 0.31 1.08 0.28
1.00 2.47 0.80 1.67 0.48 1.18 0.39 1.11 0.31 1.08 0.26
1.25 2.42 0.77 1.65 0.49 1.17 0.37 1.10 0.30 1.07 0.25
1.50 2.33 0.75 1.64 0.49 1.15 0.36 1.09 0.28 1.06 0.24
1.75 2.25 0.70 1.60 0.49 1.12 0.33 1.07 0.26 1.05 0.21
2.00 2.15 0.67 1.53 0.50 1.09 0.29 1.05 0.21 1.03 0.17
2.25 2.06 0.63 1.44 0.50 1.06 0.23 1.03 0.17 1.01 0.14
2.50 1.98 0.61 1.33 0.47 1.01 0.19 1.00 0.14 1.00 0.10
2.75 1.89 0.59 1.20 0.43 1.00 0.14 1.00 0.09 1.00 0.06
3.00 1.80 0.56 1.04 0.35 1.00 0.09 1.00 0.06 1.00 0.04

After analyzing the impact of measurement error on the efficiency of the Max-EWMA chart
and its solution in terms of multiple measurements, the comparison with other charts is also
deemed necessary to provide an idea about the preference of this chart for real life utilization.

5. Comparison with other charts


We have analyzed the efficiency of Max-EWMA chart for joint monitoring of mean and variance
and the impact of measurement error to reduce the chart performance. The impact of measure-
ment error is reduced by applying the method of multiple measurements of the same selected
unit to be used for the construction of the chart. However, it is also important to compare the
efficiency of our proposed chart with other charts to show its preference over the counterparts
for real life utilization. For this purpose, we name our Max-EWMA with measurement error con-
trol chart as Max-EWMAME chart and compare it with two counterparts i.e. Maravelakis,
Panaretos, and Psarakis (2004), and Maravelakis (2012). We calculated ARL values for compari-
son purpose with ARL0 ¼ 370 for all three types of charts. There is no change in variance i.e.
b ¼ 1, but shift in mean is shown for comparison purpose. Table 6 shows the calculated values of
ARLs for comparison between the proposed chart Max-EWMAME, and control charts proposed
by various authors during 2004 and 2012. The paper of Maravelakis, Panaretos, and Psarakis
(2004) analyzed EWMA chart with measurement error while paper of Maravelakis (2012) high-
lighted measurement error effect on the CUSUM control chart. The values for 2004 have been
taken from that paper but values for 2012 have been calculated afresh due to non-availability of
those values with ARL0¼370.
Table 6 reveals supportive results to our proposed Max-EWMAME control chart that it is
preferable as compared with other two counter parts developed in 2004 and 2012. The column
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
19

Table 5. ARLs and SDRLs of Max-EWMA chart for various values of k taking r2m  r2 ¼ 1 and B ¼ 1:
k
No error 5 10 20 50
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
0.25 0.00 4.16 0.49 5.56 1.11 3.88 0.66 2.84 0.43 2.00 0.05
0.25 4.16 0.48 5.57 1.10 2.88 0.67 2.84 0.43 2.00 0.04
0.50 4.16 0.48 5.22 0.87 3.71 0.56 2.80 0.41 2.00 0.04
0.75 4.16 0.49 4.05 0.61 2.97 0.33 2.06 0.23 1.67 0.44
1.00 4.14 0.47 3.12 0.41 2.19 0.39 1.97 0.09 1.00 0.02
1.25 3.91 0.25 2.57 0.49 2.00 0.03 1.49 0.50 1.00 0.00
1.50 3.04 0.19 2.07 0.25 1.92 0.18 1.01 0.09 1.00 0.00
1.75 2.89 0.02 2.00 0.03 1.54 0.50 1.00 0.00 1.00 0.00
2.00 2.54 0.47 1.96 0.09 1.06 0.23 1.00 0.00 1.00 0.00
2.25 2.00 0.05 1.82 0.31 1.00 0.01 1.00 0.00 1.00 0.00
2.50 2.00 0.00 1.50 0.50 1.00 0.00 1.00 0.00 1.00 0.00
2.75 2.00 0.00 1.11 0.32 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.99 0.00 1.01 0.08 1.00 0.00 1.00 0.00 1.00 0.00
0.50 0.00 7.36 1.45 7.56 1.86 5.08 1.00 3.58 0.62 2.31 0.46
0.25 7.35 1.45 7.42 1.75 5.07 0.99 3.57 0.61 2.29 0.45
0.50 7.26 1.31 5.88 1.07 4.17 0.65 3.05 0.37 2.01 0.09
0.75 6.20 0.83 4.09 0.68 2.99 0.39 2.08 0.27 1.67 0.44
1.00 4.82 0.59 3.13 0.44 2.20 0.40 1.97 0.10 1.00 0.04
1.25 3.91 0.43 2.57 0.50 2.00 0.04 1.49 0.50 1.00 0.00
1.50 3.19 0.40 2.08 0.28 1.92 0.21 1.01 0.11 1.00 0.00
1.75 2.89 0.20 2.00 0.04 1.54 0.50 1.00 0.00 1.00 0.00
2.00 2.54 0.49 1.96 0.11 1.07 0.25 1.00 0.00 1.00 0.00
2.25 2.08 0.27 1.82 0.34 1.00 0.03 1.00 0.00 1.00 0.00
2.50 2.00 0.04 1.50 0.50 1.00 0.00 1.00 0.00 1.00 0.00
2.75 2.00 0.00 1.13 0.34 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.99 0.00 1.01 0.11 1.00 0.00 1.00 0.00 1.00 0.00
0.75 0.00 19.57 7.98 15.85 6.24 10.04 3.09 6.72 1.67 4.14 0.80
0.25 17.41 6.10 11.95 3.57 8.02 1.91 5.64 1.10 3.63 0.58
0.50 10.22 2.52 6.24 1.41 4.37 0.83 3.15 0.50 2.04 0.21
0.75 6.58 1.31 4.12 0.76 3.00 0.46 2.11 0.31 1.67 0.46
1.00 4.86 0.85 3.15 0.50 2.23 0.42 1.97 0.14 1.00 0.06
1.25 3.92 0.62 2.57 0.50 2.00 0.07 1.49 0.50 1.00 0.00
1.50 3.27 0.48 2.10 0.31 1.92 0.25 1.02 0.15 1.00 0.00
1.75 2.89 0.37 2.00 0.08 1.54 0.50 1.00 0.01 1.00 0.00
2.00 2.54 0.50 1.96 0.14 1.09 0.29 1.00 0.00 1.00 0.00
2.25 2.18 0.38 1.82 0.37 1.01 0.05 1.00 0.00 1.00 0.00
2.50 2.03 0.16 1.50 0.50 1.00 0.00 1.00 0.00 1.00 0.00
2.75 2.00 0.04 1.16 0.37 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.99 0.04 1.02 0.14 1.00 0.00 1.00 0.00 1.00 0.00
0.90 0.00 90.70 71.07 59.97 42.10 33.47 19.54 19.76 9.01 10.91 3.65
0.25 25.73 12.25 14.02 5.32 9.23 2.74 6.24 1.51 3.92 0.75
0.50 10.61 3.31 6.28 1.54 4.38 0.89 3.15 0.53 2.06 0.24
0.75 6.63 1.58 4.12 0.81 3.01 0.50 2.13 0.34 1.67 0.47
1.00 4.89 1.00 3.16 0.53 2.24 0.43 1.97 0.18 1.01 0.08
1.25 3.92 0.72 2.57 0.51 2.00 0.09 1.50 0.50 1.00 0.00
1.50 3.30 0.56 2.13 0.34 1.92 0.27 1.03 0.17 1.00 0.00
1.75 2.89 0.46 2.00 0.09 1.54 0.50 1.00 0.01 1.00 0.00
2.00 2.54 0.51 1.96 0.18 1.11 0.31 1.00 0.00 1.00 0.00
2.25 2.21 0.41 1.82 0.38 1.01 0.09 1.00 0.00 1.00 0.00
2.50 2.05 0.21 1.50 0.50 1.00 0.01 1.00 0.00 1.00 0.00
2.75 2.00 0.08 1.18 0.38 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.99 0.10 1.03 0.18 1.00 0.00 1.00 0.00 1.00 0.00
1.00 0.00 370.1 349.2 370.0 350.1 370.9 353.8 370.5 354.9 369.5 355.2
0.25 26.54 14.16 14.02 5.58 9.29 2.97 6.32 1.64 3.95 0.80
0.50 10.62 3.67 6.30 1.61 4.40 0.95 3.17 0.56 2.07 0.25
0.75 6.66 1.76 4.15 0.85 3.01 0.52 2.15 0.35 1.67 0.47
1.00 4.89 1.09 3.16 0.55 2.27 0.44 1.97 0.19 1.02 0.10
1.25 3.93 0.79 2.57 0.52 2.00 0.11 1.50 0.50 1.00 0.00
(continued)
20 A. JAVAID ET AL.

Table 5. Continued.
k
No error 5 10 20 50
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
1.50 3.30 0.60 2.16 0.35 1.92 0.28 1.05 0.20 1.00 0.00
1.75 2.90 0.51 2.01 0.12 1.54 0.50 1.00 0.01 1.00 0.00
2.00 2.55 0.52 1.96 0.19 1.14 0.33 1.00 0.00 1.00 0.00
2.25 2.24 0.43 1.82 0.39 1.01 0.09 1.00 0.00 1.00 0.00
2.50 2.07 0.26 1.51 0.50 1.00 0.00 1.00 0.00 1.00 0.00
2.75 2.01 0.12 1.20 0.40 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.99 0.13 1.04 0.19 1.00 0.00 1.00 0.00 1.00 0.00
1.10 0.00 73.26 58.26 52.53 37.25 30.55 17.65 18.81 8.83 10.58 3.72
0.25 24.20 13.12 13.85 5.49 9.23 2.96 6.28 1.64 3.93 0.82
0.50 10.55 3.90 6.29 1.69 4.40 0.99 3.17 0.58 2.07 0.27
0.75 6.66 1.92 4.15 0.90 3.01 0.56 2.15 0.36 1.67 0.47
1.00 4.89 1.21 3.16 0.59 2.27 0.45 1.95 0.21 1.02 0.11
1.25 3.93 0.85 2.57 0.53 2.00 0.14 1.50 0.50 1.00 0.00
1.50 3.30 0.66 2.16 0.37 1.89 0.31 1.05 0.21 1.00 0.00
1.75 2.89 0.56 2.01 0.14 1.53 0.50 1.00 0.01 1.00 0.00
2.00 2.54 0.53 1.96 0.20 1.14 0.35 1.00 0.00 1.00 0.00
2.25 2.24 0.44 1.79 0.40 1.01 0.11 1.00 0.00 1.00 0.00
2.50 2.07 0.27 1.51 0.50 1.00 0.02 1.00 0.00 1.00 0.00
2.75 2.01 0.17 1.20 0.40 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.98 0.15 1.04 0.21 1.00 0.00 1.00 0.00 1.00 0.00
1.25 0.00 20.52 10.70 14.05 5.92 9.11 3.08 6.13 1.65 3.83 0.82
0.25 16.20 7.54 10.84 3.73 7.50 2.10 5.30 1.21 3.44 0.63
0.50 9.93 3.76 6.24 1.73 4.39 1.02 3.16 0.62 2.07 0.30
0.75 6.65 2.10 4.15 0.94 3.01 0.58 2.15 0.38 1.65 0.48
1.00 4.90 1.34 3.16 0.63 2.27 0.46 1.95 0.23 1.02 0.14
1.25 3.93 0.95 2.57 0.53 2.00 0.17 1.50 0.50 1.00 0.00
1.50 3.30 0.74 2.16 0.38 1.89 0.31 1.05 0.24 1.00 0.00
1.75 2.89 0.62 2.01 0.17 1.53 0.50 1.00 0.04 1.00 0.00
2.00 2.54 0.55 1.95 0.23 1.14 0.36 1.00 0.00 1.00 0.00
2.25 2.24 0.46 1.78 0.41 1.01 0.13 1.00 0.00 1.00 0.00
2.50 2.07 0.32 1.51 0.50 1.00 0.01 1.00 0.00 1.00 0.00
2.75 2.01 0.21 1.20 0.42 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.97 0.22 1.04 0.24 1.00 0.00 1.00 0.00 1.00 0.00
1.50 0.00 9.23 3.75 6.20 1.85 4.25 1.03 3.05 0.63 2.03 0.28
0.25 8.83 3.38 6.07 1.72 4.20 0.99 3.03 0.62 2.03 0.26
0.50 7.57 2.67 5.20 1.27 3.76 0.78 2.77 0.53 1.98 0.18
0.75 6.05 1.96 4.04 0.95 2.95 0.60 2.15 0.38 1.62 0.48
1.00 4.85 1.46 3.16 0.69 2.27 0.47 1.92 0.28 1.02 0.18
1.25 3.93 1.10 2.57 0.56 2.00 0.23 1.50 0.50 1.00 0.00
1.50 3.30 0.85 2.16 0.41 1.86 0.35 1.05 0.28 1.00 0.00
1.75 2.89 0.71 2.01 0.23 1.53 0.50 1.00 0.04 1.00 0.00
2.00 2.54 0.61 1.92 0.28 1.14 0.39 1.00 0.00 1.00 0.00
2.25 2.25 0.50 1.76 0.43 1.01 0.17 1.00 0.00 1.00 0.00
2.50 2.07 0.40 1.51 0.50 1.00 0.04 1.00 0.00 1.00 0.00
2.75 2.01 0.31 1.20 0.43 1.00 0.00 1.00 0.00 1.00 0.00
3.00 1.94 0.31 1.04 0.28 1.00 0.00 1.00 0.00 1.00 0.00
2.00 0.00 6.63 1.55 3.05 0.73 2.21 0.45 1.70 0.46 1.00 0.06
0.25 4.63 1.52 3.05 0.71 2.21 0.46 1.70 0.46 1.00 0.06
0.50 4.50 1.47 3.03 0.69 2.19 0.43 1.70 0.46 1.00 0.05
0.75 4.23 1.34 2.93 0.65 2.15 0.39 1.69 0.46 1.00 0.04
1.00 3.92 1.19 2.73 0.59 2.05 0.30 1.61 0.49 1.00 0.01
1.25 3.52 1.05 2.46 0.54 1.95 0.28 1.35 0.48 1.00 0.00
1.50 3.18 0.93 2.16 0.42 1.78 0.42 1.05 0.29 1.00 0.00
1.75 2.85 0.80 2.00 0.32 1.51 0.50 1.00 0.09 1.00 0.00
2.00 2.55 0.70 1.88 0.35 1.14 0.42 1.00 0.02 1.00 0.00
2.25 2.26 0.60 1.70 0.46 1.01 0.25 1.00 0.00 1.00 0.00
2.50 2.07 0.52 1.51 0.50 1.00 0.10 1.00 0.00 1.00 0.00
2.75 2.01 0.46 1.20 0.45 1.00 0.03 1.00 0.00 1.00 0.00
3.00 1.92 0.44 1.04 0.34 1.00 0.01 1.00 0.00 1.00 0.00
(continued)
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
21

Table 5. Continued.
k
No error 5 10 20 50
b a ARL SDRL ARL SDRL ARL SDRL ARL SDRL ARL SDRL
2.50 0.00 3.26 1.03 2.13 0.47 1.56 0.50 1.02 0.15 1.00 0.00
0.25 3.28 1.04 2.13 0.47 1.56 0.50 1.02 0.15 1.00 0.00
0.50 3.23 1.04 2.12 0.45 1.56 0.50 1.02 0.15 1.00 0.00
0.75 3.12 0.97 2.11 0.45 1.56 0.50 1.02 0.14 1.00 0.00
1.00 3.03 0.94 2.08 0.43 1.55 0.50 1.02 0.13 1.00 0.00
1.25 2.90 0.89 2.04 0.39 1.52 0.50 1.01 0.11 1.00 0.00
1.50 2.73 0.82 1.97 0.36 1.42 0.50 1.00 0.07 1.00 0.00
1.75 2.57 0.75 1.89 0.37 1.29 0.45 1.00 0.03 1.00 0.00
2.00 2.41 0.70 1.79 0.42 1.14 0.35 1.00 0.01 1.00 0.00
2.25 2.26 0.65 1.65 0.48 1.01 0.24 1.00 0.00 1.00 0.00
2.50 2.07 0.59 1.49 0.50 1.00 0.12 1.00 0.00 1.00 0.00
2.75 1.99 0.55 1.20 0.46 1.00 0.06 1.00 0.00 1.00 0.00
3.00 1.89 0.53 1.04 0.37 1.00 0.02 1.00 0.00 1.00 0.00
3.00 0.00 2.57 0.84 1.69 0.49 1.09 0.28 1.00 0.01 1.00 0.00
0.25 2.57 0.85 1.68 0.49 1.09 0.28 1.00 0.00 1.00 0.00
0.50 2.55 0.84 1.68 0.48 1.09 0.28 1.00 0.00 1.00 0.00
0.75 2.51 0.81 1.68 0.49 1.09 0.29 1.00 0.00 1.00 0.00
1.00 2.47 0.80 1.67 0.48 1.08 0.28 1.00 0.00 1.00 0.00
1.25 2.42 0.77 1.65 0.49 1.08 0.26 1.00 0.00 1.00 0.00
1.50 2.33 0.75 1.64 0.49 1.06 0.24 1.00 0.00 1.00 0.00
1.75 2.25 0.70 1.60 0.49 1.05 0.21 1.00 0.00 1.00 0.00
2.00 2.15 0.67 1.53 0.50 1.03 0.16 1.00 0.00 1.00 0.00
2.25 2.06 0.63 1.44 0.50 1.01 0.11 1.00 0.00 1.00 0.00
2.50 1.98 0.61 1.33 0.47 1.00 0.07 1.00 0.00 1.00 0.00
2.75 1.89 0.59 1.20 0.43 1.00 0.03 1.00 0.00 1.00 0.00
3.00 1.80 0.56 1.04 0.35 1.00 0.02 1.00 0.00 1.00 0.00

Table 6. Comparison among ARL values of Max-EWMAME, 2004, and 2012.


r2m  r2
No error 0.1 0.5 1.0
a Max 2004 2012 Max 2004 2012 Max 2004 2012 ARL 2004 2012
0.0 370.1 370.2 371.1 369.9 370.3 370.2 372.2 370.3 370.6 370.9 370.3 371.8
0.5 10.62 41.13 27.71 10.75 45.22 29.45 12.17 60.96 35.52 16.47 79.06 42.14
1.0 4.89 10.25 8.55 4.94 11.21 9.24 5.53 15.15 11.70 7.11 20.26 14.55
1.5 3.30 5.18 4.84 3.33 5.57 5.23 3.69 7.16 6.62 4.65 9.20 8.26
2.0 2.55 3.46 3.41 2.55 3.69 3.66 2.83 4.57 4.64 3.48 5.67 5.77
2.5 2.07 2.65 2.67 2.07 2.80 2.86 2.26 3.37 3.59 2.86 4.08 4.48
3.0 1.99 2.19 2.24 1.99 2.29 2.38 2.02 2.71 2.95 2.38 3.22 3.67

Max is for our proposed Max-EWMAME chart, column 2004 is for paper of 2004 and column
2012 is for the values of paper published in 2012. It is clear from the ARL values under all the
columns that with every shift in mean from 0.5 to 3.00, the ARLs are decreasing which indicates
the efficiency of all three charts for early detection of process shifts. However, it is also evident
that the ARL values for column under Max are lesser than the other two columns under 2004
and 2012 for every shift in mean. These results have the same trends for no error as well as for
measurement errors showing 0.1, 0.5, and 1.0 in this table. We may construe from Table 6 that
our proposed chart Max-EWMAME is preferable than the two counterparts.

6. Illustrative example
In order to study the impact of measurement error using covariate method, we use the data set
provided by Hu et al. (2015) and Costa and Castagliola (2011). It is a process of yogurt filling in
22 A. JAVAID ET AL.

Table 7. Max-EWMA statistics with measurement error and multiple measurements.


r2m  r2 ¼ 0.5
Sample Number Without Error K¼1 K¼2 UCL
1 0.1294566 0.08725757 0.06406335 0.523259
2 0.2427311 0.12713797 0.14715075 0.523259
3 0.2150009 0.21369195 0.16771890 0.523259
4 0.1706670 0.25781985 0.18784393 0.523259
5 0.2807006 0.17217325 0.13655097 0.523259
6 0.3981817 0.22550517 0.15049792 0.523259
7 0.3835686 0.28034903 0.09699982 0.523259
8 0.4929641 0.34523209 0.17655988 0.523259
9 0.4450725 0.34984632 0.26088219 0.523259
10 0.5102159 0.39281982 0.34667657 0.523259
11 0.6871738 0.39923320 0.3804086 0.523259
12 0.8172838 0.43580461 0.4824973 0.523259
13 0.9370634 0.49561864 0.7242925 0.523259
14 1.0343778 0.50072140 0.8973379 0.523259
15 1.1506582 0.51441431 1.0864112 0.523259
16 1.2255241 0.54338109 1.2328343 0.523259
17 1.2557640 0.55021596 1.3785407 0.523259
18 1.2636203 0.56548011 1.5635413 0.523259
19 1.2883652 0.59247349 1.7476345 0.523259
20 1.3569342 0.8940101 1.8519586 0.523259

Figure 1. Max-EWMA chart with measurement error and multiple measurements.

a cup with weight 125 grams as quality characteristic Y: In a long term study in phase-I, the in
control mean i.e. l0 ¼ 124.9 and r0 ¼ 0.76 as standard deviation, have been estimated. The qual-
ity control expert selected a sample of n¼ 5 after every hour and weighted each cup twice, in
order to study the production process to maintain the quality of cup filling. Data for 20 samples
of size five measuring each cup twice have been produced in these papers and we reproduce the
same 200 values here. However, we used first 100 values considering that process is in control
and other 100 values considering that the shift occurs.
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
23

124.78, 125.33, 124.19, 124.22, 124.55, 124.22, 123.98, 124.25, 124.97, 125.42, 124.54, 124.51,
124.39, 124.85, 125.00, 124.91, 124.82, 124.73, 123.81, 123.20, 124.25, 124.33, 124.48, 124.42,
123.10, 123.80, 125.74, 125.59, 123.34, 123.45, 125.25, 125.62, 124.78, 124.52, 124.52, 124.58,
125.42, 125.31, 124.89, 124.77, 124.44, 124.55, 124.74, 124.33, 125.93, 126.20, 125.43, 125.56,
125.71, 125.22, 125.39, 124.87, 125.46, 125.26, 125.67, 125.43, 125.00, 125.45, 125.13, 125.61,
124.98, 124.38, 125.06, 124.60, 125.63, 125.77, 124.42, 124.22, 124.85, 124.69, 124.17, 124.88,
126.13, 125.77, 125.50, 124.68, 125.21, 125.72, 125.37, 124.92, 125.62, 125.58, 125.95, 125.21,
125.25, 125.07, 124.14, 124.17, 125.06, 124.92, 127.00, 127.15, 124.77, 124.88, 125.32, 124.93,
125.18, 125.70, 123.47, 123.85, 124.98, 124.29, 123.08, 123.17, 124.56, 124.66, 123.89, 124.03,
124.01, 123.81, 123.54, 123.63, 123.92, 123.93, 124.66, 124.68, 124.40, 124.38, 124.98, 125.06,
122.56, 122.75, 123.94, 123.28, 122.62, 123.16, 124.41, 123.84, 123.95, 124.29, 122.94, 123.05,
125.10, 124.74, 123.33, 123.85, 124.23, 124.06, 123.24, 123.80, 123.13, 123.28, 123.19, 123.61,
122.83, 122.51, 124.66, 124.90, 122.00, 121.42, 122.70, 122.98, 122.50, 122.17, 123.03, 122.45,
121.53, 121.33, 124.05, 124.14, 124.12, 123.54, 122.81, 122.70, 122.71, 122.98, 122.87, 122.68,
123.26, 123.46, 122.14, 122.28, 124.11, 123.83, 124.05, 123.50, 123.75, 124.05, 124.20, 124.24,
124.31, 123.91, 123.46, 123.59, 123.61, 123.75, 123.55, 124.06, 123.08, 123.54, 124.90, 125.00,
125.06, 124.57, 123.70, 123.68, 123.98, 123.19, 123.07, 122.71.
On the basis of first 100 values, we arrived ARL0 ¼ 370 with L ¼ 3.549 considering a single
measurement of each cup, with parameters A ¼ 0, B ¼ 1 and k ¼ 0:05: Then we select 20
samples of size n ¼ 5 yogurt cups such that first 10 samples out of first 100 in control values and
the second 10 samples out of second 100 shifted values. After that, we calculate the maximum
exponentially weighted moving average with measurement error (Max-EWMAME) statistics as
well as UCL for each sample. Calculations are made first without introducing any measurement
error, then with measurement error r2m  r2 ¼ 0.5 having single measurement of each sample i.e.
k ¼1 and multiple measurements i.e. k¼2 for each sample. Afterward the calculated values of
UCL and Max-EWMA statistics without measurement error and then with measurement error for
different measurements i.e. k¼1 and k¼2 are plotted on the same graph. The values of statistics
and UCL are given in Table 7.
It can be seen from Table 7 that all the values of Max-EWMA statistics without error are lesser
than the UCL till the 10th sample selected from in control values, while are more than UCL from
11th sample to onwards selected from shifted values. However, by introducing measurement error
i.e. r2m  r2 ¼ 0.5, with single measurement of each sample k ¼ 1, using covariate model, the val-
ues of statistics highlight a disturbance in the process and become larger than the UCL at 16th
sample, which reveals the effect of measurement error by delaying the shift indication from 11th
to 16th sample. In order to observe the effect of multiple measurements to reduce the effect of
measurement error in covariate model, we use multiple measurements i.e. k ¼ 2 of each sample.
By taking two measurements of each sample, we observe that statistics have become greater than
UCL at the 13th sample and onward. It witnesses that multiple measurements have reduced the
effect of measurement error and detected out of control process earlier at 13th sample rather at
16th sample which was with a single measurement. The same situation is also highlighted in the
graph of Max-EWMA control charts with measurement error (Max-EWMAME) in Figure 1.
Figure 1 is a pictorial position of Table 7 which shows Max-EWMA charts with measurement
error and with multiple measurements. In this chart, UCL is depicted as a horizontal line along
with X-axis, the first line of graph Max0 is for no measurement error which shows that first 10
samples are in control while the 11th and onward statistics are above UCL. The other two graph
lines are with measurement error of 0.5, where Maxk1 line is for single measurement and Maxk2
line is for multiple measurements. The graph line Maxk1 shows that measurement error has
delayed the shift indication from 11th to 16th sample which needs to be addressed to produce
quality products. However, line Maxk2 reveals that strategy of two measurements has reduced the
error effect and process shift is indicated at 13th sample which is earlier than 16th sample.
24 A. JAVAID ET AL.

Therefore, it is clear from the graph that measurement error has an effect on the performance of
the Max-EWMAME chart which can be reduced by taking multiple measurements of the samples.

7. Main findings
Results regarding effects of measurement error on the Max-EWMA control charts using covariate
method, are presented in Tables 1 and 2 and the multiple measurements in Tables 3–5, showing
ARLs and SDRLs: The first two tables are regarding different values of error variance and coeffi-
cient of the X in the covariate model i.e. B: However, Tables 3–5 are regarding multiple measure-
ments of the sample or observation collected for the purpose. Major findings of the Max-EWMA
with measurement error (Max-EWMAME) chart, according to this study, are as below.

1. When there is no shift in the mean and variance i.e. a ¼ 0 and b¼ 1, it is the value of ARL0
for in control process.
2. When there is no error, it indicates the actual ARL values of Max-EWMA chart.
3. All the tables uncover that as the shift in mean “a” increases from 0.00 to 0.25, 0.50, 0.75,
1.00, 1.25, 1.50, 1.75, 2.00, 2.25, 2.50, 2.75 and 3.00, the ARLs and SDRLs are reduced
respectively, for any value of change in variance “b”, ratio r2m  r2 , k or B: It shows that our
chart is efficient in order to indicate the shift in mean at the earliest.
4. The same trend is also shown in all the tables regarding shift in variance “b”, that as the
value of shift in variance is deviated from unit i.e. from 0.90 to 0.25 and from 1.10 to 3.00,
the values of ARLs and SDRLs are reduced respectively for any value of change in mean “a”,
ratio r2m  r2 , k or B: It is also a clear efficiency of this chart that shift in variance is indi-
cated at the earliest.
5. As the values of ratio of error variance “r2m  r2 ” increased, even minor from 0.1 to 0.2 and
so on, the ARLs and SDRLs also increased respectively, showing that measurement error
causes a delay in indicating the shift during joint monitoring.
6. When the values of coefficient of X i.e. “B” in covariate model, increased from 1 to 5, the
values of ARLs and SDRLs are reduced with every increase in B respectively for fixed error
r2m  r2 ¼ 1: It proves that an increase in B reduces the error effect and enhances the effi-
ciency of our Max-EWMAME control chart for early detection of any shift in joint monitor-
ing of mean and variance shifts of quality characteristic during production process which is
the same as proved by Linna and Woodall (2001) and Maravelakis, Panaretos, and
Psarakis (2004).
7. The “A” component of the covariate model has no impact on the Max-EWMAME control
chart with covariate model.
8. To control the measurement error during sample or observation collection, if multiple meas-
urements of those samples or observations are taken, it will reduce the effect of measurement
error and will support to avoid the misleading results due to measurement error, for process
monitoring. However, multiple measurements of the same selected unit is a critical decision
with respect to cost and time.
9. Comparison with other charts shown in Sec. 5 indicates the superiority of our proposed
Max-EWMAME control chart in terms of efficiency for detection of process shift at
the earliest.

In the light of the aforementioned facts and results derived from the calculations shown in the
tables for ARL values, we can construe that our proposed Max-EWMAME control chart works
efficiently. The Max-EWMAME chart is eligible to detect even small shifts in the process mean
and variance jointly. There is no need to develop separate charts for monitoring of mean and
then for monitoring of variance but this chart can manage both the shifts in the process with
COMMUNICATIONS IN STATISTICS - SIMULATION AND COMPUTATIONV
R
25

single statistic. In addition our proposed chart has an advantage over the others that it also
encompasses the issue of measurement error which is a big problem for quality experts. If we can
resolve the problem of measurement error during measuring the sampled values to be used for
calculation of statistic(s) of control charts to maintain the quality of products, it will serve the
purpose of not only enhance the product quality but also to maintain the improved quality of
industrial production. Therefore, we can say that the practitioners should implement our pro-
posed Max-EWMAME chart instead of other existing charts for joint monitoring of mean and
variance to produce quality products.

8. Conclusion
In the field of quality control during production process, the researchers have developed different
types of control charts. However, most of the researchers have ignored the issue of measurement
error. When desired data are not collected due to measurement error, it can damage the results
and can mislead the decision makers. In this study we addressed the issue of measurement error
for Max-EWMA control chart and name it Max-EWMAME chart. We applied covariate model to
study the effect of measurement error in the Max-EWMA control charts, and then the same
model with multiple measurements which provided us a strategy to reduce the effects of measure-
ment error. The values of ARLs and SDRLs have been calculated to study the effects and treat-
ment of measurement error. It can be concluded that measurement error has a negative effect on
the efficiency of Max-EWMA control charts which can be improved by taking multiple measures
of the quality characteristic to detect shift at the earliest stage for joint monitoring of mean and
variance. Otherwise, control charts developed in the presence of measurement error would not
detect process shifts correctly and could lead to misleading decisions.

References
Asif, F., S. Khan, and M. Noor-ul-Amin. 2020. Hybrid exponentially weighted moving average control chart with
measurement error. Iranian Journal of Science and Technology, Transactions A: Science 44 (3):801–11. doi:10.
1007/s40995-020-00879-3.
Bennet, C. A. 1954. Effect of measurement error on chemical process control. Industrial Quality Control 10 (4):
17–20.
Chen, G., S. W. Cheng, and H. Xie. 2001. Monitoring process mean and variability with one EWMA chart. Journal
of Quality Technology 33 (2):223–33. doi:10.1080/00224065.2001.11980069.
Chen, G., S. W. Cheng, and H. Xie. 2004. A new EWMA control chart for monitoring both location and disper-
sion. Quality Technology & Quantitative Management 1 (2):217–31. doi:10.1080/16843703.2004.11673074.
Cheng, X. B., and F. K. Wang. 2018. The performance of EWMA median and CUSUM median control charts for
a normal process with measurement errors. Quality and Reliability Engineering International 34 (2):203–13. doi:
10.1002/qre.2248.
 chart. Journal
Costa, A. F. B., and P. Castagliola. 2011. Effect of measurement error and autocorrelation on the X
of Applied Statistics 38 (4):661–73.
Ghashghaei, R., M. Bashiri, A. Amiri, and M. R. Maleki. 2016. Effect of Measurement Error on Joint Monitoring
of Process Mean and Variability under Ranked Set Sampling. Quality and Reliability Engineering International
32 (8):3035–50. doi:10.1002/qre.1988.
Haq, A., J. Brown, and E. Moltchanova. 2015. New exponentially weighted moving average control charts for mon-
itoring process mean and process dispersion. Quality and Reliability Engineering International 31 (5):877–901.
doi:10.1002/qre.1646.
Hu, X., P. Castagliola, J. Sun, and M. B. C. Khoo. 2015. The Effect of Measurement Errors on the Synthetic x̄
Chart. Quality and Reliability Engineering International 31 (8):1769–78. doi:10.1002/qre.1716.
Javaid, A., M. Noor-ul-Amin, and M. Hanif. 2020. A new Max-HEWMA control chart using auxiliary information.
Communications in Statistics - Simulation and Computation 49 (5):1285–305. doi:10.1080/03610918.2018.
1494282.
Khoo, M. B. C., S. Y. Teh, and Z. Wu. 2010. Monitoring process mean and variability with one double EWMA
chart. Communications in Statistics - Theory and Methods 39 (20):3678–94. doi:10.1080/03610920903324866.
26 A. JAVAID ET AL.

Lee, P. H., and C. S. Lin. 2012. Adaptive Max charts for monitoring mean and variability. Journal of the Chinese
Institute of Industrial Engineers 29 (3):193–205. doi:10.1080/10170669.2012.673508.
Li, J., and S. Huang. 2009. Regression-based process monitoring with consideration of measurement errors. IIE
Transactions 42 (2):146–60. doi:10.1080/07408170903232563.
Linna, K. W., and W. H. Woodall. 2001. Effect of measurement error on Shewhart control charts. Journal of
Quality Technology 33 (2):213–22. doi:10.1080/00224065.2001.11980068.
Lu, S.-L., C. F. Tsai, and C. J. Huang. 2017, October 25–27. An EWMA-type control chart with variable sampling
intervals for monitoring process mean and variability. Proceedings of the World Congress on Engineering and
Computer Science, Vol-II, WCECS, San Francisco, USA.
Maravelakis, P. E. 2012. Measurement error effect on the CUSUM control chart. Journal of Applied Statistics 39
(2):323–36. doi:10.1080/02664763.2011.590188.
Maravelakis, P., J. Panaretos, and S. Psarakis. 2004. EWMA chart and measurement error. Journal of Applied
Statistics 31 (4):445–55. doi:10.1080/02664760410001681738.
Mittag, H. J., and D. Stemann. 1998. Gauge imprecision effect on the performance of the X   S control chart.
Journal of Applied Statistics 25 (3):307–17.
Montgomery, D. C. 2009. Introduction to Statistical Quality Control. 6th ed. New York: John Wiley & Sons.
Montgomery, D. C., and G. C. Runger. 1993. Gauge capability and designed experiments. Part I: Basic methods.
Quality Engineering 6 (1):115–35. doi:10.1080/08982119308918710.
Noor-ul-Amin, M., F. Arif, and M. Hanif. 2019. Joint monitoring of mean and variance using likelihood ratio test
statistic under pair ranked set sampling scheme. Iranian Journal of Science and Technology, Transactions A:
Science 43 (5):2449–60. doi:10.1007/s40995-019-00718-0.
Noor-ul-Amin, M., A. Javaid, M. Hanif, and E. Dogu. 2020. Performance of maximum EWMA control chart in
the presence of measurement error using auxiliary information. Communications in Statistics – Simulation and
Computation. Advance online publication. doi: 10.1080/03610918.2020.1772301.
Noor-ul-Amin, M., A. Riaz, and A. Safeer. 2019. Exponentially weighted moving average control chart using auxil-
iary variable with measurement error. Communications in Statistics - Simulation and Computation. Advance
online publication. doi:10.1080/03610918.2019.1661474.
Page, E. S. 1954. Continuous inspection schemes. Biometrika 41 (1–2):100–15. doi:10.1093/biomet/41.1-2.100.
Raza, S. M. M., M. H. Sial, M. Haider, and M. M. Butt. 2019. Hybrid exponentially weighted moving average
(HEWMA) control chart based on exponential type estimator of mean. Journal of Reliability and Statistical
Studies 12 (2):187–98. doi:10.13052/jrss2229-5666.12214.
Riaz, A., M. Noor-ul-Amin, A. Shehzad, and M. Ismail. 2019. Auxiliary information based mixed EWMA-CUSUM
mean control chart with measurement error. Iranian Journal of Science and Technology, Transactions A: Science
43 (6):2937–49. doi:10.1007/s40995-019-00774-6.
Roberts, S. W. 1959. Control chart tests based on geometric moving averages. Technometrics 1 (3):239–50. doi:10.
1080/00401706.1959.10489860.
Sabahno, H., and H. Amiri. 2017. The effect of measurement errors on the performance of variable sample size
and sampling interval X  control chart. International Journal of Engineering, IJE Transactions A: Basics 30 (7):
995–1004.
Salmasnia, A., M. R. Maleki, and S. T. A. Niaki. 2018. Remedial measures to lessen the effect of imprecise measure-
ment with linearly increasing variance on the performance of the MAX-EWMAMS scheme. Arabian Journal for
Science and Engineering 43 (6):3151–62. doi:10.1007/s13369-017-2896-1.
Sanders, F. F. 1995. Handling process disturbances and measurement errors. ISA Transactions 34 (2):165–73. doi:
10.1016/0019-0578(95)00010-W.
Tang, A., P. Castagliola, J. S. Sun, and X. L. Hu. 2018. The effect of measurement error on the adaptive EWMA X 
chart. Quality and Reliability Engineering International 34 (4):609–30.
Wu, C. W. 2011. Using a novel approach to assess process performance in the presence of measurement errors.
Journal of Statistical Computation and Simulation 81 (3):301–14. doi:10.1080/00949650903313761.
Xie, H. 1999. Contributions to Qualimetry. Ph.D. Thesis., University of Manitoba, Winnipeg, Canada.
Zhang, L., and G. Chen. 2005. An extended EWMA mean chart. Quality Technology & Quantitative Management
2 (1):39–52. doi:10.1080/16843703.2005.11673088.

You might also like