This month is the first in a multi-part publication on X-R charts.

This month we
introduce the chart and provide the steps in constructing an X-R chart. Next month, we
will look at a detailed example of an X-R chart. The X-R chart is a type of control chart
that can be used with variables data. Like most other variables control charts, it is
actually two charts. One chart is for subgroup averages ( X). The other chart is for
subgroup ranges (R). These charts are a very powerful tool for monitoring variation in a
process and detecting changes in either the average or the amount of variation in the
process.

Introduction to X-R Charts

Suppose you are a member of a

bowling team. You bowl three games a night once a week in a bowling league. You are
interested in determining if you are improving your bowling game. What are some
different approaches you could use? One idea is that you could plot the score from each
game. However, you are more interested in what your average score is on a given
night. So another idea is to plot the average of the three games each night. You
definitely would like to increase that average over time. You are also interested in being
more consistent, i.e., not having one great game followed by a poor one. Thus, another
idea is to keep track of the range in scores for the three games each night. In situations
such as this (when you want to monitor averages over time but still keep track of the
variation between individual results), the X-R chart is very useful.

The X-R chart is a method of looking at two different sources of variation. One source
is the variation in subgroup averages. The other source is the variation within a
subgroup. Consider the bowling example above. You have data available on a fairly
frequent basis (three games each week). You can also rationally subgroup the data.
The three individual games you bowl on one night can be used to form a subgroup.

Continuing with the bowling example, suppose that one night your three bowling scores
are 169, 155, and 189. These three scores form a subgroup. You can calculate the
range of this subgroup by subtracting the minimum score from the maximum score.
Thus the range is:
 Range = Maximum – Minimum = 189 – 155 = 34

You can plot this value on a range (R) chart. This is done for each subgroup (one night
of bowling three games). The range chart shows how much variation there is within
each subgroup, i.e., the amount of variation in your bowling scores on one night. You
would like this variation to be small and be consistent over time.

The chart for averages ( X) presents a different variation than the range chart. Using the
three scores above, you can calculate an average score for the night by taking the
average of the three individual scores. The subgroup average is:

X = (169+155+189)/3 = 171

You can plot this value on the X chart. This is done for each subgroup. The X chart
shows how much week-to-week variation there is in your weekly average bowling score.
You would like this variation to be small and be consistent over time. This permits you to
predict what your average score will be on any night, within certain limits.

The figure below is an example of the X-R chart for this bowling example. The top part
of the figure is the X chart. Each weekly average bowling score (i.e., the average of the
three individual games) is plotted. The overall average (Xdbar = X double bar) has been
calculated and plotted as a solid line. Xdbar is the average of all the subgroup
averages. Upper and lower control limits have also been calculated and plotted.
The X chart is in statistical control. The lower part of the figure is the range (R) chart.
The range is plotted for each week. The average range and control limits have been
calculated and plotted. The range is also in statistical control.

What does it mean when the X-R chart is in statistical control? It means that the
subgroup average is consistent over time and the variation within a subgroup is
consistent over time. We can predict what the process will do in the near future. In the
bowling example, this means that you can predict what the average of your three games
on any given night will be. Your average will be between about 158 and 208 with a long
term average of about 183. You can also predict what your range in bowling scores will
be on any given night. The range can be anywhere from 0 to about 62 with an average
range of about 24. As long as the process stays in control (your bowling), the results will
continue to the same.

Example
When to Use X-R Charts
X-R charts should be used when you have taken data frequently. How often you plot
points on the charts depends on your subgroup size. For example, if your subgroup size
is four, it will take four samples before you calculate the average and range and plot the
points. If you only take one sample per day, it will be four days before you can plot the
points. If the point is out of control, the reason for it could have occurred four days ago.
This often makes it difficult to find out what happened.

X-R charts should be used if you can rationally subgroup the data and are interested in
detecting differences between subgroups over time. This means there should be some
logical basis for the way the subgroups are formed. They should be formed to examine
the variation of interest to you. You might be interested in the variation from day to day.
In this case, samples from one day would be used to form a subgroup. The X chart
would examine the variation from day to day, while the R chart would examine the
variation within a day.

The R chart is a measure of the short-term variation in the process. Subgroups should
be formed to minimize the amount of variation within a subgroup. This causes
the X chart to do the work in detecting process changes.

Steps in Constructing an X-R Chart

The steps in constructing an X-R chart are given below.

1. Gather the data.

a. Select the subgroup size (n). Typical subgroup sizes are 4 to 5. The concept
of rational subgrouping should be considered. The objective is to minimize the
amount of variation within a subgroup. This helps us “see” the variation in the
averages chart more easily.
b. Select the frequency with which the data will be collected. Data should be
collected in the order in which it is generated (in most cases).
c. Select the number of subgroups (k) to be collected before control limits are
calculated. You can start with initial control limits after ten subgroups, but
recalculate the limits each time until you get to twenty subgroups.
d. For each subgroup, record the individual, independent sample results.
e. For each subgroup, calculate the subgroup average:

where n is the subgroup size.

f. For each subgroup, calculate the subgroup range:

R = Xmax – Xmin

where Xmax is the maximum individual sample result in the subgroup and Xmin
is the minimum individual sample result in the subgroup.

2. Plot the data.

a. Select the scales for the x and y axes for both the X and R charts.
b. Plot the subgroup ranges on the R chart and connect consecutive points with
a straight line.
c. Plot the subgroup averages on the X chart and connect consecutive points
with a straight line.

3. Calculate the overall process averages and control limits.

a. Calculate the average range (Rbar):

where k is the number of subgroups.

b. Plot Rbar on the range chart as a solid line and label.
c. Calculate the overall process average (Xdbar):

d. Plot X on the X chart as a solid line and label.

e. Calculate the control limits for the R chart. The upper control limit is given by
UCLr. The lower control limit is given by LCLr.

where D4, D3, are control chart constants that depend on subgroup size (see the
table below).
f. Plot the control limits on the R chart as dashed lines and label.
g. Calculate the control limits for the X chart. The upper control limit is given by
UCLx. The lower control limit is given by LCLx.
where A2 is a control chart constant that depends on subgroup size (see the
table below).

h Plot the control limits on the X chart as dashed lines and label.

4. Interpret both charts for statistical control.

a. Always consider variation first. If the R chart is out of control, the control limits
on the X chart are not valid since you do not have a good estimate of . All tests
for statistical control apply to the X chart. Points beyond the limits, number of
runs and length of runs tests apply to the R chart.

5. Calculate the process standard deviation, if appropriate.

a. If the R chart is in statistical control, the process standard deviation, s, can be

calculated as:

where d2 is a control chart constant that depends on subgroup size (see the
table below).

To calculate control limits and to estimate the process standard deviation, you
must use the control chart constants D4, D3, A2, and d2. These control chart
constants depend on the subgroup size (n). These control chart constants are
summarized in the table below. For example, if your subgroup is 4, then D4 =
2.282, A2 = 0.729, and d2 = 2.059. There is no value for D3. This simply means
that the R chart has no lower control limit when the subgroup size is 4.

Subgroup
A2 D3 D4 d2
Size (n)
2 1.880 3.267 1.128

3 1.023 2.574 1.693

4 0.729 2.282 2.059

5 0.577 2.114 2.326

6 0.483 2.004 2.534

7 0.419 0.076 1.924 2.704

8 0.373 0.136 1.864 2.847

9 0.337 0.184 1.816 2.970

10 0.308 0.223 1.777 3.078

11 0.285 0.256 1.774 3.173

12 0.266 0.284 1.716 3.258

13 0.249 0.308 1.692 3.336

14 0.235 0.329 1.671 3.407

15 0.223 0.348 1.652 3.472

16 0.212 0.364 1.636 3.532

17 0.203 0.379 1.621 3.588

18 0.194 0.392 1.608 3.640

19 0.187 0.404 1.596 3.689

 20 0.180 0.414 1.586 3.735

21 0.173 0.425 1.575 3.778

22 0.167 0.434 1.566 3.819

23 0.162 0.443 1.557 3.858

24 0.157 0.452 1.548 3.895

25 0.153 0.459 1.541 3.931

Summary
This publication has introduced the X-R chart. When you should use an X-R chart was
covered as well as the steps in constructing the chart.

Ten Things to Know When Using the

Individuals Control Charts
Home

SPC Knowledge Base

Ten Things to Know When Using the Individuals Control Charts

March 2022

(Note: all the previous SPC Knowledge Base in the variable control chart category are
listed on the right-hand side. Select this link for information on the SPC for Excel
software.)

Over the years, the use of individuals control charts has increased, particularly in non-
manufacturing situations. In fact, there are those who say there is no reason to learn
any other chart – saying you can use the individuals control chart anywhere.

The individuals control chart is empirical, and the empirical approach is always right. Is
that true? Is the only control chart we need to know the individuals? In many cases, yes,
but not all cases. And there are some other things you need to remember when using
the individuals control chart.

The individuals control chart is often denoted by X-mR or I-mR, where mR stands for
the moving range. In this publication, we will examine the X-mR chart and answer 10
questions about this chart. In this issue:

 How do you construct a X-mR chart?

 When should a X-mR chart be used?
 Does data have to be normally distributed to use a X-mR chart?
 Is the time between points important for a X-mR chart?
 Do you need to think about rational subgrouping with a X-mR chart?
 Do X-mR charts work with rare events data?
 Do X-mR charts work with chunky data?
 Do X-mR charts give false signals?
 Can you use the X-mR chart with data that are counts?
 Can you use the X-mR chart with data with varying large subgroup size?
 Summary
 Quick links
Please feel free to leave a comment at the end of the publication. You can download a
pdf copy of this publication at this link.

We will start with the basics of how to construct a X-mR chart.

How Do You Construct a X-mR Chart?

The X-mR chart monitors individual values (X) over time. The individual values are
plotted on the X chart. The moving range between consecutive individual values are
plotted on the mR chart. Suppose we take a sample once an hour from a process and
measure it for some quality characteristic. The results for the last 30 hours are shown in
Table 1.

Table 1: Process Data

Sample X Sample X

1 93.4 16 101.7

2 101.8 17 94.7

3 97.1 18 96.8

4 91.8 19 111.4
 Sample X Sample X

5 99.4 20 103.7

6 104.5 21 102.4

7 95.9 22 99.5

8 95.8 23 109.6

9 86.6 24 107.5

10 88.2 25 92.0

11 103.7 26 103.4

12 85.4 27 87.4

13 95.2 28 92.2

14 102.3 29 89.3

15 92.7 30 96.6

The first moving range value is the range between samples 1 and 2:

First Moving Range: |93.4 – 101.8| = 8.4

The overall average (X) and the average moving range (R) are then calculated using
the following formulas:

X= X/k

R= R/(k-1)

where k = the number of samples. Now you can calculate the upper control limit (UCLx)
and lower control limit (LCLx) for the X chart as follows:

UCLx =X + 2.66R
 LCLx =X – 2.66R

The upper control limit (UCLr) for the moving range chart can then be calculated.

UCLr = 3.268R

The 2.66 and 3.268 are constants based on using a moving range of 2 (between
consecutive samples) in the analysis. There is no lower control limit on the moving
range chart.

The averages and control limits can then be added to the charts. Figure 1 is the X chart
with the average being the centerline. Figure 2 is the mR chart with the average moving
range being the centerline.

Figure 1: X Chart

Figure 2: mR Chart
Both charts are in statistical control – there are no points beyond the control limits or
patterns in the data (e.g., a run below the average).

The information on constructing a X-mR chart comes from our SPC Knowledge Base
article Medians and the Individuals Control Chart. There are times when you can use
the median in place of the overall average. More on that below.

When Should a X-mR Chart be Used?

Control charts, when properly designed, monitor variation over time. Two options for
monitoring the variation in a process include monitoring the individual values as shown
above or combining samples (e.g., 4 samples) to from a subgroup and monitoring the
subgroup averages and ranges over time (i.e., using the X-R chart).

This appears to indicate that the decision to use a certain chart should include how
much data are available to represent a given situation at a given time. If you have
multiple data points at a point in time to represent the situation, then you should use one
of the subgrouping charts, like the X-R chart. If you have only one point in time to
represent the situation, you should use the X-mR chart.

For example, if you have only one data point per day, a week, or a month to represent a
situation, then you have infrequent data. You would want to use the X-mR chart. If you
are testing a stream for a contaminant daily, you probably want to use the X-mR chart.
You have only one data point to represent the situation daily and, in this case, you
would like to be able to respond immediately if the contaminant is out of control. It
doesn’t make any sense to wait four days, for example, to form a subgroup and plot the
result on the X-R chart.

So, it is best to use the X-mR chart when you have only one data point to represent a
situation at a given point in time, and it is important to respond to signals at that moment
in time.

Does Data Have to Be Normally Distributed to Use a X-Mr Chart?

The short answer is no. The longer answer involves the

zones tests.

You may have heard that data must be normally distributed before you can plot the data
using a control chart? Quite often you hear this when talking about the X-mR chart. This
is a myth. Data do not have to be normally distributed before a control chart can be
used – including the X-mR chart.

Dr. Donald Wheeler, for one, has shown extensively, in multiple publications
(www.spcpress.com), that the X-mR chart can be used to monitor the variation
regardless of the type of distribution – even heavily skewed distributions.

But there is more to the story here. Dr. Wheeler focused primarily on the ability to detect
a point beyond the control limits. He did not focus on the impact on the other out of
control tests, for example, the zones tests. Care should be given when applying the
other out of control tests to non-normal data.

Our SPC Knowledge Base article Control Charts and Non-Normal Data covers this topic
in detail. In that article, 100 points are randomly generated from an exponential
distribution. The data are heavily skewed towards the left. The X control chart for that
data is shown in Figure 3 (the mR chart will not be shown here).

Figure 3: X Control Chart for Exponential Data

The chart shows the average and UCL. There is no LCL since the exponential function
can’t be below 0. There are two red lines between the average and UCL and one red
line between the average and 0. These are used for the zones tests. The zones tests
can be used to interpret the individuals control chart. For more information, please see
our SPC Knowledge Base article Interpreting Control Charts.

The red points in Figure 3 represent out of control points. There are two points above
the UCL. There are also two runs of at least 7 below the average, one area where there
are 4 points in a row in zone B, and an area where there are two out of three
consecutive points in zone A. These all represent special causes.

With heavily skewed data, is it surprising that you get runs of 7 or more below the
average? It is not surprising – after all, the distribution is skewed that direction. The
conclusion here is that if you are plotting non-normal data on the X-mR chart, take care
in applying the zones tests. These tests are designed for a normal (or at least a
somewhat symmetrical) distribution. Using them with non-normal data can create false
signals of problems.

Removing the zones tests leaves two points that are above the UCL – out of control
points. With our knowledge of variation, we would assume there is a special cause that
occurred to create these high values. Are these false signals? You cannot assign a
probability to a point being due to a special cause or not – regardless of the data
distribution. So, are they false signals? In the real world, you don’t know. But wouldn’t
you want to investigate what generated these high values?

The amazing thing is that the individuals control chart can handle the heavily skewed
data so well – only two “out of control” points out of 100 on the X chart. This
demonstrates how robust the moving range is at defining the variation. The +/- three
sigma limits work for a wide variety of distributions.

You can also use the median in place of the average if you want to see the impact of
this on the zone tests. The median is not impacted like the average by the higher
results. This is described in more detail in the article referenced above.

Is the Time Between Points Important for a X-Mr Chart?

As stated above, control charts, when properly designed, monitor variation over time.
Time is an important but sometimes ignored part of control charts – particularly for the
X-mR charts. Remember, the X chart control limits as well as the mR chart control limits
depend on the average moving range.

Suppose you are monitoring a variable once a day and plotting those results on a X-mR
chart. You get your results for day 1 and day 2 and plot those as well as the moving
range between the 2 days. Now you wait 5 days to get the next result. How do the
potential sources of variation compare between days 1 and 2 and then, between days 2
and 7? The odds are that there are more sources during the 5 days than the 1 day. You
are not comparing apples to apples. The time frame between samples should be the
same. In one case, the moving range represents the variation between consecutive
days and the other moving range represents the variation between 5 days. They are
most likely not the same.

It is important that the time between results on the X-mR chart be the same over time,
for example, between consecutive days. In fact, if you miss a sample, don’t use the
previous sample in determining a moving range. The table below demonstrates this.

 Day 1: X =100
 Day 2: X = 102; moving range = |102 – 100| = 2
 Day 3: Missed
 Day 4: X = 98; No moving range
 Day 5: X = 103, moving range = |98 – 103| = 5
Pay attention to missed data on the X-mR chart and be sure the moving range is
handled properly.

Do You Need to Think About Rational Subgrouping with a X-mR Chart?

Yes, and it is directly related to the last question above – time. You have to remember
what you are doing when you setup a X-mR chart. The subgroup size is one. And based
on the results, you are going to calculate the control limits based on the moving range
between successive differences. You want it to be logical to compare the consecutive
differences. The moving range then captures the normal variation in the process, and
you can use the control limits to filter out the noise and see the signals.

Do X-mR Charts Work with Rare Events Data?

No, they don’t work well at all. But with some calculations, you can convert the rare
events to rates and use the X-mR chart based on those rates. Our SPC Knowledge
Base article Rare Events and X-mR Charts describes this situation in more detail.

One data set in the article monitors the number of injuries per month in a plant. There
are seldom injuries – essentially a rare event for that plant. Figure 4 shows the X chart
for the data, which is given in the above article. The mR chart is not shown.

Figure 4 is monitoring the number of injuries in a plant per month. As you can see, the
chart is not of much use.

Figure 4: X Chart with Rare Events

There are only two possible outcomes in the chart, 0 or 1. But you can change the data
you are looking at by calculating the number of days between injuries and converting
that to an injuries per year rate. Figure 5 shows the chart with the rates calculated. The
article referenced above explains how to do this calculation.

Figure 5: X Chart: Injuries per Year Rate

This chart is much more useful than the chart in Figure 4. You know what your yearly
rate is and can see from the chart if it is staying the same, improving or getting worse.
You could also chart the number of days between injuries instead of the yearly rate.

Do X-mR Charts Work with Chunky Data?

Not really. Chunky data occurs when the range between possible values becomes too
large. One example is measuring a person’s height to the nearest yard. This
measurement is too large and would obscure the variation in height from person to
person. Excessive round-off will lead to chunky data. It can also occur when the
measurement process cannot tell the difference between samples (usually indicated by
a very large gage R&R %). In this case, the measurement unit is too large (as in the
case of measuring a person’s height to the nearest yard).
Our SPC Knowledge Base article Chunky Data and X-mR Charts explores this issue in
detail. In that article, temperature values are being charted, with the temperature values
rounded to the nearest 0.1. The article shows this process to be in statistical control and
is a typical looking X-mR chart, with many different values on the mR chart. Now the
data are rounded to the nearest degree, and another X-mR chart is created. Figures 6
and 7 are the X and mR charts based on the rounded data.

Figure 6: X Chart with Chunky Data

Figure 7: mR Chart with Chunky Data

Note that there are eight points beyond the control limits on the X chart and 1 out of
control point on the moving range chart. These out of control points occurred simply
because of the way we rounded the data. They have nothing at all to do with the
process – which the article above shows was in statistical control. The out of control
points are the result of excessive rounding. This type of data can lead to false alarms.
You can get out of control points when the process is actually in statistical control.

It is easy to tell if you have chunky data. Simply count the number of possible values on
the mR chart. For X-mR charts, the data is chunky if the mR chart has three or fewer
possible values below the upper control limit.

Do X-mR Charts Give False Signals?

Yes, all control charts can give false signals. This is one potential reason not to use the
X-mR chart if there is a lot of data available. If you have a lot of data and can rationally
subgroup the data, it is best to use the X–R chart. This chart actually gives less false
signals than the X-mR chart. Please see our SPC Knowledge Base article Comparing
the X-mR and Xbar-R Control Charts for the simulation that shows this to be the case.

Can You Use The X-mR Chart with Data That Are Counts?
Yes, you can. There are two classes of data: attribute and variables data. Attributes
data are either yes/no or counting.
With yes/no data, there are only two possible outcomes for an item: either it passes, or
it fails some preset specification. Each item inspected is either defective (i.e., it does not
meet the specifications) or is not defective (i.e., it meets specifications). An example of
yes/no data includes is the product in spec or not.

With counting data, you count the number of defects. A defect occurs when something
does not meet a preset specification. It does not mean that the item itself is defective.
For example, a dresser can have a scratch (a defect) but still work properly. When
looking at counting data, you end up with whole numbers such as 0, 1, 2, 3; you can’t
have half of a defect.

Variables data consist of observations made from a continuum (such as the

temperature). That is, the observation can be measured to any decimal place you want
if your measurement system allows it. Some examples of variables data are contact
time with a customer, sales dollars, amount of time to make a delivery, height, weight,
and costs.

In the past, it was often taught that you use attributes control charts with attributes data
and variables control charts with variables data. For example:

 Attributes Data:
 Yes/No type data: use p or np chart
 Counting type data: use c or u chart
 Variables data: use X-R or X-mR charts
There are other attribute and variable control charts. Over time it has been emphasized
that certain conditions must be met to use the attributes charts, primarily because of the
assumption of the underlying distribution, – and that seldom the conditions are met. It
has been suggested more and more that the X-mR chart should be used for attributes
data.

If you compare the results for the X-mR chart with the attribute charts you will find that
the results are very similar. Please see our SPC Knowledge Base article Comparing
Individuals Charts to Attributes Charts to see the comparison between the X-mR chart
and the attribute charts.

Yes, you can use the X-mR chart in place of the p, np, c and u charts. But there is more
to the story on this. That is the last question.

Can You Use The X-mR Chart with Data with Varying Subgroup Sizes?
Yes, but there is the potential to miss signals if the large subgroup sizes vary too much.
In the past, the X-mR chart has been used to address the issue of large subgroup sizes
with attributes charts. Please refer to our SPC Knowledge Base article Laney u’ Control
Chart for details and data on this topic. The data from the article is the number of errors/
patient per week. During week 1, there were 98 errors for 6566 patients. This gives X =
98/6566 = 0.01493. If a u chart was used for these data, the control limits would be very
tight with the potential for many out of control points. The individuals chart avoids that
problem. Figure 8 is the X chart for the data in the article. It is in statistical control.

Figure 8: X Chart for Patients/Errors Data

The X chart looks good. However, it does not account for the varying subgroup sizes in
the data. The subgroup size (the number of patients per week) varies from 5114 to
10496. The Laney u’ chart – another type of attribute control chart designed to handle
large sample sizes – accounts for this issue. Figure 9 is the Laney u’ chart for the data.

Figure 9: Laney u’ Control Chart for Patients/Errors Data

Take a close look at Figure 9. A couple of items to notice. First, the control limits vary
because the subgroup size (number of patients) varies each week. There is one out of
control point with the Laney u’ chart. The X-mR chart did not have any. The reason the
Laney u’ control chart picks it up is that it is accounting for the varying subgroup size –
which the X-mR chart does not. In this situation, the Laney u’ control chart is a better
choice than the X-mR chart.

Summary
The X-mR chart has gained in popularity over the years, even to the point of some
recommending just teaching the X-mR chart – that it is all you need. Yes, it is empirical
and will work in many, many circumstances. Just not all. This publication pointed out
some of the things you should be aware of when using the X-mR chart. This publication
also looked at how to construct, and when to use the X-mR chart as well as the
importance of time when setting up the d

Comparing XBar-R and XBar-s Control

Charts
Home

SPC Knowledge Base

Comparing XBar-R and XBar-s Control Charts

 April 2021

(Note: all the previous SPC Knowledge Base in the variable control chart category are
listed on the right-hand side. Select this link for information on the SPC for Excel
software.)

When I was first introduced to control charts, the X-R chart was
used the most. This was some years ago – back in the early 1980s. The control charts
we did back then were done manually – doing the calculations using a calculator and
manually plotting the points, averages, and control limits. There is still something to be
said for teaching control charts using this manual method. We definitely tried to keep
things simple.

There were a number of control charts I was exposed to back then. One was the X-
s chart. This chart is similar to the X-R chart, except that the subgroup standard
deviation is used instead of the subgroup range for the within subgroup variation.

In this publication, we will compare the two charts to see when you use one or the other.
Traditionally, it has been suggested that you use the X-R chart with subgroup sizes of 9
or less; you use the X-s chart with subgroup sizes larger than 9. For the purposes of this
publication, the chart to use is the one that gives you the best estimate of the process
standard deviation. A simulation was developed to help do this.

The basic conclusion is that there is no reason not to use the X-s chart all the time.

In This Issue:

 Review of X-R and X-s Control Charts

 When to Use the X-R and X-s Control Charts Historically
 Estimating Sigma, the Process Standard Deviation
 Data and Simulation
 Control Charts from a Stable Process
 Simulation with a Stable Process
 Simulation with a Mixed Process
 Simulation with Out of Control Points
 Summary
 Quick Links
You may download the workbook with the simulation at this link. This workbook contains
the results of the simulation as well as the VBA code to let you run your own
simulations. The data used in the control chart figures below are also included in the
workbook. You may also download a pdf copy of this publication at this link.

Review of X-R and X-s Control Charts

These two control charts are quite similar. Each is really two charts – one to track the
between subgroup variation and the other to track the within subgroup variation. You
start with the individual values and then form subgroups of size n. For example,
suppose you are tracking the weight of bags containing sand. You are measuring the
weight for 5 consecutive bags at the start of each hour. You take those 5 weights and
form a subgroup. In this case, the subgroup size, n, is 5. The average of each subgroup
is calculated using the following:

X= Xi/n

where Xi is are the individual X results in the subgroup. This is repeated and
the X values are plotted over time. So, the X chart monitors how the subgroup averages
vary over time – this is why it is said to track the between subgroup variation.

The two charts differ in how they track the within subgroup variation. For the X-R chart,
the subgroup range is used. The range within a subgroup is defined as the largest value
minus the smallest value:

Subgroup Range = R = Xmax – Xmin

This is a measure of the variation within a subgroup. The subgroup range is plotted on
the R chart, which is used to measure the variation within a subgroup from subgroup to
subgroup.

For the X-s chart, the subgroup standard deviation (s) is used:

The value of s is a measure of the variation within a subgroup – it essentially measures

how far each individual X value in the subgroup is away from the subgroup average, X.
The s values are plotted on the s chart. Like the R chart, the s chart monitors the
variation within a subgroup from subgroup to subgroup.
Note that the calculation of s involves each individual X value in the subgroup while the
subgroup range, R, uses only two of the individual values, the maximum and the
minimum, regardless of how large the subgroup is.

Once you have sufficient data, you can begin to calculate the averages and control
limits. These calculations are given below:

X-R Chart:

X-s Chart:

where k = number of subgroups, R = the average subgroup range, s = the average

subgroup standard deviation, X= the overall average and D4, D3, A2, B4, B3 and A3 are all
constants that depend on subgroup size.

For more information on the X-R and X-s control charts, please visit our SPC
Knowledge Base.

When to Use the X-R and X-s Control Charts Historically

One reason that the X-R chart has been used more than the X-s chart is that it is easier
to explain a range then the standard deviation. Back in the calculator days, the range
was a lot easier to calculate than the standard deviation. The use of software now
makes that point rather moot.

Another reason is that as the subgroup size, n, increases, the standard deviation
becomes a better estimator of the process standard deviation – the variation in the
process. Is that true? Quite often, the suggestion is to use the X-R chart when n is less
than 8 to 10. Different sources suggest either 8, 9 or 10. Otherwise, for larger
subgroups, use the X-s chart. The rest of this publication takes a look at how accurate
this is.

Estimating Sigma – The Process Standard Deviation

One key use of a control chart is to estimate the process standard deviation, . This
value is used in other calculations such as process capability. The within subgroup
variation control chart is used to estimate the process standard deviation. The process
standard deviation is measuring the variation in the individual values. The equations for
estimating  are given below for the two chart:

X-R Chart:

 =R/d2

X-s Chart:

 =s/c4

where d2 and c4 are constants that depend on subgroup size. We will use the value
of  to help compare how well the X-R chart and X-s chart perform and when you
should use one or the other.

Data and Simulation

To determine how well the X-R chart and X-s chart perform, a simulation was run
multiple times. A “population” of 5000 randomly generated numbers was created using
the random number generator in the SPC for Excel software. The normal distribution
was used with an average of 100 and standard deviation of 10. The actual values for
the 5000 points are:

Population Average = 100.0257

Population Standard Deviation = 10.0401

These 5000 points represent all the possible outcomes from a process. The simulation
uses subgroup sizes from n = 2 to 20. The number of subgroups for each simulation is k
= 20. The simulation works in the following way:

 Sort the 5000 points in random order

 Take the first nk points
 For example, if n = 3 and k = 20, the first 3*20 = 60 points are taken to form the
subgroups and do the calculations
  For each n value (from 2 to 20)
 Form subgroups of size n
 Calculate the subgroup range (R) and subgroup standard deviation (s) for each
subgroup
 Calculate the average subgroup range (R) and the average subgroup standard deviation
(s) for the 20 subgroups
 Repeat this 1000 times
 After the 1000 times, calculate the overall average subgroup range and overall average
subgroup standard deviation for the 1000 times for that value of n
 Calculate the process standard deviation () from the overall average subgroup range
and overall average subgroup standard deviation
 Compare the results to each other and to the actual process standard deviation from the
5000 data points
The results of the simulation are described below.

Control Charts from a Stable Process

Drawing random samples from the 5000 point population described above is a stable
process. Taking samples from it and creating a control chart will produce a process that
is very stable – it is in statistical control. Below are examples of the X-R chart and X-s
chart for a random sample taken from the 5000 points with n = 5.

Figure 1: X Chart with Limits Based on the Average Subgroup Range

Figure 2: Subgroup Range Chart
Figure 3: X Chart with Limits Based on the Average Subgroup Standard Deviation
Figure 4: Subgroup Standard Deviation Chart
Look at Figures 1 and 3, which are the X charts for the X-R chart and X-s chart,
respectively. The plotted values are the same for each chart – the X values. In addition,
the overall average is the same for both charts. The only differences are the values of
the upper control limit (UCL) and the lower control limit (LCL). The control limits in
Figure 1 are based on the average subgroup range R while the control limits in Figure 3
are based on the average subgroup standard deviation s. Note that the values are
close.

Figures 2 and 4 represent the within subgroup variation as defined by either the
subgroup range or the subgroup standard deviation. These two will often follow a similar
pattern, e.g., a large subgroup range will generate a large subgroup standard deviation.
Usually, the X-R chart and X-s chart will look similar.

The control charts in Figures 1 to 4 are in statistical control – there are no points beyond
the control limits or patterns in the data. The process standard deviation can be
calculated for the two charts as shown below:

 =R/d2 =23.894/2.326=10.272
  =s/c4 =9.961/0.94=10.597

The first one is based on the average subgroup range; the second on the average
subgroup standard deviation. The two are similar. From above the “true” standard
deviation is 10.0401. Both are near that as well.

Note that is one time going through the process of determining the process standard
deviation. The simulation does this 1000 times for each subgroup size and chart and
takes the overall average of the 1000 results for each chart to estimate .

Simulation with a Stable Process

The simulation was run 1000 times to determine the average long-term value of  as a
function of subgroup size. The results are shown in Table 1.

Table 1: Simulation Results for a Stable Process

n R s d2 c4  from R from s

2 11.322 8.006 1.128 0.7979 10.0371 10.0335

3 17.087 8.946 1.693 0.8862 10.0930 10.0944

4 20.649 9.241 2.059 0.9213 10.0287 10.0299

5 23.238 9.395 2.326 0.9400 9.9906 9.9949

6 25.471 9.565 2.534 0.9515 10.0517 10.0520

7 27.163 9.633 2.704 0.9594 10.0456 10.0411

8 28.588 9.692 2.847 0.9650 10.0414 10.0437

9 29.803 9.725 2.97 0.9693 10.0346 10.0334

10 30.921 9.778 3.078 0.9727 10.0456 10.0526

11 31.905 9.814 3.173 0.9754 10.0552 10.0615

12 32.721 9.819 3.258 0.9776 10.0434 10.0444

13 33.522 9.857 3.336 0.9794 10.0485 10.0646

 n R s d2 c4  from R from s

14 34.277 9.874 3.407 0.9810 10.0608 10.0649

15 34.699 9.836 3.472 0.9823 9.9939 10.0137

16 35.394 9.862 3.532 0.9835 10.0208 10.0273

17 36.092 9.916 3.588 0.9845 10.0592 10.0725

18 36.522 9.897 3.64 0.9854 10.0335 10.0437

19 37.043 9.911 3.689 0.9862 10.0414 10.0492

20 37.498 9.915 3.735 0.9869 10.0395 10.0467

This table contains the average of the 1000 estimates for R and s for the various
subgroup sizes. It also includes the control chart constants needed to determine . The
values of  are included in the table. The values of  are plotted in Figure 5 along with
the “true” population average.

Figure 5: Values of  from the Simulation for a Stable Process

The values of  are close to the population value of 10.0401. For n values up to 9, the
estimate of  is about the same between the two charts. Sometimes the  from R is
higher than the  from s. Other times it is not.

Starting at n = 10, the value of  from s is always higher than the  from R. But, for a
stable process, both give results fairly close to the population standard deviation. In fact,
if you take the average of the  results from n = 10 to n = 20, the average based
on R is closer to the population standard deviation than the average based on s:

“True” population standard deviation = 10.0401

Based on R = 10.0402 (for n = 10 to 20)

Based on s: = 10.0492 (for n = 10 to 20)

It appears, that for a stable process, it doesn’t matter whether you use the X-R chart or
the X-s chart for subgroups n = 2 to 20.
But what about unstable processes? Two cases are considered below.

Simulation with Two Processes Mixed

To simulate a less stable process, 500 of the original 5000 points were replaced with a
distribution that had an average of 85 and a standard deviation of 10. So, 90% of this
new distribution had an average of 100 and 10% had an average of 85. Both had a
standard deviation of 10. The new distribution average and standard deviation were:

Population Average = 98.454

Population Standard Deviation = 11.0486

The simulation was run again using this distribution. The tabular results are not included
here but are in the workbook containing the simulation at this link. Figure 6 shows the
results for .

Figure 6: Values of  from the Simulation for Mixed Processes

For n = 2 to n = 5, the results are similar. But from n = 6 on, the value of  estimated
from R is considerably higher than that estimated from s. So, when there is some
additional variation added to the mix, it appears that s is a much better choice than R to
estimate .

Simulation with Out of Control Points Added

To include some out of control points, 200 points in the original distribution were deleted
and 50 values each of 60, 65, 135 and 140 were added to the distribution. These
represent out of control points since the original distribution was based on an average of
100 and a standard deviation of 10. About 4% of the points are beyond the 70 to 130
range defined by the original distribution. The new distribution had the following values:

Population Average = 100.025

Population Standard Deviation = 12.3904

The simulation was run again. The tabular results are not included here but are in the
workbook containing the simulation at this link. The results are shown in Figure 7.

Figure 7: Values of  from the Simulation with Out of Control Points Added

Figure 7 shows that the estimated process standard deviation from the two methods
diverge considerably. The value of  estimated by R is inflated while the value
of  estimated by s is deflated but does begin to approach the population standard
deviation as n increases.

If the process is out of control, then neither method is that accurate although
the s method appears to be a better choice as n increases.

Summary
This publication examined using the estimated process standard deviation () to
determine if it is better to use the X-R chart or the X-s chart. Historically, if your
subgroup size is less than 9, you would use the X-R chart; if larger than 9, you would
use the X-s chart.

A simulation was used that examined the values of  obtained by both charts using 20
subgroups with subgroup sizes (n) of 2 to 20. The simulation was run 1000 times for
each subgroup size under three different scenarios.

The first scenario was a stable process. In this scenario, there was little different
between the values of  obtained from the two different charts. It really doesn’t matter
that much which chart you used.

The second scenario used a mixed process. In this scenario, using s gave better
results. R seemed to overestimate the value of  .

The third scenario added some out of control points to the process. Neither method
predicted  very well. This is not surprising since the process is out of control.
However, it did appear that R definitely inflated the value of  .

Final conclusion: it appears to me that you might as well use the X-s chart all the time. It
works just as well as the X-R chart for the smaller subgroup sizes and it seems to
handle out of control situations better overall.

Comparing X-mR and Xbar-R Control

Charts
Home

SPC Knowledge Base

Comparing X-mR and Xbar-R Control Charts

August 2021

(Note: all the previous SPC Knowledge Base in the variable control chart category are
listed on the right-hand side. Select this link for information on the SPC for Excel
software.)

You have a set of data. You want to analyze it using a control chart.
What do you do? Many of us simply put the data into a software package and select the
control chart we want to use – and, presto, we have our control chart. Wonderful, isn’t
it? So simple. No thinking required.

There really should be some thought given to how a control chart is designed – and that
depends on what you want to obtain from the control chart. What variation do you want
the control chart to analyze/monitor? Every control chart is a movie of the variation in
your process over time. You just need to be sure it is a movie of the variation you are
interested in examining.

To explore this concept, two control charts will be examined and compared in this
publication: the individuals (X-mR) control chart and the X-R control chart. The X-R
chart, at one time, was the most used control chart. It has probably given way to the X-
mR chart as the use of SPC spread beyond manufacturing into other areas and
“frequent” data became less frequent. If you have infrequent data (like monthly), you
don’t have many options beyond the X-mR chart. Some say the X-mR chart is the only
chart you need to know because it always works. Yes, it usually works – but is it the
best choice always?

This publication compares the X-R chart with the X-mR chart to better understand the
differences – and to show that the X-mR chart actually can give more false signals than
the X-R control chart. A simulation is used to show why this is true. In this issue:

 Understanding the Variation Being Monitored

 Variation and the X-R Chart
 Variation and the X-mR Chart
 The Simulation
 Simulation Results
 Conclusions
 Summary
 Quick Links
Please feel free to leave a comment at the end of this publication. You can download a
pdf copy of the publication at this link.

Understanding the Variation Being Monitored

Control charts, when properly designed, monitor variation over time. Consider the
following scenario. You are in charge of a line that loads a product into bags. The
weight in the bags is critical. Too much and you are giving away product; too little and
you may get customer complaints that the bag is not full enough.

You want to take four samples per hour to monitor the variation. There are numerous
ways you could do this including taking four consecutive samples at the top of each
hour (Plan A). In this case, the X-R chart is the best option. Do you know why the X-mR
chart is not good for this sampling plan?
Another option is to take a sample every 15 minutes (Plan B). In this case, either the X-
R chart or the X-mR chart will work. But is one better than the other? We will see. First,
the variation examined on each chart is discussed.

The two sampling plans are shown in Figure 1.

Figure 1: Sampling Plans

Variation and the X-R chart

With the X-R chart, samples are taken, and a subgroup is formed. Remember, some
thought has to go into how you form the subgroups. The average of each subgroup is
found and plotted on the X chart. The range of each subgroup is found and plotted on
the range chart. What is the variation being monitored on the X and R chart?

The X chart monitors the variation in subgroup averages over time (often called the
between-subgroup variation). Each subgroup average is compared to the overall
average to determine if a significant change has occurred. If a subgroup average is
beyond the control limits, it is a signal that something has changed for that subgroup
average. The variation being monitored is shown in the figure below.
X is the overall average of the X values, R is the average range, UCL is the upper
control limit, LCL is the lower control limit, and A2 is a control chart constant that
depends on subgroup size.

What is the X chart monitoring in Plan A for the bag weight example?

The X chart monitors the variation in the average bag weight from hour to hour.

The range chart monitors the variation between results within a subgroup (often called
the within-subgroup variation). Each subgroup range is compared to the average range
to determine if a significant change has occurred. If a subgroup range is beyond the
control limits, it is a signal that something has changed for that subgroup range. This is
shown in the figure below.

D4 and D3 are control chart constants that depend on subgroup size.

What is the range chart monitoring in Plan A for the bag weight example?

The range chart monitors the variation in the four bag weights within a subgroup from
hour to hour.

What variation is being monitored if you used plan B with the X-R chart? The X-R chart
is still monitoring the same variations: the variation in average bag weight from hour to
hour and the variation in the four bag weights within a subgroup from hour to hour.

What is the difference between Plans A and B? In plan A, four consecutive bags are
measured in a row, while with Plan B, four bags are measured at 15 minute intervals.
What do you think will be the differences in the X-R charts for Plan A and Plan B? The
average range will probably be smaller for Plan A than Plan B. This is because Plan A
selects four bags in a row, while Plan B selects one every 15 minutes. Plan A is better
than Plan B because of this. It represents one of the tenets of rational subgrouping:
Select the subgroups so the chance for variation within a subgroup is minimized. This
forces the X chart to do the work of finding the special causes.

For more information, please see our four SPC Knowledge Base articles on rational
subgrouping.

Variation and the X-mR Chart

With the X-mR chart, the individual values (X) are plotted on the X chart, and the
moving range (mR) between consecutive points are plotted on the mR chart. The X
chart monitors the variation in individual values over time. You can refer to this as the
long-term variation for the X-mR chart. Each individual value is compared to the overall
average to see if it is significantly different. If the value is beyond the control limits, a
significant difference exists. This is shown in the figure below.

X is the average X value, and mR is the average moving range.

The mR chart monitors the variation in the range between consecutive X values over
time. You can refer to this as the short-term variation for the X-mR chart. Each moving
range value is compared to the average moving range to see if it is significantly
different. If a moving range is beyond the control limits, a significant difference exists.
This is shown in the figure below.

Why shouldn’t you use the X-mR chart for Plan A? Plan A collects four consecutive bag
weights at the start of each hour. If at all possible, you want to minimize the variation in
the mR chart, so the X chart does the work in finding out of control points. You attempt
to do this with the X-mR chart by ensuring that the time between the samples is the
same. The first three moving ranges represent these differences: |X1 – X2|, |X2 – X3|, and
|X3 – X4|. What is the fourth moving range? It is between X4 and X1 for the start of the
next hour. A large time difference compared to the first three samples. The odds of
these moving ranges being homogeneous is lower due to the sampling. Applying the X-
mR chart with Plan A will probably lead to problems – most likely missed signals.

Plan B will work with the X-mR chart. The time period between samples is the same so
the short-term variation between consecutive samples will probably be more
homogeneous.

What variation is the X chart monitoring in Plan B for the bag weight example?

The X chart is monitoring the variation in bag weights every 15 minutes.

What variation is the mR chart monitoring in Plan B for the bag weight example?

The mR chart is monitoring the variation in consecutive bag weights every 15 minutes.

Notice in the descriptions of what the X-R and X-mR charts are measuring, there is
always time mentioned. Time is always a part of the variation a control chart is
monitoring.

Plan B works with either chart. So, is one better than the other? Maybe it is the X-mR
chart. A point is plotted every 15 minutes compared to a point every hour with the X-R
chart. It is the more frequent charting of points that causes potential problems with the
X-mR chart. There are more false signals with the X-mR chart. This is shown in the
simulation described below.

The Simulation
A simulation was created to compare the X-mR chart and the X-R chart. The objective
is to compare the two charts for monitoring a process. Our process for the simulation
has an average of 100 and a standard deviation of 10. We will assume that our process
is normally distributed. A random number generator was used to generate 10,000 points
for a normal distribution.

These 10,000 points represent our “population.” This is all possible outcomes from the
process. A histogram of the 10,000 points is shown in Figure 2.

Figure 2: Population Histogram

The average of the 10,000 point is 99.999 with a standard deviation of 9.931. This
population is randomly sampled to obtain the X values. Since these 10,000 values don’t
change and the samples are randomly selected, you would expect the process to be in
statistical control.

The simulation (created in Excel) is run the following way:

 Randomly sample 120 X values from the population

 Construct the X-mR chart from the 120 X values
 Determine if there are any out of control points on either the X or mR chart
 Use the same 120 points and form subgroups of size 4 (in the order the data was
randomly selected)
 Construct the X-R from the 30 subgroups
 Determine if there are any out of control points on either the X or R chart
 Repeat this 1,000 times
 Compare the results between the two control charts
You might ask why this simulation is repeated 1000 times. It is called variation. The
simulation could be run 1 time and the results for the two control charts compared. But
will you get the same result the second time you run the simulation? Most likely not.
Figures 3 and 4 show the results for the X-mR chart from one run of the simulation.

Figure 3: X Chart from 1 Simulation Run

Figure 4: mR Chart from 1 Simulation Run

There is one out of control point on each of the X chart and the X-mR chart. Are these
true signals that the process has changed? No, they are not. The process doesn’t
change, and the X values were drawn at random from the process. These are false
signals.

The fraction of area beyond +/- 3 standard deviations for a true normal distribution
(doesn’t exist) is 0.0027 – which means you would expect 2.7 false signals per 1000
points you plot.

Figures 5 and 6 show the results for the X-R chart using the same data as in Figures 3
and 4 but with a subgroup size of 4.

Figure 5:X Chart from 1 Simulation Run

Figure 6: R Chart from 1 Simulation Run
There are no out of control points on the X-R chart, while there were two on the X-mR.
Will this always be the case? No out of control points on the X-R chart? No, there will be
false signals as well on the X-R chart. The subgroup averages are normally distributed,
so you would expect (for a true normal distribution) a false signal 2.7 times per 1000
points plotted on average.

Now, let’s see the results from running the simulation 1000 times.

Simulation Results
The simulation was run 1000 times. The calculations were done for each control chart.
The results are summarized in the table below.

Table 1: Simulation Results

 Chart X mR X R

Charts with no out of control points 758 350 931 889

% of Charts with out of control points 24% 65% 7% 11%

Points plotted 120,000 119,000 30,000 30,000

Points out of control 290 1014 73 114

% of Points out of control 0.24% 0.85% 0.24% 0.38%

The results look at two items primarily: the % of charts out of the 1000 that had at least
one point beyond a control limit and the % of points beyond the control limits.
Remember, the process is stable, so theoretically, these are false signals. When an X-
mR chart is used, 24% of the time the X chart had an out of control point. 65% of mR
charts had at least one out of control point. Both these are larger than the
corresponding numbers for the X-R chart, at 7% and 11%.

This implies that the X-mR produces more false signals than the X-R chart. Part of the
reason for this is that the X-mR plots each individual value while the X-R plots subgroup
averages – fewer points when the same data set is used in each chart. The points
plotted on the X chart for 1000 charts at 120 points a chart is 120,000. The X chart has
30 subgroup averages on each chart or 30,000 total points plotted (1/4 of the X chart
points since using a subgroup size of 4).

The % of points beyond the limits for the X and X chart is essentially the same at
0.24%. You would expect this since they both follow the normal distribution. The mR
chart though has 0.85% points out of control compared to 0.38% for the range chart.

Why does the mR chart has so many out of control points? Part of it is that the moving
range and subgroup range values are not normally distributed. The moving range chart
and the range chart are both skewed, with the moving range being more skewed.

Figures 9 and 10 show the histograms for the moving range values and the subgroup
range values.

Figure 9: Moving Range Histogram

Figure 10: Subgroup Range Histogram
The mR chart has more false signals in the simulation due to the fact it is a skewed
distribution. The subgroup range chart has more than the X chart – again because the
range distribution is skewed – but not as much as the mR chart.

Conclusions
If you are faced with a situation where you can use the X-R chart or the X-mR chart,
the X-R chart is probably the better choice. The subgrouping of data allows you to
specifically look at the variation you are interested in monitoring. In addition, for the
same amount of data, the X-R control chart gives less false signals than the X-mR
chart.

Summary
This publication compared the differences in the X-mR and X-R charts in terms of the
variation they monitor and in the number of false signals that can occur. In situations
where either can be used, it is probably best to use the X-R control chart. You plot fewer
points because you are subgrouping the data and this leads to fewer false signals
Xbar-s Control Charts: Part 1
Home

SPC Knowledge Base

Xbar-s Control Charts: Part 1

September 2008

In This Issue

 Introduction
 Understanding X-s Control Charts
 When to Use
 Steps in Construction
 Control Chart Constants
 Summary
 Quick Links
This month’s publication is the first part of a two part series on X-s charts. The X-s
chart is often overlooked in favor of the X-R chart. But, the X-s chart might actually be
the better chart to use. This month we will introduce X-s charts and describe how they
are constructed. Next month we will look at a detailed example of an X-s chart.

Introduction to X-s Control Charts

The most common control chart for years has been the X-R chart. This control chart
uses the range to measure the variation within a subgroup. For the measurements
within a subgroup, the range is the maximum – minimum value. The range is an easy
concept to understand – and to calculate. This was important when the control chart
calculations had to be done by hand or with a calculator. But with computer software,
this is no longer an issue. One problem with the X-R chart is that the range becomes a
poorer and poorer measure of within-subgroup variation as the subgroup size
increases. A different method is needed for the larger subgroup sizes. This is where
the X-s chart provides the solution. This month’s newsletter introduces the X-s chart.

This type of control chart is used with variables data – data that is taken along a
continuum. Time, density, weight, and length are examples of variables data. Like most
other variables control charts, it is actually two charts. One chart is for the subgroup
averages ( X). The other chart is for the subgroup standard deviations (s).

Understanding the X-s Control Chart

In the past, there has been reluctance to use the X-s chart in place of the X-R chart.
The standard deviation is just not as easy to understand as the range. Plus, there was
the calculation issue.
Yet, the X-s chart is very similar to the X-R chart. The major difference is that the
subgroup standard deviation is plotted when using the X-s chart, while the subgroup
range is plotted when using the X-R chart. One advantage of using the standard
deviation instead of the range is that the standard deviation takes into account all the
data, not just the maximum and the minimum. The constants used to calculate the
control limits and to estimate the process standard deviation are different for the X-s
chart than for the X-R chart. As for the X-R chart, frequent data and a method of
rationally subgrouping the data are required to use the Xbar-s chart.

The figures below are an example of an X-s chart. A company is tracking performance
of a bagging machine. Each bag should contain a minimum of 50 pounds (lbs) of sand.
Ten bags are weighed at the start of each hour. This provides frequent data as well as a
method of rationally subgrouping the data. The ten bags are used to form a subgroup,
so the subgroup size (n) is 10. The average weight of the ten bags is calculated. This is
the subgroup average ( X). The standard deviation of the ten bags is calculated. This is
the subgroup standard deviation (s).

The figure below is the X chart. The X values are plotted on this chart. Three lines are
plotted on the chart. The middle line is the overall process average; the upper line is
the upper control limit; and the lower line is the lower control limit.

The figure below is the s chart. The subgroup standard deviations are plotted on this
chart. Three lines are plotted on this chart as well. The middle line is the average
standard deviation ( s). The other two lines are the upper and lower control limits for the
subgroup standard deviations.
The bagging process represented by control charts above is in statistical control. Since
the s chart is in statistical control, this means that the variation between the individual
bag weights (the within-hour variation) is consistent over time. We don’t know what the
next s value will be, but we do know that it will be between 0.06 and 0.34 with a long
term average of 0.2. Since the X chart is in statistical control, the variation in subgroup
averages (the hour-to-hour variation) is consistent over time. The next result will be
between 50.1 and 50.49 with a long term average of 50.3. The process is consistent
and predictable in the near future.

When to Use X-s Control Charts

X-s charts are used to analyze a process operating over time. Like all control charts,
they will send a signal when a special cause of variation is present. You can use X-s
charts for any subgroup size greater than 1. This means they can be used in place
of X-R charts all the time. But, you definitely should use the X-s charts when the
subgroup size is 10 or more. The standard deviation gives a better estimate of the
variation in large subgroups than the range does.

Data has to be frequently available (multiple samples per hour, day, or week). The
subgroups should be formed with rational subgrouping in mind. In the bag weight
example above, the subgroups were formed so that the X chart examined the variation
in the average bag weight for the subgroup from hour to hour while the s chart
examined the variation within the subgroup from hour to hour. Selecting ten bags at the
start of the hour helps minimize the variation in the s chart and causes the out of control
situations to appear on the X chart.

In situations where either the X-s chart or the X-R chart can be used (small
subgroups), the choice is one of convenience more than anything. The control charts
will generally look very much alike and you will reach the same conclusions.

The Steps in Constructing an X-s Control Chart

The steps in constructing the X-s chart are given below. Most of the time you will use a
software program to generate control charts. However, it is important to understand how
the control charts are constructed and the steps in constructing them.

1. Gather the data.

Select the subgroup size (n). Typical subgroup sizes for an X-s chart are 10 or
more. However, you can use the X-s chart with any size subgroup of two or
more. The concept of rational subgroup should be considered. The objective is to
minimize the amount of variation within a subgroup. This helps you “see” the
variation in the averages chart easier.

Select the frequency with which the data will be collected. This will be part of the
rational subgrouping. Time is always important in taking data and in interpreting
the data. Data should be collected in the order in which they are generated (in
most cases).

Select the number of subgroups (k) to be collected before control limits are
calculated. You can start a control chart with only 5 subgroups. You will need to
recalculate the averages and control limits for each new subgroup until you have
at least twenty subgroups of data.

For each subgroup, record the individual sample results. For each subgroup,
calculate the subgroup average ( X): where X1, X2, etc. are the individual
sample results and n is the subgroup size:

For each subgroup, calculate the subgroup standard deviation:

2. Plot the data.

Select the scales for the x and y axes for both the X and s charts

Plot the subgroup standard deviations (s) on the s chart and connect consecutive
points with a straight line.

Plot the subgroup averages on the X chart and connect consecutive points with
a straight line.

3. Calculate the overall process averages and control limits.

Calculate the average standard deviation (s), where s1, s2, etc. are the standard
deviations for subgroups 1, 2, etc. and k is the number of subgroups:

Plot s on the s chart as a solid line and label.

Calculate the overall process average, where X1, X2, etc. are the subgroup
averages for subgroups 1, 2, etc:

Plot the overall process average on the X chart as a solid line and label.

Calculate the control limits for the s chart. The upper control limit is given by
UCLs. The lower control limit is given by LCLs. B4 and B3 are control chart
constants that depend on the subgroup size.

Plot the control limits on the chart as dashed lines and label.
Calculate the control limits for the X chart. The upper control limit is given by
UCLX. The lower control limit is given by LCLX. A3 is a control chart constant
that depends on the subgroup size.

Plot the control limits on the X chart as dashed lines and label.

4. Interpret both charts for statistical control.

Always consider variation first. If the s chart is out of control, the control limits on
the X chart are not valid since you do not have a good estimate of s. All tests
for statistical control apply to the X chart. Points beyond the control limits,
number of runs and length of runs apply to the s chart.

5. Calculate the process standard deviation, if appropriate.

If the s chart is in statistical control, the process standard deviation can be

calculated as:

c4 is a control chart constant that depends on subgroup size.

If the control charts indicate that the process is in statistical control, extend the
control limits into the future and monitor the process performance using these
control limits. If the control charts indicated that there are special causes of
variation, find the reason for the special cause of variation and remove it from the
process. Once you have 20 points in a row in statistical control, recalculate the
control limits based on that data, and use those limits in the future.

X-s Control Chart Constants

Below are the control charts constants for the X-s chart for subgroup sizes up to 25.
n A3 B3 B4 c4

2 2.659 3.267 0.7979

3 1.954 2.568 0.8862

4 1.628 2.266 0.9213

5 1.427 2.089 0.94

6 1.287 0.03 1.97 0.9515

7 1.182 0.118 1.882 0.9594

8 1.099 0.185 1.815 0.965

9 1.032 0.239 1.761 0.9693

10 0.975 0.284 1.716 0.9727

11 0.927 0.321 1.679 0.9754

12 0.886 0.354 1.646 0.9776

13 0.85 0.382 1.618 0.9794

14 0.817 0.406 1.594 0.981

15 0.789 0.428 1.572 0.9823

16 0.763 0.448 1.552 0.9835

17 0.739 0.466 1.534 0.9845

18 0.718 0.482 1.518 0.9854

19 0.698 0.497 1.503 0.9862

20 0.68 0.51 1.49 0.9869

21 0.663 0.523 1.477 0.9876

 22 0.647 0.534 1.466 0.9882

23 0.633 0.545 1.455 0.9887

24 0.619 0.555 1.445 0.9892

25 0.606 0.565 1.435 0.9896

Next month, we will continue our look at the X-s control chart.

Summary
This month’s publication has introduced the X-s control chart. Like most other variables
control charts, it is actually two charts. One chart is for the subgroup averages ( X). The
other chart is for the subgroup standard deviations (s). The X-s chart is very similar to
the X-R chart. The major difference is that the subgroup standard deviation is plotted
when using the X-s chart, while the subgroup range is plotted when using the X-R
chart. One advantage of using the standard deviation instead of the range is that the
standard deviation takes into account all the data, not just the maximum and the
minimum. An example of an X-s chart was introduced. The steps in constructing the
chart were covered and the constants used to calculate the control limits and to
estimate the process standard deviation were given.

Xbar-s Control Charts: Part 2

Home

SPC Knowledge Base

Xbar-s Control Charts: Part 2

October 2008

In this Issue:

 Bag Weight Process

 X-s Charts for the Bag Weights
 Interpreting the Charts
 Is the Process Capable?
 Comparison to X-R Charts
 Summary
 Quick Links
This month is the second publication on X-s charts. Last month we introduced the chart
and provided the steps in constructing an X-s chart. This month, we will look at a
detailed example of an X-s chart. The X-s chart is a type of control chart that can be
used with variables data. Like most other variables control charts, it is actually two
charts. One chart is for subgroup averages (X). The other chart is for subgroup standard
deviations (s). These charts are a very powerful tool for monitoring variation in a
process and detecting changes in either the average or the amount of variation in the
process.

Bag Weight Process

A process involves the filling of bags with sand. Each bag is supposed to weigh a
minimum of 50 pounds. The process has an automatic method of weighing the bags.
Once the bag reaches 50 pounds, the filling operation is supposed to stop. To
determine how much variation there is in the bag weights, an operator takes the first
four bags filled each hour and manually weighs them using a calibrated scale. The data
is shown in the table below.

Subgroup
Number 1 2 3 4
(k)

1 50.4 50.2 50.5 50.8

2 50.7 50.8 50.1 50.5

3 50.7 50.3 50.4 50.2

4 50.1 50.3 50.3 50.1

5 50.4 50.4 50.5 50.4

6 50.4 50.4 50.8 50.5

7 50.5 50.4 50.9 50.7

8 51 50.4 50.8 50.2

9 50.6 50.7 50.9 50.5

 10 50.4 50.6 50.4 50.7

11 50.2 50.3 50.2 50.4

12 50.5 50.5 50.4 50.9

13 50.2 50.4 50 50.8

14 50.2 50.4 50.7 50.6

15 50.7 50.6 50.2 50.3

16 50.6 50.6 50.3 50.5

17 50.5 50.6 50.5 50.3

18 50.9 50.6 50.5 50.7

19 50.7 50.4 50.3 50.7

20 50.3 50.2 50.6 50.6

We want to use an X-s chart to find out if this process is consistent and predictable (in
control) as well as if it is capable of producing bag weights with a minimum of 50
pounds.

The calculations needed to construct the X-s chart were covered in last month’s
newsletter. For practice, you can enter the data from this newsletter into Excel and see
if you can get the same results as shown below. The subgroup size is 4. For the data
above, the following averages and control limits can be generated.
For the X chart:

 Average = 50.49
 UCL = 50.81
 LCL = 50.16
For the standard deviation chart:

 Average = 0.2
 UCL = .45
 LCL = None

X-s Charts for the Bag Weights

The X-s charts are shown below. Are these charts in control? For information on
interpreting control charts, please see our past publications. Interpreting control charts
was covered in April 2004.

Interpreting the Charts

The X and s charts above are in statistical control. This means that the process is
consistent and predictable. There are only common causes of variation present (see the
January 2004 publication on variation for more information on common and special
causes of variation). What does it mean for an s chart to be in statistical control? It
means that variation within the subgroup is consistent from subgroup to subgroup. For
the bag weight example, it means that the standard deviation of each subgroup is the
“same” for all subgroups. There is no statistical difference between these standard
deviations. The average standard deviation will be 0.2 pounds but it can vary anywhere
from 0 to 0.45 pounds. Since the s chart is in control, you can estimate the standard
deviation of the individual values using the formula provided in last month’s publication
The standard deviation is 0.22.

What does it mean for the X chart to be in statistical control? It means that the variation
between subgroup averages is the same from subgroup to subgroup. For the bag
weight example, it means that there is no statistical difference between the subgroup
averages. As long as the process stays the same, we can predict the average bag
weights for the four samples. The long-term average will be 50.49. The average will
vary from 50.16 to 50.81.

Is the Process Capable?

Can the process meet the specification of having a minimum bag weight of 50? Be
careful here. The X chart plots the subgroup averages. The LCL on the X chart is 50.16.
This means that the average of the four bag weights will not be below 50.16 as long as
the process stays the same. However, the specification deals with individual bag
weights, not averages. Individual values will vary more than the subgroup averages.
Just because you do not have any subgroup averages below 50, does not mean that
you will not have any individual bag weights below 50.

Information on process capability is given on our website. We have a three part series
on process capability. The process capability chart for the bag weight process is shown
below. What can you conclude about the process capability?
Note that the data in the table above do not have any bag weights below 50. However,
the process capability analysis shows that the Cpk = 0.76. Since this is less than 1, it
means that there is out-of-specification material – some bags will weigh less than 50
pounds. The analysis predicts about 1% of the bags will weigh less than 50 pounds.
The only way to prevent any underweight bags from going to the customer with the
current process is to inspect 100% of the bags. Of course, the best approach would be
to improve the process – to move the average higher or reduce the amount of variation
– so no bags will be produced that weigh less than 50 pounds.

Comparison to X-R Charts

There is very little difference in the conclusions you reach if you use an X-R chart or
an X-s chart. The control limits on the X chart will essentially be the same. In 2005, we
had a multi-part series on X-R charts. If you look at the April 2005 newsletter, you will
see the same bag weight example. The results are essentially identical.

Summary
This month’s publication continued our look at X-s charts. A detailed example was
covered. The X-s chart was interpreted and explained. Being in control does not mean
being capable. In this example, the processes was not capable. The X-s and X-R
charts give very similar results.

Xbar-R Charts: Part 1

Home

SPC Knowledge Base

Xbar-R Charts: Part 1

March 2005

In this issue:
  Introduction to X-R Charts
 Example
 When to Use X-R Charts
 Steps in Constructing an X-R Chart
 Summary
 Quick Links
This month is the first in a multi-part publication on X-R charts. This month we
introduce the chart and provide the steps in constructing an X-R chart. Next month, we
will look at a detailed example of an X-R chart. The X-R chart is a type of control chart
that can be used with variables data. Like most other variables control charts, it is
actually two charts. One chart is for subgroup averages ( X). The other chart is for
subgroup ranges (R). These charts are a very powerful tool for monitoring variation in a
process and detecting changes in either the average or the amount of variation in the
process.

Introduction to X-R Charts

Suppose you are a member of a

bowling team. You bowl three games a night once a week in a bowling league. You are
interested in determining if you are improving your bowling game. What are some
different approaches you could use? One idea is that you could plot the score from each
game. However, you are more interested in what your average score is on a given
night. So another idea is to plot the average of the three games each night. You
definitely would like to increase that average over time. You are also interested in being
more consistent, i.e., not having one great game followed by a poor one. Thus, another
idea is to keep track of the range in scores for the three games each night. In situations
such as this (when you want to monitor averages over time but still keep track of the
variation between individual results), the X-R chart is very useful.

The X-R chart is a method of looking at two different sources of variation. One source
is the variation in subgroup averages. The other source is the variation within a
subgroup. Consider the bowling example above. You have data available on a fairly
frequent basis (three games each week). You can also rationally subgroup the data.
The three individual games you bowl on one night can be used to form a subgroup.
Continuing with the bowling example, suppose that one night your three bowling scores
are 169, 155, and 189. These three scores form a subgroup. You can calculate the
range of this subgroup by subtracting the minimum score from the maximum score.
Thus the range is:

Range = Maximum – Minimum = 189 – 155 = 34

You can plot this value on a range (R) chart. This is done for each subgroup (one night
of bowling three games). The range chart shows how much variation there is within
each subgroup, i.e., the amount of variation in your bowling scores on one night. You
would like this variation to be small and be consistent over time.

The chart for averages ( X) presents a different variation than the range chart. Using the
three scores above, you can calculate an average score for the night by taking the
average of the three individual scores. The subgroup average is:

X = (169+155+189)/3 = 171

You can plot this value on the X chart. This is done for each subgroup. The X chart
shows how much week-to-week variation there is in your weekly average bowling score.
You would like this variation to be small and be consistent over time. This permits you to
predict what your average score will be on any night, within certain limits.

The figure below is an example of the X-R chart for this bowling example. The top part
of the figure is the X chart. Each weekly average bowling score (i.e., the average of the
three individual games) is plotted. The overall average (Xdbar = X double bar) has been
calculated and plotted as a solid line. Xdbar is the average of all the subgroup
averages. Upper and lower control limits have also been calculated and plotted.
The X chart is in statistical control. The lower part of the figure is the range (R) chart.
The range is plotted for each week. The average range and control limits have been
calculated and plotted. The range is also in statistical control.

What does it mean when the X-R chart is in statistical control? It means that the
subgroup average is consistent over time and the variation within a subgroup is
consistent over time. We can predict what the process will do in the near future. In the
bowling example, this means that you can predict what the average of your three games
on any given night will be. Your average will be between about 158 and 208 with a long
term average of about 183. You can also predict what your range in bowling scores will
be on any given night. The range can be anywhere from 0 to about 62 with an average
range of about 24. As long as the process stays in control (your bowling), the results will
continue to the same.
Example

When to Use X-R Charts

X-R charts should be used when you have taken data frequently. How often you plot
points on the charts depends on your subgroup size. For example, if your subgroup size
is four, it will take four samples before you calculate the average and range and plot the
points. If you only take one sample per day, it will be four days before you can plot the
points. If the point is out of control, the reason for it could have occurred four days ago.
This often makes it difficult to find out what happened.

X-R charts should be used if you can rationally subgroup the data and are interested in
detecting differences between subgroups over time. This means there should be some
logical basis for the way the subgroups are formed. They should be formed to examine
the variation of interest to you. You might be interested in the variation from day to day.
In this case, samples from one day would be used to form a subgroup. The X chart
would examine the variation from day to day, while the R chart would examine the
variation within a day.

The R chart is a measure of the short-term variation in the process. Subgroups should
be formed to minimize the amount of variation within a subgroup. This causes
the X chart to do the work in detecting process changes.

Steps in Constructing an X-R Chart

The steps in constructing an X-R chart are given below.

1. Gather the data.

a. Select the subgroup size (n). Typical subgroup sizes are 4 to 5. The concept
of rational subgrouping should be considered. The objective is to minimize the
amount of variation within a subgroup. This helps us “see” the variation in the
averages chart more easily.
b. Select the frequency with which the data will be collected. Data should be
collected in the order in which it is generated (in most cases).
c. Select the number of subgroups (k) to be collected before control limits are
calculated. You can start with initial control limits after ten subgroups, but
recalculate the limits each time until you get to twenty subgroups.
d. For each subgroup, record the individual, independent sample results.
e. For each subgroup, calculate the subgroup average:

where n is the subgroup size.

f. For each subgroup, calculate the subgroup range:

R = Xmax – Xmin

where Xmax is the maximum individual sample result in the subgroup and Xmin
is the minimum individual sample result in the subgroup.

2. Plot the data.

a. Select the scales for the x and y axes for both the X and R charts.
b. Plot the subgroup ranges on the R chart and connect consecutive points with
a straight line.
c. Plot the subgroup averages on the X chart and connect consecutive points
with a straight line.

3. Calculate the overall process averages and control limits.

a. Calculate the average range (Rbar):

where k is the number of subgroups.

b. Plot Rbar on the range chart as a solid line and label.
c. Calculate the overall process average (Xdbar):

d. Plot X on the X chart as a solid line and label.

e. Calculate the control limits for the R chart. The upper control limit is given by
UCLr. The lower control limit is given by LCLr.

where D4, D3, are control chart constants that depend on subgroup size (see the
table below).
f. Plot the control limits on the R chart as dashed lines and label.
g. Calculate the control limits for the X chart. The upper control limit is given by
UCLx. The lower control limit is given by LCLx.
where A2 is a control chart constant that depends on subgroup size (see the
table below).

h Plot the control limits on the X chart as dashed lines and label.

4. Interpret both charts for statistical control.

a. Always consider variation first. If the R chart is out of control, the control limits
on the X chart are not valid since you do not have a good estimate of . All tests
for statistical control apply to the X chart. Points beyond the limits, number of
runs and length of runs tests apply to the R chart.

5. Calculate the process standard deviation, if appropriate.

a. If the R chart is in statistical control, the process standard deviation, s, can be

calculated as:

where d2 is a control chart constant that depends on subgroup size (see the
table below).

To calculate control limits and to estimate the process standard deviation, you
must use the control chart constants D4, D3, A2, and d2. These control chart
constants depend on the subgroup size (n). These control chart constants are
summarized in the table below. For example, if your subgroup is 4, then D4 =
2.282, A2 = 0.729, and d2 = 2.059. There is no value for D3. This simply means
that the R chart has no lower control limit when the subgroup size is 4.

Subgroup
A2 D3 D4 d2
Size (n)
2 1.880 3.267 1.128

3 1.023 2.574 1.693

4 0.729 2.282 2.059

5 0.577 2.114 2.326

6 0.483 2.004 2.534

7 0.419 0.076 1.924 2.704

8 0.373 0.136 1.864 2.847

9 0.337 0.184 1.816 2.970

10 0.308 0.223 1.777 3.078

11 0.285 0.256 1.774 3.173

12 0.266 0.284 1.716 3.258

13 0.249 0.308 1.692 3.336

14 0.235 0.329 1.671 3.407

15 0.223 0.348 1.652 3.472

16 0.212 0.364 1.636 3.532

17 0.203 0.379 1.621 3.588

18 0.194 0.392 1.608 3.640

19 0.187 0.404 1.596 3.689

 20 0.180 0.414 1.586 3.735

21 0.173 0.425 1.575 3.778

22 0.167 0.434 1.566 3.819

23 0.162 0.443 1.557 3.858

24 0.157 0.452 1.548 3.895

25 0.153 0.459 1.541 3.931

Summary
This publication has introduced the X-R chart. When you should use an X-R chart was
covered as well as the steps in constructing the chart.

Xbar-R Charts: Part 2

Home

SPC Knowledge Base

Xbar-R Charts: Part 2

April 2005

In this issue:

 Bag Weight Process

 X-R Chart
 Interpreting the Charts
 Is the Process Capable?
 Summary
 Quick Links
This month is the second in a multi-part publication on X-R charts. Last month we
introduced the chart and provided the steps in constructing an X-R chart. This month,
we will look at a detailed example of an X-R chart. The X-R chart is a type of control
chart that can be used with variables data. Like most other variables control charts, it is
actually two charts. One chart is for subgroup averages ( X). The other chart is for
subgroup ranges (R). These charts are a very powerful tool for monitoring variation in a
process and detecting changes in either the average or the amount of variation in the
process.

Bag Weight Process

A process involves the filling of bags with sand. Each bag is supposed to weigh a
minimum of 50 pounds. The process has an automatic method of weighing the bags.
Once the bag reaches 50 pounds, the filling operation is supposed to stop. To
determine how much variation there is in the bag weights, an operator takes the first
four bags filled each hour and manually weighs them using a calibrated scale. The data
are shown in the table.

We want to use an X-R chart to find out if this process is consistent and predictable (in
control) as well as if it is capable of producing bag weights with a minimum of 50
pounds.

The calculations needed to construct the X-R chart were covered in last month’s e-zine.
For practice, you can copy the data from this e-zine into Excel and see if you can get
the same results as shown below. The subgroup size is 4. For the data above, the
following averages and control limits can be generated.

For the X chart:

Average = 50.49
UCL = 50.81
LCL = 50.16

For the range chart:

Average = 0.44
UCL = 1.00
LCL = None

The X-R charts are shown below. Are these charts in control? For information on
interpreting control charts, please see our past publication on our website. Interpreting
control charts was covered in April 2004.

X-R Chart
Interpreting the Charts
The X and R charts above are in statistical control. This means that the process is
consistent and predictable. There are only common causes of variation present (see the
January 2004 e-zine on variation for more information on common and special causes
of variation).

What does it mean for a range chart to be in statistical control? It means that variation
within the subgroup is consistent from subgroup to subgroup. For the bag weight
example, it means that the range between the heaviest bag and lightest bag is the
“same” for all bag weights. There is no statistical difference between these ranges. The
average range will be 0.44 pounds but it can vary anywhere from 0 to 1 pounds. Since
the range chart is in control, you can estimate the standard deviation using the formula
provided in last month’s e-zine. The standard deviation is 0.21

What does it mean for the X chart to be in statistical control? It means that the variation
between subgroup averages is the same from subgroup to subgroup. For the bag
weight example, it means that there is no statistical difference between the subgroup
averages. As long as the process stays the same, we can predict the average bag
weights for the four samples. The long-term average will be 50.49. The average will
vary from 50.16 to 50.81.

Is the Process Capable?

Can the process meet the specification of having a minimum bag weight of 50? Be
careful here. The X chart plots the subgroup averages. The LCL on the X chart is
50.16. This means that the average of the four bag weights will not be below 50.16 as
long as the process stays the same. However, the specification deals with individual
bag weights, not averages. Individual values will vary more than the subgroup
averages. Just because you do not have any subgroup averages below 50, does not
mean that you will not have any individual bag weights below 50.

Information process capability is given on our website. We have a three-part series on

process capability. The process capability chart for the bag weight process is shown in
the figure above. What can you conclude about the process capability?

Note that the data in the table do not have any bag weights below 50. However, the
process capability analysis shows that the Cpk = 0.74. Since this is less than 1, it
means that there is out-of-specification material – some bags will weigh less than 50
pounds. The analysis predicts about 1% of the bags will weigh less than 50 pounds.
The only way to prevent any underweight bags from going to the customer with the
current process is to inspect 100% of the bags. Of course, the best approach would be
to improve the process – to move the average higher or reduce the amount of variation
– so no bags will be produced that weigh less than 50 pounds.

Summary
This publication showed how to create an X-R chart using bag weights as an example.
The process was in statistical control. Being in statistical control does not mean that the
process is capable. The process capability chart showed that the process was not
capable. To make it capable, the process has to be fundamentally change.
Individuals Control Charts
Home

SPC Knowledge Base

Individuals Control Charts

October 2006

In this issue:

 Introduction
 X Chart
 Moving Range Chart
 Statistical Control
 When to Use
 Steps in Constructing
 Example: Waiting in Line
 X Chart – Example
 Moving Range Chart – Example
 Summary
 Quick Links
Suppose your process generates data on a very limited frequency. Maybe you only get
data once a day, once a week, or once every two weeks. How can we apply control
charts to these types of data? If we wait to get several data points to form a subgroup,
we won’t be able to plot a point very often. Perhaps the test method used to analyze the
process is very expensive to run or takes a long time. How can we handle this type of
situation? In these instances, individuals control charts are useful. This type of chart is
useful when you have only one data point at a time to represent a given situation. The
individuals control chart is introduced in this publication.

Introduction
The individuals control chart is a type of control chart that can be used with variables
data. Like most other variables control charts, it is actually two charts. One chart is for
the individual sample result (X). The other chart is for the moving range (R) between
successive individual samples. The individuals chart is very useful for monitoring
processes where data are not available on a frequent basis. The individuals control
chart examines variation in individual sample results over time. While rational
subgrouping does not apply, thought must be given to when the results will be
measured. If the process is in statistical control, the average on the individuals chart is
our estimate of the population average. The average range will be used to estimate the
population standard deviation.

The individuals control chart is a method of looking at variation. One source of variation
is the variation in the individual sample results. This represents “long-term” variation in
the process. The second source of variation is the variation in the ranges between
successive samples. This represents “short-term” variation.

X Chart

The figure in this section is an example

of an X chart for individual results. In this example, a sample is pulled once a day from a
given process stream and tested for a certain contaminant. Since data are not obtained
very frequently, an individuals control chart was selected. On the first day the test
method indicated that there were 105 ppm of the contaminant in the product stream. On
the second day, the sample result was 93 ppm. The overall process average has been
calculated and plotted as a solid line. The upper and lower control limits have also been
calculated and plotted as dashed lines.

Moving Range Chart

The figure in this section is the moving

range chart that goes with the X chart. This chart represents the range between
successive data points. This range is often called a moving range. The range value
plotted for the second day is simply the range between day 1 and day 2 (largest minus
smallest). This range is 105 – 93 = 12. The average range has been calculated and
plotted as a solid line. The upper control limit has also been calculated and plotted as a
dashed line. There is no lower control limit on the range chart for an individuals chart.

Statistical Control
The charts above are in statistical control. What does it mean when the individuals
control chart is in statistical control? It means that the individual sample results are
consistent over time, i.e., they are not significantly different from the process average. It
also means that the difference between successive sample results is consistent over
time. We can predict what the process will make in the near future. For more
information on interpreting control charts, please see our April 2004 publication on our
website.

When to Use
Individuals control charts should be used when there is only one data point to represent
a situation at a given time. The individuals control chart allows you to plot a point on the
chart for each sample taken. This permits you to determine if the process is in statistical
control or not for each sample taken. This may seem to imply that you should always
use individuals charts. That is not necessarily true as is discussed below.

Individuals control charts are not as sensitive to changes as Xbar-R charts. In addition,
values of X and R can have significant variation (even though the process is in control)
until the number of individual data points reaches 100.

Steps in Constructing

The steps in constructing the individuals control chart are given below.

1. Gather the data.

a. Select the frequency with which the data will be collected. Data should be collected in
the order in which they are generated.

b. Select the number of data points (k) to be collected before control limits are
calculated (at least twenty). These will be tentative limits. Control limits should be
recalculated after you have collected 100 data points.

c. Record the individual sample results.

d. Calculate the moving range (Ri+1) between consecutive sample results:

Ri+1 = |Xi+1 – Xi|

where Xi+1 is the result for sample i+1 and Xi is the result for sample i. The range value
is always positive.
2. Plot the data.

a. Select the scales for the x and y axes for both the X and R charts.

b. Plot the ranges on the R chart and connect consecutive points with a straight line.

c. Plot the individual sample results on the X chart and connect consecutive points with
a straight line.

3. Calculate the overall process averages and the control limits.

a. Calculate the average moving range (Rbar):

Note there is always one less range value than individual sample results for the
individuals control chart.

b. Plot Rbar on the range chart as a solid line and label.

c. Calculate the overall process average (X):

d. Plot X on the X chart as a solid line and label.

e. Calculate the control limits for the R chart. The upper control limit is given by UCLr.
The lower control limit is given by LCLr.

f. Plot the upper control limit on the R chart as a dashed line and label.

g. Calculate the control limits for the X chart. The upper control limit is given by UCLx.
The lower control limit is given by LCLx.

h. Plot the control limits on the X chart as dashed lines and label.

4. Interpret both charts for statistical control.

a. Always consider variation first. If the R chart is out of control, the control limits on the
X chart may not be valid since you do not have a good estimate of Rbar.

b. All tests for statistical control apply to the X chart. However, the data on the range
chart are not independent. Each data point is used twice. The only test that is valid for
the range chart is points beyond the control limits.

5. Calculate the process standard deviation, if appropriate.

a. If the R chart is in statistical control, the process standard deviation,

s, can be calculated as:

Example: Waiting in Line

Waiting in line at a bank can be very frustrating. Your bank has just implemented a
promise that your wait in line will never be longer than five minutes. You decide to find
out if this is really true. Since you only go to the bank about once a week, you have
infrequent data. You decide to use an individuals/moving range chart to determine if the
bank is keeping its promise. Each time you go to the bank, you measure how long you
wait in line to the nearest 0.1 minute. The results for 24 trips to the bank are given
below.

Trip Waiting in Line

1 3.5

2 2.4

3 4.1

4 2.8

5 3

6 4.7

7 1.2

8 0.9

9 2.5
 10 3.1

11 3.6

12 4.1

13 3.8

14 2.5

15 2.8

16 4.3

17 4.1

18 3.6

19 2.4

20 4.8

21 3.5

22 2.5

23 1.3

24 4.5

The first step after collecting the data is to calculate the moving range between trips to
the bank. For example, the moving range is the range between trips 1 and 2 is:

Ri+1 = |Xi+1 – Xi|=3.5 – 2.4 = 1.1

This is done for each consecutive trips. The moving range between trips 2 and 3 is |2.4
– 4.1| = 1.7. Remember, the range is always positive.
The next step is to calculate the overall process average and the average range. The
overall process average is determined by adding up the individual results for each trip
and dividing by the number of samples (trips). In this case, the number of trips (k) is 24.
The average range is determined by adding up the moving range values and dividing by
k- 1 since there is one less range value than individual samples.

Xbar = 76/24 = 3.17

Rbar = 27.4/23 = 1.19

The next step is to calculate the control limits. The control limits are:

UCLr = 3.27 Rbar = 3.27(1.19) = 3.89

LCLr = None

UCLx = Xbar + 2.66Rbar = 3.17 + 2.66(1.19) = 6.34

LCLx = Xbar – 2.66Rbar = 3.17 – 2.66(1.19) = 0.005

If the moving range chart is in control, the standard deviation of the individual results
can be determined. The moving range chart (as shown below) is in control. The
standard deviation is then given by:

‘ = Rbar/1.128 = 1.19/1.128 = 1.05

X Chart – Example
The X chart for waiting in line is shown in this example. The moving range chart is
shown below. The process is in control. This means, that as long as the process stays
the same, you can predict, within a range, how long you will wait in line when you go to
the bank.

This means that, when you go to the bank, you will wait in line anywhere from 0 to 6.34
minutes. What does this mean about your bank’s commitment that you will not wait
more than 5 minutes in line? It means that the bank is not capable of meeting that
guarantee. You have not waited more than 5 minutes yet based on the data. But it is
just a matter of time before you do.

Moving Range Chart – Example

Summary
This publication has introduced individuals control charts. This type of chart should be
used when data are infrequently available. The X chart is examining the long-term
variation in individual sample results. The range chart is examining the short-term
variation between consecutive sample results. Like all control charts, the individuals
chart is used to determine if the process is in statistical control.