JMP®: Jump into Control
Charting
Workshop Notes
JMP®: Jump into Control Charting Workshop Notes was developed by Di Michelson. Additional
contributions were made by Ledi Trutna, Scott Wise, and Monical Beals.
JMP Statistical Discovery, SAS and all other SAS Institute Inc. product or service names are
registered trademarks or trademarks of JMP Statistical Discovery and SAS Institute Inc. in the USA
and other countries. ® indicates USA registration. Other brand and product names are trademarks of
their respective companies.
JMP®: Jump into Control Charting Workshop Notes (version 18.1)
Copyright © 2025 JMP Statistical Discovery LLC, Cary, NC, USA. All rights reserved. Printed in the
United States of America. No part of this publication may be reproduced, stored in a retrieval system,
or transmitted, in any form or by any means, electronic, mechanical, photocopying, or otherwise,
without the prior written permission of the publisher, JMP Statistical Discovery LLC.
1.1 Control Chart Builder 3
Jump into Control Charting
1.1 Control Chart Builder ................................................................................................................ 4
Demonstration: Control Chart Builder ..................................................................................... 10
1.2 X-MR Charts.............................................................................................................................. 16
1.3 Xbar-R and Xbar-S Charts ....................................................................................................... 20
Demonstration: Xbar-R and Xbar-S Charts ............................................................................ 23
1.4 Phases ....................................................................................................................................... 29
Demonstration: Using Phases in Control Chart Builder .......................................................... 31
Practice .................................................................................................................................... 37
1.5 Solutions ................................................................................................................................... 38
Solutions to Practices .............................................................................................................. 38
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
4 Jump into Control Charting
1.1 Control Chart Builder
A control chart is a time plot of data with special limits added to the chart. These limits are based on
the historical variation of the data. The centerline is at the historical process mean, and the upper
and lower control limits correspond to the edges of the historical distribution. The purpose of a
control chart is to detect when the distribution of the process has changed. A point outside of the
control limits is a signal that the process has shifted.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.1 Control Chart Builder 5
There are many different types of control charts, and JMP can create all of these. All but the last
category can be created with JMP’s Control Chart Builder platform.
The purpose of a control chart is to detect if the distribution has changed. A distribution can change
in two ways: the location can shift or the dispersion can get bigger.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
6 Jump into Control Charting
This is why control chart usually come in pairs, one for location and one for dispersion.
The big question in control charts is how to estimate sigma? We’ll use the dispersion in the data. In
control charts, you always want to make sure that you are estimating sigma using only common
cause variation – natural process variation, not any estimator that might contain special cause
variation – variation outside the normal process. That’s the reason we use short-term estimators.
You’ll never want to use the average plus or minus three times the standard deviation to calculate
control limits. The sample standard deviation of all the data can contain special cause variability, and
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.1 Control Chart Builder 7
it will inflate the control limits so you will miss true changes in your process. The type of control chart
determines the estimator of sigma, and all estimators of sigma are biased.
The bias in these estimators is known and there are bias correction factors to adjust the sigma
estimators for the bias.
For example, if you take the average moving range of individual observations, that’s a biased
estimator of sigma. But dividing by 1.128 makes it an unbiased estimator. JMP uses unbiased
estimators when it calculates control limits.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
8 Jump into Control Charting
Now we know about estimating sigma, where does three sigma come from? Walter Shewhart, the
father of statistical quality control, observed empirically that limits at three standard deviations from
the mean balances the trade-off between too many false signals and missing true process changes.
The empirical rule backs up Shewhart’s observations.
Here are some examples of the empirical rule. No matter what the shape of the data, over 98% of
the distribution falls within three standard deviations of the mean.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.1 Control Chart Builder 9
JMP’s Control Chart Builder platform helps you interactively investigate the control status of your
process. It automatically selects the type of chart based on your variables, you can immediately add
or remove variables, change the plotting statistic, format the chart, add additional charts for other
variables, and even create three-way charts for nested variables, all without leaving the platform.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
10 Jump into Control Charting
Control Chart Builder
This demonstration is an introduction to Control Chart Builder. You examine the platform in greater
detail in later demonstrations.
The critical dimension has been measured on each lot. The data are in the table Critical Dimension.
1. Open Jump into Control Charts.jrn.
2. Select Critical Dimension.
3. Click the Show Header Graphs icon.
There are 120 rows and three columns. There is one value of CD Measurement per LotID.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.1 Control Chart Builder 11
4. Select Analyze > Quality and Process > Control Chart Builder.
Control Chart Builder works similarly to Graph Builder. Variables are dragged into zones to
assign them to various roles. Use right-clicking and the Control Panel (initially empty) to change
graph options.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
12 Jump into Control Charting
5. Select CD Measurement and drag to the center of the graph area without releasing the mouse
button.
The blue rounded rectangle shows the zone where the variable will be dropped. The graph
shows a preview of the control chart. You can also drag the variable to the Y drop zone.
6. Release the mouse button.
The initial control chart plots the values of CD Measurement in row order, along with Individual
control limits. A moving range chart is placed below the Individual chart. This pair of charts is
suitable for controlling a process with subgroups of size one.
Note: The control panel on the left enables you to change the chart type and statistic being plotted,
display or remove points and lines, and change how the control limits are calculated, as well
as other options.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.1 Control Chart Builder 13
7. Click the red triangle next to Control Chart Builder to see the menu options.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
14 Jump into Control Charting
8. Right-click in the body of the graph to see additional menu options.
9. Click Done.
The control panel has been closed. The chart and limits summaries are displayed. The CD
Measurement Limit Summaries report contains the control limit values and information on how
those limits were calculated for each displayed chart.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.1 Control Chart Builder 15
10. Close the control chart report and the data table without saving.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
16 Jump into Control Charting
1.2 X-MR Charts
When you collect measurements on your process and those measurements have the continuous
modeling type, you can make a Shewhart control chart. A subgroup represents every time you
collect data. If you are collecting one observation per subgroup, the appropriate Shewhart chart is
called an X-MR chart or an Individuals and Moving Range control chart. If you are collecting multiple
observations per subgroup, the appropriate Shewhart chart is an Xbar-R or Xbar S chart.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.2 X-MR Charts 17
The purpose of a control chart is to determine whether the mean or variance of the process has
changed. When you have subgroups of size one, the individual values will tell you if the process
mean has changed and the range of successive observations will tell you if the process variance has
changed.
For each pair of consecutive observations plotted on the individuals chart, we calculate the moving
range.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
18 Jump into Control Charting
As data are being collected, the centerline and control limits update. The centerline on the chart for
individuals is the average of the observations. The centerline on the chart for moving ranges is the
average of the moving ranges. There are n – 1 moving ranges for n subgroups.
Since the average moving range is a biased estimator of sigma, we use the d2 bias correction factor,
then use three sigma limits. The moving range chart doesn’t have a lower control limit. JMP will draw
a line at zero for the lower control limit.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.2 X-MR Charts 19
Here are some values of control chart constants used in Shewhart charts for different subgroup
sizes.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
20 Jump into Control Charting
1.3 Xbar-R and Xbar-S Charts
When you have subgroups of size greater than one, the subgroup average will tell you if the process
mean has changed and the subgroup range or standard deviation will tell you if the process variance
has changed.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Xbar-R and Xbar-S Charts 21
If you use a range chart with your Xbar chart, then sigma is estimated using the average range. If
you use a standard deviation chart with your Xbar chart, then sigma is estimated using the average
standard deviation. There are bias correction factors to get you to the correct three sigma limits.
For subgroups of size two, the range and standard deviation contain equivalent information about
the true population variability.
As sample size increases, the standard deviation and range are near equivalent estimates of the
variability, until about n = 15.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
22 Jump into Control Charting
Perhaps the best recommendation is to use the range if calculations are done by hand and use the
standard deviation if calculations are done by machine.
With two charts, which chart should you interpret first? The answer is to interpret the dispersion chart
first. The control limits for both charts are based on the estimate of sigma that comes from the
centerline of the dispersion chart. If the dispersion chart is out of control, the limits on the location
chart are too wide.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Xbar-R and Xbar-S Charts 23
Xbar-R and Xbar-S Charts
Suppose you prefer to measure the critical dimension of more than one part in each lot. You have
five readings per lot and would like to plot the average and range on an Xbar-R chart.
Xbar-R Chart
1. Return to the workshop journal and select CD Xbar Charts.
2. Click the Show Header Graphs icon.
These data have five parts measured in each lot, with all the measurements coming from one
piece of equipment.
3. Select Analyze > Quality and Process > Control Chart Builder.
4. Drag and drop CD in the middle of the graph.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
24 Jump into Control Charting
The initial graph is an X-MR chart of CD. The chart does not recognize that there are five
observations per lot.
5. Drag LotID to the middle of the graph without dropping.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Xbar-R and Xbar-S Charts 25
Because LotID is a nominal variable, it is added to the Subgroup role in the graph. You can also
drag the variable to the Subgroup role.
6. Release the mouse button.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
26 Jump into Control Charting
The Xbar-R chart is displayed. The points in the top chart are the subgroup averages, that is, the
average of the critical dimension from the five measurements for each lot. The bottom chart
shows the subgroup ranges, that is, the range of the critical dimension from the five
measurements from each lot.
The calculated control limits are displayed in the CD Limits Summary report. This report
indicates that sigma is estimated by the average range of subgroups of size five.
Xbar-S Chart
Control Chart Builder enables you to create an Xbar-S control chart without having to relaunch
the platform.
1. Under Points[2], change Statistic to Standard Deviation.
2. Under Limits[1], change Sigma to Standard Deviation.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.3 Xbar-R and Xbar-S Charts 27
Notice that, under Limits[2], the value of Sigma automatically changed to Standard Deviation.
Note: These options can also be changed in the graph itself. You can right-click on the Xbar chart
and select Limits > Sigma > Standard Deviation. You can right-click in the Range chart
and select Points > Statistic > Standard Deviation.
The CD Limit Summaries report has been updated to show the limit calculation method. Save
the limits to the data table.
3. Click the red triangle next to Control Chart Builder and select Limits > Save Control Limits >
in Column.
4. Return to the data table.
5. In the column panel, click the asterisk next to CD, then select Control Limits.
The control limits for both Xbar and S have been saved.
More than one set of limits can be saved as a column property. Click the XBar drop-down to see
the other types of charts whose limits can also be saved as a column property.
6. Click Cancel.
7. Return to the Control Chart Builder window.
8. Click the red triangle next to Control Chart and select Limits > Show Limit Labels.
9. Click Done.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
28 Jump into Control Charting
The Limits Summaries report has been updated to show that the limits being used are those that
were saved as a column property. New data added to the table will not change these limits. The
report also shows that a user defined value was used for the estimate of sigma.
10. Close the control chart and the data table without saving.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.4 Phases 29
1.4 Phases
Control charts can be used in all phases of product and process development from exploration
through experimentation and launch. In early stages, the control chart limits are updated as new
data are added to the chart. After launch comes using the chart in production, and here the limits are
fixed. You might still have projects to improve or change the process, and you can update the limits
after successful changes are made, but the limits will not change as new data are added.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
30 Jump into Control Charting
You can also use the Phase role in Control Chart Builder to explore fixed factors. Maybe your
process uses different machines or gauges or operators. You can look for a difference in those
variables by using a Phase variable. If your control charts look different for each level of the fixed
factor, like over different machines, then determine why that’s happening. Then decide if you want to
monitor each machine separately or initiate an improvement project to match the machines.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.4 Phases 31
Using Phases in Control Chart Builder
The critical dimension is measured on parts manufactured on four different pieces of equipment. You
will use control charts to determine whether the equipment produces a similar distribution of
measurements. In this dataset, each lot contains measurements for nine sites measured on one
part.
1. Return to the workshop journal and select CD multiple EqID.
2. Click the Show Header Graphs icon.
There are 292 lots, four levels of EqID, and nine levels of site per lot.
3. Select Analyze > Quality and Process > Control Chart Builder.
4. Drag CD to the Y zone.
5. Drag LotID to the Subgroup zone.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
32 Jump into Control Charting
The control limits have been calculated for all the data. These limits do not look appropriate for
these data. The values on the Xbar chart have bands or stripes. Is there a difference based on
equipment ID?
6. Drag EqID to the Phase zone.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.4 Phases 33
Control limits have been calculated for each level of the phase variable. The Limits Summaries
report shows the four sets of limits for the four unique levels of EqID. The X axis has also been
sorted by LotID within EqID.
Another categorical variable in the data table is site. Is there a difference in between the multiple
sites measured in each lot?
7. Drag site to the Phase zone, replacing EqID.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
34 Jump into Control Charting
Yes, the values of site 1 are the lowest, sites 2-5 have medium values, and sites 6-9 have higher
values of CD.
8. In the control panel, open the Warnings[1] outline, then select the checkbox for Test Beyond
Limits.
9. In the column list, select EqID, then click By.
10. Click the red triangle next to Control Chart Builder and select Show Control Panel. This action
is equivalent to clicking the Done button for each level of the By variable.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.4 Phases 35
It is now easy to see that Etch 3 produced the very high value and Etch 4 produced the very low
value. Etch 2 produced a lot with a high value for each site, but it was hidden in the variation
among equipment on the other control charts. (Only Etch 1 is shown in the screen capture.)
11. Close the control chart and data table without saving.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
36 Jump into Control Charting
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.4 Phases 37
Practice
1. Creating Control Charts
In a printing process, the color of the printed output is verified once per day by printing a test
page and using a colorimeter to compare the color of the test page to a color standard. The
colorimeter measures hue wavelength. Visible light wavelengths vary from about 400nm in the
blue range to about 700nm in the red range. The orange wavelength standard has a peak
wavelength of 590nm. The orange of your test page is measured daily. Data can be found in the
Wavelength data table.
a. What pair of control charts is best to control Wavelength?
b. Create an appropriate pair of charts for Wavelength.
c. Do the measurements appear in date order? If so, why? If not, fix the chart so that time is
read from left to right.
d. Close the control chart builder window and the data table without saving.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
38 Jump into Control Charting
1.5 Solutions
Solutions to Practices
1. Creating Control Charts
In a printing process, the color of the printed output is verified once per day by printing a test
page and using a colorimeter to compare the color of the test page to a color standard. The
colorimeter measures hue wavelength. Visible light wavelengths vary from about 400nm in the
blue range to about 700nm in the red range. The orange wavelength standard has a peak
wavelength of 590nm. The orange of your test page is measured daily. Data can be found in the
Wavelength data table.
a. What pair of control charts is best to control Wavelength?
1) Return to the course journal and select Wavelength.
2) Because one measurement is collected for each day, Date is the subgroup, and an X-
MR chart is appropriate.
b. Create an appropriate pair of charts for Wavelength.
1) Select Analyze > Quality and Process > Control Chart Builder.
2) Drag Wavelength to the Y zone.
c. Do the measurements appear in date order? If so, why? If not, fix the chart so that time is
read from left to right.
1) With no variable in the Subgroup role, the data appear in row order, which is reverse
date order.
2) Drag Date to the Subgroup zone.
3) Click Done.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
1.5 Solutions 39
d. Close the control chart builder window and the data table without saving.
Copyright © JMP Statistical Discovery LLC, Cary, North Carolina, USA. ALL RIGHTS RESERVED.
