Supplement 6

Statistical Process Control

Background
Of all the mathematical concepts presented in the text, SPC charts certainly rank among the most
important to cover in an operations management course. Instructors who want to delve into some of the
statistical background and who want to cover acceptance sampling could present Supplement 6 as rather
challenging for business students. On the other hand, instructors who focus on the basics of the four SPC
charts will find that most students can grasp the techniques rather handily.

It’s important to emphasize that control limits are not the same thing as engineering specification limits.
Control limits are designed to monitor output over time to ensure that the system continues to produce
consistent output. Output produced outside of control limits sends a signal that something extraordinary
has occurred and the firm should investigate. One way to illustrate this in class is to bump into the
projector in the middle of introducing control charts. Then point out that if you did not turn around
periodically to look at the screen, you would never know that the slide is now projecting to the ceiling. So
even though the equipment was set up perfectly at the beginning, we use control charts (turn around and
look at our output) to periodically check to ensure that output still looks OK. Spec limits, on the other
hand, are fixed engineering measurements that define exactly what determines acceptable output or not.
Any output produced outside of spec limits must be rejected. It is possible for output to exceed control
limits and still be within spec limits. And, theoretically at least, it is possible for output that exceeds spec
limits be appear to be within statistical control.

With SPC charts, students can get tripped up in several places. First, they must be able to choose the
appropriate chart for the problem at hand. For variables (anything that can be measured along a number
line, such as length, temperature, volume, etc.), both and R-charts must be used. Figure S6.5 (and slides
S6-34 and S6-35) show nicely why both are necessary. Choosing the correct chart for attributes (binary
output such as percent defects) can be a bit tricky. The p-chart is used to monitor the percentage over
time, while the c-chart is used to monitor the number of occurrences over time. But what if students are
provided a set of number of errors over time, as in Example S4, but without a “Fraction Defective”
column? The answer is that they should convert those numbers to percentages and use a p-chart. What’s
the difference? A p-chart is used for output that has a specific sample size. A c-chart, on the other hand,
has no sample size. A c-chart looks at an “event,” which has no practical upper limit on the number of
occurrences. For example, students could make a c-chart to monitor the number of mistakes in the
instructor’s lectures over time. The instructor could have one mistake in a day, or five, or ten, or...there’s
really no upper limit. Hence, there is no sample size and it is impossible to convert the numbers into a
comparable fraction from day to day.

Another potential area of confusion for students is mixing up sample size and number of samples,
particularly for and R-charts, and especially if the data are presented with output from each sample
going down the columns of a table instead of across the rows. Students sometimes also do not fully
realize that -charts examine sample averages. So it is perfectly acceptable to have an individual product
with output that exceeds the control limits as long as the average for that entire sample still lies within the
limits (assuming, of course, that the individual item was still produced within spec limits). With p-charts,

122
Copyright ©2017 Pearson Education, Ltd.
123 Supplement 6

students sometimes enter the number of samples for n instead of the sample size, and they sometimes
round too much within the square root sign for σ p. They should be told to generally use four decimal
places, at least until after the square root function is applied. Finally, students sometimes are confused by
the thought that a percent defects that is below the LCL is out of control. Here, instructors can stress that
control limits are pointing out the unusual, whether unusually good or unusually bad. If, say, a firm
determined that no defects occurred on the day that the boss walked around the production floor, perhaps
the boss should walk around the production floor more often.

If instructors cover process capability, this can be a place where the distinction between control limits and
spec limits can be emphasized. It is probably worth covering the process capability ratio Cp in case
students see it at their firms; however, the process capability index Cpk is the one that should be stressed.
It is also easy to calculate but overcomes a major flaw with the Cp value.

Acceptance sampling calculations are rather complicated. They are skipped in this textbook and in most
other introductory operations texts. Interested instructors can refer to Tutorial 2 on the website for more
background.

Class Discussion Ideas

1. SPC is a set of rules designed to signal when there is a change in a process – to signal when
assignable causes are likely to be present. Too often, students focus on the math and fail to create a
link to the actual process being measured. Select a set of processes the students will be familiar with
and create control charts that show an out-of-control condition. Have the students identify possible
assignable causes for each of the processes.

2. It can be a fun (if not, humbling) exercise to ask the students to help set up a c-chart to monitor the
instructor’s mistakes during class from day to day. The students could identify the types of errors that
would be counted as a mistake, as well as possible causes for those errors when the number of
mistakes becomes out of control. An alternative exercise might set up a c-chart that would monitor
the number of mistakes displayed in student oral presentations, which could include mistakes of
content as well as presentation skills. Student-identified causes of poor presentations might be
humorous.

Active Classroom Learning Exercises

1. Dice Game for Statistical Process Control. See Other Supplementary Material below.

2. Perhaps the most common mistake students make when working SPC problems is to confuse sample
size and number of samples. Set up some simple sampling exercises and have the students identify
both values. This seems to work best when the instructor is careful to avoid using either phrase in his
or her description of the processes.

3. Deming, W. Edwards. 1986. Deming's Experiment. Out of the Crisis. MIT Center for Advanced
Engineering Studies, Cambridge, MA, 346-354. This activity is a substantial variation on the glass
bead. It is used to illustrate the impact of variation which exists within a system and the extent to
which that variation limits the effectiveness with which individuals can be evaluated. After seeing
that the variation in the proportion of red beads is similar to that in the proportion of defectives, the
students should recognize that system variation should be a primary focus of attention rather than
individual efforts.

Copyright ©2017 Pearson Education, Ltd.

124 Statistical Process Control

Company Videos
1. Frito-Lay’s Quality-Controlled Potato Chips (10:15)
Frito-Lay is committed to quality in four key areas: (1) quality ingredients, (2) strict adherence to
recipes, (3) adherence to all process parameters, and (4) twice per shift inspections that mimic
consumer inspections (e.g., bag appearance, snack taste, etc.). Continuous improvement is the heart of
the firm’s quality assurance program. Frito-Lay focuses on two key metrics: (1) customer complaints
per million bags, and (2) hitting the center line on SPC charts for various attributes such as oil
content, moisture, seasoning, salt, thickness, and weight. The plant has nine critical checkpoints in the
production process, which are all shown in the video. A significant portion of the video includes a
thorough explanation of SPC charts, including an example of how to produce an x-bar chart, with a
known population standard deviation, for the percent salt content in potato chips. The video then
shows us how the operator at Frito-Lay produces an SPC chart observation, from scooping the
samples to measuring the salt content to updating the chart. At Frito-Lay, action is taken whenever
the SPC chart displays the following: (1) an observation outside the 3σ control limits, (2) two
consecutive observations very near either control limit, (3) five consecutive observations that trend in
the same direction, or (4) five consecutive observations that fall on the same side of the mean. “Star
Fleet” teams are available from other plants to help individual plants solve some of their more
difficult production problems. All of these practices contribute to Frito-Lay’s whopping 60% market
share.

Prior to showing the video, the instructor might ask the students to guess how many and what types of
inspections occur at a plant that makes potato chips. Afterwards, the nine types of inspection shown
in the video, in addition to the “twice per shift consumer-like inspection” and the in-store inspections,
could be compared to the student guesses. Further discussion could attempt to see if students
understand the difference between inspection and statistical process control. In particular, “control
limits” are not the same as “specification limits.” A sample that fails an SPC test would not
necessarily be rejected as being an unsuitable product. SPC is an ongoing exercise that is looking for
unusual circumstances or some sort of shift/wear in the production process so that the process can be
corrected quickly and before serious output problems emerge. In other words, failure of an SPC test
would not necessarily cause a full shutdown of the line or the rejection of an entire production lot.

2. Farm to Fork: Quality at Darden Restaurants (12:14)

The primary theme of this video is that Darden works to ensure quality at the source (i.e., “farm to
fork,” “dirt to door,” or “pond to plate”), rather than inspecting for quality once the product arrives at
local restaurants. Darden sources much of its food, including 50 million pounds of seafood annually,
from Asia and Latin America. The firm employs 50 microbiologists, food scientists, and public health
professionals. Inspection teams are located in China, Singapore, Thailand, Ecuador, Chile, and
Honduras. In fact, such detailed source inspection results in the need for only a few domestic
verification labs. Darden uses most of the total quality tools described in the text, including SPC
charts, Pareto charts, flow diagrams, fishbone diagrams, and scatter charts. Within SPC, Darden
primarily uses -charts, R-charts, and capability histograms. These charts monitor items like the
thickness and weight of steaks and the thickness and length of salmon. An example is provided
showing a process that appears to be in control, but the lot would be rejected because the means show
up on the capability histogram as being skewed on the low side. Vendors are aware that Darden
conducts regular audits that are stricter than FDA standards. Darden provides extensive education for
vendors around the world to show them how to prevent contamination and take samples of product.
The firm will visit farms to inspect and train, and also to check sanitation and personal hygiene
conditions. Darden takes temperatures of fish throughout the entire supply chain journey, including

Copyright ©2017 Pearson Education, Ltd.

125 Supplement 6

right before it is served to the customer. All of these approaches to quality assurance helped Darden
win the prestigious Black Pearl Award for quality in the food industry.

Prior to showing the video, instructors might ask the students to consider what type of inspections
they expect for the food served in their favorite restaurant, and at what point in the process they
would expect those inspections to take place. Discussion following the video could cover some of
these initial impressions. Students might be surprised to know that Darden conducts so many
inspections before the products leave the respective countries of origin. Such a program clearly
involves a great deal of resources. Students could be asked to identify the pros and cons of such an
approach of attempting to ensure quality at the source. Instructors could then ask if this approach
makes sense in every industry, or is there something special about food that makes quality at the
source particularly important? Clearly not all manufacturing firms today are conducting such
extensive testing at their overseas suppliers. Can students identify examples from the news that have
described recent quality problems from overseas suppliers?

Cinematic Ticklers
1. The Simpsons, Season 4: “Duffless,” 20th Century Fox Video, 2004 (1992-1993)
Homer visits a Duff beer plant where the inspector pulls out beer bottles containing syringes and rats,
but fails to catch Hitler’s head rushing by on the conveyor belt because he was talking to a customer.

Jay, Barry, and Chuck’s OM Blog

1. OM in the News: The Meningitis Scare and Acceptance Sampling

This post provides a good example of acceptance sampling. The Wall Street Journal (Oct. 25, 2012)
reports that the sample size for the test of the drug injected for the treatment of acute back and leg
pain was too small to be meaningful and did not comply with industry guidelines.
http://heizerrenderom.wordpress.com/2012/10/30/om-in-the-news-the-meningitis-scare-and-acceptance-
sampling/

2. Teaching Tip: Using an SPC Chart to Examine American Airlines’ Pilots “Sick Out”
The WSJ (Sep. 24, 2012) reports on the high number of flight delays and cancellations at American
Airlines. The company argues that pilots intentionally pretended to be sick to disrupt operations.
Using the same data, the union argues that sick rates have not deviated from historical norms.
Students can use the data provided to plot a p-chart to see for themselves.
http://heizerrenderom.wordpress.com/2012/09/26/teaching-tip-using-an-spc-chart-to-examine-american-
airlines-pilots-sick-out/

3. Teaching Tip: Building a P-Chart Using Airline Frequent Flier Award Data
This WSJ (May 26, 2011) article provides airline frequent flier award data that can be turned into a
teaching exercise with a p-chart.
http://heizerrenderom.wordpress.com/2011/05/27/teaching-tip-building-a-p-chart-using-airline-frequent-
flier-award-data/

Copyright ©2017 Pearson Education, Ltd.

126 Statistical Process Control

4. Teaching Tip: Building an SPC Chart with Airline Safety Data

Using data on eight major airlines reported in US News and World Report (Jan. 25, 2011), students
can plot a p-chart based on the number of safety incidents. Six of the carriers are “out of control,” but
four of those are below (better than) the LCL.
https://heizerrenderom.wordpress.com/2011/01/31/teaching-tip-building-an-spc-chart-with-airline-safety-
data/

Presentation Slides
INTRODUCTION (S6-1 through S6-5)

S6-1 S6-2 S6-3

S6-4

STATISTICAL PROCESS CONTROL (SPC) (S6-5 through S6-53)

Introductory Subsection (S6-5 through S6-15)
Slides 5-8: Here, instructors can discuss the idea that variation occurs all the time. We expect it and
plan for it. A basketball player, for example, may have a good shooting percentage one
night but a poor one the next night. We use control charts, however, to try to determine
when the variation is so far beyond the norm that we think it is likely caused by something
extraordinary. For example, the basketball player may have been sick on the day that she
shot 10% from the floor, or she might have been eating a granola bar before the game on
the day that she shot 90%. In the latter case, she might try to eat a granola bar prior to
subsequent games to see if that should become her new pre-game routine.
Slides 9-13: These all come from Figure S6.1, which describes the important steps in determining
process variation. Managers are trying to determine if the shape of the sample output over
time remains stable (Slide 12) or varies (Slide 13). If it varies, we say that the process is out
of control and investigate to take possible corrective action.
Slide 14: The book covers four different charts, but no matter which is being used, the concept is
always the same. That is, output is monitored, and if it falls above or below the control
limits, the firm investigates for assignable causes.
Slide 15: The difference between control limits and spec limits can be discussed here.

Copyright ©2017 Pearson Education, Ltd.

127 Supplement 6

S6-5 S6-6 S6-7

S6-8 S6-9 S6-10

S6-11 S6-12 S6-13

S6-14 S6-15

Control Charts for Variables (S6-16 through S6-17)

Slides 16-17: Variables are essentially characteristics that can be measured on a number line, such as
length, width, weight, and temperature. The - and R-charts must be used together because
it’s possible for the mean to remain constant while the dispersion significantly varies from
sample to sample, or it's possible for the dispersion to remain constant but for the mean to
be drifting up or down from sample to sample. Both need to be in control for the process to
be considered in control.

Copyright ©2017 Pearson Education, Ltd.

128 Statistical Process Control

S6-16 S6-17

The Central Limit Theorem (S6-18 through S6-21)

Slides 18-21: These provide the statistical foundation for control chart theory.

S6-18 S6-19 S6-20

S6-21

Setting Mean Chart Limits ( -Charts) (S6-22 through S6-29)

Slides 22-26: This is the first numerical example shown in the slides (Example S1), and it is applied for
-charts when the process standard deviation is known. At this point, instructors should
discuss the implementation of the z-value. The most common practice is to draw limits at
three standard deviations from the mean (z = 3), but any other z-value is possible. The
manager must make a trade-off between Type I and Type II errors. The higher the z-value,
the more likely it is that assignable causes of variation will go undetected, while the lower
the z-value, the more likely it is that natural causes will mistakenly be designated as
assignable. Slide 26 nicely illustrates why the process in this example is out of control.
Slides 27-29: This example (Example S2) shows the formulas and an application for -charts when the
process standard deviation is not known. Table S6.1 (Slide 28) is used in this case, and it is
based on a z-value of three. Instructors could point out that the limits for -charts are
equidistant from the mean.

Copyright ©2017 Pearson Education, Ltd.

129 Supplement 6

S6-22 S6-23 S6-24

S6-25 S6-26 S6-27

S6-28 S6-29

Setting Range Chart Limits (R-Charts) (S6-30 through S6-33)

Slides 30-31: Slide 30 introduces the concept of R-charts. The formulas appear in Slide 31. Once again,
Table S6.1 (Slide 28) is used. Instructors can point out that the range (highest value minus
the lowest value in a sample) is used as a proxy for the standard deviation. Also, for small
enough sample sizes, the LCL for the range chart will be zero, suggesting that it is
impossible to have an unusually small amount of dispersion for samples that small.
Slide 32: This slide shows real SPC charts used by Darden restaurants. These also appear in the
video case study.
Slides 33: This example (Example S3) shows an application of R-charts.

S6-30 S6-31 S6-32

Copyright ©2017 Pearson Education, Ltd.

130 Statistical Process Control

S6-33

Using Mean and Range Charts (S6-34 through S6-38)

Slides 34-35: These come from Figure S6.5. They nicely illustrate why firms must monitor both the -
and R-charts when monitoring variables. (In fact, the R-chart should come first, because if
the process variability is not in control, then the formulas are not valid because the
formulas incorporate variability information.)
Slides 36-37: These two slides summarize the steps for creating - and R-charts.
Slide 38: Table S6.2 provides common z values. The most commonly used is z = 3.00.

S6-34 S6-35 S6-36

S6-37 S6-38

Control Charts for Attributes (S6-39 through S6-46)

Slide 39: Here is where the difference between p-charts and c-charts should be emphasized. As stated
above in the “Background” section, the typical determining factor should be whether or not
each sample has a sample size: if so, use a p-chart, if not, use a c-chart.
Slides 40-44: This example shows the formulas and an application for p-charts (Example S4). The default
value for z should be 3, but that does not always have to be the case. Instructors should
point out on Slide 41 that the sample size is 100, and n = 100 is not always the case. Also,
is calculated in a streamlined way on Slide 42. Students should note that this approach is
quicker and safer than summing up all of the pi values with decimal points and then taking
an average. Slide 42 shows the LCL limit as 0 for this example. Here, students should see
that while a negative LCL calculation is possible, it is just rounded up to zero in practice.
Slide 44 shows that the process in this example appears to be out of control.

Copyright ©2017 Pearson Education, Ltd.

131 Supplement 6

Slides 45-46: This example shows the formula and an application (Example S5) for c-charts. The c-chart
limits are the easiest to calculate and don’t require any sample size information. Here again,
Slide 46 shows that a negative LCL should be rounded up to 0. For this example, the
process appears to be in control.

S6-39 S6-40 S6-41

S6-42 S6-43 S6-44

S6-45 S6-46

Managerial Issues and Control Charts (S6-47 through S6-56)

Slides 47-49: These summarize the control charting process, as well as determine which chart to use.
Slides 50-55: The basic control chart decision rule says to investigate for assignable causes once any
point lies beyond the control limits. And, actually, this decision can easily be made by
comparing the output data to the control limits—an actual chart does not have to be drawn.
There are more subtle signaling techniques, however, that process control professionals
recommend, and four of these are shown in Slides 52-55 (Figure S6.7). An actual graph of
the output becomes particularly useful when looking for patterns such as these.
Slide 56: This slide describes the run test, which can be used to identify the types of abnormalities
seen in Slides 52-55.

Copyright ©2017 Pearson Education, Ltd.

132 Statistical Process Control

S6-47 S6-48 S6-49

S6-50 S6-51 S6-52

S6-53 S6-54 S6-55

S6-56

PROCESS CAPABILITY (S6-57 through S6-67)

Slide 57: Capability charts deal with spec limits, not control limits, and they more commonly occur
at the design stage to see if a process is “capable” of producing within limits most of the
time (“most” usually means 99.73%). A process deemed “incapable” needs a design fix,
which may be more severe than correcting an assignable cause as occurs with control chart
violations.
Slides 58-61: These slides provide the formula and an example (Example S6) for the process capability
ratio Cp. The biggest danger in making decisions based on C p is that the ratio cannot detect
a process that is not centered. Thus, it may signal a process as “capable” that is consistently
producing output above or below the target.

Copyright ©2017 Pearson Education, Ltd.

133 Supplement 6

Slides 62-66: These slides provide the formula and an example (Example S7) for the process capability
index Cpk. This measure is generally preferable to C p because it does a better job of
signaling “capability” on both sides of the mean, thus better picking up the effects of a
skewed distribution.
Slide 67: Figure S6.8 helps to visualize what different values of the process capability index imply.
In this slide, the first three examples are not capable; whereas, the last two are.

S6-57 S6-58 S6-59

S6-60 S6-61 S6-62

S6-63 S6-64 S6-65

S6-66 S6-67

ACCEPTANCE SAMPLING (S6-68 through S6-79)

Slides 68-69: Capability studies focus on whether or not a process is capable of producing within specs
under normal circumstances; control charts monitor output over time to make sure that
something unusual hasn’t occurred; while inspection itself determines whether completed

Copyright ©2017 Pearson Education, Ltd.

134 Statistical Process Control

items themselves are acceptable or not. Firms practice acceptance sampling to avoid having
to inspect every single incoming or finished goods item. It is assumed that the defect rate of
the sample is reflective of the defective rate of the entire lot.
Slides 70-77: The text cannot devote enough space to show how to compute everything needed to draw
an operating characteristic curve, so these slides just describe the concepts. The average
outgoing quality can be computed (Slide 76) if (1) the sampling plan replaces all defective
items encountered, and (2) the true incoming percent defective for the lot is known.
Slide 78: Certain automated inspection plans have eliminated the need for acceptance sampling.
Slide 79: The final slide (Figure S6.10) can be used to summarize the three major methods
introduced in this supplement.

S6-68 S6-69 S6-70

S6-71 S6-72 S6-73

S6-74 S6-75 S6-76

S6-77 S6-78 S6-79

Copyright ©2017 Pearson Education, Ltd.

135 Supplement 6

Additional Assignment Ideas

1. Visit the website of a company that markets SPC software. Explore its software product. Print out
examples that show the variety of charts that can be created and explain their use. What types of
services are offered by these companies to help implement SPC? (Hint: Search for one of these
products or any other product or service of your choice: "Qualitran" or "SAS" or "DataLyzer".)

Internet Resources
American Society for Quality www.asq.org
American Statistical Association www.amstat.org
BPI Consulting: SPC for Excel www.spcforexcel.com
Statistical Engineering Division of NIST http://www.nist.gov/itl/sed/index.cfm
Total Quality Engineering www.tqe.com

Other Supplementary Material

Videos
Films available from:
Society of Manufacturing Engineers
One SME Drive
P.O. Box 930
Dearborn, Michigan 48121-0930
(P) 313-425-3000
(F) 313-425-3412
http://www.sme.org
o The Cost of Poor Quality--Case studies in Merix Corp., IBM, and Copeland Corp.: Show
successful strategies for reducing poor quality costs. Order # PI-VT570-3456.
o Factory Data Collection: See how Hayes Industrial Granke, Modem Plastics Molding, S&W
Screw Products, and Tandem Computers handle data collection. Order # PI-VT287-3456.
o Linking Measurements to the Machine: Case studies show how manufacturers implement this
technology and how they benefit. Order # PI-VT514-3456.

Learning Games
o Price, B. and Zhang, X. (2007). The Power of Doing: A Learning Exercise that Brings the Central
Limit Theorem to Life. Decision Sciences Journal of Innovative Education, 5(2), 405-411.
Teaching brief that demonstrates an active learning technique for teaching the Central Limit
Theorem. Groups of students conduct experiments tossing a die in a set of 5 rolls (or 10 rolls).
Students are asked to calculate sample average.
o Fish, L. (2007). Statistical Quality Control: Developing Student’s Understanding of Variable
Control Charts using String. Decision Sciences Journal of Innovative Education, 5(1), 191-196.
Teaching brief offers a mini-demonstration for variable control charting.
o Reyes, P.M. (2006). Using a Rubber Band to Teach the Management of Quality. Decision
Sciences Journal of Innovative Education, 4(1), 123-128. Teaching brief that helps students
understand measurement system analysis and its effects on process improvement using a simple
two-phase, hands-on gage of repeatability and reproducibility (GR&R) study.

Copyright ©2017 Pearson Education, Ltd.

136 Statistical Process Control

o Wright, C. and Smith, M. (2003). Serving Up the Red Beads Experience. Decision Sciences
Journal of Innovative Education, 1(1), 127-131. Teaching brief useful when teaching quality
management, statistical process control, and the management of people within a manufacturing
setting. Uses an airplane example and incorporates many of the principles from Deming’s 14
Points.
o LaPoint, Gary. Dice Game for Statistical Process Control. (See below.)

Copyright ©2017 Pearson Education, Ltd.

137 Supplement 6

Teaching Note

Dice Game for Statistical Process Control

Gary LaPoint, School of Management, Syracuse University

This exercise works best after a normal lecture on statistical process control. It does an excellent job of
getting the students to understand the mechanics of the process. The difference in student understanding
of SPC with this exercise and without the exercise is significant.

Length: Approximately 45 minutes

Material needed: (1) A pair of dice for each team. (Usually available at the local Dollar Store or hobby
shop.); (2) Work template (see following pages.)

Exercise: Organize the class into teams of two. Give each team a pair of dice and a work template. One
student rolls the pair of dice while the other records the number on the template. Take 10 samples with a
sample size of 4. (This requires rolling the dice 40 times.)

Then have students calculate the X-bar, the Range, X-double bar, and the R-bar.

When students have completed the above, have them then calculate the UCL and LCL for X-bar and R
charts. (I have them calculate the Upper and Lower control limits using Table S6.1 in the text.) Students
cannot manually calculate the upper and lower control limits using the formula because they do not have
the standard deviation of the dice rolling process. Although if you do the calculation, the standard
deviation is 2.297.

Have the students develop the X-bar and R charts (you can provide the templates that follow or have the
students prepare them themselves). Once the students have created their graphs, ask them if their process
is in control.

Then give them upper and lower tolerance limits as defined by the customer that are lower than the
process control limits they calculated, say 7.5 and 4. Following that discussion, ask them to determine the
Cpk Index, and decide if the process is capable of operating within those tolerances. Several teams will not
be capable. Then collect the students’ charts and go over them on an overhead so the entire class can see
different types of processes.

Finally, ask what is inherently wrong with the exercise. The sharp students will say we can only get a
value between 2 and 12, which is correct. Therefore, this exercise is not perfect in simulating random
variation.

On the following pages are instructions to the students and the required tables and graphs for creating the
data set and control charts.

Copyright ©2017 Pearson Education, Ltd.

138 Statistical Process Control

Dice Game for Statistical Process Control

Student Instructions

1. Take 10 samples of sample size 4. (Roll the dice 40 times.)

2. Calculate the X-bar and the Range for each sample.
3. Calculate the X-double bar and the R-bar.
4. Calculate the UCL and the LCL for the X-bar and the R chart.
5. Plot the values for each sample.
6. Is the process in control?
7. Given the upper and lower tolerances desired by the customer, is the process in control?
8. Using the Cpk index, is the process capable of operating within those tolerances?

Record of Observations

Sample
1 2 3 4 X-bar R
Number
1

10

Average

UCLR = UCLX-bar =

LCLR = LCLX-bar =

Cpk =

Copyright ©2017 Pearson Education, Ltd.

139 Supplement 6

R-Chart

1 2 3 4 5 6 7 8 9 10
Samples

X-bar Chart

1 2 3 4 5 6 7 8 9 10
Samples

Copyright ©2017 Pearson Education, Ltd.