Академический Документы
Профессиональный Документы
Культура Документы
281-304-9504
www.spcforexcel.com
Statistics
Cp=1.34
Cpk= 0.59
Cpu= 0.59 (3.84%)
Cpl= 2.09 (0%)
Est. Sigma= 2.49
Pp=1.31
Ppk= 0.57
Ppu= 0.57 (4.36%)
Ppl= 2.04 (0%)
Sigma= 2.55
Average=75.63
Count=96
No. Out of Spec=5 (5.21%)
Kurtosis=0.62
Skewness=0.71
Capability Analysis
35
LSL=60
30
Nominal=70
USL=80
Frequency
25
20
15
10
0
57
62
67
72
77
82
87
Measurement
Thank you for selecting our software package. This program was written by Dr. William H. McNeese
and is distributed by Business Process Improvement (Cypress, Texas). This program cannot be copied or
used unless under license with Business Process Improvement. Business Process Improvement is not
liable for any decisions made based on the use of this software package.
Requirements: This program is a Microsoft Excel add-in. You must Microsoft Excel for this program
to work. This program supports any version of Excel from 2000 on.
Business Process Improvement
20314 Lakeland Falls
Cypress, TX 77433
281-304-9504
www.spcforexcel.com
Installation
The necessary files to run the program are installed when you run the installation program. The
installation file is the .exe program you downloaded or received on the CD. Running the setup.exe file
creates the following directory (there are slight differences for Excel 2007):
C:\Documents and Settings\{username}\Application Data\SPC_for_MS_Excel
The following program files are installed into this directory:
spcforexcelv3.xla
spcfor2007excelv3.xlam
installspcforexcelv3.exe
The installation process may leave uninstallation files such as unins001.exe and unins001.dat in this
directory. During operation, the program may save user preferences and other settings in one or more text
or binary files in this directory. The user is discouraged from altering any of these files and from storing
any work files in this directory.
The program also creates the following directory:
C:\Documents and Settings\{username}\My Documents\SPC for MS Excel
The following sample data files and instruction manual are installed into this directory:
Gage R&R Example Workbook.xls
SPC Example Data V3.xls
SPC for MS Excel V3.0 Instructions.pdf
The user is encouraged to use these files to learn how SPC for MS Excel works. If you also purchased the
PowerPoint training modules, they will be installed this directory as well.
The program is installed as an add-in. It will open whenever Excel is opened. In Excel 2000 to Excel
2003, SPC for Excel will appear on the Worksheet and Chart Menus next to Window. There is also a free
standing toolbar that can be placed anywhere in the window. Both are shown below.
In Excel 2007, SPC for Excel appears on the ribbon next to Home as shown below. The SPC Menu to the
right lists all the buttons.
The menu and toolbar allows you to access the various components of the program:
The data entry requirements to run each component of this program are given below. All the examples
are in the workbook SPC Example Data V3.0.xls and Gage R&R V3 Example Workbook.xls. This
instruction manual is intended to demonstrate how the program is used.
To learn more about SPC, please refer to one of the many books on the subject. The best reference is
probably Understanding Statistical Process Control by D. Wheeler and D. Chambers, SPC Inc., 1986 or
any of the later books by Dr. Wheeler.
You can also visit our website where many of these SPC tools are described in our past free e-zines. Go
to www.spcforexcel.com.
Pareto Diagrams
A Pareto diagram is a special type of bar chart that is used to separate the "vital few" from the "trivial
many." It is based on the 80/20 rule; e.g., 20% of our customers buy 80% of our products. The
horizontal (x) axis most often represents problems or causes of problems (the categories). The vertical
(y) axis most often represents frequency or cost. The problem or cause that occurs most frequently (or
costs the most) is listed first on the x axis. The second most frequently occurring problem or cause is
listed second and so on. A bar is generated for each cause or problem. The height of the bar is the
frequency with which that problem or cause occurred. A cumulative percentage line is sometimes added
to the Pareto diagram.
Pareto Diagram
50
100%
100%
96%
45
90%
40
80%
80%
35
70%
30
60%
25
25
50%
50%
20
Percent
Number of Returns
An example of a Pareto
diagram is shown to the
right.
In this Pareto
diagram, the number of
return goods by product is
analyzed. The x axis is
the different types of
products. The y axis is
how often each product
has been returned. The
bars are arranged so the
first bar (for Product B)
has the largest frequency.
The other bars are then
arranged in decreasing
frequency.
40%
15
15
30%
10
20%
10%
2
0%
B
Products
Products
A
B
C
D
Number
of Returns
15
25
8
2
select the Pareto Diagram option from the toolbar. You could also highlight both the product and number
of returns information. The data does not have to in adjacent columns. In this case, you would select the
category range, then hold down the Control key and select the frequency range. Then select the Pareto
Diagram option on the SPC toolbar. You will get the Pareto Diagram dialog which is described below
(after the data entry requirements for option 3). Once you fill the information in the dialog box and select
OK, you will get the Pareto diagram shown above (with the cumulative line option selected).
Date
2/1/03
2/4/03
2/6/03
2/9/03
2/11/03
2/14/03
2/14/03
1/2/00
2/23/03
2/24/03
3/1/03
3/2/03
Week
Product
1
1
1
1
2
2
2
2
3
3
3
3
4
4
4
4
A
B
C
D
A
B
C
D
A
B
C
D
A
B
C
D
Product
Returned
A
A
B
C
A
A
D
B
A
D
A
A
Number
of Returns
4
2
12
3
0
3
1
4
4
2
5
1
8
1
10
1
Reason
Customer Did Not Need
Broken
Wrong Quantity
Wrong Quantity
Salesman Ordered Wrong
Wrong Quantity
Wrong Quantity
Broken
Customer Did Not Need
Salesman Ordered Wrong
Wrong Quantity
Wrong Quantity
Options Page
but there are other options including count, average, and standard deviation.
Pareto on One Column? This is Option 3: Pareto Diagram Based on One Column. Select Yes
if the data are in one column. The Data Range contains the worksheet range containing the
data. The default value is the range that is selected prior to PD being selected on the toolbar.
Include Frequencies >= to is used to determine what frequencies you want to include in the
chart. For example, if you enter 3, only those items that occur three or more times will be
included in the chart.
Examples
Below are the results for Options 2 and 3 using the data given above.
Option 2: Summing the results for each week
Pareto Diagram
100%
60
100%
90%
87%
50
80%
72%
70%
60%
30
50%
28
46%
Percent
Number of Returns
40
40%
20
30%
16
20%
10
8
10%
0%
C Total
A Total
D Total
B Total
Product
100.0%
100.0%
90.0%
10
83.3%
80.0%
70.0%
8
66.7%
50.0%
50.0%
Percent
60.0%
6
40.0%
4
30.0%
20.0%
2
10.0%
0.0%
Wrong Quantity
Broken
Reason
Histograms
A histogram is a bar chart that provides a snapshot in time of the variation in a process. It tells us how
often a value or range of values occurred in a given time frame. A histogram will tell us the most
frequently occurring value (the mode), the overall range, and the shape of the distribution (e.g., bellshaped, skewed, bimodal, etc.). It is best to have 50 to 100 data points to construct a histogram, if
possible. This program will construct a histogram from the raw data. It will automatically determine the
number of classes (bars) as well as the class width. You have the opportunity to change the number of
classes. An example of a histogram is shown below.
Descriptive Stats
Mean=75.625
Standard Error=0.26
Median=75
Standard Deviation=2.547
Variance=6.489
Sum=7260
Count=96
Maximum=84
Mininum=70
Range=14
Kurtosis=0.6157
Skewness=0.7079
Histogram of Yields
20
19 (19.8%)
18
17 (17.7%)
16
14
13 (13.5%)
Frequency
12
11 (11.5%)
10 (10.4%)
10
8 (8.3%)
8
5 (5.2%)
4 (4.2%)
3 (3.1%)
2 (2.1%)
1 (1.0%) 1 (1.0%)
1 (1.0%)
1 (1.0%)
0 (0.0%)
0
69.5 to
70.5
70.5 to
71.5
71.5 to
72.5
72.5 to
73.5
73.5 to
74.5
74.5 to
75.5
75.5 to
76.5
76.5 to
77.5
77.5 to
78.5
78.5 to
79.5
79.5 to
80.5
80.5 to
81.5
81.5 to
82.5
82.5 to
83.5
83.5 to
84.5
Measurement
Data Entry
Enter the data you want to use in the histogram into a worksheet. The data
can be in any number of rows and columns. Select the cells containing the
data for the histogram as shown to the right. Then select the histogram
option (H) from the SPC toolbar.
81
77
74
75
73
77
76
76
74
79
77
76
76
75
84
73
77
74
79
80
74
75
75
76
76
82
73
76
76
81
74
77
75
76
73
80
74
75
74
78
77
73
72
76
76
77
79
76
74
75
75
79
73
72
74
77
76
75
75
75
74
74
70
74
77
79
75
72
75
75
78
78
72
74
73
74
79
77
77
81
73
74
74
78
74
74
75
75
73
79
73
72
79
71
74
75
10
When you select the histogram option (H) on the SPC toolbar, you will get the dialog box shown to the
left. Each entry is discussed below.
Enter location of Values: This is the range containing the values for the histogram. The default
range is the range selected on the worksheet before selecting the histogram option on the toolbar.
Enter Histogram Title: This is the title that goes on the histogram chart. The default value is
Histogram.
Enter Y-Axis (Vertical Label): This is the vertical axis label. The default is Frequency.
Enter X-Axis (Horizontal Label): This is the horizontal axis label. The default is Measurement.
Name of Chart: This is very important. This will be the name of the worksheet tab that contains
the chart in your workbook.
Enter Number of Integers to Right of Decimal: This is the rounding that is used in the data. For
example, if the data contains whole numbers, this value is 0 (the default value). If the data has
one decimal point to the right of the data (as shown in the data above), this value is 1. It is used
to set the class boundaries.
Dates of Data Collection: Add the starting date and ending dates of data collection. These dates
are optional. If entered, they will appear in a dialog box in the lower left-hand corner of the chart.
Include Descriptive Statistics? If you want the descriptive statistics on the chart, select Yes.
The descriptive statistics include the average, standard deviation, count, etc. The default value is
Yes. There is also the option to Select Which to Include. This option allows you to
determine which of the descriptive statistics you want to include. If you select this option, you
will see the dialog box below. Select which statistics you want to include. The statistics you
select will remain the same if you update the histogram. You can Check All or Uncheck All
if desired.
11
You also have the option to view the frequency distribution for the View/Hide Frequency Distribution
histogram. This is done by selecting the button with the caption
View/Hide Frequency Distribution. This button appears the first time the histogram is made. An
example of a histogram with the frequency distribution added is shown below. Selecting the button again
hides the frequency distribution.
Histogram of Yields
20
19 (19.8%)
18
17 (17.7%)
16
14
13 (13.5%)
Frequency
12
11 (11.5%)
10 (10.4%)
10
Descriptive Stats
Mean=75.625
Standard Error=0.26
Median=75
Standard Deviation=2.547
Variance=6.489
Sum=7260
Count=96
Maximum=84
Mininum=70
Range=14
Kurtosis=0.6157
Skewness=0.7079
8 (8.3%)
8
5 (5.2%)
4 (4.2%)
3 (3.1%)
2 (2.1%)
1 (1.0%) 1 (1.0%)
1 (1.0%)
1 (1.0%)
0 (0.0%)
0
69.5 to
70.5
70.5 to
71.5
71.5 to
72.5
72.5 to
73.5
73.5 to
74.5
74.5 to
75.5
75.5 to
76.5
76.5 to
77.5
77.5 to
78.5
78.5 to
79.5
79.5 to
80.5
80.5 to
81.5
81.5 to
82.5
82.5 to
83.5
83.5 to
84.5
Measurement
12
p Charts
A p control chart is used to examine the variation in the proportion (or percentage) of defective items in a
group of items. An item is defective if it fails to conform to some preset specification (operational
definition). The p control chart is used with "yes/no" attributes data. This means that there are only two
possible outcomes: either the item is defective or it is not defective. For example: either the phone is
answered or it is not answered.
An example of a p chart generated by this program is given below. In this example, the percentage of
telemarketing calls that result in an order each day is being examined. "n" is the subgroup size (the
number of telemarketing calls made each day). "np" is the number of "defective" items -- in this case, the
number of calls that result in an order. "p" is the proportion defective and is determined by p = np/n. For
example, on the first day there were 40 telemarketing calls made (n = 40). Of these, 5 resulted in an order
(np = 5). Thus, p = np/n = 5/40 = 0.125 or 12.5%. In the chart below, 12.5% is the point plotted on
2/1/2003.
np
n
p Control Chart
45%
40%
UCL=36.16
35%
% Defective
30%
25%
20%
Avg=19.47
15%
10%
5%
20
03
20
03
15
/
2/
20
03
14
/
2/
13
/
2/
20
03
20
03
12
/
2/
11
/
20
03
2/
00
3
10
/
2/
2/
9/
2
00
3
2/
8/
2
00
3
00
3
2/
7/
2
00
3
2/
6/
2
2/
5/
2
00
3
00
3
2/
4/
2
2/
3/
2
00
3
LCL=2.78
2/
2/
2
00
3
0%
2/
1/
2
Subgroup Number
UCL p 3
p(1 p)
n
13
LCL p 3
p(1 p)
n
The above control limits are not valid for the small np case. This occurs when n p < 5 or n(1- p )< 5.
In this case, the program automatically calculates the control limits using the binomial distribution.
Data Entry
The p chart monitors the fraction or percentage of defective
items in a group of items. Subgroup number (like the date
shown to the right), subgroup size (n) and number
nonconforming (np) are required as shown in the example
below. After entering the data, highlight the subgroup
numbers (in the example these are the dates). Then select the
attribute control chart option (Att) from the SPC toolbar and
select the p control chart option.
Date
2/1/2003
2/2/2003
2/3/2003
2/4/2003
2/5/2003
2/6/2003
2/7/2003
2/8/2003
2/9/2003
2/10/2003
2/11/2003
2/12/2003
2/13/2003
2/14/2003
2/15/2003
Number
of
Telemarketing
Calls (n)
40
63
47
52
34
59
36
71
53
50
41
48
67
45
54
Number
that
Result in an
Order (np)
5
10
12
7
3
21
12
7
11
3
12
10
5
12
18
14
Data in: Select columns or rows depending on how the data is entered into the spreadsheet. The
program selects one or the other depending on the range selected prior to selecting Att on the SPC
toolbar.
15
np Control Charts
A np control chart is used to monitor the variation in the number of defective items in a group of items.
With this chart, the subgroup size (n), the number of items in the group, must be the same each time. An
item is defective if it fails to conform to some preset specification (operational definition). The np control
chart is used with "yes/no" attributes data. This means that there are only two possible outcomes: either
the item is defective or it is not defective. For example: either the phone is answered or it is not
answered.
An example of a np chart generated by this program is shown below. In this example, the number of
defective invoices each day is being tracked. The control chart is developed by taking a random sample
of 100 invoices each day and determining the number that are defective. In this case, the subgroup size is
constant (100). np is the number of defective items. For example, on day one, there were 22 defective
invoices.
np
np
k
np
n
np Control Chart
40
UCL=37.83
35
30
Number Defective
25
Avg=24.87
20
15
LCL=11.9
10
0
1
10
11
12
13
14
15
Subgroup Number
UCL np 3 np(1 p)
LCL np 3 np(1 p)
The above control limits are not valid for the small np case. This occurs when n p < 5 or n(1- p )< 5.
In this case, the program automatically calculates the control limits using the binomial distribution.
16
Data Entry
Day
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Number
Defective
Invoices (np)
22
33
24
20
18
24
24
29
18
27
31
26
31
24
22
of
17
c Control Charts
A c control chart is used to monitor the variation in the number of defects. A defect occurs when
something does not meet a preset specification (operational definition). A c control chart is used with
counting type attributes data (e.g., 0, 1, 2, 3). These are whole numbers. You cannot have 1/2 defect. In
addition, to use a c control chart, two other conditions must be true:
With a c control chart, we are often looking at an area, not a group of items. For example, we may use a c
control chart to monitor injuries in a chemical plant. In this case, the subgroup is the plant. The
opportunities for injuries to occur is large; the actual number that occur is small relative to the
opportunity. With a c control chart, the area of opportunity for defects to occur must be constant.
c Control Chart
50
45
40
UCL=37.56
35
Number of Defects
An example of a c control
chart generated using this
program is given to the
right. In this example, the
number of returned goods
to a distributor is being
tracked.
c is the
number of returned goods
each day. For the first
date, there were 20 goods
returned.
30
25
Avg=23.13
20
15
c
k
20
03
20
03
2/
15
/
2/
14
/
20
03
20
03
2/
13
/
20
03
2/
12
/
2/
11
/
20
03
00
3
2/
10
/
9/
2
2/
00
3
00
3
8/
2
2/
00
3
7/
2
2/
6/
2
2/
00
3
5/
2
00
3
UCL c 3 c
2/
00
3
4/
2
2/
3/
2
2/
00
3
2/
2
2/
2/
1/
2
00
3
LCL c 3 c
These control limits are valid only if c > 3. The program will automatically use the Poisson Distribution
to determine the control limits if the average is less than 3.
18
Data Entry
The c control chart is used to monitor the variation in the number of defects in
a constant subgroup size. Require data include the subgroup number and the
number of defects. An example of a c chart data is given to the right. To
make a c chart, select the subgroup numbers in the worksheet (shaded area).
Then select the attribute control chart option (Att) from the SPC toolbar and
select the c control chart option.
Day
2/1/2003
2/2/2003
2/3/2003
2/4/2003
2/5/2003
2/6/2003
2/7/2003
2/8/2003
2/9/2003
2/10/2003
2/11/2003
2/12/2003
2/13/2003
2/14/2003
2/15/2003
Number of
Returned Goods
(c)
20
24
14
32
28
16
19
32
27
25
24
12
17
44
13
19
u Control Charts
A u control chart is used to monitor the variation in the number of defects. A defect occurs when
something does not meet a preset specification (operational definition). A u control chart is used with
counting type attributes data (e.g., 0, 1, 2, 3). These are whole numbers. You can not have 1/2 defect. In
addition, to use a u control chart, two other conditions must be true:
With a u control chart, we are often looking at an area of opportunity for defects to occur. A u control is
similar to a c control chart, except that the area of opportunity for defects to occur is not constant.
An example of a u chart generated by this program is given below. In this example, radiators are being
checked for leaks. Each day, the number of radiators assembled is counted. This is the area of
opportunity for leaks to occur (n). The number of leaks found when two portions of the radiator were
assembled for the first time is also counted (c). For example, on the first day, 39 radiators were hooked
up. There were 14 leaks detected. In this case, u = c/n = 14/39 = .359
c
n
u Control Chart
0.4
0.35
UCL=0.32
0.3
0.25
0.2
0.15
Avg=0.15
0.1
0.05
0
2-Jun
3-Jun
4-Jun
5-Jun
6-Jun
7-Jun
8-Jun
9-Jun
10-Jun
11-Jun
12-Jun
13-Jun
14-Jun
15-Jun
16-Jun
Subgroup Number
UCL u 3
u
n
LCL u 3
u
n
The control limits will be based on the actual subgroup size for each point. If the subgroup size varies,
the control limits will also vary (as shown in the above example). These control limits are valid only if
c > 3. The program will automatically use the Poisson Distribution to determine the control limits if the
average is less than 3.
20
Data Entry
Date
2-Jun
3-Jun
4-Jun
5-Jun
6-Jun
7-Jun
8-Jun
9-Jun
10-Jun
11-Jun
12-Jun
13-Jun
14-Jun
15-Jun
16-Jun
Number of
Radiatiors
Assembled (n)
39
45
46
48
40
58
50
50
50
50
32
50
33
50
50
Number of Leaks
(c)
14
4
5
13
6
2
4
11
8
10
3
11
1
3
6
21
X -R control charts
X -s control charts
Moving average and moving range control charts
X-MR (individuals) control charts
Table X-MR (for making multiple individuals charts at once)
Run charts
An X -R control chart is used to examine the variation in variables data. Variables data are
measurements (e.g., height, weight, time, dollars, density). This control chart is used when you have
lots of data and a method of rationally subgrouping the data. An example of an X -R chart is given on the
next page. In this example, sales per day are monitored. A subgroup is made up of the results for one
week. The subgroup size (n) in this case is 5.
The X -R chart is really two charts. The top chart is the X chart. This chart looks at the variation in
subgroup averages. The subgroup average is the average of the individual results in the subgroup. The
bottom chart is the range chart. The subgroup range is the largest result minus the smallest result in the
subgroup.
The values of X and the moving range are plotted over time. The average and control limits for both
charts are calculated using the equations below. The average is plotted as a green solid line and the
control limits are plotted as red dashed lines on both charts. For the equations below, k is the number of
subgroups. A2, D3, and D4, d2 are constants used in the calculations for charts. See the e-zines on our
website for more information about these constants.
X Chart Equations:
X
X
k
UCL X A 2 R
LCL X A 2 R
UCL D4R
LCL D3R
R
k
R
d2
The values for the average and control limits are printed on the respective charts.
22
Xbar Chart
35
34
UCL=33.8
33
Subgroup Average
32
31
Avg=30.2
30
29
28
27
LCL=26.6
26
2/1/2003
2/8/2003
2/15/2003 2/22/2003
3/1/2003
3/8/2003
4/5/2003
5/3/2003
5/10/2003
4/5/2003
5/3/2003
5/10/2003
Subgroup Number
R Chart
14.0
UCL=13.3
12.0
Subgroup Range
10.0
8.0
Avg=6.3
6.0
4.0
2.0
0.0
2/1/2003
2/8/2003
2/15/2003 2/22/2003
3/1/2003
3/8/2003
23
Data Entry
Week of
2/1/2003
2/8/2003
2/15/2003
2/22/2003
3/1/2003
3/8/2003
3/15/2003
3/22/2003
3/29/2003
4/5/2003
4/12/2003
4/19/2003
4/26/2003
5/3/2003
5/10/2003
Monday
33.0
30.0
31.8
28.5
27.2
30.5
35.4
33.6
35.8
30.4
26.9
28.0
29.1
26.4
27.4
Tuesday
33.1
30.1
29.4
34.0
27.6
30.5
31.3
33.3
34.1
32.6
32.4
28.2
31.6
30.8
26.0
Wednesday
26.4
29.2
28.0
33.6
30.6
28.1
27.8
26.4
30.1
32.5
29.0
25.5
29.0
34.0
28.2
Thursday
28.3
31.5
26.9
29.7
30.1
37.7
31.3
32.4
30.3
25.2
26.8
31.1
33.1
27.0
27.9
Friday
28.9
28.4
32.2
33.4
27.4
28.7
33.0
34.1
26.1
32.1
29.3
34.4
30.9
31.7
27.3
24
o
o
Xbar Chart Only: only the X chart is constructed. This is the default option
Xbar and R Charts Different Worksheets: Both the X and range charts are constructed
but on different worksheets.
o Xbar and R Charts Same Worksheet: Both the X and range charts are constructed on
the same worksheet.
Data in: Select columns or rows depending on how the data is entered into the spreadsheet. The
program selects one or the other depending on the range selected prior to selecting the variable
control chart option (Var) on the SPC toolbar.
Options Page
25
Enter your own limits: You may also enter your own values for the X chart control limits. An
additional two lines can also be added to the chart. The values entered here must be between the
average and the UCL entered. The program will add them automatically to the area between the
average and the LCL.
26
X -s Control Charts
A X -s chart is very similar to the X -R chart. Instead of using the subgroup range, the X -s chart uses
the subgroup standard deviation to determine process variability. It is usually used when your subgroup
is greater than or equal to 10.
For the equations below, k is the number of subgroups. A3, B3, and B4, C4 are constants used in the
calculations for the charts.
X Chart Equations:
X
X
k
UCL X A 3s
LCL X A 3s
UCL B4 s
LCL B3 s
s Chart Equation:
s
k
s
c4
The data entry requirements and the dialog boxes for these two charts are the same as the X -R control
chart. Please refer to the instructions above for the X -R control charts (page 22). The X -s charts for the
same data as used above for the X -R charts are shown below.
Xbar Chart
35
Subgroup Average
34
UCL=33.9
33
32
31
Avg=30.2
30
29
28
27
LCL=26.5
26
2/1/2003
2/8/2003
2/15/2003
2/22/2003
3/1/2003
3/8/2003
3/15/2003
3/22/2003
3/29/2003
4/5/2003
4/12/2003
4/19/2003
4/26/2003
5/3/2003
5/10/2003
4/5/2003
4/12/2003
4/19/2003
4/26/2003
5/3/2003
5/10/2003
Subgroup Number
s Chart
6.0
UCL=5.4
5.0
4.0
3.0
Avg=2.6
2.0
1.0
0.0
2/1/2003
2/8/2003
2/15/2003
2/22/2003
3/1/2003
3/8/2003
3/15/2003
3/22/2003
3/29/2003
Subgroup Number
27
UCL=124.3
Avg=103.6
20
03
20
03
18
/
6/
6/
11
/
00
3
4/
2
6/
20
03
28
/
5/
20
03
20
03
21
/
5/
5/
14
/
00
3
7/
2
5/
20
03
30
/
4/
20
03
23
/
20
03
16
/
4/
4/
4/
00
3
9/
2
00
3
2/
2
4/
20
03
20
03
26
/
3/
19
/
3/
20
03
3/
12
/
00
3
5/
2
3/
20
03
20
03
26
/
2/
2/
19
/
12
/
5/
2
2/
2/
20
03
LCL=82.9
00
3
Result
129
124
119
114
109
104
99
94
89
84
79
Sample Number
Moving Range
25
UCL=25.5
20
15
10
Avg=7.8
5
20
03
18
/
6/
20
03
6/
11
/
00
3
4/
2
6/
20
03
28
/
5/
20
03
21
/
5/
20
03
5/
14
/
00
3
7/
2
5/
20
03
30
/
4/
20
03
23
/
4/
20
03
4/
16
/
00
3
9/
2
4/
00
3
2/
2
4/
20
03
26
/
3/
20
03
19
/
3/
20
03
3/
12
/
00
3
5/
2
3/
20
03
26
/
2/
20
03
19
/
2/
12
/
2/
2/
5/
2
00
3
20
03
Sample Number
The bottom chart is the moving range chart. The moving range between consecutive points is plotted on
this chart. For example, the range between accounts receivable past due for 90 days between the week of
2/6 and 2/13 is $110,000 - $104,000 = $6,000. There is no range corresponding to the first data point on
the X chart.
The values of X and the moving range are plotted over time. The average and control limits for both
charts are calculated using the equations below. The average is plotted as a green solid line and the
control limits are plotted as red dashed lines on both charts. For the equations below, k is the number of
samples (individual X values).
28
X Chart Equations:
X
k
UCL X 2.66R
LCL X 2.66R
R
k 1
UCL 3.27R
LCL None
R
1.128
The values for the average and control limits are printed on the respective charts
Data Entry
Week of
The only data required for an individuals control chart are the sample number
and the result as shown to the right. Select the sample numbers (shaded area).
Then select the variable control chart option (Var) on the SPC toolbar and
select the X-MR (Individuals) Chart option. Select OK and you will get the
two-page dialog box for the individuals control chart. This dialog box is the
same as for the X -R control chart. Please refer to the instructions above for
the X -R control charts.
29
2/5/2003
2/12/2003
2/19/2003
2/26/2003
3/5/2003
3/12/2003
3/19/2003
3/26/2003
4/2/2003
4/9/2003
4/16/2003
4/23/2003
4/30/2003
5/7/2003
5/14/2003
5/21/2003
5/28/2003
6/4/2003
6/11/2003
6/18/2003
Accounts
Receivable
110
104
98
112
113
100
89
113
109
105
108
95
101
98
100
105
103
99
112
98
Subgroup
Number
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
2
110
104
98
112
113
100
89
113
109
105
108
95
101
98
100
105
103
99
3
104
98
112
113
100
89
113
109
105
108
95
101
98
100
105
103
99
112
Week of
98
112
113
100
89
113
109
105
108
95
101
98
100
105
103
99
112
98
2/5/2003
2/12/2003
2/19/2003
2/26/2003
3/5/2003
3/12/2003
3/19/2003
3/26/2003
4/2/2003
4/9/2003
4/16/2003
4/23/2003
4/30/2003
5/7/2003
5/14/2003
5/21/2003
5/28/2003
6/4/2003
6/11/2003
6/18/2003
Accounts
Receivable
110
104
98
112
113
100
89
113
109
105
108
95
101
98
100
105
103
99
112
98
The charts below are the output from the MA/MR chart for this data.
Moving Averge Chart
119
UCL=115.8
Moving Average
114
109
104
Avg=103.4
99
94
LCL=91
89
1
10
11
12
13
14
15
16
17
18
11
12
13
14
15
16
17
18
Subgroup Number
Moving Range
30
25
20
15
Avg=12.1
10
5
0
1
10
Subgroup Number
30
Sample
1
2
3
4
5
6
7
8
9
10
Name
Y Label
X Label
Result 1 Result 2
98.5
93.61
101.22
106.38
105.99
108.67
89.08
98.83
105.48
94.57
96.55
91.55
90.77
95.11
96.13
89.41
97.16
97.98
100.67
98.17
Chart 1
Y
X
Sample
1
2
3
4
5
6
7
8
9
10
Chart 2
Y
X
Result 1 Result 2
98.5
93.61
101.22
106.38
105.99
108.67
89.08
98.83
105.48
94.57
96.55
91.55
90.77
95.11
96.13
89.41
97.16
97.98
100.67
98.17
31
Run Charts
The data entry requirements and the dialog box for the run chart are essentially the same as the X-MR
chart. Please review to the information above on the X-MR chart (page 28).
Data
32
97.00
87.22
102.44
112.76
111.98
117.33
78.16
97.66
110.95
89.13
93.10
83.10
81.53
90.22
92.26
78.82
94.32
95.96
101.35
96.35
96.73
96.30
113.43
99.15
98.14
94.87
119.72
108.66
123.76
93.45
Process Capability
Process capability answers the question: Is the process capable of meeting specifications? Specifications
can be set by customers. Specifications could also be standards set by management for a process. For
example, the standard for days sales outstanding might be set by leadership to be less than 46 days. One
measure of process capability is the Cpk index. Another is Ppk. To determine the process capability, the
individual sample results should be normally distributed (the histogram is a bell shaped curve) and the
process should be in statistical control.
The value of Cpk is the minimum of two process capability indices. One process capability is Cpu, which
is the process capability based on the upper specification limit. The other is Cpl, which is the process
capability based on the lower specification limit. Algebraically, Cpk is defined as:
Cpk = Minimum (Cpu, Cpl)
Cpu
USL X
3 '
Cpl
X LSL
3 '
where USL = upper specification limit and LSL = lower specification limit. Both Cpu and Cpl take into
account where the process is centered. The value of Cpk is the difference between the process average
' ). This standard
( X ) and the nearest specification limit divided by three times the standard deviation (
deviation is the standard deviation estimated from a range or s chart. In determining Ppk, the standard
deviation is the actual standard deviation of the measurements.
Cpk values above 1.0 are desired. This means that essentially no product or service is being produced
above USL or below LSL. The figure below shows how the Cpk values are developed. If Cpk is less
than 1.0, this means that there is some product being produced out of specification.
LSL
USL
=
X - LSL
= ^
X -3
'
= ^
X -2
'
=
USL - X
= ^
X -1
'
=
X
^'
3
= ^
X +1
'
= ^
X +2
'
= ^
X +3
'
^'
3
The process capability feature of this program includes Cpk and Ppk. The data used in the analysis can
either be data entered into a spreadsheet for this analysis alone or data that has been used for a control
chart previously.
33
An example of the process capability analysis performed by the program is shown below. The histogram
of the data is given along with a normal curve. The specification limits are added. The statistics are
given to the right. The percentages in parentheses give the % out of specification for that metric.
Statistics
Cp=1.34
Cpk= 0.59
Cpu= 0.59 (3.84%)
Cpl= 2.09 (0%)
Est. Sigma= 2.5
Pp=1.31
Ppk= 0.57
Ppu= 0.57 (4.36%)
Ppl= 2.04 (0%)
Sigma= 2.6
Average=75.6
Min=70
Max=84
Count=96
No. Out of Spec=5 (5.21%)
Kurtosis=0.62
Skewness=0.71
Sigma Level=1.77
DPMO=394117.9
Capability Analysis
35
LSL=60
30
Nominal=70
USL=80
Frequency
25
20
15
10
0
57
62
67
72
77
82
87
Measurement
' where ' is the estimated standard deviation from a range or s chart
Cp: = (USL-LSL)/6
Cpk: the minium of Cpu or Cpl
' where X is the overall average (the
Cpu: the capability based on the USL = (USL- X )/3
number in parentheses is the theoretical % greater than the USL)
' (the number in parentheses is the
Cpl: the capability based on the LSL = ( X -LSL)/3
theoretical % less than the LSL)
'
Est. Sigma =
Pp: = (USL-LSL)/6s where s is the standard deviation of the measurements
Ppk: the minium of Ppu or Ppl
Ppu: the capability based on the USL = (USL- X )/3s where X is the overall average (the number
in parentheses is the theoretical % greater than the USL)
Ppl: the capability based on the LSL = ( X -LSL)/3s (the number in parentheses is the theoretical
% less than the LSL)
Sigma: = s
Average: = X
Count: = number of data points in the analysis
No. Out of Spec: = actual number out of specification (number in parentheses is the % out)
34
Kurtosis: a measure of the shape of the distribution. A positive value means that the distribution
has longer tails than a normal distribution; a negative value means that the distribution has shorter
tails. The normal distribution has kurtosis of 0.
Skewness: a measure of asymmetry. If skewness is 0, there is perfect symmetry (like the normal
distribution). A positive value means that the tail of the distribution is stretched on the side above
the mean. The negative values means it is stretch on the side below the mean.
Sigma Level: A statistical term that measures how much a process varies from perfection, based
on the number of defects per million units.
o One Sigma = 690,000 per million units
o Two Sigma = 308,000 per million units
o Three Sigma = 66,800 per million units
o Four Sigma = 6,210 per million units
o Five Sigma = 230 per million units
o Six Sigma = 3.4 per million units
DPMO: Defects per million opportunities
Data Entry
If you are just using data to determine process capability without using a
control chart, enter the data into the spreadsheet. An example is shown to
the right. Select the data to be used in the analysis and then select the
process capability option (Cpk) on the SPC toolbar. If you want to do a
process capability analysis for an existing chart, you do not have to select
anything on a worksheet prior to selecting the process capability option on
the SPC toolbar.
35
81
77
74
75
73
77
76
76
74
79
77
76
76
75
84
73
77
74
79
80
74
75
75
76
76
82
73
76
76
81
74
77
75
76
73
80
74
75
74
78
77
73
72
76
76
77
79
76
74
75
75
79
73
72
74
77
76
75
75
75
74
74
70
74
77
79
75
72
75
75
78
78
72
74
73
74
79
77
77
81
73
74
74
78
74
74
75
75
73
79
73
72
79
71
74
75
36
Dates of Data Collection: Add the starting date and ending dates of data collection. These dates
are optional. If entered, they will appear in a dialog box in the lower left-hand corner of the chart.
More Than One Chart? Select Yes if you want to make multiple process capability charts by
looping through the dialog box. The program assumes that the next set of data for the process
capability analysis is adjacent to the current set. Use one row or one column of data if you are
selecting this option. No is the default value.
Remove Outliers? Select Yes if you want to remove outliers from the calculations. Enter the
number of standard deviations you want to remove outliers beyond (e.g., beyond +/- 6 sigma).
The default option is No.
37
Name of Chart
LSL
Nominal
USL
Chart 1
70
100
130
Chart 2
68
100
132
Chart 3
75
97.5
120
Chart 4
60
100
140
Chart 5
65
95
125
90
92
109
89
92
97
112
95
95
102
108
101
116
110
81
94
91
108
95
100
92
112
82
91
109
107
85
99
108
99
97
93
89
118
87
101
103
97
87
96
96
101
117
91
105
112
97
101
105
85
89
111
104
102
117
97
106
106
84
113
109
102
103
116
83
76
108
115
108
112
99
83
105
98
114
116
100
108
116
114
111
99
90
117
95
105
75
94
94
101
107
106
102
99
107
95
99
87
105
112
38
Remove Outliers? Select Yes if you want to remove outliers from the calculations. Enter the
number of standard deviations you want to remove outliers beyond (e.g., beyond +/- 6 sigma).
The default option is No.
Reset Specifications Limits? Select Yes if you want the program to replace the existing
specification limits with new limits set at the value of +/- sigma you enter. This is useful if you
are trying to set specification limits, e.g., for prototype data.
Add +/- 3 Sigma Limits: In addition to the specifications, you can add the +/- three sigma limits to
the chart. The default is No. If you select Yes, you can chose sigma to the estimated sigma from
the range chart or the calculated standard deviation of all the data.
Cpk
1.03
0.87
0.54
1.16
0.85
Cpu
1.11
1.01
0.54
1.16
0.85
Pp
1.09
1.1
0.77
1.13
1.04
Ppk
1.05
1.02
0.62
1.01
0.87
Ppu
1.13
1.18
0.62
1.01
0.87
39
USL
130
132
120
140
125
Scatter Diagram
A scatter diagram is used to show the relationship between two kinds of data. It could be the relationship
between a cause and an effect, between one cause and another, or even between one cause and two others.
In the example file, there is data that relates steam usage in a plant to the atmospheric temperature. The
question being answered here is Does the atmospheric temperature have an effect on steam usage in the
plant. For 25 days, data were collected on steam usage and temperature. There are 25 sets of data point.
Each set of data points is charted on a scatter. The scatter diagram is shown below.
Scatter Diagram (Significant, p = 0)
13.3
12.3
11.3
10.3
9.3
8.3
y = b1x + b0
where y is the variable on the y axis, x is the variable on the x axis, b1 is the slope of the line and b0 is
where the line crosses the y axis. In the example above, the slope is -0.0871. This means that for each
unit increase in x (one degree of temperature in this case), the y value (steam usage in this case) decreases
by -0.0871.
The value of R2 in the chart is the % of variation in y that is explained by x. In this example, 75% of the
variation in steam usage is explained by the variation in temperature.
40
Data Entry
Sample
No.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
Temperature
(X)
35.3
29.7
30.8
58.8
61.4
71.3
74.4
76.7
70.7
57.5
46.4
28.9
28.1
39.1
46.8
Steam
Usage (Y)
10.98
11.13
12.51
8.4
9.27
8.73
6.36
8.5
7.82
9.14
8.24
12.19
11.88
9.57
10.94
41
Data in: Select columns or rows depending on how the data is entered into the spreadsheet. The
program selects one or the other depending on the range selected prior to selecting the scatter
diagram option on the SPC toolbar.
Rounding/Forecasting/Point Labels
42
Updating Charts
All charts can be updated easily after new data has been entered into the spreadsheet. There is no need to
select anything on the worksheet. The program automatically checks to see what new data has been
entered. Once you have entered the new data, select the
update option from the SPC toolbar. You will get the
dialog shown to the right.
The dialog box lists all the available charts for updating
in the workbook. You may select multiple charts at
once. You can also check the Update All Charts
option to update all the charts in the workbook.
Caution on Updating Charts:
If you change the name on the worksheet tab containing your chart, you will not be able to update the
chart because the program cant find the new name. There is a manual method that allows you to change
the name on the worksheet tab. See the section on Frequently Asked Questions in the manual.
43
You must select a point on the chart prior to selecting the single point
option from the menu. To select a single point:
Below is an example of a chart with one point selected (this is the c chart data from the example
workbook).
After selecting the point, select the single point action option (SP) from the SPC toolbar. The dialog box
above will appear. Select the option you want and then select OK. The average and limits are
recalculated based on your option and the chart is re-made. See the notes on the following page for more
information.
44
2/8/2003
2/9/2003
2/10/2003
2/11/2003
2/12/2003
2/13/2003
2/14/2003
2/15/2003
Subgroup Number
2/7/2003
2/8/2003
2/9/2003
2/10/2003
2/11/2003
2/12/2003
2/13/2003
2/14/2003
2/15/2003
Subgroup Number
45
Note: When removing all points beyond the limits, the average and limits are
recalculated. It is possible that additional points will now be beyond the
limits. You may have to run this several times to remove all points beyond
the limits.
46
Prepared by:
Measurement
People
Environment
Product out of
specifications
Machines
Methods
Materials
You can change the main category headings (measurement, people, etc). You can enter the problem or
goal at the head of the fishbone (see above). You will see a button that says Add Item. Select this
button and you will get the dialog box below.
Enter an item to add to the cause and effect diagram. For example, you might enter Not calibrated.
Then select OK. This item is then shown on the chart and you may move it to any location you want.
Cause and Effect Diagram Diagram
Date:
Prepared by:
Measurement
People
Environment
Product out of
specifications
Not calibrated
Machines
Methods
Materials
47
48
Select the first option to setup the data entry sheet based on the number of operators, parts, and trials and
then select OK. You will see the form below which has been filled in with the numbers for this example.
You must enter all the information. The number of decimal places in the measurement is VERY
IMPORTANT. It controls how the data is rounded and shown. Entering zero when you have two or
three decimal places in the data may lead to inaccurate results. After entering all the information, select
OK. This will generate the data entry sheet shown below. The number of trials and parts can range from
2 to 20; the number of appraisers can range from 1 to 25 (with one appraiser you will only get a Gage
R&R report based on the average and range method).
Enter the operator names in the upper right hand corner. The names will automatically appear in the first
column. NOTE: The program uses Microsoft Excels naming function to run. You cannot have
spaces or certain characters (e.g. /). Instead of using John Smith, use John_Smith. Enter the rest of
the information for Date, Gage Name, Gage Number, Gage Type, Product, Characteristic, Upper
Specification Limit, Lower Specification Limit, and Performed By. None of this information is required
to run the program with the possible exception of the specification limits. These are required if you are
basing the acceptability of the measurement system on the tolerances. You then enter the data from the
appraisers for each trial and each part. A completed data entry screen is shown on the next page.
49
2
Operator 1: Enter Operator 1 Name Here
Operator 2: Enter Operator 2 Name Here
Operator 3: Enter Operator 3 Name Here
Operator
Trial/Part
Enter Operator 1 Name Here
1
Enter Operator 1 Name Here
2
Enter Operator 1 Name Here
3
Enter Operator 2 Name Here
1
Enter Operator 2 Name Here
2
Enter Operator 2 Name Here
3
Enter Operator 3 Name Here
1
Enter Operator 3 Name Here
2
Enter Operator 3 Name Here
3
10
2
Operator 1: Hal
Operator 2: Beth
Operator 3: Loa
10/31/2005
Thickness Gage
T-101
Thickness
Widget
Thickness
3
-3
Bill
Trial/Part
1
2
3
1
2
3
1
2
3
2
0.29
0.41
0.64
0.08
0.25
0.07
0.04
-0.11
-0.15
3
-0.56
-0.68
-0.58
-0.47
-1.22
-0.68
-1.38
-1.13
-0.96
4
1.34
1.17
1.27
1.19
0.94
1.34
0.88
1.09
0.67
5
0.47
0.5
0.64
0.01
1.03
0.2
0.14
0.2
0.11
6
-0.8
-0.92
-0.84
-0.56
-1.2
-1.28
-1.46
-1.07
-1.45
7
0.02
-0.11
-0.21
-0.2
0.22
0.06
-0.29
-0.67
-0.49
8
0.59
0.75
0.66
0.47
0.55
0.83
0.02
0.01
0.21
9
-0.31
-0.2
-0.17
-0.63
0.08
-0.34
-0.46
-0.56
-0.49
2.26
1.99
2.01
1.8
2.12
2.19
1.77
1.45
1.87
10
-1.36
-1.25
-1.31
-1.68
-1.62
-1.5
-1.49
-1.77
-2.16
50
On the right-hand side, you have three options for additional charts that can be generated along with the
Gage R&R report. The options are:
1. All Charts: this will generate all the charts associated with the study (see the first page of this
instructional manual for the list or the figure below)
2. No Charts: only the Gage R&R report will be generated
3. Select Charts: only the charts you select will be generated
If you select Select Charts, you will get the dialog box below. You select the charts you want to
include in the output and then select OK. This returns you to the form above.
Once you have selected your options, select OK and the program will generate the Gage R&R report as
well as the charts you have selected (if any).
When finished, the program will display the Gage R&R report. The one generated from this data (using
the Total Variation Based on Parts Option) is shown below. The report contains all the information in a
classical Gage R&R study and bases the conclusion if the measurement system is acceptable based on one
of the three options selected below. Any charts that were selected are generated on separate worksheets
in the workbook. You can download a completed workbook with all the charts from our website
(www.spcforexcel.com). A summary of each chart is given below.
Stacked Averages Chart the average of each appraiser on each part is plotted by appraiser
using the part number as the index. There is one line for each appraiser. This helps
determine how consistent the operators are. The overall average and control limits are also
plotted. If the parts represent the total (true) variation in the process, at least half of theses
points should be out of control. If this not the case, the measurement system does not have
the ability to distinguish between samples (poor resolution) or the parts do not reflect the total
variation in the process.
Unstacked Averages Chart same as the stacked chart but the appraisers are plotted together,
not separately.
Stacked Range Chart used to show the range of each operators trials on a part and includes
the average range and control limits. There is one line for each appraiser. The chart is used
to determine if the process is in control. If there are out of control points, the special causes
need to be found and eliminated. Care should be taken with interpreting the Gage R&R
results if there are special causes present. Special causes occur if there are points beyond the
control limits.
Unstacked Range Chart same as the stacked range chart but the appraisers are plotted as
one line.
Run Chart by Part plots the individual readings by part for all appraisers to help see if there
are any outliers and to see the variation in the individual parts.
51
Rbar=
K1=
0.341667
0.5908
Product: Widget
Characteristic: Thickness
USL:
3
LSL:
-3
XbarDiff=
K2=
0.444667
0.5231
Date: 10/31/05
Performed by: Bill
Rp=
K3=
3.511111
0.3146
% Tolerance
% EV = 100(EV/TV)
=
17.61%
% EV = 100(EV/(USL-LSL)/6)
=
20.19%
% AV = 100(AV/TV)
=
20.04%
% AV = 100(AV/(USL-LSL)/6)
=
22.97%
% R&R = 100(R&R/TV)
=
26.68%
% R&R = 100(R&R/(USL-LSL)/6)
=
30.58%
% PV = 100(PV/TV)
=
96.38%
% PV = 100(PV/(USL-LSL)/6)
= 110.46%
ndc= 1.41(PV/R&R)
= 5.093652
Conclusion
% R&R under 10% of Total Variation: Measurement system is acceptable
**** % R&R from 10% to 30% of Total Variation: Measurement system may be acceptable based the application
% R&R over 30% of Total Variation: Measurement system needs improvement
Scatter Plot plots the individual readings by part-by-appraiser to examine how consistent
the appraisers are, to look for part-appraiser interactions, and to look for outliers.
Whiskers Charts plots the high, average, and low value by part for each appraiser to
examine how consistent the appraisers are, to look for part-appraiser interactions, and to look
for outliers (same items as for the scatter plot).
Error Charts plots the error (observed value average measurement of the part) by partappraiser to determine which operator may have bias and which operator has the most
variability.
Normalized Histograms plots the normalized value (observed value average measurement
of part) as a histogram to determine how the error is distributed by appraiser.
X-Y Plot plots the average of the readings by each appraiser against the overall part
averages to examine consistency in linearity between appraisers.
Appraiser Charts consists of three charts:
o Range charts for each appraiser to determine if each is in control
o Bias chart for all appraisers to determine if different appraisers display detectably
different average values for the parts.
o Consistency chart for all appraisers to determine if different appraisers display detectably
different standard deviations for the parts.
52
ANOVA Method
Set-up Data Entry
The set-up is the same as for the Average and Range Method. Please follow the instructions for set-up
data entry as well as data entry for the Average and Range Method.
Generating the Results
After the data has been entered into the worksheet, select the MSA icon on the SPC toolbar. You will get
the form below. Select the ANOVA Method option as shown in the form and then select OK.
The program generates two new worksheets. One (ANOVA Report) contains the Gage Repeatability and
Reproducibility ANOVA Method Report as shown below. The acceptability of the measurement system
is based on percent contribution, not on % Total Variation.
Gage Repeatability and Reproducibility ANOVA Method Report
Gage Name: Thickness Gage
Gage No.
T-101
Gage Type: Thickness
Product: Widget
Characteristic: Thickness
USL:
3
LSL:
-3
% TOTAL
VARIATION
STD DEV.
Repeatability (EV)
Reproducibility (AV)
Appraiser by Part (INT)
GRR
Part (PV)
Date: 10/31/05
Performed by: Bill
0.199933
0.226838
0
0.302372
1.042327
PERCENT
CONTRIBUTION
18.42%
20.90%
0.00%
27.86%
96.04%
3.39%
4.37%
0.00%
7.76%
92.24%
4
1.0853
The other worksheet (GRR ANOVA) contains the ANOVA table and the residuals plot as shown below.
The residuals chart plots the residual versus the average for each appraiser for each part. The residual is
the result minus that average. The points should be randomly scattered above and below zero.
53
ANOVA Results
Source
Appraiser
Part
Appraiser by Part
Equipment
Total
df
2
9
18
60
89
SS
3.16726222
88.3619344
0.35898222
2.75893333
94.6471122
MS
1.58363111
9.81799272
0.01994346
0.04598222
F
34.44
213.517
0.434
Sig
0.0000
0.0000
0.9740
Source of
Variation
Estimate of
Variance
Std. Dev.
Equipment
Appraiser
Interaction
GRR
Part
Total Variation
0.03997328
0.05145526
0
0.09142854
1.0864466
1.17787514
0.19993318
0.22683752
0
0.30237152
1.04232749
1.08529956
ndc = 4.86051648
or
% Total
Variation
%
Contribution
18.42%
20.90%
0.00%
27.86%
96.04%
100.00%
3.39%
4.37%
0.00%
7.76%
92.24%
Residuals
-2
-1.5
-1
-0.5
0.5
1.5
2.5
Average Values
54
Select the data as highlighted above. Do not select the column headings. To run the program, select the
icon on the SPC toolbar. You will see the first page of the form as usual. Select the second page tab
labeled Range/Bias/Linearity/Attribute Gage R&R.
Select the Range Gage R&R Method and select OK. The dialog below will appear.
The range that appears in the Range containing part numbers and appraiser results is the range that is
selected on the worksheet. You can save some time by selecting the data before selecting the icon on the
55
toolbar. Enter the process standard deviation. For the example, this standard deviation is 0.0777. Then
select OK. A new worksheet will be added to the workbook with the following output.
Range Method for Gage R&R
Average Range (Rbar) =
Upper Control Limit =
Lower Control Limits =
In Control?
GRR = Rbar/d2* =
Process Standard Deviation =
%GRR = 100(GRR/Process Standard Deviation =
Conclusion:
0.07
0.22869
None
Yes
0.058772
0.0777
75.6%
The measurement system needs improvement.
Range Chart
0.25
Range
0.2
0.15
0.1
0.05
0
1
Part Number
The results give the average range and control limits. The range chart is shown to check for special
causes of variation. The % GRR is calculated and the conclusion given based on the % GRR.
56
A conclusion is also presented that states if you can assume the bias is zero.
57
Measured Value
Alpha=0.05
Standard
Reference Deviation
n
Mean
Value
()
t statistic
15 6.006667
6 0.21202 0.121781
Confidence Interval
df
t value (2
tailed)
Bias
Lower
Upper
14 2.144787 0.006667 5.889254 6.124079
There is no evidence that the average is significantly different than reference value. You may assume the bias is zero.
Frequency
5
4
3
2
1
0
5.6
5.7
5.8
5.9
6.0
6.1
6.2
6.3
6.4
Measured Value
58
Once you have entered the information into the dialog box, select OK. A new worksheet is added to the
workbook with the results as shown on the next page for this example.
59
Alpha=0.05
Standard
Reference Deviation
Mean
Value
()
t statistic
101.045
100.3 0.566127 5.885146
n
Measured Value
Confidence Interval
df
t value (2
tailed)
19 2.093024
Bias
0.745
Lower
100.78
Upper
101.31
There is evidence that the average is significantly different than the reference value. The bias is not zero.
2.5
2
1.5
1
0.5
0
10
0.
1
10
0.
2
10
0.
3
10
0.
4
10
0.
5
10
0.
6
10
0.
7
10
0.
8
10
0.
9
10
1.
0
10
1.
1
10
1.
2
10
1.
3
10
1.
4
10
1.
5
10
1.
6
10
1.
7
10
1.
8
10
1.
9
10
2.
0
Frequency
Measured Value
Sample Result
X Chart
104
103
102
101
100
99
98
97
UCL=102.98
Avg=101.05
UCL=99.11
10
11
12
13
14
15
16
17
18
19
20
13
14
15
16
17
18
19
20
Sample Number
Moving Range
2.5
UCL=2.38
2
1.5
1
Avg=0.73
0.5
0
1
10
11
12
Sample Number
The output contains the same information as given in the Bias Independent Sample Method. It also
includes the control charts for the sample to help determine stability.
60
Linearity
Linearity is the difference of bias throughout the
measurement range. To determine linearity, the
samples you select must cover the expected operating
range of the measurement system. You should use at
least five samples that cover this range. One
appraiser should measure each of the parts at least ten
times. An example of the required data input is
shown to the right (page 95, Measurement Systems
Analysis, Third Edition). The data must be in
columns. The first row of data contains the part
number; the second row contains the reference values
for each part; and the remaining rows contain the
measurements.
Select the data as highlighted above. Do not select the first column. To
run the program, select the icon on the SPC toolbar. You will see the
first page of the form as usual. Select the second page tab labeled
Range/Bias/Linearity/Attribute Gage R&R. Select the Linearity
option and select OK. You will then see the dialog box to the left.
The range is the range selected on the worksheet before starting the
program. Enter the number of decimal places and alpha (default is
0.05). The Rounding in Equations determines how the best fit equation is displayed. The default value
is 3. Once you have entered the information, select OK and the program will generate a chart like the one
below.
The linearity is NOT acceptable. Ta=12.043, Tcritical=2.002
Linearity
y = -0.132x + 0.737, R Sqr = 71.4%
The bias is NOT the same for all reference values. Tb=10.158, Tcritical=2.002
1.5
Bias
0.5
Regression
Bias
Upper 95% CI
Lower 95% CI
Bias Average
0
Bias=0
-0.5
-1
2
10
Reference Values
In the upper left-hand side of the chart, the conclusion is given for linearity. In this example, there is a
prl0bem with linearity. The bias = 0 line (green on the chart) should be contained by the upper and lower
95% confidence intervals. The equation is the title is the best fit equation for the individual readings.
The R squared value gives the % of variation in the bias that is explained by the variation in reference
values.
61
1
0
Reference
Operator
Trial/Part
Enter Operator 1 Name Here
1
Enter Operator 1 Name Here
2
Enter Operator 1 Name Here
3
Enter Operator 2 Name Here
1
Enter Operator 2 Name Here
2
Enter Operator 2 Name Here
3
Enter Operator 3 Name Here
1
Enter Operator 3 Name Here
2
Enter Operator 3 Name Here
3
10
11
12
13
Enter the operator names in the upper right hand corner. The names will automatically appear in the first
column. NOTE: The program uses Microsoft Excels naming function to run. You cannot have
spaces or certain characters (e.g. /). Instead of using John Smith, use John_Smith. Enter the rest of
the information for Date, Gage Name, Gage Number, Gage Type, Product, Characteristic, Pass Value,
Fail Value, and Performed By. None of this information is required to run the program except the Pass
Value and Fail Value. Then fill in the data as shown below.
Attribute Gage R&R Study
Date:
Gage Name:
Gage Number:
Gage Type:
Product:
Characteristic:
Pass Value
Fail Value
Performed By:
Operator
A
A
A
B
B
B
C
C
C
Operator 1: A
Operator 2: B
Operator 3: C
1
0
Reference
Trial/Part
1
2
3
1
2
3
1
2
3
1
1
1
2
1
1
1
1
1
1
1
1
1
0
3
1
1
1
1
1
1
1
1
1
4
0
0
0
0
0
0
0
0
0
5
0
0
0
0
0
0
0
0
0
62
1
6
0
0
0
0
0
0
0
0
0
1
7
1
1
0
1
1
0
1
0
0
1
8
1
1
1
1
1
1
1
0
1
0
9
1
1
1
1
1
1
1
1
1
1
10
0
0
0
0
0
0
0
0
0
1
11
1
1
1
1
1
1
1
1
1
0
12
1
1
1
1
1
1
1
1
1
1
13
0
0
0
0
0
0
0
1
0
1
1
1
1
1
1
1
1
1
To generate the results, select the icon on the SPC toolbar. You will see the first page of the form as
usual. Select the second page tab labeled Range/Bias/Linearity/Attribute Gage R&R. You will see the
form below. Select the Attribute Gage R&R option, followed by the Run the Attribute Gage R&R
analysis.
There are three options: crosstabulations, kappa values, and the effectiveness table. Select the options
you want and select OK. The program will generate the following results depending on options selected.
Crosstabulations
You will get a table like this one for each of the
combinations of appraisers. The kappa value is
given. If kappa is above 0.75 there is good
agreement between the appraisers. If it is less
than 0.40, there is poor agreement. The count
information is as follows:
A and B both rate as fail
A rates as fail and B rates as pass
A rates as pass and B rates as fail
A and B both rate as pass
A * B Crosstabulation
B
Fail Pass Total Kappa
A
Fail Count
44
6
50 0.86
Expected Count 15.7 34.3 50.0
Pass Count
3
97 100
Expected Count 31.3 68.7 100
Total
Count
47 103 150
Expected Count 47.0 103.0
63
A
0.86
0.78
B
0.86
0.79
C
0.78
0.79
-
There is good to excellent agreement since all kappa values are greater than 0.75
Kappa
A
0.88
B
0.92
C
0.77
There is good to excellent agreement since all kappa values are greater than 0.75
Effectiveness Table
Attribute Gage R&R Effectiveness
Gage Name:
Gage No.
Gage Type:
Product:
Characteristic:
% Appraiser1
Source
A
B
C
Total Inspected
50
50
# Matched
42
45
False Negative (Appraiser biased toward rejection)
False Positive (Apprasier biased toward acceptance)
Mixed (Appraiser accepts and rejects the same part)
95% UCI
92.8%
96.7%
Calculated Score
84.0%
90.0%
95% LCI
70.9%
78.2%
Total Inspected
# in Agreement
95% UCI
Calculated Score
95% LCI
A
50
40
90.0%
80.0%
66.3%
Date:
Performed by:
% Score vs Attribute2
B
C
50
50
50
42
45
40
0
0
0
0
0
0
8
5
10
92.8%
96.7%
90.0%
84.0%
90.0%
80.0%
70.9%
78.2%
66.3%
Notes:
(1) Appraiser agrees with him/herself on all trials
(2) Appraiser agrees on all trails with the known reference
(3) All appraisers agreed within and between themselves
(4) All appraisers agreed with and between themselves and agreed with the reference
(5) UCI and LCLI are the upper and lower confidence interval bounds respectively
If the calculated score for each appraiser falls within the confidence interval of the other appraisers, the
effectiveness of the appraisers is the same.
64
65
Regression
The program contains a multiple regression component under the miscellaneous button on the SPC
toolbar. This is used to determine which independent variables (the Xs) have a significant impact on the
dependent variable Y. An example of the data entry is shown below. In this example, we want to find
out if attendance (Y) at major league baseball games is impacted by the team batting average, the number
of home runs hit by the team, the teams earned run average, winning percentage of total payroll. Data
for the 2005 baseball season is given below.
Team
Arizona Diamondbacks
Atlanta Braves
Baltimore Orioles
Boston Red Sox
Chicago Cubs
Chicago White Sox
Cincinnati Reds
Cleveland Indians
Colorado Rockies
Detroit Tigers
Florida Marlins
Houston Astros
Kansas City Royals
Los Angeles Angels
Los Angeles Dodgers
Milwaukee Brewers
Minnesota Twins
New York Mets
New York Yankees
Oakland Athletics
Philadelphia Phillies
Pittsburgh Pirates
San Diego Padres
San Francisco Giants
Seattle Mariners
St. Louis Cardinals
Tampa Bay Devil Rays
Toronto Blue Jays
Washington Nationals
Texas Rangers
Average
0.256
0.265
0.269
0.281
0.27
0.262
0.261
0.271
0.267
0.272
0.272
0.256
0.263
0.27
0.253
0.259
0.259
0.258
0.276
0.262
0.27
0.259
0.257
0.261
0.256
0.27
0.274
0.267
0.265
0.252
Home
Runs
191
184
189
199
194
200
222
207
150
168
128
161
126
147
149
175
134
175
229
155
167
139
130
128
130
170
157
260
136
117
Earned
Run Avg.
4.84
3.98
4.56
4.74
4.19
3.61
5.15
3.61
5.13
4.51
4.16
3.51
5.49
3.68
4.38
3.97
3.71
3.76
4.52
3.69
4.21
4.42
4.13
4.33
4.49
3.49
5.39
4.96
4.06
3.87
Winning
%
0.475
0.556
0.457
0.586
0.488
0.611
0.451
0.574
0.414
0.438
0.512
0.549
0.346
0.586
0.438
0.5
0.512
0.512
0.586
0.543
0.543
0.414
0.506
0.463
0.426
0.617
0.414
0.488
0.494
0.5
Total Payroll
$62,329,166
$86,457,302
$73,914,333
$123,505,125
$87,032,933
$75,178,000
$61,892,583
$41,502,500
$48,155,000
$69,092,000
$60,408,834
$76,779,000
$36,881,000
$97,725,322
$83,039,000
$39,934,833
$56,186,000
$101,305,821
$208,306,817
$55,425,762
$95,522,000
$38,133,000
$63,290,833
$90,199,500
$87,754,334
$92,106,833
$29,679,067
$55,849,000
$45,719,500
$48,581,500
Attendance
2,059,331
2,521,534
2,624,804
2,813,354
3,100,262
2,342,834
1,943,157
1,973,185
1,915,586
2,024,505
1,823,388
2,762,472
1,371,181
3,404,686
3,603,680
2,211,023
2,013,453
2,782,212
4,090,440
2,109,298
2,665,301
1,794,237
2,832,039
3,140,781
2,689,529
3,491,837
1,124,189
2,486,925
1,977,949
2,692,123
To run the regression program, select the shaded area as shown. You MUST select the column labels
(data can be rows also). Then select the regression option under the MISC icon on the SPC toolbar. You
will get the dialog box shown below.
66
67
Regression Output
Regression Summary for Attendance
df
Regression
Residual
Total
5
24
29
Coefficient
Intercept
AVG
HR
ERA
Winning %
Totalpayroll
8099237.433
-18922620.41
689.429779
-355703.5027
-863435.1508
0.016759643
ANOVA Table
SS
9.79271E+12
3.23131E+12
1.3024E+13
MS
Signif. F
1.95854E+12
1.34638E+11
14.54672883
1.33582E-06
p Value
95% Lower
Coefficients
Standard Error
t Stat
2667080.801
3.03674243
12537980.63 -1.509223931
2600.899626 0.265073582
270091.4996 -1.316974074
2564965.184 -0.33662646
0.00246773 6.791523367
95% Upper
Regression Statistics
Multiple R
R Square
Adjusted R Square
Standard Error
Observations
Durbin-Waston Statistic
0.867119393
0.751896042
0.700207717
366930.4105
30
1.655584043
Predicted
1,290,000
1,790,000
2,290,000
2,790,000
3,290,000
3,790,000
4,290,000
4,790,000
Actual
68
69
Miscellaneous
Selecting the miscellaneous option (Misc) on
the SPC toolbar generates the dialog box
shown to the right. Select the option you
want. The various options are described
below.
Descriptive Statistics
This option displays certain statistics on a number of samples. For example,
suppose you have measured the moisture content of a powder over time and
have fifteen sample results. The data is entered into the spreadsheet as shown
to the right. Select the data only (shaded area), then select the miscellaneous
option (Misc) on the SPC toolbar and then the descriptive statistics option.
You will get the dialog below.
Observation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
% Moisture
0.08
0.13
0.12
0.17
0.1
0.16
0.14
0.14
0.13
0.14
0.10
0.17
0.14
0.12
0.13
0.131333
Standard Error
0.006537
Median
0.13
Mode
0.14
Standard Deviation
0.025317
Sample Variance
0.000641
Kurtosis
-0.0048
Skewness
-0.29422
Range
0.09
Minimum
0.08
Maximum
0.17
Sum
1.97
Count
15
70
Observation
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
% Moisture
0.08
0.13
0.12
0.17
0.1
0.16
0.14
0.14
0.13
0.14
0.10
0.17
0.14
0.12
0.13
Select the shaded data as shown. Select the confidence interval around a mean
option after selecting the miscellaneous option (Misc) from the SPC toolbar.
You will get the dialog box below.
Enter data range (no headings): The default range is the range
selected on the spreadsheet. You can change this if necessary.
Alpha: The default value for alpha is 0.05. This represents the chance
of what we find out in this sample is not representative of the population. It is typically 0.05.
Enter the known standard deviation (if known): In
this example, it is known and is 0.02.
Output Options: You have two output options: on
the current worksheet (select the cell location) or a
new worksheet. Select OK and the program
generates the output shown below.
Confidence Interval Around a Mean
Mean
0.131333
Standard Deviation
0.02
Count
15
Degrees of Freedom
14
Alpha
0.05
t Value
1.959964
Upper Confidence Level
0.141455
Lower Confidence Level
0.121212
Since the range of the lower to upper confidence levels (0.121 to 0.141) does not include 0.15, you
conclude that the change has made a significant difference.
71
The confidence interval for the average is between 0.117 and 0.145.
72
Sample
Number
1
2
3
4
5
6
7
8
9
10
Bulk
Density
0.697
0.698
0.673
0.657
0.710
0.702
0.680
0.709
0.670
0.671
73
Process 2
n2
Sample average
Y1
Y2
s1
s2
Sample variance
s1
Sample size
s2
These sample statistics will be used to compare the variances and averages of the two processes. The
variances will be compared by taking the ratio of the two variances and using the F distribution to
determine if there is a significant difference between the two. The averages will be compared by
constructing a confidence interval around the difference in two averages. If the confidence interval
contains zero (i.e., the difference could be zero), we will conclude that there is no evidence that the two
processes are operating at different averages. If the confidence interval does not include zero, we will
conclude that there is evidence that the processes are operating at different averages.
For example, consider two batch reactors that make the "same" product.
Questions have arisen about whether the reactors really do make the same
product. One indication of the completeness of reaction is the residual
catalyst remaining after reaction. Ten observations were taken from each
reactor. The results are given to the right.
Select the cells in the shaded area for the two processes (do not have to be
adjacent on spreadsheet) and then select the confidence interval for the
difference in two means option after selecting the miscellaneous option
(Misc) from the SPC toolbar. You will get the dialog box shown below.
Reactor 1
450
423
443
476
490
436
457
421
485
491
Reactor
2
482
422
463
492
445
483
476
462
479
495
Enter range for variable 1: The default range is the first range
selected on the spreadsheet. You can change this if necessary.
Enter range for variable 2: The default range is the second range selected on the spreadsheet.
You can change this if necessary.
Alpha: The default value for alpha is 0.05. This
represents the chance of what we find out in this
sample is not representative of the population. It is
typically 0.05.
Enter the known standard deviation (if known): In
this example, it is known and is 0.02.
74
Output Options: You have two output options: on the current worksheet (select the cell location)
or a new worksheet. Select OK and the program generates the output shown below.
Difference Betwen Two Means Confidence Interval
Variable 1
Variable 2
Mean
457.2
469.9
Standard Deviation
26.93119
22.56078
Variance
725.2889
508.9889
Observations
10
10
Pooled Variance
617.1389
Variance Same?
Yes
t Statistic
1.143134
Degrees of Freedom
18
Alpha
0.05
Critical t value
2.100922
Upper Confidence Level
10.64084
Lower Confidence Level
-36.0408
P(T<t)
0.26796
Conclusion
No statistical difference in the means.
The program will provide the conclusion for you. In this case, there is no statistical difference in the
means. Note that the program also tells you if the variances in the two processes are the same. In this
case, the variances are the same also.
75
Alpha: The default value for alpha is 0.05. This represents the chance of what we find out in this sample
is not representative of the population. It is typically 0.05. Select OK. The program will insert a new
worksheet with the output. The output from the data above is shown on the next page. The following are
calculated:
76
Furnace 1
Mean
Sigma
Variance
Observations
Average
Sigma
UCLs
LCLs
Pooled
Variance
Interval Alpha
T Value
Process
Furnace 1
Furnace 2
67
67
64
65
64
65
67
65.57143
1.397276
1.952381
7
Furnace
2
64
65
67
64
65
68
64
65.28571
1.603567
2.571429
7
Furnace
3
68
65
67
64
67
65
65
65.85714
1.46385
2.142857
7
Furnace
4
69
72
70
68
67
68
70
69.14286
1.676163
2.809524
7
1.535214
2.889273
0.181155
2.369048
0.008
2.892495
Lower
Confidence
-2.094
Upper Confidence
2.665432 No statistical difference in the means.
Furnace 1
Furnace 3
-2.66543
2.094004
Furnace 1
Furnace 4
-5.95115
-1.19171
Furnace 2
Furnace 3
-2.95115
1.808289
Furnace 2
Furnace 4
-6.23686
-1.47743
Furnace 3
Furnace 4
-5.66543
-0.906
Each possible pair of averages is checked. The program tells you if the means are different (see italics
above).
77
Observation
Number
Sieve
Test
Particle
Size
Analyzer
1
2
3
4
5
6
7
8
9
10
245
196
205
226
213
234
229
192
233
204
242
190
208
215
203
235
220
193
224
195
Fill in the dialog box and select OK. The program will
generate the output below.
The program will provide the conclusion for you. In this
case, there is a difference between the two processes. The
confidence interval defined by the lower and upper
confidence limits do not contain zero.
78
5.2
5.266245
27.73333
10
9
0.05
2.262157
8.967245
1.432755
Means are different
Analysis of Means
Analysis of means is a graphical and statistical way of comparing k treatments means with the overall
mean. The method used in this program is described in the book Advanced Topics in Statistical Control
by Dr. Donald J. Wheeler (www.spcpress.com). The example is from the book.
The maximum number of treatments is 25. The maximum subgroup size for each treatment is also 25.
Suppose you are studying five different methods A
(treatments) of applying a coating and are 250
measuring the weight of the coating. The data 260
entry requirement for this option is shown to the 230
right. The treatments are A E. The weight for 270
each treatment is given in the column under the treatment letter.
B
310
330
280
360
C
250
230
220
260
D
340
270
300
320
E
250
240
270
290
To run the Analysis of Means, select the treatment headings and the data
(the shaded area above). Then select the Analysis of Means option after
selecting the miscellaneous option (Misc) from the SPC toolbar. The
dialog box to the right appears.
Select OK and the output and charts on the next page are generated. The UDL and LDL are the upper
decision limit and the lower decision limits. Any points beyond these limits on the top chart represent
significant differences from the overall treatment average. The range chart is a classical range chart and
compares the variation within treatments to see if they are the same.
79
Treatment
Average
Maximum
Minimum
Range
Treatment Variance
Treatment DF
Overall Average
Average Range
Est. V(X)
d2*
Est SD(X)
Est SD(Xbar)
Degrees of Freedom
H
ANOM Upper Limit
ANOM Lower Limit
A
250
260
230
270
252.5
270
230
40
291.6666667
3
276.5
56
N/A
2.096
26.71755725
13.35877863
14
2.266666667
306.7798982
246.2201018
B
310
330
280
360
320
360
280
80
1133.333333
3
C
250
230
220
260
240
260
220
40
333.3333333
3
D
340
270
300
320
307.5
340
270
70
891.6666667
3
E
250
240
270
290
262.5
290
240
50
491.6666667
3
Treatment Average
336
316
UDL
296
Avg
276
256
LDL
236
216
A
Treatment Number
Treatment Range
140
UCL
120
100
80
60
Avg
40
20
0
LCL
A
C
Treatment Number
80
Correlation Coefficients
The linear correlation coefficient, R, is a measure of the association between two
variables. The data to the right shows the lines picked per hour in a warehouse and
the overtime hours. The value of R will measure the degree of association
between these two variables. The maximum value for R is + 1. The minimum
value for R is - 1. In both these cases, all sample points fall on a straight line.
As R approaches +1 or -1, the stronger the correlation between x and y. The
square of this coefficient (R2) indicates the fraction of variation in y that is
associated with x.
To determine the correlation coefficient, select the data including the headings
(shaded area to the right) and then select the Correlation Coefficients options after
selecting the miscellaneous option (Misc) from the SPC toolbar. The dialog box
below appears.
Lines
Picked
599
658
699
738
791
685
656
570
614
684
749
608
653
650
671
606
648
758
712
611
671
651
Hours
Overtime
23.5
28.5
29
30.5
31.5
28
28
24.5
26
29.5
30
24.5
25.5
27
29
24
25.5
31
30
25.5
26
27
Lines Picked
Hours
Overtime
Lines Picked
1.00
Hours
Overtime
0.92
0.92
1.00
The value of R is 0.92 indicating a strong relationship between lines picked and overtime hours.
In most cases, you will have more than two variables that you are examining. The program works the
same for multi-variables and will generate a correlation matrix as shown below (for an example using 6
variables, A F.
A
B
C
D
E
F
A
1.00
0.14
0.37
-0.03
-0.15
-0.04
B
0.14
1.00
0.30
0.08
-0.34
0.19
C
0.37
0.30
1.00
-0.17
-0.35
-0.05
81
D
-0.03
0.08
-0.17
1.00
0.09
0.12
E
-0.15
-0.34
-0.35
0.09
1.00
0.13
F
-0.04
0.19
-0.05
0.12
0.13
1.00
Status
Detection
No.
What is the
What are the Potential Effects of
Potential Failure
the Failure Mode
Mode?
Occurrence
What is the
Process?
Date Prepared:
Severity
Prepared by:
RPN
Due
Date
Improved
RPN
82
Jan
Feb
Mar
Apr
May
June
July
Aug
Sept
Oct
Nov
Dec
Select the data including the headings (the shaded area to the
right) and then select the Box and Whisker plots option after
selecting the miscellaneous option (Misc) from the SPC
toolbar. You will see the dialog box below.
Select the range: The default range is the range that
is selected on the worksheet.
Treatment Data in Columns/Rows: The data can be in
columns or rows.
Box and Whisker Title and Labels
o Title: The title that will appear on the chart
o Y and Axis Label: The label that will appear on the
y axis.
Seattle,
WA
46
51
54
58
64
70
74
74
69
60
52
46
San
Antonio
TX
61
66
74
80
85
92
95
95
89
82
72
64
New
York, NY
38
40
50
61
72
80
85
84
76
65
54
43
Selecting OK will generate the Box and Whisker plot shown on the
next page. The graph below shows the values for each part of the
plot. The maximum value is presented by the top line (whisker).
Box and Whisker Chart
The top part of the box is the 75% quartile; the bottom part is the
25% quartile. The minimum is represented by the bottom line
100
(whisker).
90
Maximum
80
75% quartile
Subgroup Average
Median
60
30
Minimum
New York, NY
40
Seattle, WA
50
25% quartile
San Antonio TX
70
20
10
83
90
80
40
30
New York, NY
50
San Antonio TX
60
Seattle, WA
Subgroup Average
70
20
10
84
The sample size calculator is available under the MISC option on the
SPC toolbar. When you select this option, you will see the dialog
boxes to the right. The first page is for variables data. The second
page is for attributes data.
For variables data, enter the following:
Confidence interval
Standard deviation
Measurement error
Then select Calculate Sample Size, and the sample size needed
will appear in the bottom box.
For attributes data, enter the following:
Confidence interval
Estimated percentage of successes
Measurement error
Population size (if known)
Then select Calculate Sample Size, and the sample size needed
will appear in the bottom box.
85
6
6
5
5
Frequency
Female
Male
61-70
71-80
0
0-10
11-20
21-30
31-40
41-50
51-60
81-90
91-100
Rating
Class
0-10
11-20
21-30
31-40
41-50
51-60
61-70
71-80
81-90
91-100
Female
Male
1
3
1
0
2
4
3
3
3
4
1
6
0
1
5
4
2
2
3
0
86
Female
41
99
78
96
99
58
69
85
54
79
65
7
52
23
15
63
41
78
12
87
96
13
55
85
Male
18
89
56
78
88
45
51
65
45
60
47
4
36
14
12
45
15
55
11
76
87
12
45
64
87
2
Y
4.2
-2
-4
-6
time (sec)
The data for this chart must be in columns with the X variable in
the first column. The Y variables must be in adjacent columns.
The data used to generate the above chart is shown to the right.
When you select the Plot Multiple Y Variables Against One X
Variable from the Misc icon on the SPC toolbar, you will get
the form shown below. Each entry is discussed below.
The first entry is for the range containing the data (including the
headings as shown to the right). The default value is the range
selected on the worksheet. Enter the title for the chart as well as
the axis labels. You can enter the dates of data collection
(optional).
88
time
(sec)
Experimental
Intensity
(amps)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
2.2
2.4
2.6
2.8
3
3.2
3.4
3.6
3.8
4
4.2
6
5.8
5.8
5.6
5.5
5
4.7
4.5
4
3.4
2.9
2.3
1.7
1
0.5
0
-0.05
-1.25
-1.75
-2.25
-3
-3.5
Theoretical
Intensity
(amps)
6
5.967131372
5.868885604
5.706339098
5.481272746
5.196152423
4.854101966
4.458868953
4.014783638
3.526711514
3
2.440419858
1.854101966
1.247470145
0.62717078
-9.64723E-16
-0.62717078
-1.247470145
-1.854101966
-2.440419858
-3
-3.526711514
Select Cells
You can save some time with some of the tools in this program by selecting the appropriate cells prior to
selecting the icon on the SPC toolbar. For example, if you are making a p control chart, it is beneficial to
select the subgroup identifiers before selecting the attribute control chart icon from the SPC toolbar. This
can be a little cumbersome if you have lots of subgroups. The Select Cells (SC) on the SPC toolbar helps
make this easier.
Selecting One Cell Only
If you have the cursor in one cell only (as shown by the
shaded cell in the table to the right) and you select the SC
icon on the SPC toolbar, the program will select the cells
directly below the shaded cell (including the shaded cell)
as shown the table below.
Date
2/1/2003
2/2/2003
2/3/2003
2/4/2003
2/5/2003
2/6/2003
Number of
Telemarketing
Calls (n)
Number that
Result in an
Order (np)
40
63
47
52
34
59
Date
2/1/2003
2/2/2003
2/3/2003
2/4/2003
2/5/2003
2/6/2003
Number of
Telemarketing
Calls (n)
40
63
47
52
34
59
Number that
Result in an
Order (np)
5
10
12
7
3
21
5
10
12
7
3
21
Date
2/1/2003
2/2/2003
2/3/2003
2/4/2003
2/5/2003
2/6/2003
Number of
Telemarketing
Calls (n)
40
63
47
52
34
59
Date
2/1/2003
2/2/2003
2/3/2003
2/4/2003
2/5/2003
2/6/2003
Number that
Result in an
Order (np)
5
10
12
7
3
21
89
Number of
Telemarketing
Calls (n)
40
63
47
52
34
59
Number that
Result in an
Order (np)
5
10
12
7
3
21
90
How can I change the title or the x and y labels on an existing chart?
To change the chart title or labels on an existing chart, you must go select the Options icon on the SPC
toolbar and change the title or labels there. If you make the changes directly on the chart, they will not be
saved if the chart is updated.
.
Any other questions?
Please feel free to e-mail us with your questions or suggestions for improvement (bill@spcforexcel.com)
or visit our website (www.spcforexcel.com) for all the SPC information we have in our e-zines and
articles.
91