Академический Документы
Профессиональный Документы
Культура Документы
Use tests for special causes to determine which observations you may need to investigate and to
identify specific patterns in your data.
In This Topic
Test 2: Nine points in a row on the same side of the center line
Test 2 evaluates the pattern of variation for stability. If small shifts in the process are of concern, Test 2 can be used to
supplement Test 1 to create a control chart with greater sensitivity.
Test 5: Two out of three points more than 2 from the center line (same side)
Test 5 evaluates the pattern of variation for small shifts in the process.
Test 6: Four out of five points more than 1 from center line (same side)
Test 6 evaluates the pattern of variation for small shifts in the process.
Test 8: Eight points in a row more than 1 from center line (either side)
Test 8 detects a mixture pattern. A mixture pattern occurs when the points tend to avoid the center line and instead
fall near the control limits.
Which tests should I use to detect specific patterns of special-
cause variation?
Apply certain tests based on your knowledge of the process. If it is likely that your data might contain
particular patterns, you can look for them by choosing the appropriate test. Adding more tests makes
the chart more sensitive, but may also increase the chance of getting a false signal. When you use
several tests together, the chance of obtaining a signal for lack-of-control increases.
Variables charts
If you are not sure which tests apply in your specific situation, you might try using Tests 1, 2, and 7
when you first establish the control limits based on your data. After the control limits are established,
you should use the known values of those limits and Test 7 is no longer needed.
Test 1 (a point outside the control limits) detects a single out-of-control point.
Test 2 (9 points in a row on one side of the center line) detects a possible shift in the process.
Test 7 (too many points in a row within 1 standard deviation of the center line) detects control limits
that are too wide. Wide control limits are often caused by stratified data, which occur when you have a
systematic source of variation within each subgroup.
Attributes charts
If you are not sure which tests apply in your specific situation, you might try using Tests 1 and 2 when
you first establish the control limits based on your data. After the control limits are established, you
should use the known values of those limits.
Test 1 (a point outside the control limits) detects a single out-of-control point.
Test 2 (9 points in a row on one side of the center line) detects a possible shift in the process.
To detect high rates of an event on a G chart, Minitab also includes the Benneyan test. The
minimum data value for a G chart is 0. In most cases, the lower control limit for a G chart is also
0. Thus, in most cases, no points on a G chart can be below the lower control limit. The
Benneyan test fails if there are too many consecutive points that equal 0.
T charts
Experts recommend that you use both Test 1 and Test 2 when you create a T chart because the
T chart may be slow to detect small to moderate decreases in the average time between events.
Which tests are available with my control chart?
Tests 18 are available for most variables control charts. Note that only tests 14 are available for
R, S, and moving range charts.
On time-weighted control charts, Minitab only performs a test for points that go beyond the
control limits. The other seven tests assume that the points are independent. Because the plotted
points on time-weighted charts combine information from previous subgroups, the points are not
independent.
1. Choose Stat > Control Charts > Variables Charts for Subgroups > Xbar.
2. Complete the dialog box as usual.
3. Click Xbar Options, then click the Tests tab.
4. Select the following:
The control limits will be drawn at 2.5 instead of 3. Points will fail test 1 if they are more
than 2.5 sigma from the center line.
1. Choose Tools > Options > Control Charts and Quality Tools > Tests.
2. From the drop-down list, select Perform all tests for special causes.
Minitab will now perform all applicable tests when you create a control chart.
3. Next to 1 point more than K standard deviations from center line, enter 2.5.
The control limits for control charts will be drawn at 2.5 by default. Points will fail test 1 if
they are more than 2.5 from the center line.
NOTE
This procedure will not undo any changes that you have made in the Tests tab for specific graphs
in existing projects. For example, if you previously modified the parameters for an Xbar chart in
the current project, Minitab will remember your settings for that chart. The new defaults will not
be applied to Xbar charts in the current project.
1. 1 Point > 3.0 from the center line
Test 1 evaluates the pattern of variation for stability. Test 1 provides the strongest
evidence of lack of control. If small shifts in the process are of interest, Tests 2, 5, and 6
can be used to supplement Test 1 to create a control chart with greater sensitivity