Example 42.1: Applying Tests for Special Causes

XCHART Statement

Example 42.1: Applying Tests for Special Causes

See SHWTEST in the SAS/QC Sample Library

This example illustrates how you can apply tests for special causes to make $\bar{X}$ charts more sensitive to special causes of variation.

The following statements create an $\bar{X}$ chart for the gap width measurements in the data set PARTS at "Creating Charts for Means from Subgroup Summary Data" and tabulate the results:

   title 'Tests for Special Causes Applied to Gap Width Data';
   symbol v=dot c=yellow;
   proc shewhart history=parts;
      xchart partgap*sample/ tests    = 1 to 5
                             ltests   = 20
                             tabletests
                             tablecentral
                             tablelegend
                             zonelabels
                             nolegend
                             cframe   = vigb
                             cinfill  = vlib
                             ctests   = red
                             cconnect = yellow;
   run;

The $\bar{X}$ chart is shown in Output 42.1.1 and the printed output is shown in Output 42.1.2. The TESTS= requests Tests 1, 2, 3, 4, and 5, which are described in Chapter 48, "Tests for Special Causes." The TABLECENTRAL option requests a table of the subgroup means, control limits, and central line. The TABLETESTS option adds a column indicating which subgroups tested positive for special causes, and the TABLELEGEND option adds a legend describing the tests that were signaled.

The ZONELABELS option displays zone lines and zone labels on the chart. The zones are used to define the tests. The LTESTS= option specifies the line type used to connect the points in test patterns that were signaled. The NOLEGEND option suppresses the subgroup sample size legend that is displayed by default in the lower left corner of the chart.

Output 42.1.1 and Output 42.1.2 indicate that Test 5 was positive at sample 14, signaling a possible shift in the mean of the process.

Output 42.1.1: Tests for Special Causes Displayed on an $\bar{X}$ Chart

Output 42.1.2: Tabular Form of $\bar{X}$ Chart

Tests for Special Causes Applied to Gap Width Data

The SHEWHART Procedure

Means Chart Summary for partgap
sample	Subgroup Sample Size	3 Sigma Limits with n=5 for Mean				Special Tests Signaled
sample	Subgroup Sample Size	Lower Limit	Subgroup Mean	Average Mean	Upper Limit	Special Tests Signaled
1	5	242.08741	270.00000	259.66667	277.24592
2	5	242.08741	258.00000	259.66667	277.24592
3	5	242.08741	248.00000	259.66667	277.24592
4	5	242.08741	260.00000	259.66667	277.24592
5	5	242.08741	273.00000	259.66667	277.24592
6	5	242.08741	260.00000	259.66667	277.24592
7	5	242.08741	259.00000	259.66667	277.24592
8	5	242.08741	248.00000	259.66667	277.24592
9	5	242.08741	260.00000	259.66667	277.24592
10	5	242.08741	255.00000	259.66667	277.24592
11	5	242.08741	268.00000	259.66667	277.24592
12	5	242.08741	253.00000	259.66667	277.24592
13	5	242.08741	273.00000	259.66667	277.24592
14	5	242.08741	275.00000	259.66667	277.24592	5
15	5	242.08741	257.00000	259.66667	277.24592
16	5	242.08741	269.00000	259.66667	277.24592
17	5	242.08741	249.00000	259.66667	277.24592
18	5	242.08741	264.00000	259.66667	277.24592
19	5	242.08741	258.00000	259.66667	277.24592
20	5	242.08741	248.00000	259.66667	277.24592
21	5	242.08741	248.00000	259.66667	277.24592

Test Descriptions
Test 5	Two out of three points in a row in Zone A or beyond

Chapter Contents
Previous
Next
Top