Example 13.31 Applying Tests for Special Causes

[See SHWUEX1 in the SAS/QC Sample Library]This example illustrates how you can apply tests for special causes to make charts more sensitive to special causes of variation.

A textile company inspects rolls of fabric for defects. The rolls are one meter wide and 30 meters long. The following statements create a SAS data set named Fabric3, which contains the number of fabric defects for 20 rolls of fabric:

data Fabric3;
   input Roll Defects @@;
   datalines;
 1   6    2   9    3  14    4  17
 5   3    6   8    7   9    8   2
 9  14   10   1   11   3   12   5
13   6   14   9   15  10   16  12
17  11   18   4   19   9   20   4
;
run;

The following statements create a chart and tabulate the information on the chart. The chart and tables are shown in Output 13.31.1 and Output 13.31.2.

ods graphics on;
title1 'u Chart for Fabric Defects';
title2 'Tests=1 to 4';
proc shewhart data=Fabric3;
   uchart Defects*Roll / subgroupn = 30
                         tests     = 1 to 4
                         odstitle  = title
                         odstitle2 = title2
                         tabletests
                         zonelabels;
run;
ods graphics off;

The TESTS= option requests Tests 1, 2, 3, and 4, which are described in Tests for Special Causes Only Tests 1, 2, 3, and 4 are recommended for charts. The ZONELABELS option requests the zone lines, which are used to define the tests, and displays labels for the zones. The TABLETESTS option requests a table of the values of and the control limits, together with a column indicating the subgroups at which the tests are positive.

Output 13.31.1 and Output 13.31.2 indicate that Test 1 is positive for Roll 4 and Test 3 is positive at Roll 15.

Output 13.31.1 Tests for Special Causes Displayed on Chart

Output 13.31.2 Tabular Form of Chart

u Chart for Fabric Defects

Tests=1 to 4

The SHEWHART Procedure