PROC CAPABILITY and General Statements

Note: See Computing Summary Stats and Capability Indices in the SAS/QC Sample Library.

The fluid weights of 100 drink cans are measured in ounces. The filling process is assumed to be in statistical control. The measurements are saved in a SAS data set named Cans.

data Cans;
   label Weight = "Fluid Weight (ounces)";
   input Weight @@;
   datalines;
12.07  12.02  12.00  12.01  11.98  11.96  12.04  12.05  12.01  11.97
12.03  12.03  12.00  12.04  11.96  12.02  12.06  12.00  12.02  11.91
12.05  11.98  11.91  12.01  12.06  12.02  12.05  11.90  12.07  11.98
12.02  12.11  12.00  11.99  11.95  11.98  12.05  12.00  12.10  12.04
12.06  12.04  11.99  12.06  11.99  12.07  11.96  11.97  12.00  11.97
12.09  11.99  11.95  11.99  11.99  11.96  11.94  12.03  12.09  12.03
11.99  12.00  12.05  12.04  12.05  12.01  11.97  11.93  12.00  11.97
12.13  12.07  12.00  11.96  11.99  11.97  12.05  11.94  11.99  12.02
11.95  11.99  11.91  12.06  12.03  12.06  12.05  12.04  12.03  11.98
12.05  12.05  12.11  11.96  12.00  11.96  11.96  12.00  12.01  11.98
;

You can use the PROC CAPABILITY and VAR statements to compute summary statistics for the weights.

title 'Process Capability Analysis of Fluid Weight';
proc capability data=Cans normaltest;
   var Weight;
run;

The input data set is specified with the DATA= option. The NORMALTEST option requests tests for normality. The VAR statement specifies the variables to analyze. If you omit the VAR statement, all numeric variables in the input data set are analyzed.

The descriptive statistics for Weight are shown in Figure 5.1. For instance, the average weight (labeled Mean) is 12.0093. The Shapiro-Wilk test statistic labeled W is 0.987876, and the probability of a more extreme value of W (labeled Pr < W) is 0.499. Compared to the usual cutoff value of 0.05, this probability (referred to as a p-value) indicates that the weights are normally distributed.