The MCHART statement creates a chart for subgroup medians, which is used to monitor the central tendency of a process.

You can use options in the MCHART statement to

• compute control limits from the data based on a multiple of the standard error of the plotted medians or as probability limits

• tabulate subgroup sample sizes, subgroup medians, control limits, and other information

• save control limits in an output data set

• save subgroup sample sizes and subgroup medians in an output data set

• read preestablished control limits from a data set

• apply tests for special causes (also known as runs tests and Western Electric rules)

• specify one of several methods for estimating the process standard deviation

• specify whether subgroup standard deviations or subgroup ranges are used to estimate the process standard deviation

• specify a known (standard) process mean and standard deviation for computing control limits

• create a secondary chart that displays a time trend removed from the data (see Displaying Trends in Process Data)

• display distinct sets of control limits for data from successive time phases

• add block legends and symbol markers to reveal stratification in process data

• superimpose stars at points to represent related multivariate factors

• clip extreme points to make the charts more readable

• display vertical and horizontal reference lines

• control axis values and labels

• control layout and appearance of the chart

You have three alternatives for producing medians charts with the MCHART statement:

• Traditional graphics are produced by default.

• ODS Graphics output is produced when you specify the ODS GRAPHICS statement prior to the PROC SHEWHART statement.

• Legacy line printer charts are produced when you specify the LINEPRINTER option in the PROC SHEWHART statement.

See Chapter 3, SAS/QC Graphics, for more information about producing these different kinds of graphs.

Note:When analyzing variables data, you should examine the variability of the process as well as the mean level. You can use the MRCHART statement in the SHEWHART procedure to monitor both the mean level and variability.