The next area of cleanup
that we will tackle is excessive statistic labels. Every time you
display a statistic in a table, by default PROC TABULATE generates
a row or column heading for that statistic.
If you have two nested
row variables and then you add a statistic to the row dimension of
the table, you get row headings that are three items deep. Similarly,
if you have two nested column headings and an analysis variable, and
then you add a statistic, you will end up with column headings that
are six items deep!
In the example here, we have a table with a single row variable and two nested column variables. The statistic that we have opted to display is the mean of the analysis variable income. ...