The UNIVARIATE Procedure

Example 4.15 Creating a One-Way Comparative Histogram

This example illustrates how to create a comparative histogram. The effective channel length (in microns) is measured for 1225 field effect transistors. The channel lengths (Length) are stored in a data set named Channel, which is partially listed in Output 4.15.1:

Output 4.15.1: Partial Listing of Data Set Channel

The Data Set Channel

Lot	Length
Lot 1	0.91
.	.
Lot 1	1.17
Lot 2	1.47
.	.
Lot 2	1.39
Lot 3	2.04
.	.
Lot 3	1.91

The following statements request a histogram of Length ignoring the lot source:

title 'Histogram of Length Ignoring Lot Source';
ods graphics off;
proc univariate data=Channel noprint;
   histogram Length;
run;

The resulting histogram is shown in Output 4.15.2.

Output 4.15.2: Histogram for Length Ignoring Lot Source

To investigate whether the peaks (modes) in Output 4.15.2 are related to the lot source, you can create a comparative histogram by using Lot as a classification variable. The following statements create the histogram shown in Output 4.15.3:

title 'Comparative Analysis of Lot Source';
ods graphics off;
proc univariate data=Channel noprint;
   class Lot;
   histogram Length / nrows = 3;
run;

The CLASS statement requests comparisons for each level (distinct value) of the classification variable Lot. The HISTOGRAM statement requests a comparative histogram for the variable Length. The NROWS= option specifies the number of rows per panel in the comparative histogram. By default, comparative histograms are displayed in two rows per panel.

Output 4.15.3: Comparison by Lot Source

Output 4.15.3 reveals that the distributions of Length are similarly distributed except for shifts in mean.

A sample program for this example, uniex09.sas, is available in the SAS Sample Library for Base SAS software.