Descriptive Statistics using SAS

Advanced PROC UNIVARIATE

See www.stattutorials.com/SASDATA for files mentioned in this tutorial © TexaSoft, 2006

 

These SAS statistics tutorials briefly explain the use and interpretation of standard statistical analysis techniques for Medical, Pharmaceutical, Clinical Trials, Marketing or Scientific Research. The examples include how-to instructions for SAS Software.

 

 

Evaluating more than one category of a variable

 

Suppose you have several groups that you are comparing and you want to examine the distribution of the variable by group. The following example provides examples of how you could create histograms by RACE_CATEGORY using PROC UNIVARIATE. (PROCUNI2.SAS)

 

PROC UNIVARIATE DATA=SASDATA2.SBPDATA NOPRINT;

  CLASS RACE_CATEGORY;

  VAR SBP;

  HISTOGRAM /NORMAL (COLOR=RED W=5) NROWS=3;

RUN;

 

In this example data is from a trauma data set (SBPDATA extracted from the National Trauma Data set, 2004). The new statements used in this example include:

 

• NOPRINT – since we’re only interested in producing the graph, this option suppress other output

• CLASS RACE_CATEGORY -- This statement indicates that the data is to be examined for each category (classification) of the RACE_CATEGORY variable.

• ROWS=3 -- Since we know that there are three categories (BLACK, WHITE and OTHER), we add the option “NROWS=3” to the HISTOGRAM statement to indicate how many graphs to put on a singe page.

 

The following plot is created:

 

Notice that the three histograms are for the three values of RACE_CATEGORY which are BLACK,”“OTHER,” and “WHITE.” This graph is helpful in comparing the distribution of data in two or more groups. In this case, there is visual agreement that SBP is similarly distributed for all races.

 

Graph by two factors

 

Suppose you have two grouping variables and you want to produce a series of histograms to compare distributions.

 

The following program (PROCUNI3.SAS) produces a series of histograms by GENDER and WOUND type. Since this is a more detailed program the parts are annotated and described below:

 

uPROC FORMAT;

VALUE FMTWOUND 0="NONPENETRATE"

               1="PENETRATE";

RUN;

vTITLE 'HISTOGRAMS of SBP by GENDER and WOUND TYPE';

w PROC UNIVARIATE DATA=SASDATA2.SBPDATA NOPRINT;

  CLASS WOUND GENDER;

  VAR SBP;

  xHISTOGRAM / NROWS=2 NCOLS=2 CFILL=BLUE PFILL=M3N45;

  yINSET N='N:' (4.0) MIN='MIN:' (4.1) MAX='MAX:' (4.1)

               / NOFRAME POSITION=NE HEIGHT=2;

  FORMAT WOUND FMTWOUND.;

RUN;

 

 

u PROC FORMAT – this procedure creates a format for the WOUND variable to describe the coded 0,1 variables. Using this format allows you to display the groups in the graph by clearer category names (PENETRATE and NONPENETRATE) than by the cryptic 0 and 1. (See Chapter 3 for more information on PROC FORMAT.)

 

v TITLE statement – this places a title at the top of the graph. If you use other title statements such as TITLE2, the subsequent titles will be smaller by default than the first title (unless you change that in code.) (See chapter 3 for more information on titles.)

 

w CLASS statement – In this example there are two grouping variables indicated in the CLASS statement.

 

CLASS WOUND GENDER;

 

x HISTOGRAM STATEMENT -- The options within the HISTOGRAM statement define how the graph will appear. The columns and rows: The statements

 

NROWS=2 NCOLS=2

 

produce 2 histograms per row (for WOUND – first item in the CLASS statement) and 2 histograms for per COL (for GENDER or 2^nd item in the CLASS statement)

 

The histogram bar colors are specified by the CFILL (color fill) statement:

 

CFILL=BLUE

 

In this case, the bars will be blue. Some of the colors available in SAS (there are thousands to choose from) include

 

BLACK      WHITE      RED        GREEN      BLUE       PURPLE     

VIOLET     ORANGE     YELLOW     PINK       CYAN       MAGENTA    

BROWN      GOLD       LIME       GRAY       LILAC      MAROON     

SALMON     TAN        ROSE       CREAM      

The default color is black.

 

The pattern for the bars is specified by the PFILL (Pattern fill) statement

 

PFILL=M3N45

 

You can select from a number of available patterns. The default pattern is solid. Here are some of the other patterns you can select:

 

 

y INSET option – this defines an inset or key to the graph. This example illustrates several of the options:

 

INSET N='N:' (4.0) MIN='MIN:' (4.1) MAX='MAX:' (4.1)

               / NOFRAME POSITION=NE HEIGHT=2;

 

The statement

 

N='N:' (4.0) MIN='MIN:' (4.1) MAX='MAX:' (4.1)

 

defines which statistics will be included in the inset. In this case N (the sample size) will be designated with “N:” and will be displayed using the SAS output format 4.0. The MIN and MAX are similarly defined.

 

The remaining options

 

/ NOFRAME POSITION=NE HEIGHT=2;

 

specify that there be

• no frame around the inset

• that its position will be in the NE =  North-East corner of the graph

• and that the height of the characters will be set at 2 units.

 

When this SAS code is run, it produces the following graphs:

 

Exercise: Experiment with the colors, patterns and inset to see how they effect the graph.

1. Make the histogram color Green
2. Add the option MEAN='MEAN:' (4.1) to the inset option.
3. Add the NORMAL (COLOR=BROWN W=3)statement to superimpose a normal plot
4. How does this change the plot?

Exercise: Using the SBPDATA create the following histograms:

1. Create a matrix of histograms with RACE_CATEGORY (3 categories) using the pattern M3XO and CFILL=RED.
2. Place the key on the upper left corner (NW).
3. Add MEAN='MEAN:' (4.1) to the list of statistics reported.
4. Put your name in a TITLE2 statement.
5. Redo the plot using a solid blue bars.
6. Capture the output using ODS PDF and print the results.

The resulting graphs should look like this:

 

 

End of tutorial

See http://www.stattutorials.com/SAS

Get the SAS BeSmartNotes Quick Reference Guide

Order

| Send comments | Back to Tutorial Menu | TexaSoft |

© Copyright TexaSoft, 1996-2007

This site is not affiliated with SAS(r) or SAS Institute