The correct bibliographic citation for this manual is as follows: Cody, Ron. 1996. The SAS® Workbook.

Cary,
NC: SAS Institute Inc.
The SAS® Workbook
Copyright © 1996, SAS Institute Inc., Cary, NC, USA
ISBN 978-1-55544-757-1 (Hardcopy)
ISBN 978-1-62959-661-7 (PDF)
All rights reserved. Produced in the United States of America.
For a hard-copy book: No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, or otherwise, without the prior
written permission of the publisher, SAS Institute Inc.
For a web download or e-book: Your use of this publication shall be governed by the terms established by the
vendor at the time you acquire this publication.
The scanning, uploading, and distribution of this book via the Internet or any other means without the permission
of the publisher is illegal and punishable by law. Please purchase only authorized electronic editions and do not
participate in or encourage electronic piracy of copyrighted materials. Your support of others’ rights is
appreciated.
U.S. Government License Rights; Restricted Rights: The Software and its documentation is commercial
computer software developed at private expense and is provided with RESTRICTED RIGHTS to the United
States Government. Use, duplication or disclosure of the Software by the United States Government is subject
to the license terms of this Agreement pursuant to, as applicable, FAR 12.212, DFAR 227.7202-1(a), DFAR
227.7202-3(a) and DFAR 227.7202-4 and, to the extent required under U.S. federal law, the minimum restricted
rights as set out in FAR 52.227-19 (DEC 2007). If FAR 52.227-19 is applicable, this provision serves as notice
under clause (c) thereof and no other notice is required to be affixed to the Software or documentation. The
Government's rights in Software and documentation shall be only those set forth in this Agreement.
SAS Institute Inc., SAS Campus Drive, Cary, North Carolina 27513-2414.
January 2015
SAS provides a complete selection of books and electronic products to help customers use SAS® software
to its fullest potential. For more information about our offerings, visit sas.com/store/books or call
1-800-727-0025.
SAS® and all other SAS Institute Inc. product or service names are registered trademarks or trademarks of SAS
Institute Inc. in the USA and other countries. ® indicates USA registration.
Other brand and product names are trademarks of their respective companies.
 iii

Table of Contents

Preface . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .v

Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .vii

SECTION 1 • Data Step Programming

1 Inputting Raw Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3

2 Reading and Writing from External Files . . . . . . . . . . . . . . . . . . . . . .15

3 Using Logical Structures: IF-THEN/ELSE, WHERE,

and SELECT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17

4 Combining SAS® Data Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21

5 Using Numerical Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31

6 Using Character Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39

7 Working with Dates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45

8 Using Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49

9 Looking Back and Ahead Across Observations . . . . . . . . . . . . . . . . .57

10 Working with Longitudinal Data: Multiple Observations

per Subject . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .63

11 Writing Simple DATA Step Reports . . . . . . . . . . . . . . . . . . . . . . . . . .69

SECTION 2 • Procedures (Non-statistical Problems)

12 The FORMAT Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .75

13 The PRINT Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .83

14 The SORT Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .89

15 The FREQ Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .93

16 The MEANS Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .99

17 The TABULATE Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .103

18 The CHART Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .117

19 The PLOT Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .125

iv The SAS Workbook

SECTION 3 • SAS® Programming Techniques

20 Efficient Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .131

21 Longer and More Advanced Problems . . . . . . . . . . . . . . . . . . . . . . .141

SECTION 4 • Statistics Problems

22 Basic Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .147

23 Tests of Proportions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .151

24 Comparing Means: Two Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . .157

25 Analysis of Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .161

26 Parametric and Nonparametric Correlations . . . . . . . . . . . . . . . . .173

27 Simple, Multiple, and Logistic Regression . . . . . . . . . . . . . . . . . . . .177

28 Random Assignment and Random Selection of Subjects . . . . . . . .181

29 More Advanced Statistics Projects . . . . . . . . . . . . . . . . . . . . . . . . . .185

SECTION 5 • SAS® Puzzles to Test Your Skills

30 SAS® Puzzles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .197

SECTION 6 • Appendix

Appendix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .205

Index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .217
 v

Preface

The SAS Workbook and The SAS Workbook Solutions are collections of SAS programming
problems and solutions, respectively, grouped by topic. The problems should be useful for
individuals wanting to improve their skills, for classes in SAS programming, and for those
attending problem sessions associated with SAS workshops.
Although books, manuals, and workshops are all useful mechanisms for learning and improving
your SAS programming skills, one of the best ways to transfer this information into actual
programming skills is by writing lots and lots and lots and lots (OK, you get the idea) of programs.
The problems in this book have been carefully selected to include many of the same techniques
that are needed in typical SAS applications such as cleaning "dirty" data, using arrays, or working
with SAS dates. I include problems that I have encountered in my over 17 years of SAS
programming.
Each problem is organized with a Tools list that suggests SAS statements or procedures that
may be useful in solving the problem. Several references that you may find useful include
• SAS Procedures Guide, Version 6, Third Edition
• SAS Language: Reference , Version 6, First Edition
• SAS Programming by Example
• SAS Language and Procedures: Usage, Version 6, First Edition
• SAS Language and Procedures: Usage 2, Version 6, First Edition
• SAS Technical Report P-222, Changes and Enhancements to Base SAS Software, Release 6.07
All of these reference sources are available from SAS Institute in Cary, North Carolina.
Following the Tools list, each problem provides you with sample data or refers you to a data
file or a SAS data set described in the Appendix.
The problems in the Statistics Problems section are intended to be used by students in a basic or
intermediate level statistics course or by researchers wishing to improve their skills in solving
statistical problems. These problems have been kept separate so as not to frighten those who
use the SAS System for mostly non-statistical applications.
Once you have worked out, or at least attempted, a problem, it's time to look at the book's
solution. You may or may not have been able to solve the problem. If you did solve the problem,
your solution may differ. If not, you will benefit from seeing the solution now that you have
spent some time in trying to solve it yourself. Don't think that the book's solutions are necessarily
superior to yours. The SAS language has so many powerful programming tools that most
problems can be solved in many different ways.
vi The SAS Workbook

The solutions listed in The SAS Workbook Solutions assume that all external raw data files are
located in a subdirectory called WORKBOOK on the C: drive of a microcomputer. You may
substitute the appropriate file description for the operating environment you are using. Descriptions
of these files, listings of their contents, and a program to create the SAS data set CLINICAL
are located in the Appendix.
You may wonder why there is a separate solutions book. It is the nature of SAS programming
(actually most computer programming) that once you are shown the solution to a problem, the
learning potential of the problem is lost. You think that the solution is obvious once you see it,
even though it would have taken considerable effort for you to have written it yourself. And it
is through this effort that you become a better programmer.
I therefore removed the temptation for students to flip to the back of the book and take a quick
peek at the solutions. Instructors may also feel more comfortable in assigning problems without
the solutions too close at hand.
The SAS Workbook Solutions contains one or more solutions to each of the problems. Some of
the lines are commented, but most of the solutions do not contain extensive explanations. The
two books together should be useful as a reference tool as well.
I hope you improve your programming skills and have fun in the process.
RPC
Spring 1996
 vii

Acknowledgments

This is the fun part. The book is almost finished and I get to thank all the people who helped. Since
this is my second book published by SAS Institute, I must enjoy working with them--and I do.
I can’t say enough good things about the publications group at SAS Institute. They are terrific.
They maintain a high level of professionalism yet are gentle and kind with authors like me who
make lots of silly mistakes.
I wish to thank all the folks at SAS Institute who helped transform this work from manuscript
to book in record time. As always, special thanks to David Baggett for his encouragement and
good cheer. Although I don’t know all the names, I wish to thank the reviewers for their
constructive suggestions and for catching a number of programming errors. Caroline Brickley
deserves special credit for coordinating the first edit pass. Thanks to Joan M. Stout for her very
careful detail editing and for finding some very subtle mistakes. I have not spoken with all the
production staff but there are numerous tasks that go on “behind the scenes” that are vital.
Therefore, my thanks to the production specialists Blanche W. Phillips, Denise T. Jones, and
Candy R. Farrell. Also, for the creative artist who designed the cover, thanks to Ellen Hood for
a great job.
Finally, thanks to my wife, Jan, and my two boys, Russ and Preston, for their support and
encouragement.

Ron Cody
Spring, 1996
viii
 1

SECTION 1 DATA Step Programming

CHAPTERS
1 Inputting Raw Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3

2 Reading and Writing from External Files . . . . . . . . . . . . . . . . . . . . . .15

3 Using Logical Structures: IF-THEN/ELSE, WHERE,

and SELECT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .17

4 Combining SAS Data Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21

5 Using Numerical Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .31

6 Using Character Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .39

7 Working with Dates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45

8 Using Arrays . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49

9 Looking Back and Ahead Across Observations . . . . . . . . . . . . . . . . .57

10 Working with Longitudinal Data: Multiple Observations

per Subject . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .63

11 Writing Simple DATA Step Reports . . . . . . . . . . . . . . . . . . . . . . . . . .69

2
 3

CHAPTER 1 Inputting Raw Data

PROBLEMS
1 Reading Data Values Separated by Spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4
2 Reading Data Values, Including Missing Values, Separated by Spaces . . . . . . . . . . . . . .4
3 Reading Data Values Separated by Spaces with a Character Value Greater than
8 Bytes Long . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5
4 Reading Data Values Where Missing Values Are Represented by Periods except at
the End of Short Data Lines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .5
5 Reading Data Values Separated by Commas Where Missing Values Are Represented
by Two Adjacent Commas and Some of the Character Values Are Placed in
Double Quotes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6
6 Reading Data Values Using Starting and Ending Column Numbers . . . . . . . . . . . . . . . .7
7 Reading Data Values with Missing Character and Numeric Values Using Starting
and Ending Column Numbers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7
8 Reading Data Values Using Pointers and Informats . . . . . . . . . . . . . . . . . . . . . . . . . . . .8
9 Reading Mixed Record Types in One DATA Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8
10 Creating a Single Observation from More than One Line of Raw Data . . . . . . . . . . .10
11 Creating More than One Observation from a Single Data Line . . . . . . . . . . . . . . . . . .11
12 Creating More than One Observation from a Single Data Line (Two Variables) . . . .11
13 Creating More than One Observation from a Single Data Line (Two Variables:
One Character, One Numeric) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
14 Reading “Free-form” Data, Creating More than One Observation from a Single
Data Line . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .12
15 Using Variable Lists and Informat Lists to Make the INPUT Statement More
Compact . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13
16 Using Variable Lists, Informat Lists, and Relative Column Pointers to Read
Data Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .13

INTRODUCTION
This chapter contains problems on reading raw data with an INPUT statement. All the problems
read “instream” data, that is, data that you submit as part of your program. However, all the
techniques used here can also be applied to reading data from external files (discussed in Chapter
2). For those “old-timers” who like to use a CARDS statement, you may still do so. The newer
term DATALINES is equivalent and is used throughout this book.
4 The SAS Workbook

PROBLEM 1
Reading Data Values Separated by Spaces

Tools
INPUT statement
PROC PRINT

Data
You have collected some data on a group of students. The data values, separated by one or more
spaces, represent the variables F_Name (first name), ID, Gender, GPA (grade point average),
Height and Weight.
Hector 123 M 3.5 59 155
Nancy 328 F 3.7 52 99
Edward 747 M 2.4 62 205
Michelle 778 F 3.0 54 115
Sampson 289 M 3.5 60 180

Directions
Create a temporary SAS data set called CLASS from these lines of data. Include the lines of
data “instream” in the program. Display the contents of this data set.

Notes
1. All the first names are 8 characters or less.
2. ID is to be stored as character data.

PROBLEM 2
Reading Data Values, Including Missing Values, Separated by Spaces

Tools
INPUT statement

Data
Raw data similar to Problem 1 with some missing values
Hector 123 M 3.5 . 155
. 328 F 3.7 52 99
Edward 747 . . . .
Michelle 778 F 3.0 54 .
Sampson 289 M 3.5 60 180
 CHAPTER 1 Inputting Raw Data 5

Directions
Repeat Problem 1 with the modified data. Note that there are several missing values (both for
numeric and character variables) which are represented by periods.

PROBLEM 3
Reading Data Values Separated by Spaces with a Character Value Greater than 8 Bytes
Long

Tools
INPUT statement
Colon format modifier (:) or
LENGTH statement or
INFORMAT statement

Data
Raw data similar to Problem 1 with one name greater than 8 characters in length
Hector 123 M 3.5 59 155
Nancy 328 F 3.7 52 99
Edward 747 M 2.4 62 205
Michelle 778 F 3.0 54 115
Washington 289 M 3.5 60 180

Directions
Repeat Problem 1 with the new data. Note that one of the names is now longer than 8 characters
long.

PROBLEM 4
Reading Data Values Where Missing Values Are Represented by Periods except at the
End of Short Data Lines

Tools
INFILE statement
MISSOVER option
INPUT statement
6 The SAS Workbook

Data
Raw data similar to Problem 1 with some missing values and short records
Hector 123 M 3.5 . 155
. 328 F 3.7 52 99
Edward 747
Michelle 778 F 3.0 54
Sampson 289 M 3.5 60 180

Directions
Repeat Problem 1 with the new data. Note that there are several missing values (both for numeric
and character variables) and there are some short records.

PROBLEM 5
Reading Data Values Separated by Commas Where Missing Values Are Represented by
Two Adjacent Commas and Some of the Character Values Are Placed in Double Quotes

Tools
INFILE statement
DSD option
INPUT statement

Data
Raw data similar to Problem 1 with comma delimiters
Hector,123,M,3.5,,155
,328,"F",3.7,52,99
"Edward",747,,,,,
Michelle,778,F,3.0,54,,
Sampson,289,M,3.5,60,180

Directions
Repeat Problem 1 using comma-delimited data. Note that some of the character values are
enclosed in double quotes and that consecutive commas (without any space between) represent
missing values.
 CHAPTER 1 Inputting Raw Data 7

PROBLEM 6
Reading Data Values Using Starting and Ending Column Numbers

Tools
INPUT statement, with column numbers

Data
Raw data similar to Problem 1 lined up in columns
1 2 3 4 5
123456789012345678901234567890123456789012345678901
---------------------------------------------------
Hector 123 M 3.5 59 155
Nancy 328 F 3.7 52 99
Edward 747 M 2.4 62 205
Michelle 778 F 3.0 54 115
Sampson 289 M 3.5 60 180

Directions
Use the same data as in Problem 1, but this time write the INPUT statement using starting and
ending column numbers. The data are listed with column numbers displayed.

PROBLEM 7
Reading Data Values with Missing Character and Numeric Values Using Starting and
Ending Column Numbers

Tools
INFILE statement
PAD option
INPUT statement, with column numbers
Data
Raw data similar to Problem 1 lined up in columns with some missing values and short records
1 2 3 4 5
123456789012345678901234567890123456789012345678901
---------------------------------------------------
Hector 123 M 3.5 155
328 F 3.7 52 99
Edward 747
Michelle 778 F 3.0 54
Sampson 289 M 3.5 60 180
8 The SAS Workbook

Directions
Repeat Problem 6 with the modified data. Note that there are now blanks representing both
character and numeric missing values. Notice that lines 3 and 4 are short records that are not
padded with blanks.

PROBLEM 8
Reading Data Values Using Pointers and Informats

Tools
INPUT statement, with column pointers (@) and informats

Data
The same raw data as in Problem 6

Directions
Instead of using starting and ending column numbers for the data values (as in Problem 6), use
column pointers and informats to read the same data values and create a SAS data set called
CLASS. Use the data description that follows to help you write the INPUT statement:
Variable Starting Column Length Type
-----------------------------------------------
F_NAME 1 8 Char
ID 13 3 Char
GENDER 22 1 Char
GPA 31 3 Numeric
HEIGHT 39 2 Numeric
WEIGHT 49 3 Numeric

PROBLEM 9
Reading Mixed Record Types in One DATA Step

Tools
INPUT statement, single trailing at sign (@)
 CHAPTER 1 Inputting Raw Data 9

Data
Survey data as shown
1 2
12345678901234567890
--------------------
001MRY 3 1994
00923FDY 1 1995
012FDN 2 1994
00518MRN 2 1995
003MDY 4 1994

Directions
You have data from a survey administered in 1994 and 1995. In 1995, it was decided to record
the age of each person (which was not done in 1994). Age information was not added at the end
of the raw data. Instead, it was placed right after the subject ID, creating two completely different
data layouts for the two years. Fortunately, the year the survey was administered was also
entered into the data file. Create a temporary SAS data set called SURVEY which correctly
reads these mixed records. The data description follows:

1994 Data Description

Starting Ending Format

Variable Description Column Column
------------------------------------------------------------
ID Subject ID 1 3 Char
GENDER Subject Gender 4 4 Char
PARTY Political Party 5 5 Char
VOTE Did you vote in
the last election? 6 6 Char
NUM_TV Number of TV's 7 8 Numeric
YEAR Survey year 15 18 Char

1995 Data Description

Starting Ending Format

Variable Description Column Column
------------------------------------------------------------
ID Subject ID 1 3 Char
AGE Subject's age 4 5 Numeric
GENDER Subject Gender 6 6 Char
PARTY Political Party 7 7 Char
VOTE Did you vote in
the last election? 8 8 Char
NUM_TV Number of TV's 9 10 Numeric
YEAR Survey year 15 18 Char
10 The SAS Workbook

PROBLEM 10
Creating a Single Observation from More than One Line of Raw Data

Tools
INPUT statement, line and column pointers (# and @), informats

Data
Car survey data as follows
1 2 3 4 5
12345678901234567890123456789012345678901234567890
--------------------------------------------------
12310/21/46 NJ 08822
123 2 Ford Oldsmobile
23711/01/55 NY 11518
237 1 Chevy

Directions
A survey was conducted and the data were recorded using two lines of data for each subject.
Use the data description and sample lines of data to create a SAS data set called SURVEY, with
one observation per subject. The file description follows:
Starting Ending
Variable Description Column Column Type
---------------------------------------------------------
Line 1 SUBJECT Subject number 1 3 Char
DOB Date of Birth 4 11 MM/DD/YY
STATE State where living 25 26 Char
ZIP_CODE Zip Code 40 44 Character

Line 2 SUBJECT Subject number 1 3 Char

NUMBER Number of cars 5 5 Numeric
CAR1 Make of Car 1 11 20 Char
CAR2 Make of Car 2 21 30 Char

Notes
1. You may use either pointers and informats for all of your variables, or you may use column
specifications for all of your variables except for DOB for which you will need either a
pointer and an informat or an INFORMAT statement.
2. For this problem, do not read the SUBJECT value in line 2 of the data. In a more sophisticated
program, you might want to read this as a different variable and check that it is the same as
the subject number in line 1.
 CHAPTER 1 Inputting Raw Data 11

PROBLEM 11
Creating More than One Observation from a Single Data Line

Tools
INPUT statement, double trailing at sign (@@)

Data
Twenty temperatures as shown
21 23 29 33 19 28
33 39 43 44 28 21 24 27 29
37 32 31 33 29

Directions
A researcher collected 20 temperatures (one for each day) and entered them on several lines as
shown in the data section.
As you can see, there are several temperatures per line (not always the same number) and there
are one or more spaces between each number. Create a temporary SAS data set called TEMPER
from these data. Print out the contents of this data set.

PROBLEM 12
Creating More than One Observation from a Single Data Line (Two Variables)

Tools
INPUT statement, double trailing at sign (@@)

Data
Twenty pairs of days of the month and temperatures as shown
5 21 6 23 7 29 8 33 9 19 10 28
11 33 12 39 13 43 14 44 15 28 16 21 17 24 18 27 19 29
20 37 21 32 22 31 23 33 24 29

Directions
This time the researcher recorded the day of the month along with each of the temperatures and
entered them in pairs (day temperature). Create a temporary SAS data set called TEMP_DAY
containing two variables (DAY and TEMP), using these data. List the contents of this data set.
12 The SAS Workbook

PROBLEM 13
Creating More than One Observation from a Single Data Line (Two Variables: One
Character, One Numeric)

Tools
INPUT statement, double trailing at sign (@@)

Data
Rat data (GROUP and WEIGHT) in pairs with a varying number of pairs per line as shown
A 34 B 58 A 28 C 55
C 56 A 27 B 52 C 58 A 21 B 62

Directions
A researcher treated three groups of rats (Groups A, B, and C) and recorded the weight of each
rat after one week. The data were arranged with each GROUP and WEIGHT in pairs, with a
varying number of pairs on each line.
Write a SAS DATA step to read these data and create a temporary data set called RATS. Print
out the contents of this data set.

PROBLEM 14
Reading “Free-form” Data, Creating More than One Observation from a Single Data
Line

Tools
RETAIN statement
INPUT statement, double trailing at sign (@@)

Data
Rat data similar to Problem 13
A 34 28 B 58 52
62 C 55 A 27 21
C 56 58

Directions
The same data values as found in Problem 13 were entered differently. This time, the researcher
entered a GROUP value (A, B, or C) followed by one or more weights as shown.
Write a SAS DATA step to read these data. Print out the contents of the data set. Again, call
the data set RATS.
 CHAPTER 1 Inputting Raw Data 13

Hint
Look carefully at the tools needed for this problem.

PROBLEM 15
Using Variable Lists and Informat Lists to Make the INPUT Statement More Compact

Tools
INPUT statement, variable lists, and informat lists

Data
Instream data lines in the program to be rewritten

Directions
Rewrite the DATA step below, substituting an INPUT statement which uses a variable list and
an informat list to make it more compact. Print out the contents of the data set.
DATA VARLIST;
INPUT @1 Q1 2. @3 Q2 2. @5 Q3 2. @7 Q4 2. @9 Q5 2.
@15 DATE1 MMDDYY8.
@23 DATE2 MMDDYY8.
@31 DATE3 MMDDYY8.
@50 X1 $1. @51 X2 $1. @52 X3 $1.
@53 Y1 $1. @54 Y2 $1. @55 Y3 $1.;
DATALINES;
1122334455 10/21/4611/13/4206/05/48 123456
9672347656 01/01/9501/02/9501/03/95 987654
;

PROBLEM 16
Using Variable Lists, Informat Lists, and Relative Column Pointers to Read Data Values

Tools
INPUT statement, variable lists, and informat lists; relative column pointers (+)

Data
Instream data lines in the program to be rewritten

Directions
Rewrite the DATA step below, substituting an INPUT statement which uses a variable list, an
informat list, and relative column pointers (+). Print out the contents of the data set.
14 The SAS Workbook

Hint
Read all the X’s first, then the Y’s, and finally, the Z’s.
DATA POINTER;
INPUT @1 X1 2.
@3 Y1 2.
@5 Z1 $3.
@8 X2 2.
@10 Y2 2.
@12 Z2 $3.
@15 X3 2.
@17 Y3 2.
@19 Z3 $3.;
DATALINES;
0102AAA0304BBB0506CCC
2837ABC9676DEF8765GHI
;
 15
Reading and Writing from
CHAPTER 2 External Files

PROBLEMS
1 Reading Raw Data from an External File . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .15
2 Reading More than One Raw Data File to Create a Single SAS Data Set . . . . . . . . . .16
3 Reading More than One Raw Data File Using the FILEVAR= Option to
Create a Single SAS Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .16

INTRODUCTION
Problems in this chapter ask you to read data from external files. Since the specification for
these files may vary from one operating environment to another, feel free to specify the location
of the files relevant to your operating environment. The solutions listed in The SAS Workbook
Solutions assume that all external raw data files are located in a subdirectory called WORKBOOK
on the C: drive of a microcomputer. You may substitute the appropriate file description for the
operating environment you are using.
Descriptions of these files, listings of their contents, and a program to create the SAS data set
CLINICAL are located in the Appendix.

PROBLEM 1
Reading Raw Data from an External File

Tools
INFILE statement
PAD option (optional)
FILENAME statement (optional)

Data
Raw data file CARS.DTA stored in the subdirectory WORKBOOK on your C: drive

Directions
Write a program that reads this external file and creates a temporary SAS data set called CARS.
List the contents of this data set.
16 The SAS Workbook

PROBLEM 2
Reading More than One Raw Data File to Create a Single SAS Data Set

Tools
INFILE statement
END= option

Data
Two raw data files DEMOG1.DTA and DEMOG2.DTA

Directions
Create a single temporary SAS data set called DEMOG_12 that includes all the data from both
of the raw data files.

PROBLEM 3
Reading More than One Raw Data File Using the FILEVAR= Option to Create a Single
SAS Data Set

Tools
INFILE statement
FILEVAR= and END= options

Data
The same as Problem 2

Directions
Repeat Problem 2 using the INFILE option FILEVAR= and read the two file names,
DEMOG1.DTA and DEMOG2.DTA, from instream data in the DATA step.
 17
Using Logical Structures:
CHAPTER 3 IF-THEN/ELSE,WHERE, and SELECT

PROBLEMS
1 Recoding Age Values Using IF-THEN/ELSE Statements . . . . . . . . . . . . . . . . . . . . . . . . .17
2 Using Logical Statements to Compute a New Data Set Variable . . . . . . . . . . . . . . . . .18
3 Using the SELECT Statement to Compute a New Data Set Variable . . . . . . . . . . . . .19
4 Creating a New Variable Based on Logical Combinations of Existing Data . . . . . . . .19

INTRODUCTION

Problems in this chapter require you to use logical statements in the DATA step to create new
variables and to group values of an existing variable to create new, recoded variables. Since
there are several ways to solve each of these problems, use the Tools list to guide you in
selecting appropriate logical statements.

PROBLEM 1
Recoding Age Values Using IF-THEN/ELSE Statements

Tools
IF-THEN/ELSE statements

Data
Raw data file DEMOG1.DTA

Directions
Using the DEMOG1.DTA data file, create a SAS data set called RECODE that includes the
subject’s age as of January 1, 1996 (rounded to the nearest year) and a variable called
AGEGROUP, created with IF-THEN/ELSE statements in the DATA step, with age groups
defined as follows:
AGEGROUP Description
-----------------------------------------------
1 Between 0 and 20 inclusive
2 Between 21 and 40 inclusive
3 Between 41 and 60 inclusive
4 Greater than 60
18 The SAS Workbook

Hint
Use the statement
AGE = ROUND (('01JAN96'D - DOB)/365.25);

to compute age. This is an approximation and may be incorrect on a person’s birthday, depending
on whether the current year is a leap year and on how many leap years have occurred between
the date of birth and the current date.

PROBLEM 2
Using Logical Statements to Compute a New Data Set Variable

Tools
IF-THEN/ELSE statements
Boolean AND, OR operators

Data
Raw data file CARS.DTA

Directions
A consumer magazine wants to create an index of cars based on the data in the file CARS.DTA.
This index is a function of the car size, gas mileage (MILEAGE), and reliability (RELIABLE).
The variable SIZE has values SMALL, COMPACT, and MID-SIZED. The rules are as follows:
For SMALL Cars:

Mileage Reliability Index

---------------------------------------
0 to 20 1 to 3 1
0 to 20 4 to 5 2
21 to 50 1 to 3 3
21 to 50 4 to 5 4

For COMPACT Cars:

Mileage Reliability Index

---------------------------------------
0 to 15 1 to 3 1
0 to 15 4 to 5 2
16 to 50 1 to 3 3
16 to 50 4 to 5 4
 CHAPTER 3 Using Logical Structures: IF-THEN/ELSE, WHERE, and SELECT 19

For MID-SIZED Cars:

Mileage Reliability Index

---------------------------------------
0 to 12 1 to 3 1
0 to 12 4 to 5 2
13 to 50 1 to 3 3
13 to 50 4 to 5 4

Write a SAS DATA step to create a SAS data set called CAR_INX that contains all the variables
described in the CARS.DTA file plus the INDEX variable. Be sure to set INDEX to a missing
value if either a mileage or reliability value is missing.

PROBLEM 3
Using the SELECT Statement to Compute a New Data Set Variable

Tools
SELECT statement

Data
Raw data file CARS.DTA

Directions
Solve Problem 2 using a SELECT statement to test the car size. You may also use it for the
other logical tests.

PROBLEM 4
Creating a New Variable Based on Logical Combinations of Existing Data

Tools
IF-THEN/ELSE statements
Boolean AND, OR operators

Data
SAS data set CLINICAL (created by running the program CLINICAL.SAS in the Appendix)
20 The SAS Workbook

Directions
An organization has created standards of blood pressure for men and women of certain ages.
It considers a person to be hypertensive if his or her diastolic blood pressure (DBP) or systolic
blood pressure (SBP) is above a critical value. The critical values for DBP and SBP are determined
by gender and age range as follows:
Gender Age Range DBP SBP
-----------------------------------
Male 0 - 30 88 152
Male 31 - 65 92 162
Male 66 + 94 166
Female 0 - 30 86 150
Female 31 - 65 88 158
Female 66 + 92 164

Using the SAS data set CLINICAL and the criteria for hypertension described here, create a
new data set called HYPERTEN that contains the variables GENDER, DBP, and SBP from
data set CLINICAL, the patient’s age at the time of the blood pressure test (drop any fractional
part of a year), and a variable HYPER (hypertension) which has values of Y or N depending on
the criteria listed above. If either the DBP or SBP is in the hypertensive range, set HYPER = Y
(even if the other blood pressure value is missing); if one of the values is missing and the other
is in the normal range, assign a missing value to the variable HYPER; if both DBP and SBP are
missing, assign a missing value to the variable HYPER.
You can use the SAS statement
AGE = INT((VISIT - DOB) / 365.25);

to compute the patient’s approximate age at the time of his or her blood pressure test.
 21

CHAPTER 4 Combining SAS Data Sets ®

PROBLEMS
1 Combining Similar Data from Two Different SAS Data Sets . . . . . . . . . . . . . . . . . . . .22
2 Taking a Subset of Observations from a SAS Data Set . . . . . . . . . . . . . . . . . . . . . . . . .22
3 Creating Multiple SAS Data Sets from a Single SAS Data Set . . . . . . . . . . . . . . . . . . .23
4 Selecting Observations from a SAS Data Set Based on Multiple Conditions . . . . . . .23
5 Selecting Observations Using Wildcards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
6 Adding Variables from Two SAS Data Sets Based on an Identifying
Variable (Match Merging) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24
7 Adding Summary Information to Each Observation in a
SAS Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25
8 Adding the Information in One SAS Data Set (Containing One Observation)
to Every Observation in Another SAS Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26
9 Updating Values in a SAS Data Set . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26
10 Combining Variables from Two SAS Data Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .27
11 Merging SAS Data Sets with Multiple BY Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . .27
12 Merging Data from Multiple SAS Data Sets and Selecting Observations . . . . . . . . . .28
13 Using the MERGE Statement to Perform a Table Lookup . . . . . . . . . . . . . . . . . . . . . .29

INTRODUCTION
The information you need for a report or analysis may be contained in more than one raw data
file or SAS data set. For example, an employee file may contain some fixed information such
as name, social security number, and date of birth. Another file may contain variable information
such as the number of sales calls and the dollar amount for each month. A typical report might
require you to combine information from these files.
This chapter contains problems related to combining SAS data sets in various ways. The SET,
MERGE, and UPDATE statements are all used in this chapter.
22 The SAS Workbook

PROBLEM 1
Combining Similar Data from Two Different SAS Data Sets

Tools
SET statement

Data
The SAS data set PAY1995 contains payroll data for the year 1995. The variables are ID,
LEVEL, SALARY, and GENDER. Data set PAY1996 contains identical information on some
new employees. Run the program below to create the two SAS data sets PAY1995 and PAY1996:
DATA PAY1995;
LENGTH ID $ 3 GENDER $ 1;
INPUT ID LEVEL SALARY GENDER;
DATALINES;
A23 32 68000 M
A24 35 75000 F
A30 44 97000 M
A13 28 27000 F
;
DATA PAY1996;
LENGTH ID $ 3 GENDER $ 1;
INPUT ID LEVEL SALARY GENDER;
DATALINES;
A25 29 35000 F
A26 36 88000 F
;

Directions
Create a new SAS data set PAY95_96 which contains all the observations from PAY1995 and
PAY1996.

PROBLEM 2
Taking a Subset of Observations from a SAS Data Set

Tools
SET statement
Subsetting IF statement or WHERE statement

Data
The SAS data set PAY1995 from Problem 1
 CHAPTER 4 Combining SAS Data Sets 23

Directions
From the observations in the SAS data set PAY1995, create a new data set called FEM_1995,
containing information on only female employees.

PROBLEM 3
Creating Multiple SAS Data Sets from a Single SAS Data Set

Tools
DATA statement (more than one data set name)
SET, Subsetting IF, and OUTPUT statements

Data
The SAS data set PAY1995 from Problem 1

Directions
From the SAS data set PAY1995, create two SAS data sets, one containing data for females
(FEM_1995), the other for males (MAL_1995). Accomplish this in a single DATA step.

PROBLEM 4
Selecting Observations from a SAS Data Set Based on Multiple Conditions

Tools
SET statement
WHERE statement
WHERE= data set option
Date literal (’01JAN70’D)

Data
The SAS data set CLINICAL created by running the program CLINICAL.SAS in the Appendix

Directions
Starting with the SAS data set CLINICAL, create a new data set called OLD_HYP that contains
data on all patients born before January 1, 1970 and considered to be hypertensive. For this
problem, hypertensive patients are defined as those who have a systolic blood pressure (SBP)
greater than 140 or whose diastolic blood pressure (DBP) is greater than 90. Create this data
set two ways: using a WHERE statement and using the WHERE= data set option.
24 The SAS Workbook

PROBLEM 5
Selecting Observations Using Wildcards

Tools
WHERE statement
LIKE operator
Wildcard characters percent sign (%) and underscore (_)

Data
The SAS data set CARS created from the raw data file CARS.DTA

Directions
From the SAS data set CARS, create the following data sets:
A. A data set of all car models which have names beginning with the letter C. Call this data set
BEGIN_C.
B. A data set of all car models which have names beginning with the letter C and which have five
or fewer characters in the model name. Call this data set BEGIN_C5.

PROBLEM 6
Adding Variables from Two SAS Data Sets Based on an Identifying Variable (Match
Merging)

Tools
MERGE statement
PROC SORT

Data
The SAS data set PAY1995 (from Problem 1) and the SAS data set MER1995 created by running
the program below:
DATA MER1995;
LENGTH ID $ 3 MERIT $ 1;
INPUT ID MERIT;
DATALINES;
A28 Y
A23 Y
A24 N
;
 CHAPTER 4 Combining SAS Data Sets 25

Directions
Using the SAS data set PAY1995, add merit information, saved in SAS data set MER1995, to
each employee. Call the new SAS data set PAYM1995.

Note
There is no merit information for employee A30 in data set MER1995 and the observations are
not in ID order.

PROBLEM 7
Adding Summary Information to Each Observation in a SAS Data Set

Tools
MERGE statement
PROC MEANS
KEEP= data set option

Data
The SAS data set PAY1995 from Problem 1

Directions
Starting from the SAS data set PAY1995, create a new SAS data set called SAL_PER, containing
a new variable, PERCENT. For male employees, PERCENT is computed as the yearly salary
(SALARY) expressed as a percentage of the mean salary for all males. For female employees,
PERCENT is computed as the yearly salary expressed as a percentage of the mean salary for
all females. First, to compute the male and female mean salary and to output this information
to a SAS data set (MEAN_SAL), run the program below. The SAS data set MEAN_SAL will
contain variables GENDER and M_SALARY.
PROC MEANS DATA=PAY1995 NWAY NOPRINT;
CLASS GENDER;
VAR SALARY;
OUTPUT OUT=MEAN_SAL (KEEP=GENDER M_SALARY)
MEAN=M_SALARY;
RUN;

You need to combine the information from MEAN_SAL (two observations) with the information
in PAY1995. Then create a new variable (PERCENT) which is 100*SALARY/M_SALARY.
26 The SAS Workbook

PROBLEM 8
Adding the Information in One SAS Data Set (Containing One Observation) to Every
Observation in Another SAS Data Set

Tools
Conditional SET statement
_N_ automatic variable
PROC MEANS

Data
The SAS data set PAY1995 from Problem 1

Directions
Although this problem is similar to Problem 7, the solution is quite different. This time, you
compute the mean salary for all employees from the SAS data set PAY1995 and use this number
to compute all salaries as a percentage of this mean. Your first step will be to compute the mean
salary of all employees (call it GRAND) by running the program below:
PROC MEANS DATA=PAY1995 NOPRINT;
VAR SALARY;
OUTPUT OUT=MEAN (KEEP=GRAND)
MEAN=GRAND;
RUN;

PROBLEM 9
Updating Values in a SAS Data Set

Tools
PROC SORT
UPDATE statement

Data
The SAS data set PAY1995 from Problem 1

Directions
You have some new values for the employees in data set PAY1995. Employee A23 has a new
salary of 72,000; employee A24 is now level 36 with a salary of 77,000; employee A13 was
mistakenly entered as a female (F) instead of male (M). Create a SAS data set with this new
information and update data set PAY1995, creating a new SAS data set PAY1995C (C for
“corrected”) which reflects the updated information.
 CHAPTER 4 Combining SAS Data Sets 27

PROBLEM 10
Combining Variables from Two SAS Data Sets

Tools
MERGE statement
IN= and KEEP= data set options

Data
The SAS data set PAY1995 from Problem 1 and a SAS data set called BALL, created by running
the program below
DATA BALL;
LENGTH ID $ 3.;
INPUT ID HEIGHT;
DATALINES;
A24 65
A13 66
A23 72
;

Directions
Some of the employees in data set PAY1995 (the original, not the updated one) want to form
a basketball team (albeit, a small one). A SAS data set called BALL contains the ID and height
(in inches) of these individuals. Create a new SAS data set, TEAM, which includes the employee
number (ID), GENDER, and HEIGHT. Include only those employees who are in data set BALL.

PROBLEM 11
Merging SAS Data Sets with Multiple BY Variables

Tools
MERGE statement
Data
Medical allowance data for each gender and level as follows
Gender Level Allowance
---------------------------------
F 28 2,000
M 28 1,800
F 32 2,100
M 32 1,900
F 35 2,200
M 35 2,000
F 44 4,000
M 44 3,600
28 The SAS Workbook

Directions
For each value of GENDER and LEVEL for employees in the PAY1995 data set, you have an
allowance limit as shown in the Data section. Create a SAS data set with this information. Then
create a new SAS data set called PAY1995H which contains the original variables from PAY1995
plus the health allowance information. This new data set should not contain health allowance
information for any combinations of GENDER and LEVEL that are not represented in data set
PAY1995.

Hint
Read in the allowance values using a COMMA5. informat.

PROBLEM 12
Merging Data from Multiple SAS Data Sets and Selecting Observations

Tools
MERGE statement
IN= data set option
Subsetting IF statement
Data
The SAS data set CLINICAL, plus the two SAS data sets TREAD and FAT which can be created
by running the following SAS program
DATA TREAD;
INPUT ID $ 1-3
MINUTES 4-5;
DATALINES;
123 10
811 12
586 14
278 11
193 12
;
DATA FAT;
INPUT ID $ 1-3
BODY_FAT 5-6;
DATALINES;
444 14
123 23
919 18
278 20
444 24
811 34
193 30
;
 CHAPTER 4 Combining SAS Data Sets 29

Directions
Some of the patients from the SAS data set CLINICAL entered two studies. In one study, the
length of time on a treadmill (MINUTES) was measured and stored in data set TREAD. In the
other study, the subject’s percent of body fat (BODY_FAT) was measured and stored in data
set FAT. Combine all the information from data sets CLINICAL, TREAD, and FAT and create
a new SAS data set called STUDY. Keep only those subjects who have both the treadmill data
and the percent body fat measured.

PROBLEM 13
Using the MERGE Statement to Perform a Table Lookup

Tools
MERGE statement
BY statement, with multiple BY variables

Data
SAS data set CLINICAL and blood pressure limits as described in Chapter 3, Problem 4

Directions
Solve Problem 4 in Chapter 3 by first creating a SAS data set of upper limits for systolic and
diastolic blood pressure by age group and gender. Then merge this data set with a new SAS
data set created from CLINICAL that contains a variable called AGEGROUP corresponding
to the age groupings in Chapter 3, Problem 4, along with GENDER, DBP, and SBP. Use this
new merged data set to create the variable HYPER with values of Y or N as defined.
30
 31

CHAPTER 5 Using Numerical Functions

PROBLEMS
1 Using Numerical Functions to Create New Variables . . . . . . . . . . . . . . . . . . . . . . . . . .31
2 Choosing Every Third Observation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .32
3 Computing a Moving Average . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33
4 Computing Averages within an Observation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .33
5 Computing Differences between Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .34
6 Computing Differences and Sums among Observations . . . . . . . . . . . . . . . . . . . . . . . .35
7 Performing Character to Numeric Conversion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .35
8 Performing Some Simple Trigonometric Computations . . . . . . . . . . . . . . . . . . . . . . . .36
9 Dropping the Lowest Grade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .37

INTRODUCTION
Problems in this chapter involve many of the SAS numerical functions. Chapter 6 is devoted
to character functions. Even if you don’t normally use mathematical functions such as LOG or
SIN, there are many other numerical functions that have uses such as counting the number of
missing or non-missing values in a list of variables or obtaining values of variables from previous
observations. Here are some problems that demonstrate some of the SAS numerical functions.

PROBLEM 1
Using Numerical Functions to Create New Variables

Tools
ROUND, LOG, LOG10, INT, MIN, MAX, MEAN, SUM, and N functions

Data
Raw data values for variables X, Y, and Z
X Y Z
-----------------
1 2 3
4 . 6
2.33 5 .
2.5 2.6 2.7
32 The SAS Workbook

Directions
Given the data for variables X, Y, and Z, write a SAS DATA step to create a SAS data set called
XYZ which contains the following new variables:
New Variable Description
------------------------------------------------------------
ROUND_X X rounded to the nearest tenth
LOG_X The base e log of X
LOG_10X The base 10 log of X
WHOLE_X The integer part of X
SMALL The smallest value of X, Y, or Z
BIG The largest value of X, Y, or Z
AVE The mean of the non-missing values of X, Y, and Z
SUM The sum of the non-missing values of X, Y, and Z
NONMISS The number of non-missing values of X, Y, and Z

PROBLEM 2
Choosing Every Third Observation

Tools
MOD function
_N_ automatic variable

Data
SAS data set WHOLE created by running the following program
DATA WHOLE;
DO SUBJECT = 1 TO 12;
***Generate uniform random numbers from 1 to 100;
SCORE = INT(100 * RANUNI(0) + 1);
OUTPUT;
END;
RUN;

Directions
Use the automatic variable _N_ (sometimes thought of as an observation counter but more
accurately it actually counts iterations of the DATA step) along with the MOD function to read
every third observation from an existing SAS data set WHOLE, starting with the first observation.
The new data set will contain the first, fourth, seventh, and so on, observation from data set
WHOLE.
 CHAPTER 5 Using Numerical Functions 33

PROBLEM 3
Computing a Moving Average

Tools
LAG and LAGn functions
_N_ automatic variable

Data
Month and cost figures as shown
MONTH COST
---------------
JAN 125
FEB 120
MAR 130
APR 100
MAY 140
JUN 180
JUL 200

Directions
Compute a moving average of COST using the value for the current month and the value from
the two previous months. For example, the moving average of COST for March would be the
mean of COST for January, February, and March. For the first two months where you don’t
have two previous months of data, set the moving average to missing.

Hint
The moving average for JAN and FEB should be missing. For MAR it should be the mean of 125,
120, and 130. For APR it should be the mean of 120, 130, and 100, and so on.

PROBLEM 4
Computing Averages within an Observation

Tools
N and MEAN functions
34 The SAS Workbook

Data
Sample survey data as follows
ID QUES1 QUES2 QUES3 QUES4 QUES5 QUES6 QUES7 QUES8 QUES9 QUES10
----------------------------------------------------------------
1 3 4 3 2 . 5 5 4 4 3
2 . . . 2 1 1 1 2 1 2
3 5 4 5 3 3 4 5 4 . 5

Directions
A survey asking questions about the environment was administered to some children. Compute
an overall score by taking the mean of the ten questions. However, the overall score should only
be computed if eight or more of the ten questions are answered. Write a SAS DATA step to
compute this overall score.

PROBLEM 5
Computing Differences between Observations

Tools
DIF and DIFn functions

Data
SAS data set WHOLE from Problem 2

Directions
Using the SAS data set WHOLE, write a program that will compute
A. The difference between the value of SCORE for the present subject and the value of SCORE
for the previous subject. (The difference is missing for SUBJECT=1.)
B. The difference between the value of SCORE for the present subject and the value of SCORE
for a subject two observations prior. For example, the first non-missing difference will be
the value of SCORE for subject three minus the value of SCORE for subject one.
 CHAPTER 5 Using Numerical Functions 35

PROBLEM 6
Computing Differences and Sums among Observations

Tools
LAG, LAGn, DIF, DIFn, and MEAN functions
SET and BY statements
LAST.variable temporary variable

Data
SAS data set MULTIPLE created by running the following SAS program
DATA MULTIPLE;
INPUT SUBJECT 1-2 X 4 Y 6 Z 8;
DATALINES;
01 1 2 3
01 4 5 6
01 7 8 9
02 8 7 6
02 5 4 3
02 2 1 0
;

Directions
Using data set MULTIPLE which has three observations per subject, create a new data set called
ONE which has one observation per subject and contains the following
A. The difference between the third value of X, Y, and Z and the second value for each subject.
B. The difference between the third value of X, Y, and Z and the first value for each subject.
C. The mean of X, Y, and Z for each subject.

Note
Do not use PROC MEANS to solve part C.

PROBLEM 7
Performing Character to Numeric Conversion

Tools
INPUT function
RENAME= data set option (if you take the challenge)
36 The SAS Workbook

Data
SAS data set CHAR created by running the following program
DATA CHAR;
INPUT AGE $ HEIGHT $ WEIGHT $;
DATALINES;
23 68 160
44 72 200
55 . 180
;

Directions
Given the data set CHAR, convert AGE, HEIGHT, and WEIGHT to numeric values. If you
would like a bit more of a challenge, give the numeric variables in the new data set the same
name as the original names in data set CHAR.

PROBLEM 8
Performing Some Simple Trigonometric Computations

Tools
SIN, COS, TAN, ARSIN, ARCOS, and ATAN functions

Data
Various trigonometric values as follows
DEGREE RADIAN SINE COSINE TANGENT
------------------------------------
30 1 .5 .5 .5
45 3.14159 1 1 1
0 .5708 .707 .707 0
390 10 .003 .003 .003

Directions
You want to read in some raw data representing angles (some in degrees, others in radians) and
some trigonometric values for which you want to compute angles (both in degrees and radians).
Compute the sine, cosine, and tangent of DEGREE and RADIAN, and find the angles in degrees
and radians corresponding to the SINE, COSINE, and TANGENT values. That is, find the
arcsine of SINE, the arccosine of COSINE, and the arctangent of TANGENT, and express these
angles in degrees and radians.
 CHAPTER 5 Using Numerical Functions 37

PROBLEM 9
Dropping the Lowest Grade

Tools
N and MEAN functions
ARRAY statement (optional)

Data
Student quiz grades as follows
STUDENT QUIZ1 QUIZ2 QUIZ3 QUIZ4 QUIZ5
----------------------------------------------
Baggett 4 2 6 2 3
Ginn 9 9 10 . 9
Cody 10 10 9 10 10
Smith . . 2 3 4

Directions
Being the nice teacher that you are, you decide to drop the lowest of five quiz grades if a student
took all five quizzes. Write a program that will compute the mean quiz grade based on this
decision. If the student took fewer than five quizzes, compute the mean of the non-missing
quizzes.
38
 39

CHAPTER 6 Using Character Functions

PROBLEMS
1 Reading a Mixture of Numbers and Letters from Raw Data . . . . . . . . . . . . . . . . . . . .39
2 Checking for Invalid Data Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40
3 "Unpacking" a Character String . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .40
4 Substituting One Character for Another in a String . . . . . . . . . . . . . . . . . . . . . . . . . . .41
5 Cleaning Data and Substituting Characters . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42
6 Creating a Simple Substitution Cipher . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .42
7 Deciphering a Simple Substitution Cipher . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .43

INTRODUCTION
SAS software provides a powerful assortment of character functions that allow you to perform
a variety of useful string operations. Because of the large number and flexibility of these functions,
there are usually several solutions to the problems presented here. No single solution is the
correct solution. You may want to take a hint from the functions listed in the Tools list but feel
free to find other innovative solutions as well. To accommodate alternative solutions, the Tools
list may also include more functions than you need.

PROBLEM 1
Reading a Mixture of Numbers and Letters from Raw Data

Tools
LENGTH, SUBSTR, INPUT, INDEX, and COMPRESS functions
LENGTH statement
Double trailing at sign (@@)

Data
A single line of temperature values
32F 42C 137F 84F 20C

Directions
Temperatures in degrees Celsius and degrees Fahrenheit are mixed together in an input data
stream. Celsius temperatures appear in the form nnnC, where nnn is a 1 to 3 digit number.
40 The SAS Workbook

Fahrenheit temperatures are entered as nnnF. Write a program to read the sample data and store
all temperatures in degrees Celsius. The conversion from Fahrenheit to Celsius is
C = (F - 32)*5/9 where C = temperature in degrees Celsius and
F = temperature in degrees Fahrenheit

PROBLEM 2
Checking for Invalid Data Values

Tools
VERIFY function

Data
IDs and multiple choice (A - E) responses
001ABCDEABCDE
002XCCBEBBABC
003A BACECBAA
004abcdeabcde

A student ID is stored in columns 1-3, and the item responses are stored in columns 4-13.
Directions
Answers to a ten-item multiple choice test are entered in a file. Valid responses are A, B, C, D,
or E (upper case only) and blank. Write a program that will write the student ID and the invalid
data to a SAS data set called INVALID for any student having one or more invalid data values
for the ten items. You may place the data lines in a data file and have your program read it or you
may place these data lines “instream,” following a DATALINES statement, to test your program.

Hint
You can read all ten items into a single ten-byte character string.
Note
The data set INVALID should contain the observations for ID numbers 002 (the X for item 1)
and 004 (all lowercase responses).

PROBLEM 3
“Unpacking” a Character String

Tools
SUBSTR and INPUT functions
ARRAY statement (optional)
 CHAPTER 6 Using Character Functions 41

Data
SAS data set PACK created by running the following program
DATA PACK;
INPUT TEN $10.;
DATALINES;
0123456789
1 2 3 4 56
3428645889
;

Directions
You are given a SAS data set PACK that contains a ten-byte character variable (TEN) where
each byte is a numeral 0 to 9 or a blank (representing a missing value). Write a program that
will read this data set and create a new data set called UNPACK that contains ten numeric
variables (X1-X10) from the ten-byte character variable.

PROBLEM 4
Substituting One Character for Another in a String

Tools
TRANSLATE function
ARRAY statement (optional)

Data
Five responses to a multiple choice test
12345
3 414
54321

Directions
Data values for five multiple choice questions were entered as digits 1-5 (or blank) rather than
the letters A-E. Write a program that will read the digits (you may read them as character
values), convert them to the corresponding letters, and create a SAS data set called CONVERT
that contains five character variables. For example, 1 would be converted to A, 2 to B, and
so on.
42 The SAS Workbook

PROBLEM 5
Cleaning Data and Substituting Characters

Tools
UPCASE, TRANSLATE, and VERIFY functions (suggested possibilities)
ARRAY statement (optional)

Data
Five responses to a multiple choice test
12345
aBcDe
xY73E
3 E w

Directions
This problem is similar to the previous one except that prior to the substitution of A for 1, B for
2, and so on, you have to do some data cleaning. As in Problem 4, you have five responses to
test questions in columns 1 through 5, and you want to create five variables QUES1- QUES5.
The raw data consist of some numerals, some data already entered as A-E, some data errors
(numerals above 5 and letters other than A-E), and a mixture of upper- and lowercase. Convert
all letters to uppercase and substitute the letters A through E for the numerals 1 through 5.
Convert any remaining data values (letters other than A-E or numerals not in the range of 1 to
5) to a character missing value. You may want to use an array to make this program more
compact.

PROBLEM 6
Creating a Simple Substitution Cipher

Tools
TRANSLATE and UPCASE functions
FILE and PUT statements (optional)
DATA _NULL_ (optional)

Data
Lines of “plain text” as follows
This is TEST
Line TWO of the Message
 CHAPTER 6 Using Character Functions 43

Directions
This problem and the one that follows deal with a simple substitution cipher. You may have
some fun giving a cipher to a friend and seeing if he or she can decipher it by using letter
frequencies.
Now for the problem. You want to encode plain text into a simple substitution cipher. That is,
you want to substitute one letter for another to create a simple coded message. Have your program
read lines of plain text (in upper- or lowercase) and write the encoded message (all in uppercase)
to a print file (or to the LOG).
The correspondence key is as follows
A=Q B=W C=E D=R E=T F=Y G=U H=I I=O J=P K=L L=K M=J
N=H O=G P=F Q=D R=S S=A T=Z U=X V=C W=V X=B Y=N Z=M

Use the two lines of data in the Data section to test your program.
The output should look as follows
ZIOA OA Q ZTAZ
KOHT ZVG GY ZIT JTAAQUT

PROBLEM 7
Deciphering a Simple Substitution Cipher

Tools
TRANSLATE function
FILE and PUT statements (optional)
DATA _NULL_ (optional)

Data
The encoded message from Problem 6
ZIOA OA Q ZTAZ
KOHT ZVG GY ZIT JTAAQUT

Directions
This problem is the reverse of Problem 6. Here, you want a program that will read the encoded
message and turn it back into plain text. Use the reverse letter substitutions to do this. For example,
Q=A, W=B, and so on.
44
 45

CHAPTER 7 Working with Dates

PROBLEMS
1 Reading Dates in a Variety of Configurations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45
2 Reading Dates in Non-standard Forms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .46
3 Computing Approximate Age in Years . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
4 Computing the Day of the Week, Month of the Year, and Year for
a Given Date . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .47
5 Creating Day of the Week and Month of the Year Variables without
Using a User-created Format . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .48

INTRODUCTION
Date values represent an interesting challenge in most programming languages. SAS software
has an extensive set of tools for working with dates. This section allows you to practice with
some of these tools.

PROBLEM 1
Reading Dates in a Variety of Configurations

Tools
Date informats and formats

Data
Date values in a variety of styles
1 2 3 4
123456789012345678901234567890123456789012345
---------------------------------------------
102146 10/21/46 21OCT46 46294 211046 10211946
122596 12/25/96 25DEC96 96360 251296 12251996

Directions
You are given raw data which contains dates in a variety of styles (MM/DD/YY, Julian, and so
on) as in the Data section. Write a program to read these dates and create a SAS data set called
IN_DATE. List the contents of IN_DATE, printing all the dates in MM/DD/YY format. The
date formats are
46 The SAS Workbook

Variable Starting Column Date Format

--------------------------------------------
DATE1 1 MMDDYY
DATE2 8 MM/DD/YY
DATE3 17 DDMONYY
DATE4 25 YYNNN (Julian)
DATE5 31 DDMMYY
DATE6 38 MMDDYYYY

PROBLEM 2
Reading Dates in Non-standard Forms

Tools
MDY and ROUND functions

Data
Month, day, and year data scattered about
1 2 3 4
123456789012345678901234567890123456789012345
---------------------------------------------
17 09 1990 04 96
30 11 1991 05 95

Directions
You inherited some interesting data which has date information (month, day, and year) scattered
about in different locations. In addition, one of the date values only has year and month (day is
missing), and you want to create a SAS date using the 15th of the month in place of the missing
day. The information to create the first date (DATE1) is
Day columns 3-4
Month columns 10-11
Year columns 15-18

For the second date (DATE2)

Day is missing
Month columns 20-21
Year columns 23-24

Using the sample data, create a SAS data set called IN_DATE2, containing the variables DATE1
and DATE2 as SAS date values, and compute the number of years (rounded to the nearest year)
between DATE1 and DATE2. Print out the contents of IN_DATE2 using the format
WORDDATEw. for the variables DATE1 and DATE2.
 CHAPTER 7 Working with Dates 47

PROBLEM 3
Computing Approximate Age in Years

Tools
INT, ROUND, and TODAY functions
KEEP statement or KEEP= data set option
Date literal

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, create a new SAS data set called AGE that contains the
variables AGE1-AGE3, where these variables are the person’s approximate age in years computed
as follows:
AGE1 Age as of 1/1/96 rounded to the nearest year
AGE2 Age at the time of the VISIT, with any fractional part of a
year dropped (e.g. 25.9 years would be changed to 25 years)
AGE3 Age, rounded to the nearest half year, as of the date
the program is run (i.e. today's date).

Include in this new data set the variables ID, DOB, and VISIT (from CLINICAL) and the three
computed variables AGE1-AGE3. Note that there is a data error in data set CLINICAL (put
there on purpose for a later problem) where the VISIT date is earlier than the date of birth. Just
ignore this when doing this problem.

PROBLEM 4
Computing the Day of the Week, Month of the Year, and Year for a Given Date

Tools
WEEKDAY, MONTH, and YEAR functions
KEEP statement or KEEP= data set option
PROC FORMAT
PROC FREQ

Data
SAS data set CLINICAL
48 The SAS Workbook

Directions
Using the variable VISIT (date of visit) from the data set CLINICAL, create a new data set
(CLINDATE) containing the variables ID and VISIT from CLINICAL plus three new variables,
DAY_WEEK, MONTH, and YEAR which represent the day of the week, month of the year,
and year, respectively. Create formats for DAY_WEEK and MONTH. Count frequencies for
DAY_WEEK, MONTH, and YEAR.

Note
The solution shown uses an option to omit the cumulative frequencies. You may choose to do
this or not.

PROBLEM 5
Creating Day of the Week and Month of the Year Variables without Using a
User-created Format

Tools
PUT function
WEEKDATEw. and MONNAMEw. formats

Data
Same as Problem 4

Directions
This problem is similar to the previous one except you want the values for DAY_WEEK to be
character values of MON, TUE, WED, and so on, and values of MONTH to be JAN, FEB, and so
on, without using any user-defined formats. Compute frequencies for DAY_WEEK and MONTH.
 49

CHAPTER 8 Using Arrays

PROBLEMS
1 Using an Array to Set Values of 9, 99, and 999 to Missing for a Series of
Numeric Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .49
2 Converting a Value of “NA” to Missing for Character Variables . . . . . . . . . . . . . . . . .50
3 Converting Fahrenheit Temperatures to Celsius for Lots of Variables . . . . . . . . . . . . .51
4 Creating Several Observations from a Single Observation . . . . . . . . . . . . . . . . . . . . . .52
5 Creating One Observation from Several Observations . . . . . . . . . . . . . . . . . . . . . . . . .53
6 Choosing Upper and Lower Bounds for an Array . . . . . . . . . . . . . . . . . . . . . . . . . . . . .54
7 Using a Temporary Array to Hold Passing Scores on a Test . . . . . . . . . . . . . . . . . . . . .55

INTRODUCTION
Use explicit arrays to solve all the problems in this chapter. Although the older implicit arrays
can be used to solve some of the problems, the newer implicit form of the ARRAY statement is
preferred and is used in all the solutions. If you are having trouble solving a problem, try writing
out a few lines of code without using an array. Then create your array containing the variables
in the repeated lines. Finally, using one of the sample lines as a template, substitute the array
name for the variable name and, if necessary, place the statement in a DO loop.

PROBLEM 1
Using an Array to Set Values of 9, 99, and 999 to Missing for a Series of Numeric
Variables

Tools
ARRAY statement (numeric variables)

Data
SAS data set LOTS_9 created by running the following program
DATA LOTS_9;
INPUT X1-X5 A B C D Y1-Y5 Z1-Z3;
DATALINES;
1 0 1 0 1 2 2 2 2 1 2 3 4 5 3 3 3
9 0 0 0 9 99 99 99 7 999 999 4 5 6 999 999 999
;
50 The SAS Workbook

Directions
You have a SAS data set called LOTS_9 where values of 9, 99, and 999 are to be converted to
missing values as follows
Variables Value to be Converted to Missing
---------------------------------------------------
X1-X5 9
A,B,C,D 99
Y1-Y5, Z1-Z3 999

Create a new data set LOTSMISS, replacing the appropriate values of 9, 99, and 999 with a
SAS missing value. Be sure to convert only the appropriate values 9, 99, or 999 to missing for
the variables in question. For example, a value of 9 is NOT a missing value for variables A, B,
C, D, Y1- Y5, or Z1-Z3.

PROBLEM 2
Converting a Value of “NA” to Missing for Character Variables

Tools
ARRAY statement (character variables)
UPCASE and DIM functions (optional)

Data
SAS data set NOTAPPLY created by running the following program
DATA NOTAPPLY;
LENGTH A B C D E $ 2;
INPUT ID A $ B $ C $ D $ E $ X Y Z;
DATALINES;
001 Y N N Y Y 1 2 3
002 na NA Y Y Y 3 4 5
003 NA NA NA na na 8 9 10
;

Directions
In the SAS data set NOTAPPLY, a value of either NA or na was used in place of a missing
value for all character variables. Create a new SAS data set NEW where these values are converted
to a character missing value.

Hint
Be careful to test for both upper- and lowercase values of NA. You may want to have two tests
or use the UPCASE function.
 CHAPTER 8 Using Arrays 51

PROBLEM 3
Converting Fahrenheit Temperatures to Celsius for Lots of Variables

Tools
ARRAY statement

Data
SAS data set TEMPER created by running the following program
DATA TEMPER;
ARRAY F[30]; *** Equivalent to ARRAY F[30] F1-F30;

DO OBS = 1 TO 5;
DO I = 1 TO 30;
***The line below will randomly generate temperatures
from 32 to 212;
F[I] = INT(RANUNI(0)*181 + 32);
END;
OUTPUT;
END;

KEEP F1-F30;
RUN;

Directions
A SAS data set called TEMPER contains variables F1-F30 which represent 30 temperatures in
degrees Fahrenheit. Create a new data set called TEMPER2 that contains only the
variables C1-C30 which are these temperatures converted to degrees Celsius. The Fahrenheit
to Celsius conversion is
Degrees Celsius = 5/9 x (Degrees Fahrenheit - 32)
52 The SAS Workbook

PROBLEM 4
Creating Several Observations from a Single Observation

Tools
ARRAY and OUTPUT statements

Data
SAS data set DIAG created by running the following program
DATA DIAG;
INFILE DATALINES PAD;
INPUT @1 ID $3.
@5 (DX1-DX5)($2. + 1);
DATALINES;
001 11 12
002 01 02 03 04 05
003 12
004 05 06
005 AA BB CC DD EE
;

Directions
The data set DIAG contains a patient ID and up to five diagnosis codes (DX1-DX5) per
observation. To help you visualize the data set, here is a listing of its contents.
Listing of Data Set DIAG

Patient ID DX1 DX2 DX3 DX4 DX5

----------------------------------------
001 11 12
002 01 02 03 04 05
003 12
004 05 06
005 AA BB CC DD EE

Using arrays, create a new data set called DIAG2 that contains up to five observations per
patient. Each observation should contain the patient ID and a single variable (DX) which is also
a two-byte character variable and is equal to DX1, DX2, and so on, for each of the one to five
possible observations per patient in DIAG2. Do not output an observation in DIAG2 if a diagnosis
is missing. For example, the first seven observations in data set DIAG2 should look like this
 CHAPTER 8 Using Arrays 53

ID DX
--------
001 11
001 12
002 01
002 02
002 03
002 04
002 05

PROBLEM 5
Creating One Observation from Several Observations

Tools
ARRAY and RETAIN statements
FIRST.variable and LAST.variable temporary variables
Data
SAS data set MANY created by running the following program
DATA MANY;
INPUT ID TIME X Y Z;
DATALINES;
1 1 1 2 3
1 2 4 5 6
1 3 7 8 9
2 1 10 20 30
2 3 40 . 50
3 1 15 . .
3 2 25 26 27
3 3 35 36 37
;

Directions
Write a program to restructure the data set MANY, which contains from one to three observations
per ID, to a new data set ONEPER, which contains only one observation per ID. To help you
visualize data set MANY, here is a listing of its contents
ID TIME X Y Z
1 1 1 2 3
1 2 4 5 6
1 3 7 8 9
2 1 10 20 30 Note: TIME = 2 is missing for
2 3 40 . 50 ID = 2
3 1 15 . .
3 2 25 26 27
3 3 35 36 37
54 The SAS Workbook

The variables in the new data set ONEPER are ID, X1-X3, Y1-Y3, and Z1-Z3. They represent
the values of X, Y, and Z at times 1, 2, and 3, respectively. Data set ONEPER should look like
this
ID X1 X2 X3 Y1 Y2 Y3 Z1 Z2 Z3
1 1 4 7 2 5 8 3 6 9
2 10 . 40 20 . . 30 . 15
3 15 25 35 . 26 36 . 27 37

Note
Check your ONEPER data set to be sure that you have missing values for X2, Y2, and Z2 for
ID = 2.

PROBLEM 6
Choosing Upper and Lower Bounds for an Array

Tools
ARRAY statement (upper and lower boundary specified)

Data
SAS data set PAYROLL created by running the following program
DATA PAYROLL;
INPUT AMT1990-AMT1996;
DATALINES;
50000 55000 57000 62000 66000 70000 72000
40000 43000 44000 55000 65000 69000 73000
;

Directions
A SAS data set called PAYROLL contains the variables AMT1990 through AMT1996. Using
an array with lower and upper bounds equal to 1990 and 1996 respectively, write a SAS DATA
step to create a new data set called TAXES that contains the original AMT variables plus the
variables TAX1990 through TAX1996. Each TAX variable is obtained by multiplying the
corresponding AMT variable by .25.
 CHAPTER 8 Using Arrays 55

PROBLEM 7
Using a Temporary Array to Hold Passing Scores on a Test

Tools
Temporary array

Data
Student names and test scores as shown
STUDENT TEST1 TEST2 TEST3 TEST4 TEST5
-----------------------------------------------
Bob 45 68 80 65 60
Michelle 80 90 95 90 87
Clifton 59 69 79 89 99

Directions
The passing scores on five tests are 65, 70, 65, 70, and 75. Using a temporary array to hold the
passing scores, count the number of tests passed by each of the students.
56
 57
Looking Back and Ahead
CHAPTER 9 Across Observations

PROBLEMS
1 "Copying" Values from One Observation to Another Observation . . . . . . . . . . . . . . .57
2 Computing a Moving Average . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .58
3 Looking Back . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .59
4 Looking Ahead . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .60

INTRODUCTION
This chapter contains problems that require you to either look back or look ahead across
observations. This is especially important when you have more than one observation per subject
in a data set. You may want to compare a value in the present observation to a value from a
previous observation for the same subject.
Looking back usually requires such tools as the RETAIN statement, the LAG function,
FIRST.variable and LAST.variable. Sometimes, you can look back by restructuring the data
set so that what was originally multiple observations is now contained in a single observation.
Looking ahead is more difficult and is most easily accomplished by using more than one SET
statement with the FIRSTOBS= data set option set to different values.

PROBLEM 1
“Copying” Values from One Observation to Another Observation

Tools
RETAIN statement
FIRST.variable and LAST.variable temporary variables (possibly)

Data
Raw data file DIALYSIS.DTA

Directions
You have a longitudinal data file of visits to a dialysis center called DIALYSIS.DTA. The first
record for each patient includes an ID, a visit number (VISIT), a date of birth (DOB), GENDER,
heart rate (HR), systolic blood pressure (SBP), and diastolic blood pressure (DBP). Subsequent
records for the same patient only contain data for the ID, VISIT, HR, SBP, and DBP. That is, the
DOB and GENDER information is only entered on the first record for each patient. There are a
variable number of visits for each patient.
58 The SAS Workbook

Create a data set containing the values for DOB and GENDER included in every observation.
You may assume that there are no missing IDs. The DOB and GENDER may be missing for
all visits for some patients. Solve this problem using a RETAIN statement to “remember” the DOB
and GENDER. You may want to try this “on the fly” by reading the raw data (file
DIALYSIS.DTA) directly or you may choose to create a SAS data set first and then process
this in a subsequent DATA step.

PROBLEM 2
Computing a Moving Average

Tools
LAG function
_N_ automatic variable

Data
Daily stock prices as follows
DATE PRICE
-----------------
01/01/95 23.00
01/02/95 25.00
01/03/95 24.00
01/04/95 29.00
01/05/95 26.00
01/06/95 23.00
01/07/95 24.00

Directions
You have a raw data file which contains the date and the closing price for a particular stock.
You want to compute a moving average which is the mean of today’s price and the price for
the previous two days. Create a data set that contains the date, the current price, and the moving
average, starting with day three. Provide a listing of the data set (use an appropriate format for
the dollar amounts). Plot the daily price and the three-day moving average versus the date on
one set of axes, using a separate plotting symbol for each of the two plots.

Note
Problem 3 in Chapter 5 is similar to this problem and also involves computing a moving average.
 CHAPTER 9 Looking Back and Ahead Across Observations 59

PROBLEM 3
Looking Back

Tools
LAG function

Data
Raw data consisting of three records per subject as follows
ID TRIAL X Y Z
-------------------------
1 1 1 2 3
1 2 2 3 6
1 3 4 1 2
2 1 4 5 6
9 2 5 4 3
2 3 1 2 1
3 1 2 2 2
3 2 3 3 3
3 3 4 5 6

Directions
You conducted an experiment where each subject underwent three trials. For each trial, an ID
number, a trial number (NUMBER), and three variables (X,Y, and Z) were measured. Each
subject has exactly three records. Write a program that checks if the IDs are the same in all three
records. If they are, write out the observations to data set OK; if not, write out the three
observations to data set NOT_OK.

Hint
The first three and the last three records should end up in data set OK; the middle three records
should end up in data set NOT_OK.
60 The SAS Workbook

PROBLEM 4
Looking Ahead

Tools
SET statement
FIRSTOBS=, RENAME=, and DROP= data set options

Data
SAS data set STREP created by running the following program
DATA STREP;
INFILE DATALINES PAD;
INPUT @1 ID $1
@3 VISIT MMDDYY8.
@12 DOCTOR $3.;
DATALINES;
1 11/01/95 ABC
1 12/01/95 XYZ
2 12/01/95 JBD
2 12/07/95 RPC
3 01/05/96 ABC
3 05/05/96 JBD
3 07/01/96 XYZ

Directions
You have a SAS data set called STREP which has information on patients visiting a clinic who
have a diagnosis of strep throat. Each observation contains a patient ID, visit date (VISIT), and
DOCTOR initials. If the same patient returns for a visit within 90 days for strep throat, a variable
called OUTCOME is to be set to the value FAILURE. If the patient returns more than 90 days
later or has no further visits, the variable OUTCOME is to be set to a value of SUCCESS. You
may assume that there are no missing values for visit date or doctor initials.
You want to assign the responsibility (blame?) for this success or failure to the doctor who
originally saw the patient. You could “remember” the previous value of DOCTOR with a retained
variable or use a LAG function. However, for this problem, you are to look ahead (see the Tools
list for a hint) to see if the next visit is within 90 days and if it is the same patient. Write a SAS
program that looks ahead to the next observation, creates a new data set called BLAME, and
sets the value of OUTCOME for the visit where the patient is seen by the doctor who will be
assigned the responsibility.
 CHAPTER 9 Looking Back and Ahead Across Observations 61

The data set BLAME should have seven observations as shown below
OBS ID VISIT DOCTOR OUTCOME

1 1 11/01/95 ABC FAILURE

2 1 12/01/95 XYZ SUCCESS
3 2 12/01/95 JBD FAILURE
4 2 12/07/95 RPC SUCCESS
5 3 01/05/96 ABC SUCCESS
6 3 05/05/96 JBD FAILURE
7 3 07/01/96 XYZ SUCCESS

Hint
The blank record at the end of the raw data will make your program a bit easier to write.
62
 63

Working with Longitudinal Data:

CHAPTER 10 Multiple Observations per Subject

PROBLEMS
1 Selecting the First or Last Observation per Subject . . . . . . . . . . . . . . . . . . . . . . . . . . . .63
2 Selecting All Patients with N Observations (DATA Step Approach) . . . . . . . . . . . . . . .64
3 Selecting All Patients with N Observations (Using a Procedure
to Do Some of the Work) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .64
4 Copying Data from the First Record to Replace Missing Values
on Subsequent Records (Method 1) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .65
5 Copying Data from the First Record to Replace Missing Values
on Subsequent Records (Method 2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .66
6 Computing Differences Across Observations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .66
7 Counting Code Values in a DATA Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .67

INTRODUCTION
Problems in this chapter deal with the special challenges associated with longitudinal data, that
is, data with multiple observations per subject. Many of the same tools you used in the last
chapter will be used here. In addition some of the projects in Chapter 21 require the techniques
used here.

PROBLEM 1
Selecting the First or Last Observation per Subject

Tools
FIRST.variable and LAST.variable temporary variables

Data
Raw data file CLIN_X.DTA

Directions
Create two data sets from the CLIN_X.DTA raw data file. The first data set (call it FIRST)
should consist of the first visit for each patient. The other data set (call it LAST) should contain
the last visit for each patient. If any patient has only one visit, include this visit in both data sets.
Try to create both data sets in a single DATA step.
64 The SAS Workbook

Hint
Be careful. The raw data file is not in order by ID or by date of visit.

PROBLEM 2
Selecting All Patients with N Observations (DATA Step Approach)

Tools
FIRST.variable and LAST.variable temporary variables
SUM and MERGE statements
IN= data set option
WHERE or IF statement or WHERE= data set option

Data
Raw data file CLIN_X.DTA

Directions
Create a data set of all patients in the CLIN_X.DTA data file that have exactly four visits. For
this problem, accomplish the task by first counting the number of visits for each patient using a
DATA step. Create a data set containing data on only those patients with four visits. This data
set should have one observation per patient and a single variable (ID). Combine this information
with a data set created from the original data to produce the final data set. The resulting data
set should contain four observations for each of two patients.

PROBLEM 3
Selecting All Patients with N Observations (Using a Procedure to Do Some of the Work)

Tools
PROC MEANS or PROC FREQ
MERGE statement
IN= data set option
WHERE or IF statement or WHERE= data set option

Data
Raw data file CLIN_X.DTA
 CHAPTER 10 Working with Longitudinal Data: Multiple Observations per Subject 65

Directions
This problem is the same as Problem 2, except that you should first use PROC MEANS or
PROC FREQ to create a data set containing all patients with exactly four visits. From that point,
proceed as you did before.

Hint
If you choose to use PROC MEANS, pick any numeric variable and use the _FREQ_ variable
(the number of observations, missing or non-missing) to count the number of visits per patient.

PROBLEM 4
Copying Data from the First Record to Replace Missing Values on Subsequent Records
(Method 1)

Tools
RETAIN statement
FIRST.variable temporary variable

Data
Raw data file BASKETBA.DTA (see the Appendix)

Directions
You are given the data file BASKETBA.DTA. Notice that values for ID, GENDER, HEIGHT,
and DATE are entered only on the first record for each player while the number of points scored
is entered on every record. Create a SAS data set called BASKET that has values for ID,
GENDER, HEIGHT, DATE, and POINTS for every observation. For this problem, use the
FIRST.variable temporary variable and a RETAIN statement to accomplish the task.

Note
A. This problem is similar to Problem 1 from Chapter 9.
B. It might have been smarter to have two data sets, one containing ID, GENDER, and HEIGHT
(demographic data) and another containing ID, DATE, and POINTS. However, the data
set was constructed this way to give you some programming practice.

Hint
Be especially careful since GENDER and HEIGHT are missing for a few players.
66 The SAS Workbook

PROBLEM 5
Copying Data from the First Record to Replace Missing Values on Subsequent Records
(Method 2)

Tools
FIRST.variable temporary variable
DROP= and KEEP= data set options
MERGE statement

Data
Raw data file BASKETBA.DTA

Directions
Solve Problem 4, except create one data set containing the first observation for each player with
the variables ID, GENDER, and HEIGHT. Merge this with the original data set (dropping
GENDER and HEIGHT), thus adding the missing information to the other observations.

PROBLEM 6
Computing Differences Across Observations

Tools
FIRST.variable and LAST.variable temporary variables
RETAIN statement

Data
Raw data file CLIN_X.DTA

Directions
Create a SAS data set from the CLIN_X.DTA data file. Compute the difference between the
heart rate, the systolic blood pressure, and the diastolic blood pressure from the first visit to the
last visit for all patients who had two or more visits (that is, the value at the last visit minus the
value at the first visit).
 CHAPTER 10 Working with Longitudinal Data: Multiple Observations per Subject 67

PROBLEM 7
Counting Code Values in a DATA Step

Tools
ARRAY (optional) and RETAIN statements
PROC SORT
FIRST.variable and LAST.variable temporary variables

Data
SAS data set DRUG created by running the following program
DATA DRUG;
INPUT ID DATE : MMDDYY8. RX1-RX3;
FORMAT DATE MMDDYY8.;
DATALINES;
1 10/21/95 1 0 0
1 10/22/95 2 0 1
1 10/23/95 0 0 1
2 09/02/95 1 1 1
2 09/03/95 1 1 1
3 11/11/95 0 2 1
3 11/15/95 0 0 2
;

Directions
Each time a patient comes in for a clinic visit, information is recorded on each of three drugs. A
code of 0 indicates the drug is not being taken; a code of 1 indicates the drug is being started; a
code of 2 indicates that the drug is being discontinued. For each visit, a patient ID, visit date
(DATE), and status on each of the three drugs (RX1-RX3) is recorded. Write a SAS program that
will count the number of patients who were ever on each of the three drugs (a code of 1 or 2 on
any visit). Try using only DATA step operations (plus PROC SORT). Use a PUT statement to
write the number of patients on the three drugs to a print or log file.
68
 69

Writing Simple DATA

CHAPTER 11 Step Reports

PROBLEMS
1 Writing a Simple Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .69
2 Adding Header Information and Spacing the Data Values . . . . . . . . . . . . . . . . . . . . . .70
3 Customizing a Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .71
4 Creating a Multi-column (Phone Book Style) Report . . . . . . . . . . . . . . . . . . . . . . . . . .72

INTRODUCTION
This chapter contains some basic DATA step reporting problems. PROC PRINT and PROC
REPORT can be used to create a variety of reports, but sometimes you require more flexibility,
and you need to resort to the keyword DATA _NULL_ and PUT statements to get the job done.
Here are some reporting problems where you are asked to use the DATA step to solve
the problem.

PROBLEM 1
Writing a Simple Report

Tools
DATA _NULL_
FILE and PUT statements

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, create a simple report like the one that follows. Use a TITLE
statement to create the title and write the report to the OUTPUT window or output device. The
three columns of numbers represent ID, systolic blood pressure (SBP), and diastolic blood
pressure (DBP).
70 The SAS Workbook

Simple "No Frills" Report

123 136 76
278 104 64
444 128 62
756 150 96
811 166 74
193 112 68
98 54 Note: There is a missing ID for this observation
978 98 62
586 162 96
919 58 50
more observations.

PROBLEM 2
Adding Header Information and Spacing the Data Values

Tools
DATA _NULL_
FILE PRINT statement
HEADER= option
PUT statement
Data
SAS data set CLINICAL
Directions
Improve the report in Problem 1 by adding a header which includes the title line and column
headings, and space out the data values so that they line up under the headings. The report should
look approximately like the one that follows. Do not use a TITLE statement to write the header
information. Don’t worry about getting the spacing exactly the same as shown here.
Listing of Systolic and Diastolic Blood Pressure
ID Systolic Diastolic
Blood Pressure Blood Pressure
------------------------------------------------
123 136 76
278 104 64
444 128 62
756 150 96
811 166 74
193 112 68
98 54
978 98 62
586 162 96
more observations
 CHAPTER 11 Writing Simple DATA Step Reports 71

PROBLEM 3
Customizing a Report

Tools
DATA _NULL_
FILE PRINT statement
HEADER= option
PUT statement
Either a trailing at sign (@) or SUBSTR pseudo function

Data
SAS data set CLINICAL

Directions
Create a report similar to the one in Problem 2 except place an asterisk (*) after any systolic
blood pressure greater than 160 and after any diastolic blood pressure greater than 90. A sample
report follows.
Two solutions to this problem are presented in The SAS Workbook Solutions. One uses a trailing
at sign (@) in the PUT statement along with logical statements to print an asterisk (*) where
needed. The other solution creates character variables from SBP and DBP and uses the SUBSTR
pseudo function to place an asterisk (*) at the end of the string, if needed. Try to solve this
problem both ways.
Listing of Systolic and Diastolic Blood Pressure
(Pressures marked with an asterisk (*) are hypertensive)
ID Systolic Diastolic
Blood Pressure Blood Pressure
------------------------------------------------
123 136 76
278 104 64
444 128 62
756 150 96*
811 166* 74
193 112 68
98 54
978 98 62
586 162* 96*
etc.

Hint
To use the SUBSTR pseudo function you place the function on the left side of the equal
sign (=). For example, SUBSTR(STRING,4,1) = ‘*’.
72 The SAS Workbook

PROBLEM 4
Creating a Multi-column (Phone Book Style) Report

Tools
DATA _NULL_
FILE PRINT statement
HEADER= and N=PAGESIZE options
PUT statement

Data
SAS data set CLINICAL

Directions
Again, using the information from the CLINICAL data set, create a report like the one that
follows. To help you with alignment, column rulers are placed above the report.
1 2 3 4 5 6 7
1234567890123456789012345678901234567890123456789012345678901234567890
----------------------------------------------------------------------
Listing of Systolic and Diastolic Blood Pressure

ID Systolic Diastolic ID Systolic Diastolic

BP BP BP BP
--------------------------------------------------------------------
123 136 76 919 58 50
278 104 64 529 114 76
444 128 62 324 200 100
756 150 96 012 184 98
811 166 74 812 134 82
193 112 68 338 106 70
98 54 959 176 98
978 98 62 007 106 64
586 162 96 291 132 86
more observations
 73
Procedures
SECTION 2 (Non-statistical Problems)

CHAPTERS
12 The FORMAT Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .75

13 The PRINT Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .83

14 The SORT Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .89

15 The FREQ Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .93

16 The MEANS Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .99

17 The TABULATE Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .103

18 The CHART Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .117

19 The PLOT Procedure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .125

74 The SAS Workbook
 75

CHAPTER 12 The FORMAT Procedure

PROBLEMS
1 Creating Simple Numeric and Character Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . .75
2 Creating Character and Numeric Formats Using Several
Methods to Specify Ranges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .77
3 Creating Permanent Formats . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .78
4 Using an Existing SAS Data Set to Create a Control Data Set
and a User-defined Format (Character Format) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .79
5 Using an Existing SAS Data Set to Create a Control Data Set
and a User-defined Format (Numeric Format) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .80
6 Using an Enhanced Numeric Informat to Read a Combination of
Character and Numeric Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .80
7 Using a Format to Count the Number of Missing Values . . . . . . . . . . . . . . . . . . . . . . .81

INTRODUCTION
This chapter covers simple numeric and character formats and informats as well as more
complicated problems with control data sets. One of the problems requires you to create a
permanent format library and to use the FMTSEARCH= system option indicate where to look
for these formats. Other problems use formats as an alternative to writing statements in a DATA
step to recode values when you create class variables.

PROBLEM 1
Creating Simple Numeric and Character Formats

Tools
PROC FORMAT
VALUE statement (discrete ranges)

Data
Raw data described as follows
76 The SAS Workbook

Starting
Variable Description Column Length Type
---------------------------------------------------
ID Subject ID 1 3 Char
GENDER Gender 4 1 Char
AGEGROUP Age Group 5 1 Numeric
QUES1 Question 1 6 1 Char
QUES2 Question 2 7 1 Char
QUES3 Question 3 8 1 Char

Sample Data
-----------
001M1123
002F3452
003M2421
004F4531

Directions
Given the file description in the Data section, create a SAS data set called SURVEY. Supply
formats for GENDER, AGEGROUP, QUES1, QUES2, and QUES3.
Format Information for Variables
GENDER M = Male
F = Female

AGEGROUP 1 = 0 to 20
2 = 21 to 40
3 = 41 to 60
4 = Greater than 60

QUES1-QUES3 1 = Strongly Disagree

2 = Disagree
3 = No Opinion
4 = Agree
5 = Strongly Agree
 CHAPTER 12 The FORMAT Procedure 77

PROBLEM 2
Creating Character and Numeric Formats Using Several Methods to Specify Ranges

Tools
PROC FORMAT
VALUE statement
LOW and HIGH range keywords

Data
Raw data described as follows
Starting
Variable Description Column Length Type
----------------------------------------------------
ID Subject ID 1 3 Char
AGE Age 5 5 Numeric
CODE Insurance Code 10 1 Numeric
PAY Payment Code 12 1 Char

Sample Data
------------
001 23.0 1 1
002 55.9 2 3
003 60.1 3 2
004 12.5 1 5
005 19.1 5 X

Directions
Again, you want to supply formats for some variables and create a SAS data set called
SURVEY_2. The format information is described next.
AGE Low to <20 = Less than 20
20 to <40 = 20 to less than 40
40 to <60 = 40 to less than 60
60 to high = 60 and above
Note: Age is expressed to the nearest tenth of a year.
78 The SAS Workbook

CODE 1 = Not Insured

2 = Gold Star Insurance
3 = Blue Star Insurance
4 = State Insurance
5 = Medicare
PAY 1,3,5 = Bill Paid
2 = 30 Days Overdue
X,Y = Error in Billing
4 = Collection Agency

PROBLEM 3
Creating Permanent Formats

Tools
PROC FORMAT
LIBRARY= option
FMTSEARCH= system option

Data
Same as Problem 1

Directions
Rewrite Problem 1 so that the format you create is made permanent and stored in a subdirectory
called C:\WORKBOOK (or another subdirectory of your choice). Run a separate SAS session
and submit the appropriate LIBNAME statement so that you can use your permanent formats.
When you want to use your permanent formats, either use the libref LIBRARY (in which case
the system will find the formats automatically), or use a libref of your choice and tell the system
where to find the formats. Use the FMTSEARCH= system option to indicate where to find your
formats. Use these permanent formats in a separate SAS session with the data set SURVEY
created in Problem 1.

Hint
The two SAS statements that follow create a libref called WORKBOOK and tell the SAS System
to search for user-defined formats in this library
LIBNAME WORKBOOK 'C:\WORKBOOK';
OPTIONS FMTSEARCH=(WORKBOOK);
 CHAPTER 12 The FORMAT Procedure 79

PROBLEM 4
Using an Existing SAS Data Set to Create a Control Data Set and a User-defined
Format (Character Format)

Tools
PROC FORMAT
CNTLIN= and FMTLIB options

Data
SAS data set CODES created by running the following program
DATA CODES;
INPUT @1 DX_CODE $2.
@4 DESCRIP $16.;
DATALINES;
01 Cold
02 Flu
03 Break or Fracture
04 Routine Physical
05 Heart Problem
06 Lung Disorder
07 Abdominal Pain
08 Laceration
09 Immunization
10 Lyme Disease
11 Ear Ache
;

Directions
You have a SAS data set called CODES with variables DX_CODE and DESCRIP. DX_CODE
is a 2-byte diagnosis code and DESCRIP is a 16-byte description. Use this data set to create a
control data set which can then be used to create a format called $DXFMT. The format associates
the DX codes with the descriptions. Run PROC FORMAT to create the format. Use the FMTLIB
option to check your format.
80 The SAS Workbook

PROBLEM 5
Using an Existing SAS Data Set to Create a Control Data Set and a User-defined
Format (Numeric Format)

Tools
PROC FORMAT
CNTLIN= option
Data
SAS data set ZIP created by running the following program
DATA ZIP;
INPUT ZIP_CODE LOCATION & $20.;
DATALINES;
08822 Flemington, NJ
11518 East Rockaway, NY
08903 New Brunswick, NJ
;

Directions
You have a data set called ZIP with the variables ZIP_CODE (numeric) and LOCATION
(character). Use this data set to create a user-defined format (ZIPFMT) by first creating a control
data set. Use the FMTLIB option to check your format.

PROBLEM 6
Using an Enhanced Numeric Informat to Read a Combination of Character and
Numeric Data

Tools
PROC FORMAT
INVALUE statement (character and numeric ranges)

Data
Blood lead levels as follows
5.6 A 2.0 2.1 C 9 B 3.7 X 10

Directions
You want to read some blood lead levels from several laboratories. Each laboratory has a
detection limit below which they declare the value as a “non-detect.” The data are entered with
the letters A, B, and C representing the non-detect level for three laboratories. You want to read
 CHAPTER 12 The FORMAT Procedure 81

the blood lead levels and substitute the non-detect level if an A, B, or C is read. The levels
corresponding to the letters A through C are
Letter Code Blood Lead Level
-----------------------------------
A .1
B .05
C .5

Use an “enhanced numeric informat” to read the combination of numbers and letters in the
sample line of data and substitute the non-detect levels for the letters A through C. Convert
any letters other than A through C to a numeric missing value.

Note
See SAS Technical Report P-222, Changes and Enhancements to Base SAS Software,
Release 6.07, pp 215-216 for details on the use of the enhanced numeric informat.

PROBLEM 7
Using a Format to Count the Number of Missing Values

Tools
PROC FORMAT
PROC FREQ
TABLES statement
MISSING option
_NUMERIC_, _CHARACTER_, and _ALL_ special SAS variables
Data
SAS data set MISSING created by running the following program
DATA MISSING;
INFILE DATALINES PAD;
INPUT @1 GROUP $1.
@3 GENDER $1.
@5 (X Y Z)(1.+1);
DATALINES;
A M 1 2 3

B F 1 6
A 9 3
M 5 6 7
a f 5 6 7
* $ 9 9 9
;
82

Directions
You have a data set called MISSING. Create two formats, one for the numeric variables and
another for the character variables, that you can use with PROC FREQ to count the number of
missing and non-missing observations for each of the numeric and character variables.

Note:
1. All upper- and lowercase letters plus the asterisk (*) and the dollar sign ($) are to be considered
as valid non-missing character values.
2. Line 2 is blank.
 83

CHAPTER 13 The PRINT Procedure

PROBLEMS
1 Preparing a Simple "No-frills" Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .83
2 Using Variable Labels as Column Headings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .84
3 Adding Titles and Footnotes to a Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .84
4 Producing a Report Using a BY Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .85
5 Using the Value of a BY Variable in a Title . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .85
6 Adding Sums and Counts to a Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .86

INTRODUCTION
You might think that PROC PRINT is too simple to merit a chapter of its own. However, there
are some interesting options and statements that give you quite a bit of control over the appearance
of the output. Here are some problems that give you some practice using PROC PRINT.

PROBLEM 1
Preparing a Simple “No-frills” Report

Tools
PROC PRINT
OBS= data set option
VAR and ID statements

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, produce a listing of the variables ID, DOB, VISIT, and
PRIM_DX for the first ten observations in the data set. Use the variable ID as the leftmost
column instead of the default OBS column.
84 The SAS Workbook

PROBLEM 2
Using Variable Labels as Column Headings

Tools
PROC PRINT
LABEL= and NOOBS options
OBS= data set option
VAR, LABEL, and FORMAT statements

Data
SAS data set CLINICAL

Directions
Repeat Problem 1 but substitute the following labels for the column headings.
ID = 'Patient ID'
DOB = 'Date of Birth'
VISIT = 'Date of Visit'
PRIM_DX = 'Primary DX'

Print out both DOB and VISIT using the DATE7. format. Use the NOOBS option, instead of
the ID statement, to remove the OBS column from the listing.

Note
Data set CLINICAL already has labels associated with most of the variables, but go ahead and
write your own LABEL statement for this problem.

PROBLEM 3
Adding Titles and Footnotes to a Report

Tools
PROC PRINT
OBS= data set option
VAR statement
TITLE and FOOTNOTE statements

Data
SAS data set CLINICAL
 CHAPTER 13 The PRINT Procedure 85

Directions
Repeat Problem 1 with the addition of this three-line title
Sample Report from the Indian Point Clinic
(Sample listing of the first ten cases)
------------------------------------------

and this single-line footnote

*** Prepared by Data Systems of Greater Flemington ***

PROBLEM 4
Producing a Report Using a BY Variable

Tools
PROC PRINT
BY statement

Data
Raw data file BASKET.DTA

Directions
Create a SAS data set from the raw data file BASKET.DTA. Produce a report separately for
males and females, listing all the variables except GENDER (obviously not needed since this
is the BY variable). Supply a one-line title of your choosing. Omit the OBS column in the listing.

PROBLEM 5
Using the Value of a BY Variable in a Title

Tools
PROC PRINT
BY and FORMAT statements
TITLE statement using #BYVAL (varname)
NOBYLINE system option

Data
Raw data file BASKET.DTA
86 The SAS Workbook

Directions
Produce the same report as requested in Problem 4, except add the value of the BY variable
(GENDER) to the title and remove the BY line from the report using the NOBYLINE system
option. Supply a format for GENDER so that your title reads either
Report on the Female Basketball Players
or
Report on the Male Basketball Players

PROBLEM 6
Adding Sums and Counts to a Report

Tools
PROC PRINT
N and LABEL options
TITLE statement using #BYVAL (varname)
VAR, BY, ID, SUM, LABEL, and FORMAT statements
NOBYLINE, NODATE, and NONUMBER system options

Data
Raw data file BASKET.DTA

Directions
Again, using the data from BASKET.DTA, produce a report as follows
• Include the variables GENDER, ID, DATE, and POINTS in the report.
• Print a count of the number of observations for each gender and overall.
• Print the sum of points for males, females, and overall.
• Include the gender information in the title line (and omit the BY line).
• Omit the page number and date from the report using system options.
• Use the WORDDATE. format for the variable DATE.
• Within GENDER, list the IDs in ascending order.
• Use variable labels as column headings as shown in the following report.
 CHAPTER 13 The PRINT Procedure 87

The ABC Company Basketball Team Roster

Data for the Female Players
--------------------------------------

Employee Points
Number Date per Game

002 October 21, 1995 14

005 December 12, 1995 10
005 December 15, 1995 18
--------
GENDER 42

N = 3
The ABC Company Basketball Team Roster
Data for the Male Players
--------------------------------------

Employee Points
Number Date per Game

001 December 1, 1995 10

001 December 3, 1995 12
001 December 10, 1995 8
003 December 12, 1995 20
003 December 16, 1995 22
004 November 12, 1995 6
004 November 15, 1995 8
006 October 30, 1995 2
006 November 4, 1995 6
006 November 6, 1995 8
007 November 11, 1995 9
007 November 16, 1995 7
--------
GENDER 118
========
160

N = 12
Total N = 15
88
 89

CHAPTER 14 The SORT Procedure

PROBLEMS
1 Sorting in Increasing or Decreasing Order of a Variable . . . . . . . . . . . . . . . . . . . . . . . .89
2 Using PROC SORT to Create a New Sorted Data Set . . . . . . . . . . . . . . . . . . . . . . . . .90
3 Removing Records with Duplicate Keys or Duplicate Records . . . . . . . . . . . . . . . . . . .90
4 Demonstrating the NOEQUALS Option . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .92

INTRODUCTION
PROC SORT is one of the most frequently used SAS procedures. Like PROC PRINT, it is
relatively simple. Since it is used so often and since sorting can take up so much of your system’s
resources, a brief look at some of its features is very worthwhile. Efficiency matters concerning
PROC SORT are discussed in Chapter 20.

PROBLEM 1
Sorting in Increasing or Decreasing Order of a Variable

Tools
PROC SORT
BY statement
DESCENDING option

Data
Raw data file CARS.DTA

Directions
Using the raw data file CARS.DTA, create a SAS data set called CARS which contains all the
variables described in the file but arranged in three different sort orders, as follows.
A. in order of increasing reliability (RELIABLE)
B. in order of decreasing reliability
C. by manufacturer and decreasing size.
For part C, the observations should be ordered by manufacturer (alphabetically). For each
manufacturer, the observations should be in decreasing size order. For each of the three sort
orders, list the contents of the data set.
90 The SAS Workbook

PROBLEM 2
Using PROC SORT to Create a New Sorted Data Set

Tools
PROC SORT
OUT= option
KEEP= , WHERE=, and LABEL= data set options

Data
SAS data set CLINICAL

Directions
For all the parts below, use only PROC SORT (plus PROC PRINT to list the newly created
data set) in your solution. Do not write any DATA steps. As an optional task, use the LABEL
data set option to label the newly created data sets.
A. Use PROC SORT to create a new data set called SORTED which contains all the
observations and variables in the data set CLINICAL in ascending ID order.
B. Create a new SAS data set called SUBSET1 which contains all the variables in
the data set CLINICAL in ascending ID order, but only those patients with SBP above 160.
C. Create a new SAS data set called SUBSET2 which contains all the observations in the
data set CLINICAL in ascending ID order, but only the variables ID, SBP, and DBP.
D. Create a new SAS data set called SUBSET3 which contains only patients with SBP above
160. The only variables in SUBSET3 should be ID, SBP, and DBP. This data set should
be sorted in increasing order of ID. Provide the label Clinical data: SBP> 160,
ID, and DBP for the data set SUBSET3.

PROBLEM 3
Removing Records with Duplicate Keys or Duplicate Records

Tools
PROC SORT
NODUP and NODUPKEY options
 CHAPTER 14 The SORT Procedure 91

Data
SAS data set DUP1 and DUP2 created by running the following program
DATA DUP1;
INPUT NAME $ X Y;
DATALINES;
ADAM 2 3
ADAM 2 3
CHARLES 4 7
ADAM 4 5
DAVID 8 9
CHARLES 9 9
;
DATA DUP2;
INPUT NAME $ X Y;
DATALINES;
ADAM 2 3
CHARLES 4 7
ADAM 4 5
ADAM 2 3
DAVID 8 9
CHARLES 9 9
;

Notice that data sets DUP1 and DUP2 contain the same records but in different order.

Directions
Data sets DUP1 and DUP2 contain several observations with the same name and two observations
that are identical (that is, they contain the same value for all the variables). Using only PROC
SORT (plus PROC PRINT to obtain a listing), accomplish the following tasks.
A. You want only one observation for each name in data set DUP1 (the first observation with
that name). Do this using PROC SORT with the appropriate option.
B. This time, you want to remove identical adjacent observations in data set DUP1. That is,
one of the two identical observations with NAME=ADAM, X=2, and Y=3 should be
removed. Use PROC SORT with the appropriate option to accomplish this.
C. Run the same program as in part B using data set DUP2 instead of DUP1. Notice that
after sorting by NAME, the two observations that are identical are not adjacent. Notice
which observations (if any) are removed.
D. Figure out a way to remove the non-adjacent identical observations in data set DUP2.
92 The SAS Workbook

PROBLEM 4
Demonstrating the NOEQUALS Option

Tools
PROC SORT
NOEQUALS option

Data
SAS data set TEST_EQ created by running the following program
DATA TEST_EQ;
INPUT A B @@;
DATALINES;
1 8 1 5 2 5 1 7
;

Directions
Sort the data set TEST_EQ by the variable A, creating a new data set called TMP. List the
observations in TMP. Sort data set TEST_EQ again using the NOEQUALS option. Again, create
an output data set TMP and list the observations. How do the two output data sets compare?
 93

CHAPTER 15 The FREQ Procedure

PROBLEMS
1 Producing Simple One-way Frequencies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .93
2 Producing Simple One-way Frequencies Omitting Percentages and
Cumulative Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .94
3 Demonstrating how Missing Values Are Handled . . . . . . . . . . . . . . . . . . . . . . . . . . . . .94
4 Demonstrating the Effects of the ORDER= Option . . . . . . . . . . . . . . . . . . . . . . . . . . .94
5 Producing Two-way Frequency Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .95
6 Requesting Multiple Two-way Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .95
7 Creating an Output Data Set Using PROC FREQ . . . . . . . . . . . . . . . . . . . . . . . . . . . . .96
8 Producing a Three-way Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .96
9 Using Formats to Group Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .97
10 Using a SAS Format to Group Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .97

INTRODUCTION
The problems in this chapter include some simple one way frequency tables, two- and n-way
tables, and problems requiring an output data set from PROC FREQ. For more statistically
oriented problems involving PROC FREQ, see Chapter 23 (“Tests of Proportions”) in the
statistics section of the book.

PROBLEM 1
Producing Simple One-way Frequencies

Tools
PROC FREQ
TABLES statement

Data
Raw data file CARS.DTA

Directions
Using the raw data file CARS.DTA, create a SAS data set and use PROC FREQ to count the
number of small, medium and large cars (SIZE) and the frequencies of different manufacturers
(MANUFACT).
94 The SAS Workbook

PROBLEM 2
Producing Simple One-way Frequencies Omitting Percentages and Cumulative Values

Tools
PROC FREQ
TABLES statement
NOPERCENT and NOCUM options

Data
Raw data file CARS.DTA

Directions
Repeat Problem 1 except include the options to omit percentages and cumulative values.

PROBLEM 3
Demonstrating how Missing Values Are Handled

Tools
PROC FREQ
TABLES statement
MISSING option

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, compute one-way frequencies for the variable SEC_DX.
Do this first without the option MISSING on the TABLES statement and then with this option.
Compare the two results.

PROBLEM 4
Demonstrating the Effects of the ORDER= Option

Tools
PROC FREQ
ORDER= option
TABLES statement
 CHAPTER 15 The FREQ Procedure 95

Data
SAS data set CLINICAL

Directions
Again, using the SAS data set CLINICAL, compute one-way frequencies for the variable
PRIM_DX using the ORDER= option with four different values (DATA, FORMATTED, FREQ,
and INTERNAL). Compare the results.

PROBLEM 5
Producing Two-way Frequency Tables

Tools
PROC FREQ
TABLES statement
LIST option

Data
Raw data file DEMOG1.DTA

Directions
Using the data from the file DEMOG1.DTA, count the number of males and females (GENDER)
that are employed and unemployed (EMPLOYED). Run this with and without the LIST option
on the TABLES statement and compare the results.

PROBLEM 6
Requesting Multiple Two-way Tables

Tools
PROC FREQ
TABLES statement

Data
Raw data file CPR.DTA

Directions
Using the data from the file CPR.DTA, create three 2 X 2 tables as follows: V_FIB by SURVIVE,
RESP by SURVIVE, and AGEGROUP by SURVIVE. Do this with one TABLES statement,
making the statement as compact as possible.
96 The SAS Workbook

PROBLEM 7
Creating an Output Data Set Using PROC FREQ

Tools
PROC FREQ
TABLES statement
OUT= and NOPRINT options

Data
SAS data set CLINICAL

Directions
Use PROC FREQ to create a new data set called COUNTS containing the frequencies of
PRIM_DX and SEC_DX from the data set CLINICAL. Provide a listing of this data set.

PROBLEM 8
Producing a Three-way Table

Tools
PROC FREQ
TABLES statement

Data
SAS data set DEMOG1 created in Problem 5

Directions
Using the SAS data set DEMOG1 created in Problem 5, generate a three-way table showing
the number of employed (EMPLOYED) males and females (GENDER) for each state (STATE).
 CHAPTER 15 The FREQ Procedure 97

PROBLEM 9
Using Formats to Group Variables

Tools
PROC FREQ
PROC FORMAT

Data
SAS data set CARS created in Problem 1

Directions
Using the SAS data set CARS created in Problem 1, produce a table showing the car size (SIZE)
by mileage range defined as 0 to 9, 10 to 19, 20 to 29, 30 and above. Create a format for these
ranges and add a format statement to PROC FREQ. Do not use a separate data set to create a
variable for mileage range.

PROBLEM 10
Using a SAS Format to Group Variables

Tools
PROC FREQ
MONTH. MONNAME3. (WEEKDATE3. or DOWNAME3.) Formats

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, compute frequencies for the number of visits per month
(try listing the month as a number from 1 to 12 or as a three-letter abbreviation) and the number
of visits for each of the days of the week, displayed as a three-letter abbreviation. Do this without
using any new DATA steps!

Hint
You will have to run PROC FREQ each time with a different FORMAT statement.
As a special challenge, see if you can list the days of the week (displayed as three-letter
abbreviations) in order from Sunday to Saturday. You may need a DATA step to meet this
challenge.
98
 99

CHAPTER 16 The MEANS Procedure

PROBLEMS
1 Producing Simple Summary Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .99
2 Producing Summary Statistics Broken Down by a Single Variable . . . . . . . . . . . . . . .100
3 Creating a Summary Output Data Set with PROC MEANS: One Class Variable . . .100
4 Creating a Summary Output Data Set with PROC MEANS: Two Class Variables . .101
5 Counting the Number of Missing and Non-missing Values . . . . . . . . . . . . . . . . . . . . .102

INTRODUCTION
This chapter contains non-statistical applications of PROC MEANS. This procedure is useful
for counting the number of missing and non-missing observations as well as the more usual
computing of means and sums. The problems presented here involve some simple uses of PROC
MEANS to produce output listings as well as the slightly more complicated use of PROC
MEANS to produce summary output data sets. Now, for the problems.

PROBLEM 1
Producing Simple Summary Statistics

Tools
PROC MEANS
MAXDEC= option
N, MEAN, and NMISS statistics keywords
VAR statement

Data
Raw data file CARS.DTA

Directions
Using the raw data file CARS.DTA, create a SAS data set called CARS and compute the number
of non-missing observations, the number of missing observations, and the mean for the variables
RELIABLE and MILEAGE. Have PROC MEANS print the means to one decimal place.
100 The SAS Workbook

Note
You may want to include all the variables in the CARS.DTA file when you create your data set
since they are needed for other problems in this chapter.

PROBLEM 2
Producing Summary Statistics Broken Down by a Single Variable

Tools
PROC MEANS
MAXDEC= option
N, MEAN, and NMISS statistics keywords
VAR and CLASS statements

Data
SAS data set CARS created in Problem 1

Directions
Rerun Problem 1, except compute the requested statistics for each car SIZE. Use a CLASS
statement instead of first sorting the data and using a BY statement.

PROBLEM 3
Creating a Summary Output Data Set with PROC MEANS: One Class Variable

Tools
PROC MEANS
NWAY and NOPRINT options
OUTPUT statement
OUT=, MEAN=, and N= options
DROP= data set option

Data
SAS data set CARS created in Problem 1
 CHAPTER 16 The MEANS Procedure 101

Directions
Use PROC MEANS to produce an output data set called CAR_AVE, containing the mean and
number of non-missing observations for the variables reliability (RELIABLE) and MILEAGE
for each SIZE of car. You only want the _TYPE_ = 1 observations in this data set (i.e., not the
mean for all sizes of cars). A PROC MEANS option accomplishes this. Also, use the DROP=
data set option to drop the variables _TYPE_ and _FREQ_ from the output data set. Use the
variable names M_RELIAB and M_MILE for the means of reliability and mileage. Use the
variable names N_RELIAB and N_MILE for the number of non-missing values. Display the
contents of this output data set.

PROBLEM 4
Creating a Summary Output Data Set with PROC MEANS: Two Class Variables

Tools
PROC MEANS
NWAY and NOPRINT options
OUTPUT statement
OUT=, MEAN=, and N= options

Data
SAS data set CARS created in Problem 1

Directions
Repeat Problem 3, except create a SAS data set called CAR_MEAN which contains the means
and N for every combination of manufacturer (MANUFACT) and SIZE. Use the same variable
names for the means and counts as in Problem 3. The output data set should contain only the
_TYPE_ = 3 observations. Use a data set option to drop the variables _TYPE_ and _FREQ_
from the output data set. Display the contents of this output data set.
102 The SAS Workbook

PROBLEM 5
Counting the Number of Missing and Non-missing Values

Tools
PROC MEANS
NOPRINT option
OUTPUT statement
OUT=, N, and NMISS options
DROP= and RENAME= data set options
PROC PRINT
LABEL and NOOBS options

Data
SAS data set MISSING created in Chapter 12, Problem 7

Directions
Repeat Problem 7 in Chapter 12, counting the number of missing and non-missing values for
the numeric variables only. Use PROC MEANS, instead of PROC FORMAT, and PROC FREQ
to do this. First create an output data set and then use PROC PRINT to print out the results in a
listing similar to the following one.
Number of Missing and Non-missing Values

X X Y Y Z Z
Total Non-missing Missing Non-missing Missing Non-missing Missing

7 5 2 5 2 6 1

Note
The variable _FREQ_ has been renamed to TOTAL (and labeled Total). The number of
non-missing and missing values for the variables X, Y, and Z have been labeled as shown in
the listing in the Directions section.
 103

CHAPTER 17 The TABULATE Procedure

PROBLEMS
1 Creating a Simple One-way Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .104
2 Creating a Simple Two-way Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .104
3 Computing Counts and Percentages in a Two-way Table . . . . . . . . . . . . . . . . . . . . .105
4 Selecting a Denominator for a Percentage Calculation . . . . . . . . . . . . . . . . . . . . . . .106
5 Creating a Three-way Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .107
6 Creating a Three-way Table with Percentages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .109
7 Generating Simple Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .111
8 Producing Simple Descriptive Statistics Broken Down by a Single Variable
(Statistics Nested within the Row Variable) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .112
9 Producing Simple Descriptive Statistics Broken Down by a Single Variable
(Statistics Nested within the Column Variable) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .113
10 Producing Simple Descriptive Statistics Broken Down by a Single Variable
(Breakdown Variable Nested within the Computation Variable) . . . . . . . . . . . . . . . .114
11 Producing Statistics Broken Down by Two Variables . . . . . . . . . . . . . . . . . . . . . . . . .116

INTRODUCTION
The powerful TABULATE procedure can be used to create multi-way tables of counts,
percentages, means, and other statistics. This procedure is thought by some to be difficult to
learn. One useful approach is to draw the outlines of the table you want and work backwards
to the code. By working your way through the problems that follow, you will gain some
proficiency with this extremely useful and not-so-frightening procedure.
All of the problems in this chapter use data from the SAS data set CLINICAL that can be
generated by running the program CLINICAL.SAS in the Appendix. Remember that this data
set has user-defined formats and labels for many of the variables. For some of the problems,
you need to create a new data set which adds the variables MONTH and DAY (of the week),
corresponding to the visit date, to the original CLINICAL data set. Finally, the tables shown
here are centered on the page with the page numbers and dates omitted from the title line. You
may choose to use SAS system options to accomplish this in your output or to use other system
options that you prefer.
104 The SAS Workbook

PROBLEM 1
Creating a Simple One-way Table

Tools
PROC TABULATE
CLASS and TABLE statements

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, generate this table.
Simple One-way Table
---------------------------
| Gender |
|-------------------------|
| Female | Male |
|------------+------------|
| N | N |
|------------+------------|
| 15.00| 16.00|
---------------------------

PROBLEM 2
Creating a Simple Two-way Table

Tools
PROC TABULATE
CLASS and TABLE statements
ALL class variable

Data
SAS data set CLINICAL
 CHAPTER 17 The TABULATE Procedure 105

Directions
Using the SAS data set CLINICAL, generate this table.
Two-way Table
----------------------------------------------------------
| | Pt. Taking Vitamins? | |
| |-------------------------| |
| | No | Yes | ALL |
| |------------+------------+------------|
| | N | N | N |
|-----------------+------------+------------+------------|
|Gender | | | |
|-----------------| | | |
|Female | 6.00| 9.00| 15.00|
|-----------------+------------+------------+------------|
|Male | 10.00| 6.00| 16.00|
|-----------------+------------+------------+------------|
|ALL | 16.00| 15.00| 31.00|
----------------------------------------------------------

Note
The column variable is VITAMINS which was given the label Pt. Taking Vitamins in
the DATA step that created CLINICAL.

PROBLEM 3
Computing Counts and Percentages in a Two-way Table

Tools
PROC TABULATE
FORMAT= option
CLASS, KEYLABEL, and LABEL statements
TABLE statement
RTSPACE= option
N and PCTN keywords
ALL CLASS variable

Data
SAS data set CLINICAL
106 The SAS Workbook

Directions
Using the SAS data set CLINICAL, generate the table below. Note that the column variable is
VITAMINS, and it has been assigned a label. The default table format is 7.0, and the RTSPACE=
is set to 25. The KEYLABEL statement assigned the labels All, Number, and Percent to the
terms ALL, N, and PCTN, respectively. Notice also that although the variables GENDER and
VITAMINS had labels assigned to them in the CLINICAL data set, the table shown here has
substituted other labels for these two variables (just to give you practice using a LABEL statement
within a procedure).
Counts and Percentages in a Two-way Table
-------------------------------------------------------------------------
| | Taking Vitamins? | |
| |-------------------------------| |
| | No | Yes | All |
| |---------------+---------------+---------------|
| |Number |Percent|Number |Percent|Number |Percent|
|-----------------------+-------+-------+-------+-------+-------+-------|
|Patient Gender | | | | | | |
|-----------------------| | | | | | |
|Female | 6| 19| 9| 29| 15| 48|
|-----------------------+-------+-------+-------+-------+-------+-------|
|Male | 10| 32| 6| 19| 16| 52|
|-----------------------+-------+-------+-------+-------+-------+-------|
|All | 16| 52| 15| 48| 31| 100|
-------------------------------------------------------------------------

PROBLEM 4
Selecting a Denominator for a Percentage Calculation

Tools
PROC TABULATE
FORMAT= option
CLASS, KEYLABEL, and LABEL statements
TABLE statement
RTSPACE= option
N and PCTN keywords
ALL class variable
Brackets (< >) operator (specifies denominator definition)

Data
SAS data set CLINICAL
 CHAPTER 17 The TABULATE Procedure 107

Directions
Using the SAS data set CLINICAL, create the following table. The default table format is 7.0,
and the RTSPACE= option is set to 20. The percentages add up to 100 for the people taking
vitamins and the people not taking vitamins. The KEYLABEL statement assigned the labels
All, Number, and Percent to the keywords ALL, N, and PCTN, respectively.
Selecting a Denominator for a Percentage
------------------------------------------------------------
| | Taking Vitamins? | |
| |-------------------------------| |
| | No | Yes | All |
| |---------------+---------------+-------|
| |Number |Percent|Number |Percent|Number |
|------------------+-------+-------+-------+-------+-------|
|Gender | | | | | |
|------------------| | | | | |
|Female | 6| 40| 9| 60| 15|
|------------------+-------+-------+-------+-------+-------|
|Male | 10| 63| 6| 38| 16|
|------------------+-------+-------+-------+-------+-------|
|All | 16| 52| 15| 48| 31|
------------------------------------------------------------

PROBLEM 5
Creating a Three-way Table

Tools
PROC FORMAT
PROC TABULATE
CLASS, LABEL, and KEYLABEL statements
TABLE statement
PRINTMISS, BOX=, and MISSTEXT= options
MONTH and WEEKDAY functions

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, create the two following tables. You will need to create a
new data set from CLINICAL that contains the month of the year and the day of the week
corresponding to the visit date. You will also need to create a format for the month and day
108 The SAS Workbook

abbreviations. The label in the upper left box of each table is the value of the variable GENDER.
The keyword N has been labeled Count. The labels Month of Visit and Day of Visit
have been assigned to the variables MONTH and DAY, respectively. Missing values are displayed
as blanks (not the default period).
Three-way Table - Visits by Gender, Month, and Day of Visit
-------------------------------------------------------------
|Gender Female | Day of Visit |
| |-----------------------------------------|
| | Sun | Mon | Tue | Wed | Thu | Fri | Sat |
| |-----+-----+-----+-----+-----+-----+-----|
| |Count|Count|Count|Count|Count|Count|Count|
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Month of Visit | | | | | | | |
|-----------------| | | | | | | |
|Feb | | 1| | | | | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Mar | | | | | | | 1|
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|May | | 2| 1| | | | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Jun | | | 1| | 1| | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Jul | | | | | | | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Aug | | | 1| 1| | | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Sep | | 3| | | | | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Oct | | | | | 1| | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Nov | | | | 1| | | 1|
-------------------------------------------------------------
 CHAPTER 17 The TABULATE Procedure 109

Three-way Table - Visits by Gender, Month, and Day of Visit

-------------------------------------------------------------
|Gender Male | Day of Visit |
| |-----------------------------------------|
| | Sun | Mon | Tue | Wed | Thu | Fri | Sat |
| |-----+-----+-----+-----+-----+-----+-----|
| |Count|Count|Count|Count|Count|Count|Count|
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Month of Visit | | | | | | | |
|-----------------| | | | | | | |
|Feb | | | | | | | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Mar | | | | | | | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|May | | 1| | | 2| | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Jun | 1| | | | | | 1|
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Jul | 2| | | 1| | 1| 1|
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Aug | | | | 1| | | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Sep | | 1| | | | | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Oct | 1| | 1| 1| | | |
|-----------------+-----+-----+-----+-----+-----+-----+-----|
|Nov | | | | | | | 1|
-------------------------------------------------------------

PROBLEM 6
Creating a Three-way Table with Percentages

Tools
PROC FORMAT
PROC TABULATE
CLASS, LABEL, and KEYLABEL statements
TABLE statement
PRINTMISS, BOX=, and MISSTEXT= options
N and PCTN keywords
Brackets (< >) operator (specifies denominator definition)
MONTH and WEEKDAY functions
110 The SAS Workbook

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, create the two following tables. They are similar to the
tables in Problem 5 with the addition of the percent columns. You will need to create a new
data set from CLINICAL (or use the one you created in Problem 5) that contains the month of
the year and the day of the week corresponding to the visit date. You will also need to create a
format for the month and day abbreviations. The label in the upper left box of each table is the
value of the variable GENDER. The keyword N has been labeled N and the keyword PCTN has
been labeled %. The labels Month of Visit and Day of Visit have been assigned to
the variables MONTH and DAY, respectively. The default table format is 3.0. Missing values
are displayed as blanks (not the default period). Note that the percentages add up to 100% for each
month.
Three-way Table - Visits by Gender, Month, and Day of Visit
Percentages Added
---------------------------------------------------------------------------
|Gender Female | Day of Visit |
| |-------------------------------------------------------|
| | Sun | Mon | Tue | Wed | Thu | Fri | Sat |
| |-------+-------+-------+-------+-------+-------+-------|
| | N | % | N | % | N | % | N | % | N | % | N | % | N | % |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Month of Visit | | | | | | | | | | | | | | |
|-----------------| | | | | | | | | | | | | | |
|Feb | | | 1|100| | | | | | | | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Mar | | | | | | | | | | | | | 1|100|
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|May | | | 2| 67| 1| 33| | | | | | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Jun | | | | | 1| 50| | | 1| 50| | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Jul | | | | | | | | | | | | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Aug | | | | | 1| 50| 1| 50| | | | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Sep | | | 3|100| | | | | | | | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Oct | | | | | | | | | 1|100| | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Nov | | | | | | | 1| 50| | | | | 1| 50|
---------------------------------------------------------------------------
 CHAPTER 17 The TABULATE Procedure 111

Three-way Table - Visits by Gender, Month, and Day of Visit

Percentages Added
---------------------------------------------------------------------------
|Gender Male | Day of Visit |
| |-------------------------------------------------------|
| | Sun | Mon | Tue | Wed | Thu | Fri | Sat |
| |-------+-------+-------+-------+-------+-------+-------|
| | N | % | N | % | N | % | N | % | N | % | N | % | N | % |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Month of Visit | | | | | | | | | | | | | | |
|-----------------| | | | | | | | | | | | | | |
|Feb | | | | | | | | | | | | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Mar | | | | | | | | | | | | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|May | | | 1| 33| | | | | 2| 67| | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Jun | 1| 50| | | | | | | | | | | 1| 50|
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Jul | 2| 40| | | | | 1| 20| | | 1| 20| 1| 20|
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Aug | | | | | | | 1|100| | | | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Sep | | | 1|100| | | | | | | | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Oct | 1| 33| | | 1| 33| 1| 33| | | | | | |
|-----------------+---+---+---+---+---+---+---+---+---+---+---+---+---+---|
|Nov | | | | | | | | | | | | | 1|100|
---------------------------------------------------------------------------

PROBLEM 7
Generating Simple Descriptive Statistics

Tools
PROC TABULATE
VAR, TABLE, KEYLABEL, and LABEL statements
Specifying individual formats for table elements
N, NMISS, MEAN, STD, MIN, and MAX statistics keywords

Data
SAS data set CLINICAL
112 The SAS Workbook

Directions
Using the SAS data set CLINICAL, create the following table. The row variables are HR, SBP,
and DBP. The default table format is 8.0, and formats for N, MEAN, and STD are 7.0, 6.1, and
6.2, respectively. The keywords N, NMISS, MEAN, STD, MIN, and MAX have been labeled
Number, Missing, Mean, S.D., Minimum, and Maximum, respectively. The RTSPACE=
option is set to 20.
Basic Descriptive Statistics
---------------------------------------------------------------------
| |Number |Missing | Mean | S.D. |Minimum |Maximum |
|------------------+-------+--------+------+------+--------+--------|
|Heart Rate | 30| 1| 78.5| 10.17| 58| 92|
|------------------+-------+--------+------+------+--------+--------|
|Systolic BP | 30| 1| 134.0| 39.91| 54| 220|
|------------------+-------+--------+------+------+--------+--------|
|Diastolic BP | 30| 1| 82.9| 27.50| 36| 180|
---------------------------------------------------------------------

PROBLEM 8
Producing Simple Descriptive Statistics Broken Down by a Single Variable (Statistics
Nested within the Row Variable)

Tools
PROC TABULATE
VAR, CLASS, TABLE, KEYLABEL, and LABEL statements
Specifying individual formats for table elements
N, MEAN, STD, and STDERR statistics keywords

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, create the following table. The row variables are HR, SBP,
and DBP. No formats are specified. The keywords N, MEAN, STD, STDERR, and the class
variable ALL have been labeled Number, Mean, S.D., S.E., and All, respectively. The
RTSPACE= option is set to 30. The labels Heart Rate, Systolic BP, Diastolic BP,
and Gender have been assigned to the variables HR, SBP, DBP, and GENDER, respectively.
 CHAPTER 17 The TABULATE Procedure 113

Simple Statistics Broken Down by a Single Variable

---------------------------------------------------------------------
| | Gender | |
| |-------------------------| |
| | Female | Male | All |
|----------------------------+------------+------------+------------|
|Heart Rate |Number | 15.00| 15.00| 30.00|
| |--------------+------------+------------+------------|
| |Mean | 81.33| 75.67| 78.50|
| |--------------+------------+------------+------------|
| |S.D. | 8.44| 11.21| 10.17|
| |--------------+------------+------------+------------|
| |S.E. | 2.18| 2.89| 1.86|
|-------------+--------------+------------+------------+------------|
|Systolic BP |Number | 15.00| 15.00| 30.00|
| |--------------+------------+------------+------------|
| |Mean | 130.53| 137.47| 134.00|
| |--------------+------------+------------+------------|
| |S.D. | 46.44| 33.42| 39.91|
| |--------------+------------+------------+------------|
| |S.E. | 11.99| 8.63| 7.29|
|-------------+--------------+------------+------------+------------|
|Diastolic BP |Number | 15.00| 15.00| 30.00|
| |--------------+------------+------------+------------|
| |Mean | 74.13| 91.60| 82.87|
| |--------------+------------+------------+------------|
| |S.D. | 21.32| 30.79| 27.50|
| |--------------+------------+------------+------------|
| |S.E. | 5.50| 7.95| 5.02|
---------------------------------------------------------------------

PROBLEM 9
Producing Simple Descriptive Statistics Broken Down by a Single Variable (Statistics
Nested within the Column Variable)

Tools
PROC TABULATE
VAR, CLASS, TABLE, KEYLABEL, and LABEL statements
Specifying individual formats for table elements
N, MEAN, STD, and STDERR statistics keywords
114 The SAS Workbook

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, create the following table. Notice that the information
contained in this table is similar to that in Problem 8, except that the standard error is omitted and
the statistics are now displayed within each value of GENDER. Formats for the statistics N,
MEAN, and STD are 6.0, 5.1, and 5.2, respectively. The labels for statistics and variables are
identical to the labels in Problem 8. The RTSPACE= option is set to 15.
Simple Statistics Broken Down by a Single Variable
------------------------------------------------------------------------
| | Gender | |
| |-------------------------------------| |
| | Female | Male | ALL |
| |------------------+------------------+------------------|
| |Number|Mean |S.D. |Number|Mean |S.D. |Number|Mean |S.D. |
|-------------+------+-----+-----+------+-----+-----+------+-----+-----|
|Heart Rate | 15| 81.3| 8.44| 15| 75.7|11.21| 30| 78.5|10.17|
|-------------+------+-----+-----+------+-----+-----+------+-----+-----|
|Systolic BP | 15|130.5|46.44| 15|137.5|33.42| 30|134.0|39.91|
|-------------+------+-----+-----+------+-----+-----+------+-----+-----|
|Diastolic BP | 15| 74.1|21.32| 15| 91.6|30.79| 30| 82.9|27.50|
------------------------------------------------------------------------

PROBLEM 10
Producing Simple Descriptive Statistics Broken Down by a Single Variable (Breakdown
Variable Nested within the Computation Variable)

Tools
PROC TABULATE
VAR, CLASS, KEYLABEL, and LABEL statements
TABLE statement
RTSPACE=, PRINTMISS, and BOX= options
Specifying individual formats for table elements
N, MEAN, NMISS, MEAN, STD, MIN, and MAX statistics keywords

Data
SAS data set CLINICAL
 CHAPTER 17 The TABULATE Procedure 115

Directions
Using the SAS data set CLINICAL, create the following table. The keywords N, NMISS, MEAN,
STD, MIN, and MAX have been assigned labels Number, Missing, Mean, S.D., Minimum,
and Maximum, respectively. The text Place Your Ad Here is inserted in the upper
left-hand box. The default table format is 8.0, and formats for the statistics N, MEAN, and STD
are 7.0, 6.1, and 6.2, respectively. The RTSPACE= option is set to 28.
Basic Descriptive Statistics
Broken Down by Gender
-----------------------------------------------------------------------------
|Place Your Ad Here |Number |Missing | Mean | S.D. |Minimum |Maximum |
|--------------------------+-------+--------+------+------+--------+--------|
|Heart Rate |Gender | | | | | | |
| |-------------| | | | | | |
| |Female | 15| 0| 81.3| 8.44| 68| 92|
| |-------------+-------+--------+------+------+--------+--------|
| |Male | 15| 1| 75.7| 11.21| 58| 88|
| |-------------+-------+--------+------+------+--------+--------|
| |Total | 30| 1| 78.5| 10.17| 58| 92|
|------------+-------------+-------+--------+------+------+--------+--------|
|Systolic BP |Gender | | | | | | |
| |-------------| | | | | | |
| |Female | 15| 0| 130.5| 46.44| 54| 200|
| |-------------+-------+--------+------+------+--------+--------|
| |Male | 15| 1| 137.5| 33.42| 98| 220|
| |-------------+-------+--------+------+------+--------+--------|
| |Total | 30| 1| 134.0| 39.91| 54| 220|
|------------+-------------+-------+--------+------+------+--------+--------|
|Diastolic BP|Gender | | | | | | |
| |-------------| | | | | | |
| |Female | 15| 0| 74.1| 21.32| 36| 110|
| |-------------+-------+--------+------+------+--------+--------|
| |Male | 15| 1| 91.6| 30.79| 62| 180|
| |-------------+-------+--------+------+------+--------+--------|
| |Total | 30| 1| 82.9| 27.50| 36| 180|
-----------------------------------------------------------------------------
116 The SAS Workbook

PROBLEM 11
Statistics Broken Down by Two Variables

Tools
PROC TABULATE
CLASS, VAR, KEYLABEL, and LABEL statements
TABLE statement
RTSPACE= and PRINTMISS options
Data
SAS data set CLINICAL
Directions
Using the SAS data set CLINICAL, create the following table. The labels Number, Mean, and
Both have been assigned to statistics N, MEAN, and ALL, respectively. The variables HR,
SBP, and DBP are labeled Heart Rate, Systolic BP, and Diastolic BP, respectively.
The row variables are VITAMINS and GENDER. The format for N is 6.0, and the format for
MEAN is 5.1.
Simple Statistics Broken Down by Two Variables
----------------------------------------------------------------------------
| | Heart Rate |Systolic BP |Diastolic BP|
| |------------+------------+------------|
| |Number|Mean |Number|Mean |Number|Mean |
|-----------------------------------+------+-----+------+-----+------+-----|
|Taking Vitamins? |Gender | | | | | | |
|-----------------+-----------------| | | | | | |
|No |Female | 6| 82.0| 6|150.0| 6| 84.0|
| |-----------------+------+-----+------+-----+------+-----|
| |Male | 10| 75.1| 10|142.6| 10| 98.8|
| |-----------------+------+-----+------+-----+------+-----|
| |Both | 16| 77.7| 16|145.4| 16| 93.3|
|-----------------+-----------------+------+-----+------+-----+------+-----|
|Yes |Gender | | | | | | |
| |-----------------| | | | | | |
| |Female | 9| 80.9| 9|117.6| 9| 67.6|
| |-----------------+------+-----+------+-----+------+-----|
| |Male | 5| 76.8| 5|127.2| 5| 77.2|
| |-----------------+------+-----+------+-----+------+-----|
| |Both | 14| 79.4| 14|121.0| 14| 71.0|
|-----------------+-----------------+------+-----+------+-----+------+-----|
|Both |Gender | | | | | | |
| |-----------------| | | | | | |
| |Female | 15| 81.3| 15|130.5| 15| 74.1|
| |-----------------+------+-----+------+-----+------+-----|
| |Male | 15| 75.7| 15|137.5| 15| 91.6|
| |-----------------+------+-----+------+-----+------+-----|
| |Both | 30| 78.5| 30|134.0| 30| 82.9|
----------------------------------------------------------------------------
 117

CHAPTER 18 The CHART Procedure

PROBLEMS
1 Creating a Vertical Bar Chart to Display Frequencies (Character Variables) . . . . . . .117
2 Creating a Horizontal Bar Chart to Display Frequencies (Character Variables) . . . .118
3 Creating a Bar Chart Displaying Percentages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .118
4 Creating a Bar Chart for a Numeric Variable (Discrete Values) . . . . . . . . . . . . . . . . .118
5 Creating a Bar Chart for a Numeric Variable Using a Variety of Methods for
Choosing Midpoints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .119
6 Creating a Bar Chart Displaying Sums of One Variable for Different Values
of Another Variable . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .119
7 Representing Two Variables on a Single Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .120
8 Using the Value of a Variable as a Charting Symbol . . . . . . . . . . . . . . . . . . . . . . . . . .121
9 Creating a Three-dimensional Chart . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .122

INTRODUCTION
Although you can make fancy, publication-quality charts using SAS/GRAPH software, PROC
CHART still serves a useful purpose. It is fast, easy to use, and it produces useful displays of one
or more variables.

PROBLEM 1
Creating a Vertical Bar Chart to Display Frequencies (Character Variables)

Tools
PROC CHART
VBAR statement

Data
Raw data file CARS.DTA
Directions
Using the raw data file CARS.DTA (or the SAS data set created in Chapter 2, Problem 1), create
two vertical bar charts showing the frequencies for the variables SIZE and MANUFACT.
118 The SAS Workbook

PROBLEM 2
Creating a Horizontal Bar Chart to Display Frequencies (Character Variables)

Tools
PROC CHART
HBAR statement
NOSTAT option

Data
Raw data file CARS.DTA or SAS data set CARS from Problem 1

Directions
Repeat Problem 1, but this time produce horizontal bar charts. Do this first with the default
statistics and then with the NOSTAT option to delete the statistics.

PROBLEM 3
Creating a Bar Chart Displaying Percentages

Tools
PROC CHART
VBAR statement
TYPE=PERCENT option

Data
Raw data file CARS.DTA or SAS data set CARS from Problem 1

Directions
Using data from the raw data file CARS.DTA or the SAS data set created in Problem 1, create
a single vertical bar chart showing the percentages of cars of each SIZE.

PROBLEM 4
Creating a Bar Chart for a Numeric Variable (Discrete Values)

Tools
PROC CHART
VBAR statement
DISCRETE option
 CHAPTER 18 The CHART Procedure 119

Data
Raw data file CARS.DTA or SAS data set CARS from Problem 1

Directions
Using data from the raw data file CARS.DTA or the SAS data set created in Problem 1, produce
a vertical bar chart showing the frequencies of reliability ratings (RELIABLE). The values for
the variable RELIABLE should be the integers 1 to 5.

PROBLEM 5
Creating a Bar Chart for a Numeric Variable Using a Variety of Methods for Choosing
Midpoints

Tools
PROC CHART
VBAR statement
MIDPOINTS= and LEVELS= options

Data
Raw data file CARS.DTA or SAS data set CARS from Problem 1

Directions
Using data from the raw data file CARS.DTA or the SAS data set created in Problem 1, produce
vertical bar charts for the variable MILEAGE that employ
A. system-chosen (default) midpoints
B. the MIDPOINTS= option to select the following midpoints: 15, 20, 25, 30, 35, and 40
C. the LEVELS= option to generate five bars.

PROBLEM 6
Creating a Bar Chart Displaying Sums of One Variable for Different Values of
Another Variable

Tools
PROC CHART
VBAR statement
SUMVAR= and TYPE= options
Data
Raw data file BASKET.DTA
120 The SAS Workbook

Directions
Using the BASKET.DTA data file, create a SAS data set and display the total number of POINTS
for males and females in a vertical chart.

PROBLEM 7
Representing Two Variables on a Single Chart

Tools
PROC CHART
VBAR statement
GROUP= and DISCRETE options
Data
Raw data file CARS.DTA or SAS data set CARS from Problem 1
Directions
Using data from the CARS.DTA data file or the SAS data set created in Problem 1, create a
vertical bar chart showing the distribution of the reliability index (RELIABLE) for each size of
car (SIZE). Your chart should look like the following one.
Reliability by Car Size

Frequency

3 + ** **
| ** **
| ** **
| ** **
| ** **
2 + ** ** ** **
| ** ** ** **
| ** ** ** **
| ** ** ** **
| ** ** ** **
1 + ** ** ** ** ** ** **
| ** ** ** ** ** ** **
| ** ** ** ** ** ** **
| ** ** ** ** ** ** **
| ** ** ** ** ** ** **
--------------------------------------------------------------
1 2 3 5 1 2 3 5 1 2 3 5 RELIABLE

|-- COMPACT -| MID-SIZED |--- SMALL --| SIZE

 CHAPTER 18 The CHART Procedure 121

Notes
1. There are no cars with a value of 4 for reliability. Treat the values of reliability (RELIABLE)
as discrete values in your procedure.
2. System options PS=48 and LS=74 were used to produce this chart and the remaining charts
in this chapter.

PROBLEM 8
Using the Value of a Variable as a Charting Symbol

Tools
PROC CHART
VBAR statement
SUBGROUP= option

Data
Raw data file CARS.DTA or SAS data set CARS from Problem 1

Directions
Using data from the CARS.DTA data file or the SAS data set created in Problem 1, display the
frequency distribution showing the number of cars for each manufacturer. Use the first character
of the variable SIZE to draw the bars. Your chart should look like the following one.
Distribution by Manufacturer and Car Size

Frequency

3 + SSSS
| SSSS
| SSSS
| SSSS
| SSSS
2 + SSSS MMMM SSSS SSSS
| SSSS MMMM SSSS SSSS
| SSSS MMMM SSSS SSSS
| SSSS MMMM SSSS SSSS
| SSSS MMMM SSSS SSSS
122 The SAS Workbook

1 + CCCC CCCC CCCC SSSS CCCC MMMM CCCC CCCC MMMM

| CCCC CCCC CCCC SSSS CCCC MMMM CCCC CCCC MMMM
| CCCC CCCC CCCC SSSS CCCC MMMM CCCC CCCC MMMM
| CCCC CCCC CCCC SSSS CCCC MMMM CCCC CCCC MMMM
| CCCC CCCC CCCC SSSS CCCC MMMM CCCC CCCC MMMM
---------------------------------------------------------
B C C D F H P P T
U H H O O O L O O
I E R D R N Y N Y
C V Y G D D M T O
K R S E A O I T
O L U A A
L E T C
E R H
T

MANUFACT

Symbol SIZE Symbol SIZE Symbol SIZE

C COMPACT M MID-SIZED S SMALL

PROBLEM 9
Creating a Three-dimensional Chart

Tools
PROC CHART
BLOCK statement
GROUP=, SUMVAR=, and TYPE= options

Data
Raw data file CARS.DTA or SAS data set CARS from Problem 1

Directions
Using data from the CARS.DTA data file or the SAS data set created in Problem 1, create a
three-dimensional chart showing the average (mean) gas mileage (z-axis) for each level of
reliability (x-axis) and SIZE (y-axis). The block chart should look like the following one.
 CHAPTER 18 The CHART Procedure 123

Three-Dimensional Block Chart

Mean of MILEAGE by RELIABLE grouped by SIZE

___
___ /_ /|
/_ /| |**| |
|**| | |**| |
|**| | |**| |
|**| | |**| |
|**| | |**| |
-----------------------------|**| |--------|**| |----
/ / / |**| | / |**| |
/ / / |**| | ___ |**| |
SMALL / / ___ |**| | /_ /| |**| |
/ / /_ /| |**|/ |**| | |**|/ /
/ / |**| | |**| | /
/ / |**| | 27 |**| | 31 /
/-------------/-------------/|**| |-------/|**| |-------/
/ / / |**| | / |**| | /
___ ___ / |**| | / |**| | /
MID-SIZED /_ /| /_ /| ___ |**| | / |**| | /
|**| | |**| | /_ /| |**|/ / |**|/ /
|**| | |**| | |**| | / /
|**| | |**| | |**| | 20 / 25 /
SIZE /|**| |-------/|**| |-------/|**| |-------/-------------/
/ |**| | / |**| | / |**| | / /
/ |**| | / |**| | / |**| | / /
COMPACT / |**| | / |**| | / |**| | / /
/ |**|/ / |**|/ / |**|/ / /
/ / / / /
/ 24 / 25 / 23 / /
/-------------/-------------/-------------/-------------/

1 2 3 5

RELIABLE
124
 125

CHAPTER 19 The PLOT Procedure

PROBLEMS
1 Producing a Simple X-Y Plot: Selecting Plotting Symbols . . . . . . . . . . . . . . . . . . . . . .125
2 Requesting Multiple Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .126
3 Labeling Points on a Scatter Plot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .126
4 Labeling Points on a Scatter Plot and Using the Value of a Variable as a
Plotting Symbol . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .127

INTRODUCTION
If you need high-quality, camera-ready graphs, you should consider using a SAS/GRAPH
procedure. However, PROC PLOT is useful and easy to use. It can give you a quick plot showing
relationships between variables. With a few simple statements, you can select plotting symbols
and label points on the plot.

PROBLEM 1
Producing a Simple X-Y Plot: Selecting Plotting Symbols

Tools
PROC PLOT
PLOT statement
Plotting symbol

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, do the following
A. Plot SBP (y-axis) by HR (x-axis) using the default plotting symbol.
B. Rerun (A) using a lowercase o as the plotting symbol.
C. Rerun (A) using the value of GENDER as the plotting symbol.
126 The SAS Workbook

PROBLEM 2
Requesting Multiple Plots

Tools
PROC PLOT
PLOT statement
OVERLAY option

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, produce the following plots.
A. Two separate plots: one of SBP (y-axis) by HR (x-axis), the other DBP (y-axis) by HR
(x-axis). Use the default plotting symbol.
B. The same two plots as in (A), but plot both graphs on the same set of axes. Use the letter
S as the plotting symbol for the SBP by HR plot and the letter D as the plotting symbol
for the DBP by HR plot.

PROBLEM 3
Labeling Points on a Scatter Plot

Tools
PROC PLOT
PLOT statement
Label variable

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, produce a plot of SBP (y-axis) by HR (x-axis). Label each
point with the patient’s diagnosis (PRIM_DX), and use a lowercase o as the plotting symbol.
Use SAS system options to set the pagesize to 48 and the linesize to 64.

Note
See page 230 in SAS Technical Report P-222, Changes and Enhancements to Base SAS Software,
Release 6.07.
 CHAPTER 19 The PLOT Procedure 127

PROBLEM 4
Labeling Points on a Scatter Plot and Using the Value of a Variable as a Plotting Symbol

Tools
PROC PLOT
PLOT statement
Labeling feature
PAGESIZE= , LINESIZE= , and PAGENO= system options

Data
SAS data set LEAD created by running the following program
DATA LEAD;
INPUT LAB $ SUBJECT $ LEVEL_1 LEVEL_2;
LABEL LEVEL_1 = 'Lead at Time 1'
LEVEL_2 = 'Lead at Time 2';
DATALINES;
Carter 1 3 3
Carter 2 6 6.5
Carter 3 9 9.5
H&K 1 3.5 3
H&K 2 7 7
H&K 3 10 10
Boston 1 1 2
Boston 2 5 6
Boston 3 7 6
;

Directions
A study was conducted to test the consistency of laboratories in measuring blood lead levels.
Each of three subjects had blood drawn. Two samples were sent to each of three different
laboratories, one immediately after being drawn and the other one a week later. Using the SAS
data set LEAD created in the Data section, produce a scatter plot of LEVEL_2 (y-axis) by
LEVEL_1 (x-axis) with the subject number as the plotting symbol. Label the points with the
value of the variable LAB. Set the pagesize to 24, the linesize to 64, and the page number to 1.
To help you understand what is required for this problem, see the following plot.
128 The SAS Workbook

Lead Levels at Time 1 and Time 2 by Lab and Subject

Plot of LEVEL_2*LEVEL_1$LAB. Symbol is value of SUBJECT.

L 10 + H&K 3
e 3 Carter
a
d
Carter 2 2 H&K
a 2 Boston 3 Boston
t 5 +

T Carter 1 1 H&K
i
m 1 Boston
e
0 +
2 --+-----+-----+-----+-----+-----+-----+-----+-----+-----+--
1 2 3 4 5 6 7 8 9 10

Lead at Time 1
 129

SECTION 3 SAS Programming Techniques

®

CHAPTERS
20 Efficient Programming . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .131

21 Longer and More Advanced Problems . . . . . . . . . . . . . . . . . . . . . . .141

130
 131

CHAPTER 20 Efficient Programming

PROBLEMS
1 Avoiding Unnecessary DATA Steps . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .133
2 Avoiding Unnecessary Sorts . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .134
3 Sorting Only What You Need . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .134
4 Using KEEP Statements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .135
5 Using DATA _NULL_ when a Data Set Is Not Needed . . . . . . . . . . . . . . . . . . . . . . . .135
6 Using LENGTH Statements where Possible (Numeric Example) . . . . . . . . . . . . . . . .136
7 Using LENGTH Statements to Control the Length of Character Variables . . . . . . . .137
8 Using PROC FORMAT to Regroup Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .137
9 Using a KEEP= Data Set Option instead of a KEEP Statement . . . . . . . . . . . . . . . .138
10 Using an Index to Increase Efficiency . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .139
11 Reading Only What You Need to from Raw Data . . . . . . . . . . . . . . . . . . . . . . . . . . .139
12 Arranging the Order of IF and IF-THEN/ELSE Statements . . . . . . . . . . . . . . . . . . . . .140

INTRODUCTION
Each of the problems in this chapter can be written more efficiently. There are times when a
“quick and dirty” program works fine, but learning and practicing the basic concepts of efficient
programming can reduce CPU time, disk access time, or both.
Some of the problems use the SAS data set CLINICAL. However, since it is such a small data
set, you may not notice any difference in execution time between the inefficient and efficient
versions of the programs. Therefore, I have provided the SAS code to generate two larger data
sets, EFF1 (10,000 observations and 5 variables) and EFF2 (1,000 observations and 100 variables).
To test programs that read raw data, I have also provided a program to create raw data file
RAWDATA with 10,000 observations and 10 variables.
Remember that some of the “rules” for efficient programming change with newer releases of
the SAS System and may depend on which computer or operating environment you are using.
See the Problem title and the Tools list for hints on how to approach each problem. Finally, if
you plan to “benchmark” each of the inefficient programs in this chapter with your more efficient
version, you may need to turn off, or take into account, any disk-caching software you may be
running (i.e., SMARTDISK with Microsoft Windows).
132 The SAS Workbook

Here are the programs to create data sets EFF1 and EFF2 and the raw data file RAWDATA.
*--------------------------------------------*
| Program to create SAS data set EFF1 |
*--------------------------------------------*;
DATA EFF1;
LENGTH X Y P1 P2 4 GROUP $ 1;
DO I = 1 TO 10000;
X = RANUNI(1357);
Y = RANUNI(2468);
P1 = (X LT .5);
P2 = (Y LT .001);
IF X > .8 THEN GROUP = 'A';
ELSE IF X > .6 THEN GROUP = 'B';
ELSE IF X > .4 THEN GROUP = 'C';
ELSE GROUP = 'D';
OUTPUT;
END;
DROP I;
RUN;

*--------------------------------------------*
| Program to create SAS data set EFF2 |
*--------------------------------------------*;
DATA EFF2;
ARRAY X[100] X1-X100;
DO I = 1 TO 1000;
DO J = 1 TO 100;
X[J] = RANUNI(1357);
END;
OUTPUT;
END;
DROP I J;
RUN;

*--------------------------------------------*
| Program to create data file RAWDATA |
*--------------------------------------------*;
DATA _NULL_;
FILE 'C:\WORKBOOK\RAWDATA'; ***Use a subdirectory of your choice
here;
 CHAPTER 20 Efficient Programming 133

LENGTH GROUP $ 1;
ARRAY X[10] X1-X10;
DO I = 1 TO 10000;
DO J = 1 TO 10;
X[J] = RANUNI(1357);
END;
IF X1 > .99 THEN GROUP = 'A';
ELSE GROUP = 'B';
PUT @1 (X1-X10) (5.3 + 1) @65 GROUP $1.;
END;
RUN;

PROBLEM 1
Avoiding Unnecessary DATA Steps

Tools
DATA= option
WHERE statement or WHERE= data set option

Data
SAS data set EFF1

Directions
Rewrite the following program to make it more efficient. Do not use a DATA step.
DATA NEW;
SET EFF1; *** or libref.EFF1 if you made it permanent;
WHERE P2 = 1;
RUN;

PROC MEANS N MEAN MAXDEC=2 DATA=NEW;

TITLE 'Descriptive Statistics on X and Y';
VAR X Y;
RUN;
134 The SAS Workbook

PROBLEM 2
Avoiding Unnecessary Sorts

Tools
PROC MEANS
CLASS statement

Data
SAS data set EFF1

Directions
Rewrite the following program, eliminating the need to sort the data set.
PROC SORT DATA=EFF1 OUT=TMP;
BY GROUP;
RUN;

PROC MEANS DATA=TMP N MEAN;

BY GROUP;
RUN;

PROBLEM 3
Sorting Only What You Need

Tools
PROC SORT
KEEP= and WHERE= data set options
Data
SAS data set EFF2
Directions
Rewrite the following program to make it more efficient.
PROC SORT DATA=EFF2 OUT=OUT1;
BY X1;
RUN;

DATA SUBSET;
SET OUT1;
WHERE X1 > .9;
KEEP X1;
RUN;
 CHAPTER 20 Efficient Programming 135

PROBLEM 4
Using KEEP Statements

Tools
KEEP statement
MEAN function

Data
SAS data set EFF2

Directions
Rewrite the following program to make it more efficient.
DATA DESCRIP;
SET EFF2;
MEAN = MEAN (OF X1-X100);
RUN;

Note
Only the variable MEAN is needed in the new data set.

PROBLEM 5
Using DATA_NULL_ when a Data Set Is Not Needed

Tools
DATA _NULL_
FILE statement

Data
SAS data set CLINICAL

Directions
Rewrite the following program to make it more efficient.
LIBNAME WORKBOOK 'C:\WORKBOOK';
OPTIONS FMTSEARCH=(WORKBOOK);
136 The SAS Workbook

DATA REPORT;
SET WORKBOOK.CLINICAL;
FILE PRINT;
IF SBP > 160 OR DBP > 90 THEN
PUT ID= SBP= DBP=;
RUN;

Note
You have no plans to use data set REPORT for anything.

PROBLEM 6
Using LENGTH Statements where Possible (Numeric Example)

Tools
LENGTH statement

Data
SAS data set EFF2

Directions
Rewrite the following program to make it more efficient. For this example, assume that a precision
of 4 bytes for the numeric variables is sufficient.
DATA SUBSET;
SET EFF2;
WHERE X1 > .5;
RUN;

Note
Some SAS users recommend never using numeric lengths of fewer than 8 bytes. In the right
situation, with proper care, you may be fine using shorter lengths. For more information, see
the section on numeric values in Chapter 3, “Components of the SAS Language,” in SAS
Language: Reference, Version 6, First Edition.
 CHAPTER 20 Efficient Programming 137

PROBLEM 7
Using LENGTH Statements to Control the Length of Character Variables

Tools
LENGTH statement

Data
Located in the inefficient program example in the Directions section

Directions
Rewrite the following program so that the length of character variables are as short as possible.
(The length of the variables A, B, C, and LETTER should be one byte and GROUP should be
set to seven bytes.) You should run the inefficient program first and then run PROC CONTENTS
to see the lengths assigned to these variables.
DATA LONGCHAR;
INPUT A $ B $ C $ GROUP $;
LETTER = SUBSTR(GROUP,1,1);
DATALINES;
1 0 1 CONTROL
0 0 0 TREAT
1 1 1 CONTROL
1 0 1 TREAT
0 1 1 TREAT
;

PROBLEM 8
Using PROC FORMAT to Regroup Variables

Tools
PROC FORMAT
PROC FREQ

Data
SAS data set EFF1

Directions
Rewrite the following program, using PROC FORMAT to regroup the values of X and using
a FORMAT statement in PROC FREQ to accomplish your goal. Do not use a DATA step at
all. This may not be efficient if you plan to use these groupings more than once and want to
138 The SAS Workbook

store them in a permanent data set. However, for this problem, assume that your disk space is
limited and that you do not want to create a new data set.
DATA GROUP;
SET EFF1;
IF 0 LE X LT .2 THEN RANGE = 1;
ELSE IF .2 LE X LT .4 THEN RANGE = 2;
ELSE IF .4 LE X LT .6 THEN RANGE = 3;
ELSE IF .6 LE X LT .8 THEN RANGE = 4;
ELSE IF .8 LE X LE 1 THEN RANGE = 5;
RUN;

PROC FREQ DATA=GROUP;

TABLES RANGE;
RUN;

PROBLEM 9
Using a KEEP= Data Set Option instead of a KEEP Statement

Tools
KEEP= data set option
KEEP statement

Data
SAS data set EFF2

Directions
Rewrite the following program to make it more efficient.
DATA SUBSET;
SET EFF2;
IF 0 LE X1 LE .5 THEN PROP = 0;
ELSE PROP = 1;
KEEP PROP;
RUN;

Hint
Data set EFF2 contains 100 variables, X1-X100, and you only need to have a single variable,
PROP, in the data set SUBSET.
 CHAPTER 20 Efficient Programming 139

PROBLEM 10
Using an Index to Increase Efficiency

Tools
PROC DATASETS
INDEX CREATE statement
PROC FREQ
WHERE statement

Data
SAS data set EFF1

Directions
Make the following program more efficient by using PROC DATASETS to create an index for
the variable P2 before running PROC FREQ. When you are finished, use PROC DATASETS
to remove the index.
PROC FREQ DATA=EFF1;
TITLE 'Where Processing - No Index Used';
WHERE P2 = 1;
TABLES GROUP;
RUN;

Note
An index will not always improve the efficiency of your program. You may want to run some
tests to see if there is improvement. In preparing this problem, some programs ran slower with
an indexed variable than without. You may also want to consider the time it takes to create the
index, although the index only needs to be created once.

PROBLEM 11
Reading Only What You Need to from Raw Data

Tools
Trailing at sign (@)

Data
Located in the inefficient example in the Directions section
140 The SAS Workbook

Directions
The following program reads the data file RAWDATA and creates a subset of the observations.
Rewrite this program to make it more efficient.
DATA SUBSET;
INFILE 'C:\WORKBOOK\RAWDATA'; *** The location of RAWDATA;
INPUT @1 (X1-X10) (5.3 + 1)
@65 GROUP $1.;
IF GROUP = 'A';
RUN;

PROBLEM 12
Arranging the Order of IF and IF-THEN/ELSE Statements

Tools
IF-THEN/ELSE statement

Data
SAS data set EFF1

Directions
If you know the relative frequency of values in a data set, you can arrange the IF and
IF-THEN/ELSE statements so that the statements more likely to be true appear earlier in the
list. In data set EFF1, the value of P2 is 0 for most observations. Rewrite the following program
so that it runs more efficiently.
DATA NEW;
SET EFF1;
IF P2 = 1 THEN NEWVAR = X;
ELSE NEWVAR = Y;
RUN;
 141
Longer and More
CHAPTER 21 Advanced Problems

PROBLEMS
1 Working with Longitudinal Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .141
2 Converting Letters to Morse Code . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .142
3 Cleaning Data and Writing a Simple Report . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .143

INTRODUCTION
Problems in this chapter are longer than problems elsewhere in the book and may require quite
a few DATA step techniques and SAS procedures. The Tools lists do not include all the things
you need to solve the problem; they do list suggestions on some techniques that may be useful.

PROBLEM 1
Working with Longitudinal Data

Tools
FIRST.variable and LAST.variable temporary variables
PROC MEANS
OUTPUT statement

Data
Raw data file CLIN_X.DTA

Directions
A. Using the raw data file CLIN_X.DTA, create a permanent SAS data set called CLIN_X.
Locate the data set in a subdirectory on your hard disk, on a floppy disk, or in another location
appropriate to your version of SAS software and operating environment. Provide permanent
formats (to be stored in the same library as the permanent data set) and labels.
B. Create a temporary SAS data set (LAST) which contains only the last visit for each patient.
C. For the variables HR, SBP, and DBP in the LAST data set, compute the number of non-
missing observations, the mean, and the standard deviation. Compute these statistics to two
decimal places.
142 The SAS Workbook

D. Using the CLIN_X data set, compute the mean HR, SBP, and DBP for each patient and
output these values (along with the patient ID) to a SAS data set called MEANS. Use this
MEANS data set to compute the mean HR, SBP, and DBP. That is, compute the mean of
the patient means.
E. Compute frequencies for diagnosis codes (DX) and treatments (RX_1 and RX_2). Remember
that each patient has only one diagnosis but can have two treatments. To compute frequencies
of treatments, first create a new data set called TREATMENT that has either one or two
observations per patient, depending on whether or not the patient had one or two treatments.
Omit the cumulative statistics from the frequency tables.

PROBLEM 2
Converting Letters to Morse Code

Tools
ARRAY statement
RANK function
User-defined informat or format

Data
A line of text of your choosing

Directions
Write a SAS program that reads a line of text, converts the letters to Morse code, and prints out
the code to the OUTPUT window or to your output device using FILE and PUT statements.
Do this two ways:
A. Use the elements of an array to hold the Morse code values and compute which array element
corresponds to a specific letter using one of the string functions.
B. Use a user-defined format or informat to provide the correspondence between the letters and
the Morse code equivalents.
For those of you who are a bit rusty, the Morse code equivalents for the letters A to Z are
A=.- B=-... C=-.-. D=-.. E=. F=..-. G=--. H=.... I=..
J=.--- K=-.- L=.-.. M=-- N=-. O=--- P=.--. Q=--.- R=-.-
S=... T=- U=..- V=...- W=.-- X=-..- Y=-.-- Z=--..
 CHAPTER 21 Longer and More Advanced Problems 143

PROBLEM 3
Cleaning Data and Writing a Simple Report

Tools
LAG and INT functions
WHERE= data set option
FILE and PUT statements
DATA _NULL_
IN statement
Data
SAS data set CLINICAL

Directions
This project includes data cleaning and simple reporting. Write a SAS program to accomplish
the following tasks.
A. Check if the value of the variable ID is missing for any observation. If so, print out the
message ID is missing for the subject following, and print out the ID of
the previous ID in the data set. If the ID that is missing is the first ID in the file, print a
special message, The first ID is missing. Also check if any visit date is earlier
than the date of birth. Print out an error message giving ID, DOB, and the visit date for these
patients. If either the visit date or the DOB is missing, do not print an error message.
B. Search for values of SBP below 100 or above 200 and for values of DBP below 50 or above
180. Create a separate data set called BP_HI_LO with any observations that meet these
criteria. Be careful not to include an observation in BP_HI_LO when the SBP or DBP is
missing. Use PROC PRINT to list out these observations.
C. Prepare a report of ID, HR, SBP, and DBP for all patients who are older than 50 (as of May
1, 1996). You may use an approximate age calculation. Provide a title, and label the columns
of the report as shown here.
Report for All Patients Older than Fifty Years Old
--------------------------------------------------
Subject Heart Systolic Diastolic
Rate Blood Blood
Pressure Pressure
144 The SAS Workbook

D. For all patients with a primary diagnosis (PRIM_DX) of 01, 02, or 04, compute frequencies
for the variables GENDER, VITAMINS, and PREGNANT. Do not create a separate data
set for this part.
E. Create a data set (PREG_35) containing all observations in CLINICAL for women who are
pregnant and over 35 years of age (as of May 1, 1996). List the ID, AGE (as of the last
birthday), and vitamin status of these women.
 145

SECTION 4 Statistics Problems

CHAPTERS
22 Basic Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .147

23 Tests of Proportions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .151

24 Comparing Means: Two Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . .157

25 Analysis of Variance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .161

26 Parametric and Nonparametric Correlations . . . . . . . . . . . . . . . . .173

27 Simple, Multiple, and Logistic Regression . . . . . . . . . . . . . . . . . . . . .177

28 Random Assignment and Random Selection of Subjects . . . . . . . .181

29 More Advanced Statistics Projects . . . . . . . . . . . . . . . . . . . . . . . . . .185

146
 147

Chapter 22 Basic Descriptive Statistics

PROBLEMS
1 Producing Basic Descriptive Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .147
2 Computing Descriptive Statistics, Broken Down by More than One Variable . . . . . .148
3 Generating Frequency Distributions and Histograms for Categorical Variables . . . .148
4 Generating Histograms for Numeric Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .149
5 Computing Means where There Is More than One Observation per Subject . . . . . .149

INTRODUCTION
The problems in this section cover topics ranging from simple descriptive statistics to more
advanced repeated measures analysis of variance. You can use them to sharpen your statistical
programming skills, and you can use the solutions as model programs for similar statistical
problems.
You may notice that the Tools lists do not go into as much detail (such as statements and options)
as those in previous chapters. The reason for this is that too much detail in the Tools list would
give too many hints on how to solve the problems and ruin your fun.
These problems test your ability to generate basic descriptive statistics such as means and
standard deviations; test distributions for normality; and produce frequency distributions,
histograms, and two-way tables.

PROBLEM 1
Producing Basic Descriptive Statistics

Tools
PROC MEANS
BY or CLASS statements
PROC UNIVARIATE

Data
SAS data set CLINICAL

Directions
Compute the following
148 The SAS Workbook

A. The number of non-missing observations (n), the mean, standard deviation, and standard
error (all to three decimal places) for the variables HR (heart rate), SBP (systolic blood
pressure), and DBP (diastolic blood pressure) for all patients.
B. The same statistics listed in A, except compute them separately for males and females.
C. The median HR, SBP, and DBP.
D. A stem-and-leaf plot for SBP. Test if the distribution is significantly different from a normal
distribution (use α = .05).

PROBLEM 2
Computing Descriptive Statistics, Broken Down by More than One Variable

Tools
PROC MEANS
CLASS or BY statement

Data
SAS data set CLINICAL

Directions
Compute the N, mean, and standard deviation of HR, SBP, and DBP, for each level of GENDER
and age group, where two age groups are defined as
1 = less than 65
2 = greater than or equal to 65

Compute age as the age at the time of the visit, dropping any fractional part of a year. Express
all statistics to three decimal places.

PROBLEM 3
Generating Frequency Distributions and Histograms for Categorical Variables

Tools
PROC FREQ
PROC CHART

Data
SAS data set CLINICAL
 CHAPTER 22 Basic Descriptive Statistics 149

Directions
Produce the following
A. One-way frequency tables for the variables GENDER, PRIM_DX, SEC_DX, VITAMINS,
and PREGNANT.
B. Vertical bar charts for the variables GENDER and PRIM_DX.

PROBLEM 4
Generating Histograms for Numeric Variables

Tools
PROC CHART

Data
SAS data set CLINICAL

Directions
Produce the following charts
A. Construct a histogram for the variable HR (heart rate). Let PROC CHART select the midpoints
and the number of bars for the histogram.
B. Rerun part A, but specify that the midpoints for the bars should run from 40 to 100 and be
spaced 10 apart.
C. Rerun part A, but specify 10 bars.

PROBLEM 5
Computing Means where There Is More than One Observation per Subject

Tools
PROC MEANS
CLASS and OUTPUT statements

Data
Raw data file CLIN_X.DTA
150 The SAS Workbook

Directions
If you look at the contents of the CLIN_X.DTA data file you will see that there are from one
to five observations for each ID. If you want to compute the mean HR, SBP, and DBP for your
patients, you can first compute a mean for each ID and then take the mean of the means, or you
can select a single observation for each patient (such as the first or last).
Using the CLIN_X.DTA data file, create a SAS data set and compute an overall mean for HR,
SBP, and DBP using both of these methods. For the latter method, use the last visit for each
patient. Compute all statistics to three decimal places.
 151

CHAPTER 23 Tests of Proportions

PROBLEMS
1 Computing a Chi-square and Fisher's Exact Test from Raw Data . . . . . . . . . . . . . . .151
2 Computing a Chi-square from Frequency Counts . . . . . . . . . . . . . . . . . . . . . . . . . . . .152
3 Computing Two-way Frequencies: Chi-square and Fisher's Exact Test . . . . . . . . . . .152
4 Computing Several Chi-square Statistics from Frequency Counts . . . . . . . . . . . . . . .153
5 Computing a Chi-square on a 2 X 3 Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .154
6 Testing Differences in Proportions for Paired Samples: McNemar's Test
(Raw Data) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .154
7 Testing Differences in Proportions for Paired Samples: McNemar's Test
(Frequency Data) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .155

INTRODUCTION
This chapter contains problems based on testing differences in proportions. Fisher’s Exact Test,
chi-square tests, and McNemar’s test are covered.

PROBLEM 1
Computing a Chi-square and Fisher’s Exact Test from Raw Data

Tools
PROC FREQ

Data
Raw data file CPR.DTA

Directions
Using the raw data in the CPR.DTA file, test if the proportion of patients surviving a cardiac
arrest is related to the three independent variables ventricular fibrillation (V_FIB), patient on a
respirator (RESP), and age greater than 70 (AGEGROUP). Compute both the chi-square statistic
and the Fisher’s exact test probability.
152 The SAS Workbook

PROBLEM 2
Computing a Chi-square from Frequency Counts

Tools
PROC FREQ
WEIGHT statement

Data
Data were collected on GENDER (M and F) and voting preference (candidate A and candidate B).
The data appear as follows.
Gender Candidate Count
--------------------------------
M A 35
F A 45
M B 70
F B 50

Directions
Using the frequency count data in the Data section, compute chi-square.

PROBLEM 3
Computing Two-way Frequencies: Chi-square and Fisher’s Exact Test

Tools
PROC FREQ
Data
SAS data set CLINICAL

Directions
A. Using the SAS data set CLINICAL, create a two-way table of the variable VITAMINS by
the variable GENDER. The table should look like the following one.
GENDER
Female Male
-----------------
| | |
No | | |
VITAMINS -----------------
| | |
Yes | | |
-----------------
 CHAPTER 23 Tests of Proportions 153

Compute the chi-square and Fisher’s exact test probability for this table. What are your conclusions
if you set α at .05?
B. Write a single TABLES statement to generate two 2-way tables: the variables VITAMINS
by GENDER and the variables VITAMINS by PRIM_DX.
C. Are pregnant women more likely to take vitamins than non-pregnant women based on this
sample (α =.05)?

Note
Be sure to compute this table only for females.

PROBLEM 4
Computing Several Chi-square Statistics from Frequency Counts

Tools
PROC FREQ
BY statement or three-way table
WEIGHT statement

Data
Frequency counts for three 2 X 2 tables
10 20 30 40
20 25 30 35
200 250 300 350

Directions
You have collected frequency counts for several 2 X 2 tables and want to enter the four counts
for each table on a separate line and compute a chi-square statistic for each table. The counts
are in order (A, B, C, and D; that is, top left, top right, lower left, and lower right).
Make your program general enough so that it can process any number of table requests.
154 The SAS Workbook

PROBLEM 5
Computing a Chi-square on a 2 X 3 Table

Tools
PROC FREQ

Data
Raw data file CARS.DTA

Directions
Using the data from CARS.DTA,
S.DTA, test if the proportion of reliability ratings below 3 are related
to the size of car.

PROBLEM 6
Testing Differences in Proportions for Paired Samples: McNemar’s Test (Raw Data)

Tools
PROC FREQ (Release 6.10 and above)
TABLES statement
AGREE option

Data
Opinion data as follows
SUBJECT Opinion Before Opinion After
OP_BEFOR OP_AFTER
--------------------------------------------
1 Anti Anti
2 Anti Pro
3 Pro Anti
4 Pro Anti
5 Pro Pro
6 Anti Pro
7 Pro Anti
8 Pro Anti
9 Pro Anti
10 Pro Anti
 CHAPTER 23 Tests of Proportions 155

Directions
You have a group of subjects who had either pro- or anti-smoking opinions both before and
after an educational program. Test if there is a difference in smoking attitude as a result of the
educational program.
If you are running Release 6.09 or lower of base SAS software, see the solution based on creating
an output data set from PROC FREQ and then computing the McNemar statistic in a subsequent
DATA step, included in The SAS Workbook Solutions. Even if you have Release 6.10 or above,
you may want to try your hand at the DATA step solution to give you some more programming
practice.

Notes
1. A Release 6.10 solution is also included in The SAS Workbook Solutions.
2. The McNemar test on these data should actually use a correction for continuity, but go ahead
and run it anyway. If you tried the DATA step solution, you can compute both the corrected
and uncorrected chi-square values and the corresponding p-values.

PROBLEM 7
Testing Differences in Proportions for Paired Samples: McNemar’s Test
(Frequency Data)

Tools
PROC FREQ (Release 6.10 and above)

Data
Opinion data expressed as frequencies
Opinion Before Opinion After Count
-----------------------------------------
Anti Anti 70
Anti Pro 10
Pro Anti 40
Pro Pro 50

Directions
Compute a McNemar’s chi-square as you did in Problem 6. This time you have frequency counts
instead of raw data values.
For this problem, only the Release 6.10 and above solution is presented in The SAS Workbook
Solutions.
156
 157

CHAPTER 24 Comparing Means:Two Groups

PROBLEMS
1 Performing an Unpaired t-test (Existing SAS Data Set) . . . . . . . . . . . . . . . . . . . . . . .157
2 Performing an Unpaired t-test (Raw Data) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .158
3 Performing an Unpaired Nonparametric Comparison (Wilcoxon Rank Sum Test) . .158
4 Performing a Paired t-test . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .159
5 Performing a Paired Nonparametric Comparison (Wilcoxon Signed Rank Test) . . .159
6 Performing a Paired t-test with Longitudinal Data . . . . . . . . . . . . . . . . . . . . . . . . . . .160

INTRODUCTION
Problems in this chapter involve comparisons between two groups. The problems cover parametric
and nonparametric tests as well as paired and unpaired tests.

PROBLEM 1
Performing an Unpaired t-test (Existing SAS Data Set)

Tools
PROC TTEST

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, compare the mean HR (heart rate), SBP (systolic blood
pressure), and DBP (diastolic blood pressure) between those patients who take vitamins and
those who don’t take vitamins.
158 The SAS Workbook

PROBLEM 2
Performing an Unpaired t-test (Raw Data)

Tools
PROC TTEST

Data
Blood lipid data as follows
Group A Group B
CHOL HDL TRIG CHOL HDL TRIG
--------------------- ---------------------
220 45 120 155 40 130
180 60 70 120 30 170
240 75 100 112 35 160
285 50 150 126 35 120
288 55 102 133 48 80
302 70 130

Directions
You have measured the total cholesterol (CHOL), high-density lipids (HDL), and triglycerides
(TRIG) in two groups of patients. Group A was given a placebo, and group B was given a
medication believed to have an effect on blood lipids.
Write a program to read these data and conduct a t-test on the three variables CHOL, HDL, and
TRIG. For each of the variables, determine if the assumption of equal variance should be used.

PROBLEM 3
Performing an Unpaired Nonparametric Comparison (Wilcoxon Rank Sum Test)

Tools
PROC NPAR1WAY
WILCOXON option

Data
Animal weights on two different diets
Diet 1: 45 48 57 73 63

Diet 2: 80 72 130 65 220 200

 CHAPTER 24 Comparing Means: Two Groups 159

Directions
Animal weights were recorded for each of two diets. Write a program to read these data and
conduct a Wilcoxon rank sum test. You might want to use a Wilcoxon rank sum table to compute
a more exact solution than the normal approximation values produced by PROC NPAR1WAY.

PROBLEM 4
Performing a Paired t-test

Tools
PROC MEANS

Data
Blood pressure data as follows
Subject Systolic Blood Pressure Diastolic Blood Pressure
Placebo Drug Placebo Drug
-------------------------------------------------------------
1 180 160 90 80
2 220 170 110 96
3 190 140 88 78
4 182 142 108 84
5 160 154 88 86
6 190 182 88 82

Directions
An experiment was conducted in which each subject had his blood pressure measured under
two different conditions: after taking a placebo for one week and after taking a blood pressure
medication for one week. Using the Data in the Data section, conduct a paired t-test to see if
there is a difference between the two conditions.

PROBLEM 5
Performing a Paired Nonparametric Comparison (Wilcoxon Signed Rank Test)

Tools
PROC UNIVARIATE

Data
The same as in Problem 4
160 The SAS Workbook

Directions
Using the same data as Problem 4, conduct a Wilcoxon signed rank test. Compare your answer
to Problem 4.

PROBLEM 6
Performing a Paired t-test with Longitudinal Data

Tools
PROC MEANS
FIRST.variable and LAST.variable temporary variables
LAG function (one possible solution)

Data
Raw data file CLIN_X.DTA

Directions
Using the CLIN_X.DTA raw data file, compare the heart rate (HR) and systolic blood pressure
(SBP) from the first visit to the last visit for all patients who had at least two visits. Use a paired
t-test for the analysis. There are only four patients with two or more visits, but go ahead and
run the t-test procedure, even though it is probably inappropriate for these data.
 161

CHAPTER 25 Analysis of Variance

PROBLEMS
1 Performing a One-way ANOVA (Balanced Design) . . . . . . . . . . . . . . . . . . . . . . . . . . .162
2 Performing a One-way ANOVA (Unbalanced Design) . . . . . . . . . . . . . . . . . . . . . . . .162
3 Performing a Two-way ANOVA (Balanced Design) with an Interaction Plot . . . . . .163
4 Performing a Two-way ANOVA (Unbalanced Design) . . . . . . . . . . . . . . . . . . . . . . . .164
5 Performing a Three-way ANOVA (Balanced Design) . . . . . . . . . . . . . . . . . . . . . . . . .165
6 Performing a Three-way ANOVA (Unbalanced Design) . . . . . . . . . . . . . . . . . . . . . . .166
7 Performing a One-way ANOVA (Repeated Measures Design) . . . . . . . . . . . . . . . . . .167
8 Performing a Two-way ANOVA (Repeated Measure on One Factor) . . . . . . . . . . . .168
9 Performing a Two-way ANOVA (Repeated Measure on Both Factors) . . . . . . . . . . .169
10 Performing a Three-way ANOVA (Repeated Measure on One Factor) . . . . . . . . . . .170
11 Performing a Three-way ANOVA (Repeated Measure on Two Factors) . . . . . . . . . .170
12 Performing a Three-way ANOVA (Repeated Measures on All Factors) . . . . . . . . . . .171

INTRODUCTION
This chapter covers a wide variety of analysis of variance problems from simple one-way
balanced designs to multi-way repeated measures, unbalanced designs.
I should say a few words about terminology before letting you loose on these problems. My use
of the term “repeated” is based on a common use in the medical field and described in a text,
Statistical Principles in Experimental Design, Second Edition, by B. J. Winer (1971). I use the
term “repeated” in this chapter to mean any factor where each subject is measured on every
level of that factor. For example, if a runner is timed running on two different types of track
surfaces, I am considering “surface” as a repeated measure factor. Other authors use the term
“repeated” to refer only to factors that cannot be assigned in random order, such as time. If you
use the latter meaning of “repeated”, feel free to substitute your design terminology in these
problems.
162 The SAS Workbook

PROBLEM 1
Performing a One-way ANOVA (Balanced Design)

Tools
PROC ANOVA

Data
Tomato and fertilizer data as follows
Fertilizer Tomato Weight
--------------------------------------
5-10-5 .85
5-10-5 .88
5-10-5 .87
5-10-5 .83
5-10-10 .92
5-10-10 .97
5-10-10 .86
5-10-10 .88
20-5-5 .64
20-5-5 .50
20-5-5 .48
20-5-5 .42

Directions
An experiment was conducted in which the weight of tomatoes was measured after administration
of three types of fertilizer. Conduct a one-way ANOVA comparing the fertilizers. Include a
request for a Student-Newman-Keuls multiple comparison test (α=.05).

PROBLEM 2
Performing a One-way ANOVA (Unbalanced Design)

Tools
PROC ANOVA
Data
Raw data file CARS.DTA

Directions
Using the data from CARS.DTA, conduct a one-way ANOVA comparing gas mileage among
the three sizes of cars. Include a request for a Duncan multiple-range test conducted at α=.05
and a Scheffé test conducted at α=.10.
 CHAPTER 25 Analysis of Variance 163

Hint
Remember that PROC ANOVA can be used for one-way unbalanced designs.

PROBLEM 3
Performing a Two-way ANOVA (Balanced Design) with an Interaction Plot

Tools
PROC ANOVA
PROC MEANS
PROC PLOT
Concatenation operator (DATA step)

Data
Activity scores on control and ADD children as follows
GROUP DRUG ACTIVITY
--------------------------------------
ADD PLACEBO 90
ADD PLACEBO 88
ADD PLACEBO 95
CONTROL PLACEBO 60
CONTROL PLACEBO 62
CONTROL PLACEBO 66
ADD RITALIN 72
ADD RITALIN 70
ADD RITALIN 64
CONTROL RITALIN 86
CONTROL RITALIN 86
CONTROL RITALIN 82

Directions
Two groups of children, one with attention deficit disorder (ADD) and a control group of children
without ADD, were randomly given either a placebo or the drug Ritalin. A measure of activity
was made on all the children with the results shown in the Data section (higher numbers indicate
more activity).
164 The SAS Workbook

A. Conduct a two-way ANOVA (GROUP by DRUG) on these data. Run a Duncan multiple-
range test on any significant main effects. Draw an interaction plot like the following one.
Plot of ACTIVITY*DRUG. Symbol is value of GROUP.

ACTIVITY |
90 + A
|
| C
|
80 +
|
|
|
70 +
| A
|
| C
60 +
---+-----------------------------------+--
PLACEBO RITALIN

DRUG

B. Write the SAS statements to conduct a t-test between drugs for the ADD and control children
separately. You will find a highly significant interaction term in this two-way model.
C. An alternative approach is to run a one-way ANOVA with each combination of GROUP
and DRUG as levels of the independent variable. For example, the four levels would be
ADD-PLACEBO, ADD-RITALIN, CONTROL-PLACEBO, and CONTROL-RITALIN.
Write a DATA step that creates a new variable (call it LEVEL) that has each of these four
values. Then rerun a one-way ANOVA using this new variable as the independent (CLASS)
variable. Include a request for a Duncan multiple-range test.

PROBLEM 4
Performing a Two-way ANOVA (Unbalanced Design)

Tools
PROC GLM

Data
SAS data set CLINICAL
 CHAPTER 25 Analysis of Variance 165

Directions
Using the SAS data set CLINICAL, conduct a two-way ANOVA on systolic blood pressure
(SBP) with GENDER and VITAMINS as the independent (CLASS) variables.

PROBLEM 5
Performing a Three-way ANOVA (Balanced Design)

Tools
PROC ANOVA
DO loops in the DATA step (optional)

Data
Headache data as follows
Drug Gender Dose Cure Time
--------------------------------
A M Low 22
A M High 20
A F Low 18
A F High 16
B M Low 38
B M High 34
B F Low 34
B F High 30
A M Low 21
A M High 18
A F Low 17
A F High 14
B M Low 37
B M High 33
B F Low 34
B F High 29

Directions
A drug study was conducted to measure how long it took for two different medications to cure
headaches. Men and women subjects were given either a high or low dose of either drug A or drug
B. The time in minutes it took for their headaches to be cured was recorded and is shown in the
Data section.
Create a SAS data set using these data, and write the SAS statements to conduct a three-way
ANOVA with CURETIME as the dependent variable and DRUG, GENDER, and DOSE as the
independent variables.
166 The SAS Workbook

As an interesting exercise, see if you can read the data arranged like this
22 20 18 16 38 34 34 30
21 18 17 14 37 33 34 29

The first number is the cure time for DRUG=A, GENDER=M, and DOSE=Low; the second
number is the cure time for DRUG=A, GENDER=M and DOSE=High, and so on.

PROBLEM 6
Performing a Three-way ANOVA (Unbalanced Design)

Tools
PROC GLM

Data
Third and fourth grade spelling data as follows
Grade 3 Grade 4
Phonetic Memory Phonetic Memory
---------------------------------------------------
35 30 45 40
Boys 37 . 48 42
34 37 47 41
34 29 44 39
---------------------------------------------------------
37 32 44 42
Girls 39 38 50 .
37 38 49 44
36 39 45 40

Directions
Boys and girls in the third and fourth grade were trained to spell in two ways. One way (we’ll
call it phonetic) taught the children phonetic rules; the other way (we’ll call it memory) taught
spelling by repetition and memorization. The children were asked to spell 50 words with the
number of words spelled correctly recorded as shown in the Data section. (Notice the two missing
values, making this an unbalanced design.)
Run an analysis of variance to test this factorial model. Remember that each number in the table
represents a different student. Try writing your program so that you can enter the data as shown
here, with the DATA step creating the appropriate GRADE, METHOD, and GENDER variables.
 CHAPTER 25 Analysis of Variance 167

35 30 45 40
37 . 48 42
34 37 47 41
34 29 44 39
37 32 44 42
39 38 50 .
37 38 49 44
36 39 45 40

Hint
A couple of nested DO loops would be nice. Also remember that you can have loops such as
DO GENDER = ‘Boys ‘,’Girls’;

PROBLEM 7
Performing a One-way ANOVA (Repeated Measures Design)

Tools
PROC ANOVA
REPEATED statement (optional)
CONTRAST keyword

Data
Number of problems answered correctly under different noise conditions as follows
--- Noise Level ---
Subject None Medium High
---------------------------------
1 20 15 8
2 10 8 4
3 35 25 23
4 9 5 2
5 28 24 20

Directions
Each of five subjects was asked to do mental arithmetic under three noise conditions (none,
medium, and high). The number of problems answered correctly was recorded.
Keeping in mind that this is a repeated measures design (each subject is measured under each of
the three noise conditions), run a repeated measures ANOVA. If there is a significant main
effect, use an SNK multiple comparison test to detect pairwise differences. As an interesting
exercise, what would be the result if you ignored the repeated aspect of this design and ran a
one-way ANOVA on noise level?
168 The SAS Workbook

Hint
You may want to rearrange the data (by rearranging the lines of data following the DATALINES
statement, by writing a DATA step, or by using PROC TRANSPOSE) so that they look like
this
Subject Level Number
------------------------
1 None 20
1 Medium 15
1 High 8
2 None 10
2 Medium 8
2 High 4
more data observations

You may also want to try using the original data structure and the REPEATED statement of
PROC ANOVA.

PROBLEM 8
Performing a Two-way ANOVA (Repeated Measure on One Factor)

Tools
PROC ANOVA
TEST statements specifying error terms

Data
Reaction times under different levels of alcohol consumption as follows
----- Condition -------
Subject Gender Control 1 oz. 3 oz.
------------------------------------------
1 M .4 .9 1.2
2 M .6 .6 .8
3 M .8 1.1 1.3
4 F .9 1.1 1.4
5 F 1.0 1.5 2.0
6 F .8 1.0 1.6

Directions
Male and female subjects are tested for reaction time under three different conditions: control,
1 oz. alcohol, and 3 oz. alcohol. The results are summarized in the Data section.
 CHAPTER 25 Analysis of Variance 169

Create a SAS data set containing the variables SUBJECT, GENDER, COND (condition), and
REACT (reaction time) with three observations per subject. Run a two-way ANOVA (GENDER
by COND) with COND treated as a repeated measures factor. (You can also think of this as
subjects nested within GENDER.) Run a multiple comparison test on COND to see which
pairwise comparisons are significant. Use the Duncan multiple-range test to do this.

PROBLEM 9
Performing a Two-way ANOVA (Repeated Measure on Both Factors)

Tools
PROC ANOVA
REPEATED statement

Data
Times to run 100 yards under different conditions as follows
Track
Cinder Rubber
Brand of Shoe A B A B
-------------------------------------------
Runner 1 11.4 11.5 10.9 10.9
2 10.8 10.7 10.2 10.0
3 10.9 11.1 10.7 10.4
4 10.4 11.2 10.5 10.7

Directions
Each of four runners was timed running 100 yards on a cinder track and on a composite rubber
track, each with two different brands of shoes (Brand A and Brand B). Each runner therefore
ran on two types of tracks and wore two different brands of shoes for a total of four runs. The
times are listed in the Data section.
Realizing that this is a two-way design with both factors repeated, run an analysis of variance
testing for the main effects and the interaction term. Try running this with and without the
REPEATED statement of PROC ANOVA.
170 The SAS Workbook

PROBLEM 10
Performing a Three-way ANOVA (Repeated Measure on One Factor)

Tools
PROC ANOVA
TEST statement

Data
Time to complete a puzzle by young school children as follows
Age 3 Age 4
Trial 1 2 3 1 2 3
---------------------------------------------
52 48 46 45 40 30
Pre-school 51 47 45 44 38 27
53 49 47 46 41 32
---------------------------------------------------
60 58 56 52 48 40
No Pre-school 59 57 55 51 47 39
61 59 57 53 49 41

Directions
School children, ages three and four were asked to complete a simple puzzle. Some of the
children attend preschool and others do not. Each child was given three tries to complete the
puzzle with the time to completion recorded each time (in seconds). The results are listed in the
Data section.
Test the main effects and the two- and three-way interactions. You may want to read in the data
as in Problem 6. Since you want to conduct a multiple comparison test on the variable TRIAL,
run a model that does not use the REPEATED statement (although this could be accomplished
using CONTRAST statements).

PROBLEM 11
Performing a Three-way ANOVA (Repeated Measure on Two Factors)

Tools
PROC ANOVA
REPEATED or TEST statements
Data
The number of correctly identified license plate letters identified under several different conditions
as follows
 CHAPTER 25 Analysis of Variance 171

---------------- Color Combination ----------------

White on Blue Blue on White Black on Yellow
Font Helv Times Helv Times Helv Times
----------------------------------------------------------
21 25 22 23 18 19
Policeman 21 25 23 24 22 24
24 22 25 20 19 21
27 20 25 18 20 20
-----------------------------------------------------------------
19 21 20 21 17 16
Lay Public 21 21 22 22 20 20
20 20 24 18 17 19
24 19 25 17 20 18

Directions
An experiment was conducted to determine which combinations of colors and fonts would be most
readable for car license plates by policemen and the public. Three color combinations (white
on blue, blue on white, and black on yellow) were each tried with a sans serif font (Helvetica)
and a serif font (Times Roman). The number of correctly identified letters or numbers was
recorded after a .2 second look at each of several plates. Each policeman and each lay observer
was tested on all three colors and both fonts with the results listed in the Data section.
You may want to read in these data as in Problem 6, but that is your choice. Run a three-way
ANOVA (COLOR x FONT x GROUP (Police/Lay)) with both COLOR and FONT treated as
repeated measures factors. The solutions both with and without the REPEATED statement are
found in The SAS Workbook Solutions.

PROBLEM 12
Performing a Three-way ANOVA (Repeated Measures on All Factors)

Tools
PROC ANOVA
TEST statement

Data
Time on a treadmill by volunteer subjects under several conditions as follows
172 The SAS Workbook

Drug A Drug B
Morning Night Morning Night
Trial 1 2 3 1 2 3 1 2 3 1 2 3
------------------------------------------------------------------------
Subject 1 8 7 9 10 12 13 12 13 15 15 17 16
2 9 10 11 12 13 14 15 16 17 18 19 20
3 7 7 9 9 8 10 9 9 11 11 13 15
4 8 9 9 8 9 9 9 10 10 10 11 12
5 8 8 11 8 9 12 10 12 14 10 13 15

Directions
Five volunteers were asked to run on a treadmill as long as they could. They did this three times
after taking each of two drugs. The treadmill test was also given either the first thing in the
morning or at night. Thus, each volunteer completed 3 x 2 x 2 = 12 runs on the treadmill. The
maximum time, in minutes, on the treadmill was recorded and is displayed in the Data Section.
Realizing that this is a completely repeated measures design, run the three-way design, testing
for the three main effects (DRUG, TIME, and TRIAL) as well as the two- and three-way
interactions.
 173
Parametric and
CHAPTER 26 Nonparametric Correlations

PROBLEMS
1 Creating a Simple Correlation Matrix . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .173
2 Correlating a Single Variable against a List of Variables . . . . . . . . . . . . . . . . . . . . . . .174
3 Computing Spearman Rank Order Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . .175
4 Correlating Ranked Data and Comparing the Results to the
Spearman Correlations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .175

INTRODUCTION
A valuable step before running multiple regression models is to compute the correlations among
the variables in your model. Independent variables that are strongly correlated may create
problems in multiple regression models. You may want to create new variables from sets of
highly correlated variables before running your regression models.

PROBLEM 1
Creating a Simple Correlation Matrix

Tools
PROC CORR
NOSIMPLE option
VAR statement

Data
SAS data set CLINICAL

Directions
Generate a matrix of Pearson correlation coefficients for the variables HR (heart rate),
SBP (systolic blood pressure), and DBP (diastolic blood pressure) from the SAS data set
CLINICAL. Include the NOSIMPLE option to eliminate the listing of descriptive statistics that
precede the correlation matrix.
174 The SAS Workbook

PROBLEM 2
Correlating a Single Variable against a List of Variables

Tools
PROC CORR
ALPHA option
VAR and WITH statements

Data
Student quiz data as follows
The student ID is in columns 1-3 and the ten scored items are in columns 5-14. A one indicates
a correct answer, and a zero indicates an incorrect answer.
001 1101111011
002 1111111111
003 1110101010
004 0001010001
005 1110101010
006 0101010101
007 1111111011
008 1110111011
009 0001000100
010 1111011110

Directions
You have test score data for some students who took a ten-question multiple choice test. For
each student, you have an ID and a score (0 or 1) for each of the ten questions.
First, compute a raw score (the number of items correct) for each student. Next, compute the
correlation of each of the ten questions with the raw score (this correlation is called a point-
biserial correlation coefficient and is a measure of item quality). Finally, compute Cronbach’s
coefficient alpha for the ten questions.

Note
Since the items are dichotomous, Cronback’s coefficient alpha will be equivalent to the popular
Kuder-Richardson Formula 20.

Hint
The ALPHA option generates Cronbach’s alpha. This option can only be used with variables
listed in the VAR statement and cannot be used if there is a WITH statement present.
 CHAPTER 26 Parametric and Nonparametric Correlations 175

PROBLEM 3
Computing Spearman Rank Order Correlations

Tools
PROC CORR
PEARSON, SPEARMAN, and NOSIMPLE options
PROC PLOT
Data
X, Y, and Z values for several subjects
X Y Z
-------------
1 5 3
2 5 8
5 4 11
4 9 15
5 18 13
20 25 55

Directions
Given the values for X, Y, and Z in the Data section, compute both Pearson and Spearman
correlations. Omit the descriptive statistics. Generate plots of X versus Y, X versus Z, and
Y versus Z. Use the author’s favorite plotting symbol o (lowercase letter o) for your plots.
Compare the Pearson and Spearman coefficients and examine the scatter plots, especially for
the variable pairs where the coefficients differ the most.

PROBLEM 4
Correlating Ranked Data and Comparing the Results to the Spearman Correlations

Tools
PROC RANK
PROC CORR
Data
The same as in Problem 3
Directions
Given the sample data in Problem 3, use PROC RANK to create ranks for X, Y, and Z. Generate
a Pearson correlation between the ranked values and compare the results to the Spearman
coefficients in Problem 3.
176
 177
Simple, Multiple, and
CHAPTER 27 Logistic Regression

PROBLEMS
1 Performing Simple Linear Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .177
2 Plotting 95% Confidence Intervals . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .178
3 Using the Simple Regression Equation to Predict Values . . . . . . . . . . . . . . . . . . . . . .178
4 Performing Multiple Regression (Fixed Model) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .179
5 Performing Multiple Regression Using Several Selection Methods . . . . . . . . . . . . . . .179
6 Performing Logistic Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .180

INTRODUCTION
This chapter presents a variety of regression problems, starting with simple linear regression
and ending with multiple stepwise regression and logistic regression.

PROBLEM 1
Performing Simple Linear Regression

Tools
PROC REG
PLOT statement

Data
SAS data set CLINICAL

Directions
Using the variables SBP (systolic blood pressure) and DBP (diastolic blood pressure) from the
SAS data set CLINICAL, regress SBP on DBP (SBP on the y-axis, DBP on the x-axis). What
is the slope and intercept of the least squares regression line? Plot the original data and the points
on the regression line in one plot (use separate plotting symbols for the original data points and
the predicted values) and the residuals versus DBP on another. Use the PLOT statement of
PROC REG to construct these plots. Do not use an OUTPUT statement to create a data set and
PROC PLOT to do the plotting.
178 The SAS Workbook

PROBLEM 2
Plotting 95% Confidence Intervals

Tools
PROC REG
PLOT statement

Data
SAS data set CLINICAL

Directions
Using the blood pressure data from Problem 1, create a plot showing the predicted y-values
(systolic blood pressure) and two types of 95% confidence intervals: one for individual data
points, the other for the confidence limits for the mean of Y. Use a lowercase o for the predicted
value, a dash (-) for the upper and lower limits of the 95% confidence interval for individual
points, and an equal sign (=) for the upper and lower 95% confidence interval for the mean of Y.

PROBLEM 3
Using the Simple Regression Equation to Predict Values

Tools
PROC REG
MODEL statement
P, R, CLI, and CLM options

Data
Values of X and Y as follows
X Y
--------
1 4
2 7
5 12
7 15

Directions
Using the values from the Data section, generate a least squares regression line and predict the
y-value for x-values of 10 and 20. Also create printed output showing the predicted value and
the 95% confidence limits (both types) for all values of DBP (diastolic blood pressure).
 CHAPTER 27 Simple, Multiple, and Logistic Regression 179

Hint
Use the MODEL statement options P, R, CLI, and CLM to print the requested values. Include
x-values of 10 and 20 (with missing values for Y) in your data set to compute predicted y-values.

PROBLEM 4
Performing Multiple Regression (Fixed Model)

Tools
PROC REG

Data
Raw data file CARS.DTA

Directions
Using the raw data file CARS.DTA, predict gas mileage (MILEAGE) based on car SIZE and
reliability (RELIABLE).
A. Use the values 1, 2, and 3 to represent SMALL, COMPACT, and MID-SIZED cars,
respectively.
B. Create two dummy variables to specify the three car sizes.

PROBLEM 5
Performing Multiple Regression Using Several Selection Methods

Tools
PROC REG
MODEL statement
SELECTION= option

Data
SAS data set CLINICAL

Directions
Using the SAS data set CLINICAL, generate a multiple regression equation that predicts systolic
blood pressure (SBP) using gender, age, heart rate (HR), and diastolic blood pressure (DBP) as
predictor variables. First, run a model with all of these variables included. Next, run the
STEPWISE procedure to fit a model. Finally, use the RSQUARE selection method to run all
one-, two-, three-, and four-variable models.
180 The SAS Workbook

Note that you need to compute AGE in a DATA step and create a dummy variable based on
GENDER (GENDER is stored as a character variable with values of M, and F). There are really
too few observations to include this many variables in a regression, but solve the problem for
practice.

PROBLEM 6
Performing Logistic Regression

Tools
PROC LOGISTIC
DESCENDING option
MODEL statement
SELECTION=, CTABLE, and RL options

Data
Raw data file CPR.DTA

Directions
Create a SAS data set from the raw data file CPR.DTA. Run a forward stepwise logistic regression
with SURVIVE (survival) as the dependent variable and V_FIB (is the patient in ventricular
fibrillation?), RESP (was the patient on a respirator prior to the cardiac arrest?), and AGEGROUP
(is age greater than 70?) as the independent variables.
Note that for the dependent variable, a value of 1 (yes) represents survival; and a value of 0 (no)
represents no survival (death). Your equation should predict the odds ratios for survival. The
default for PROC LOGISTIC is to predict the odds ratio for the lower value of the dependent
variable (in this case a 0). Thus, you need to create a new variable with the lower value being
survival or, more simply, use the DESCENDING option (as suggested in the Tools list). For
the variables V_FIB and RESP, a 0 indicates that the patient did not have this condition and a
1 indicates that he or she did. For the variable AGEGROUP, a value of 1 indicates the patient
is older than 70; a 0 indicates he or she is 70 or younger.
Include in your output, a classification table and the 95% confidence limits on the odds ratios.
 181
Random Assignment and
CHAPTER 28 Random Selection of Subjects

PROBLEMS
1 Assigning Subjects into Two Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .181
2 Assigning Subjects into Three Groups . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .182
3 Assigning Subjects into Groups, Balanced in Blocks . . . . . . . . . . . . . . . . . . . . . . . . . .182
4 Taking a Random Subset of Subjects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .183

INTRODUCTION
This chapter includes problems on randomly assigning subjects to groups and random selection
of samples from larger data sets.

PROBLEM 1
Assigning Subjects into Two Groups

Tools
RANUNI function
PROC RANK
GROUPS= option

Data
Generated in a DATA step

Directions
You want to assign each of 50 subjects to either group A or group B by either
A. Using a randomly generated value to assign subjects to either group A or group B with
an equal probability that the subject will be assigned to either A or B. The number of subjects
assigned to the groups may not be equal.
B. Using a combination of random numbers and PROC RANK (with the GROUPS=2 option)
to assign subjects randomly to either group A or group B with the requirement that there be
an equal number of subjects (25) in the two groups.
For both methods, provide a listing of the subjects and their group assignment and a count of
the number of subjects in each of the two groups.
182 The SAS Workbook

Note
Use a fixed seed for the random number function rather than the clock seed. For example,
RANUNI(1357).

PROBLEM 2
Assigning Subjects into Three Groups

Tools
RANUNI function
PROC RANK
GROUPS= option

Data
Generated in a DATA step

Directions
Repeat parts A and B from Problem 1, but this time you want a total of 30 subjects assigned to
either group A, B, or C. For part B, you will have ten subjects in each of the three groups.

PROBLEM 3
Assigning Subjects into Groups, Balanced in Blocks

Tools
RANUNI function
PROC RANK
GROUPS= option
Nested DO loops

Data
Generated in a DATA step

Directions
This problem is similar to Problem 1 (part B) except that you want to avoid long runs of either
group A or group B assignments. You want to assign 50 subjects to either group A or group B,
and you want an equal number of subjects assigned to A and B every ten subjects. That is, for
each block of ten subjects, there will be five A’s and five B’s.
 CHAPTER 28 Random Assignment and Random Selection of Subjects 183

PROBLEM 4
Taking a Random Subset of Subjects

Tools
RANUNI function

Data
SAS data set BIG created by running the following program
DATA BIG;
DO SUBJECT = 1 TO 100;
X = INT (RANUNI(0)*100 + 1);
OUTPUT;
END;
RUN;

Directions
You have a data set BIG with 100 observations and you want a 5% random sample. Call the
5% random sample data set SMALL.
A. Solve this problem with an approximate solution using a test for the value of a random
variable.
B. Write a program what will always generate a sample of five observations at random from
data set BIG.
184
 185
More Advanced
CHAPTER 29 Statistics Projects

PROBLEMS
1 Creating a SAS Data Set Complete with Formats and Labels, Descriptive Statistics,
Simple Hypothesis Tests, Correlations, and Multiple Regression . . . . . . . . . . . . . . . .185
2 Computing a Two-way ANOVA with a Significant Interaction Term . . . . . . . . . . . . .187
3 Performing an Unpaired t-test Using a Variety of Input Data Arrangements . . . . . .189
4 Creating New Variables, Regrouping Data, Producing Descriptive Statistics,
and Performing Significance Tests on Means and Proportions . . . . . . . . . . . . . . . . . .191
5 Working with Paired Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .192

INTRODUCTION
The problems in this chapter are longer and, perhaps, more difficult than those in the earlier
chapters. They are more typical of real world problems and may be a little more vague than
previous problems. If you are a student in a research course or a researcher wanting to practice
your statistical programming skills, they should be good practice.

PROBLEM 1
Creating a SAS Data Set Complete with Formats and Labels, Descriptive Statistics,
Simple Hypothesis Tests, Correlations, and Multiple Regression

Tools
PROC PRINT
PROC UNIVARIATE
PROC FREQ
PROC TTEST
PROC NPAR1WAY
PROC CORR
PROC GLM

Data
All sections of this problem use the SAS data set CLINICAL
186 The SAS Workbook

Directions
A. Create a new data set from CLINICAL which includes the subject’s approximate age in
years as of the last clinic visit. Truncate this value (i.e., compute age as of the last birthday
at the time of the visit).
B. Provide a listing of this data set. Use the variable labels as column headings. Remember
that this data set already has labels and formats associated with the variables. If you made the
original CLINICAL data set permanent, remember to let the program know where to find
the user-defined formats.
C. Compute descriptive statistics, including a test for normality and stem-and-leaf plots, for
the variables HR, SBP, DBP, and AGE.
D. Generate the following tables:
1. One-way frequencies for GENDER, PRIM_DX, SEC_DX, and VITAMINS.
2. Crosstab of GENDER by VITAMINS including a Fisher’s exact test.
E. Generate frequencies for the variable PREGNANT. Do this for females only since this
variable is set to 0 (no) for male patients.
F. Investigate the relationship between vitamin use and pregnancy by comparing the proportion
of pregnant and non-pregnant females using vitamins, again being sure this is computed
only for females.
G. Compute a t-test with GENDER as the independent variable and HR, SBP, DBP, and AGE
as dependent variables. Indicate which t- and p-value to use for each test, based on the
assumption of homogeneity of variance being met or not.
H. Repeat the comparisons in part G using a nonparametric test (Wilcoxon rank sum test) instead
of a t-test.
I. Generate a correlation matrix among the variables HR, SBP, DBP, and AGE. Omit simple
statistics and compute both Spearman and Pearson correlations.
J. In preparing to run an analysis of covariance with SBP and DBP as dependent variables,
GENDER as the independent variable, and AGE as a covariate, test if the assumption of
homogeneity of slope is violated. That is, test if the slope of AGE by SBP and AGE by DBP
is significantly different (p < .05) for males versus females.
K. For any dependent variables in part J where the assumption of homogeneity of slope is valid,
run an analysis of covariance and compute the adjusted means.
L. Run a multiple regression with SBP as the dependent variable and GENDER, DBP, and
AGE as predictor (independent) variables. Use a forward stepwise selection method. Since
GENDER is a character variable with values of M and F, create a new variable (SEX) which
is numeric and has a value of 0 for males and 1 for females.
 CHAPTER 29 More Advanced Statistics Projects 187

M. Compute frequencies for all diagnoses, treating primary and secondary diagnoses equally. That
is, if one patient has code 01 as a primary diagnosis and another patient has code 01 as a
secondary diagnosis, you would count two people as having diagnosis 01. To accomplish this
task, create a new data set called DX_FREQ which has as many observations as there are
diagnosis codes in the original data set (PRIM_DX codes plus SEC_DX). If a patient has a
primary diagnosis and a secondary diagnosis, that patient will contribute two observations to
DX_FREQ.

PROBLEM 2
Computing a Two-way ANOVA with a Significant Interaction Term

Tools
PROC ANOVA
PROC MEANS
OUTPUT statement
Concatenation operator
PROC TTEST
PROC PLOT

Data
Values of GROUP, GENDER, and X as follows
GROUP GENDER X
-----------------
A M 5
A M 6
A M 7
A F 11
A F 13
A F 14
B M 15
B M 10
B M 11
B F 4
B F 6
B F 7
188 The SAS Workbook

Directions
A. Given the data described, conduct a two-way ANOVA for the variable X with GROUP and
GENDER as independent variables. You will find that the GROUP*GENDER term is
significant (and the main effects are not).
B. Use PROC MEANS to create an output data set with the mean value of X for each combination
of GROUP and GENDER.
C. Use this data set to produce an interaction plot as shown here.
Plot of X*GROUP. Symbol is value of GENDER.

X |
12.5 + F
| M
|
|
10.0 +
|
|
|
7.5 +
|
| M
| F
5.0 +
---+--------------------------------------+--
A B

GROUP

D. Conduct a t-test with GROUP as the independent variable for each level of GENDER.
E. Create a data set which contains a new variable (CATEGORY) which has values of A-M,
A-F, B-M, and B-F (that is, a value reflecting both the GROUP and GENDER). Using this
new variable, conduct a one-way ANOVA with X as the dependent variable and CATEGORY
as the independent variable. Follow this ANOVA with a Student-Newman-Keuls multiple
comparison test.
 CHAPTER 29 More Advanced Statistics Projects 189

PROBLEM 3
Performing an Unpaired t-test Using a Variety of Input Data Arrangements

Tools
PROC TTEST
A variety of input methods and DATA step programming

Data
Respiration data as follows
Subject Group Respiration Rate (RESP)
--------------------------------------------
1 A 16
2 A 20
3 A 18
4 B 24
5 B 22
6 B 28
7 B 18

Directions
Subjects were assigned to either group A or group B, and the number of breaths per minute
(respirations) was recorded for each subject.
For each of the sections that follow, the data values are arranged differently. Write the SAS
statements necessary to create a SAS data set called BREATH containing the variables SUBJECT,
GROUP, and RESP. Follow this with a request for an unpaired t-test comparing the mean
respiration rates for groups A and B.
A. The data arrangement for this part is
A 16
A 20
A 18
B 24
B 22
B 28
B 18

Notice that there is no subject number in the data, and you want a SUBJECT variable in the
data set.
B. Similar to part A except the data values are all on one line.
A 16 A 20 A 18 B 24 B 22 B 28 B 18
190 The SAS Workbook

C. The data are arranged as follows

A 16 20 18
B 24 22 28 18

You are told that there are three values for group A and four values for group B, but this
number is not to be placed in the data. You need to program the number of scores to read
for each group in the DATA step.
D. The data are arranged as follows
16 20 18 24 22 28 18

You are told that the first three numbers are for group A and the next four numbers are for
group B.This time the value of GROUP and the number of scores to read for each group
must be programmed in the DATA step.
E. The data are arranged as follows
3 16 20 18 4 24 22 28 18

The number of subjects is entered, followed by the respiration values. You are told that the
first set of numbers are from group A and the second set are from group B.
F. The data are arranged as follows
A 16 B 24 22 A 20 18 B 28 18

That is, a group letter is followed by one or more scores. Have the program read each data
value and decide if it is a group letter or a score. Write the program in a general way such that
it will work for any arrangement of groups and scores. For example, the program should be
able to read the following data line without making any changes
A 16 20 B 24 B 22 28 18 A 18

It is not necessary for the subject number to match the other data sets. Just make sure that each
observation has a sequential subject number.
G. The data are arranged as follows
A 3 16 20 18
B 4 24 22 28 18

The number of subjects in each group follows the group code, followed by all the data values.
 CHAPTER 29 More Advanced Statistics Projects 191

PROBLEM 4
Creating New Variables, Regrouping Data, Producing Descriptive Statistics,
and Performing Significance Tests on Means and Proportions

Tools
Date literal
ROUND and LOG functions
PROC FREQ
PROC UNIVARIATE
WHERE statement
PROC TTEST
PROC NPAR1WAY
PROC ANOVA

Data
SAS data set CLINICAL

Directions
A. Create a new data set called HOSPITAL that contains the following new variables:
AGE Age as of 1/1/96 rounded to the nearest year
AGE_GRP Where ages 0-40 = 1; 41-65 = 2; >65 = 3
LOG_SBP Natural (base e) LOG of Systolic Blood Pressure (SBP)

B. Compute Fisher’s exact test to determine if the proportion of males is equal for the people
who take vitamins compared to people who do not take vitamins.
C. Compute the mean and standard deviation of SBP and LOG_SBP for each gender. Test if
the means are significantly different using a parametric and a nonparametric test.
D. Find the median HR for females (GENDER=F).
E. Test if the means for SBP and LOG_SBP are different as a function of AGE_GRP (as defined
in part A above). Conduct a Duncan multiple comparison test.
192 The SAS Workbook

PROBLEM 5
Working with Paired Data

Tools
PROC FREQ
PROC NPAR1WAY
RENAME= data set option
PROC SORT
MERGE statement
PROC UNIVARIATE

Data
Case-control study data as follows
SUBJ_ID MATCH_ID GROUP A B C X
-----------------------------------------
1 11 CASE 1 0 1 23
2 . CASE 1 1 1 33
3 12 CASE 1 1 0 28
4 14 CASE 1 1 1 29
5 18 CASE 0 1 1 33
6 23 CASE 1 0 0 77
11 1 CONT 0 0 0 41
12 3 CONT 1 0 0 29
13 . CONT 0 1 1 55
14 4 CONT 0 0 0 56
18 5 CONT 0 0 0 32

Directions
A researcher conducted a case-control study where each case was matched against a single
control, based on gender and age (plus or minus five years). The collected data is displayed in
the Data section.
A. Start out by creating a data set called UNPAIRED from the raw data and conducting an
unpaired analysis using all available data. Compute a Fisher’s exact test for the variables A,
B, and C versus GROUP and a Wilcoxon rank sum test for the variable X, with GROUP as
the independent variable.
B. Next, select all cases and controls who are paired and construct a data set called PAIRED,
where the cases and their matched controls are in the same observation. For example, the
first few observations in this new data set would be
 CHAPTER 29 More Advanced Statistics Projects 193

SUBJ_ID CASE_A CASE_B CASE_C CONT_A CONT_B CONT_C CASE_X CONT_X

----------------------------------------------------------------
1 1 0 1 0 0 0 23 41
3 1 1 0 1 0 0 28 29
more observations

One way to accomplish this goal is to place all case data in one data set and all control data
in another, eliminating all subjects where the MATCH_ID variable is missing. You can then
rename the MATCH_ID variable in one of the data sets to SUBJ_ID, rename the remaining
variables in both data sets (A becomes CASE_A in the case data set, and so on), sort both data
sets by SUBJ_ID, and then merge the two data sets by SUBJ_ID, being sure to keep only
subjects who form a pair.
C. Compute a McNemar chi-square for the variables that were originally called A, B, and C
and conduct a Wilcoxon signed rank test for the variable X with GROUP as the independent
variable.

Note
Not all cases and controls are paired.
194
 195

SECTION 5 SAS Puzzles to Test Your Skills

®

CHAPTERS
30 SAS Puzzles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .197
196
 197

Chapter 30 SAS Puzzles

®

This chapter contains some interesting puzzles that are fun to try and give you some insight into
the internal workings of the SAS language. Unlike the other chapters, this chapter does not
contain Problem headings or Tools lists.
Most of the problems center around understanding details of the DATA step. You can check
your answer on almost all the problems by running the programs and checking the results. You
may need to include some PUT statements in the DATA step or run the PRINT procedure
following the DATA step to check your answer. Some of the problems involve writing a few
lines of SAS code. For these, you will need to check The SAS Workbook Solutions (or drive a
colleague crazy) if you get stuck.

PROBLEM 1
After running the following program:

DATA PUZZLE1;
DO I = 1 TO 6;
INPUT X @@;
SUBJECT = SUBJECT + 1;
END;
DATALINES;
1 2 3
4 5 6
;

A. What is the value of I in observation 1?

B. How many observations are in the data set PUZZLE1?
C. What is the value of SUBJECT in the last observation?

PROBLEM 2
In the following program:

DATA PUZZLE2;
DO I = 1 TO 5;
INPUT X;
COUNTER = _N_;
OUTPUT;
END;
198 The SAS Workbook

DROP I;
DATALINES;
1
2
3
;

What is the value of COUNTER for each observation?

PROBLEM 3
In the following program:

DATA PUZZLE3;
INPUT X @@;
IF X GE 5 THEN Y = LAG(X);
DATALINES;
1 8 2 10 15
;

What are the values of Y for each of the five observations?

PROBLEM 4
In the following program:

DATA PUZZLE4;
DO I = 1 TO 5;
I + 1;
OUTPUT;
END;
RUN;

What are the values of I for each of the observations, and how many observations are there?
 CHAPTER 30 SAS Puzzles 199

PROBLEM 5
In the following program:

DATA PUZZLE5;
INPUT X @@;
IF X = 1 THEN X = 5;
IF X = 2 THEN X = 4;
IF X = 4 THEN X = 2;
IF X = 5 THEN X = 1;
DATALINES;
1 3 5
;

What is the value of X in each of the three observations?

PROBLEM 6
In the following program:

DATA PUZZLE6;
INPUT A B @@;
DATALINES;
1 1 1 2 1 3 2 1 2 2 2 3 3 1 3 2 3 3
;
PROC SORT DATA=PUZZLE6;
BY A B;
RUN;

DATA P6;
SET PUZZLE6;
BY A B;
FIRST_A = FIRST.A;
FIRST_B = FIRST.B;
LAST_A = LAST.A;
LAST_B = LAST.B;
RUN;

What are the values of FIRST_A, FIRST_B, LAST_A, and LAST_B for each of the nine
observations?
200 The SAS Workbook

PROBLEM 7
In the following program:

DATA PUZZLE7;
RETAIN X (9);
INPUT X @@;
DATALINES;
1 . 2 . 3
;

What are the values of X in each of the five observations?

PROBLEM 8
After running the following program:

DATA PUZZLE8;
INPUT X Y @;
DATALINES;
1 2 3 4 5 6 7 8
;

How many observations are there in data set PUZZLE8?

PROBLEM 9
You have a SAS data set called CLIFTON that contains GENDER (M or F) and AGE
(rounded to the nearest year). You want to compute a new variable.

DATA PUZZLE9;
SET CLIFTON;
IF GENDER = 'M' THEN NEWVAR = AGE + 20;
ELSE IF GENDER = 'F' THEN NEWVAR = AGE + 30;
RUN;

Can you rewrite this program so that NEWVAR is computed in one SAS statement?
 CHAPTER 30 SAS Puzzles 201

PROBLEM 10
You are given lines of text (all less than or equal to 80 bytes) consisting of words in
upper- and lowercase letters. There is at least one upper- or lowercase letter on each
text line. There may be punctuation in a line (period, comma, question mark, or excla-
mation mark). Write a compact DATA step that counts the number of letters in each
line and does not count spaces or punctuation.

Sample data to test your program

This is a test, this is only a TEST!!! (count = 26)
aaa bbb ...,,,xxx (count = 9)

Can you compute this number in one statement?

PROBLEM 11
In the program below:

DATA CHAR;
LENGTH A $ 1 C $ 10;
INPUT A $ B $ C $;
D = REPEAT(A,20);
E = SUBSTR(C,3,2);
F = A || B || C;
DATALINES;
X YY CCCCC
;

What are the lengths of the variables A, B, C, D, E, and F?

PROBLEM 12
In the following program:

DATA TEST;
DO I = 1 TO 3;
INPUT X @@;
N = _N_;
OUTPUT;
END;
DATALINES;
1 2 3 4 5 6
7 8 9 10 11 12
;

What is the value of N in the last observation?

202 The SAS Workbook

PROBLEM 13
You have a group variable with values of A, B, or C.

You want to compute a new variable TAX according to the following rule
For group A: TAX = .15 * INCOME
For group B: TAX = .27 * INCOME
For group C: TAX = .35 * INCOME

Assuming you have a data set with the variables GROUP and INCOME, write a single DATA
step statement that will compute TAX.

PROBLEM 14
After running the following program:

DATA PUZZLE14;
X = 3;
IF X = 1 OR 2 THEN LOGIC = 'TRUE ';
ELSE LOGIC = 'FALSE';
RUN;

What is the value of LOGIC?

PROBLEM 15
You have a dichotomous character variable SMOKE with values of Y or N (assume
there are no invalid or missing data). You want to create a numeric variable X_SMOKE
with values of 1 or 0, corresponding to the values Y and N. Two ways to do this are

IF SMOKE = 'Y' THEN X_SMOKE = '1';

ELSE X_SMOKE = '0';

or

X_SMOKE = INPUT (TRANSLATE(SMOKE,'10','YN'),1.);

Can you compute X_SMOKE using one statement without using any functions?
 Appendix 203

SECTION 6

APPENDIX
File Descriptions and Data Program to Create SAS Data
Set CLINICAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .205
204
 205

File Descriptions and Data Program

Appendix to Create SAS Data Set CLINICAL
®

A number of raw data files and SAS data sets are used throughout this workbook. This appendix
contains a description of the raw data files, a listing of the actual data files, and a program for
creating a SAS data set called CLINICAL from raw data file CLINICAL.DTA. All the programs
and solutions in this workbook assume that the raw data files and the permanent SAS data sets
are to be stored in a subdirectory called C:\WORKBOOK. Make the appropriate adjustments if
you are using another subdirectory or if you are running SAS software in a different operating
environment.
You can download the data files and the program to create the SAS data set CLINICAL using
the instructions located on the inside back cover of this book. If you don’t have a modem, you
may want to create the raw data files by typing in the data listed in the appendix. You can also
enter the program CLINICAL.SAS and run it.
Data files are listed alphabetically by name.

BASKET.DTA Data File Description

Starting
Variable Description Column Length Type
------------------------------------------------------------------
ID Employee ID 1 3 Char
GENDER Gender 5 1 Char (M or F)
HEIGHT Height in inches 7 2 Numeric
DATE Date of game 10 8 MM/DD/YY
POINTS Number of points 19 2 Numeric

Listing of Raw Data from BASKET.DTA

1 2
12345678901234567890
--------------------
001 M 72 12/01/95 10
001 M 72 12/03/95 12
001 M 72 12/10/95 8
002 F 68 10/21/95 14
003 M 74 12/12/95 20
003 M 74 12/16/95 22
004 M 74 11/12/95 6
004 M 74 11/15/95 8
005 F 67 12/12/95 10
206 The SAS Workbook

005 F 67 12/15/95 18
006 M 69 10/30/95 2
006 M 69 11/04/95 6
006 M 69 11/06/95 8
007 M 69 11/11/95 9
007 M 69 11/16/95 7

BASKETBA.DTA Data File Description

Starting
Variable Description Column Length Type
------------------------------------------------------------------
ID Employee ID 1 3 Char
GENDER Gender 5 1 Char (M or F)
HEIGHT Height in inches 7 2 Numeric
DATE Date of game 10 8 MM/DD/YY
POINTS Number of points 19 2 Numeric

Note
The file BASKETBA.DTA is the same as the file BASKET.DTA except that GENDER and
HEIGHT are entered in only the first record for each player in the BASKETBA.DTA file, and
GENDER and HEIGHT are missing for some players.

Listing of Raw Data from BASKETBA.DTA

1 2
12345678901234567890
--------------------
001 M 72 12/01/95 10
001 12/03/95 12
001 12/10/95 8
002 F 68 10/21/95 14
003 M 74 12/12/95 20
003 12/16/95 22
004 11/12/95 6
004 11/15/95 8
005 F 67 12/12/95 10
005 12/15/95 18
006 M 69 10/30/95 2
006 11/04/95 6
006 11/06/95 8
007 11/11/95 9
007 11/16/95 7
 Appendix 207

CARS.DTA Data File Description

Starting
Variable Description Column Length Type
----------------------------------------------------------
SIZE Car Size 1 9 Char
MANUFACT Manufacturer 11 9 Char
MODEL Model 22 9 Char
MILEAGE Gas Mileage 38 2 Numeric
RELIABLE Reliability 50 1 Numeric

Listing of Raw Data from CARS.DTA

1 2 3 4 5
12345678901234567890123456789012345678901234567890
--------------------------------------------------
SMALL CHEVROLET GEO PRIZM 33 5
SMALL HONDA CIVIC 29 5
SMALL TOYOTA COROLLA 30 5
SMALL FORD ESCORT 27 3
SMALL DODGE COLT 34
COMPACT FORD TEMPO 24 1
COMPACT CHRYSLER LE BARON 23 3
COMPACT BUICK SKYLARK 21 3
COMPACT PLYMOUTH ACCLAIM 24 3
COMPACT CHEVROLET CORSICA 25 2
COMPACT PONTIAC SUNBIRD 24 1
MID-SIZED TOYOTA CAMRY 24 5
MID-SIZED HONDA ACCORD 26 5
MID-SIZED FORD TAURUS 20 3

CLIN_X.DTA Data File Description

Note
1. Multiple observations per patient
2. Not sorted by ID or visit date
3. Contains some short records (not padded with blanks)
208 The SAS Workbook

Starting
Variable Description Column Length Format Codes
---------------------------------------------------------------------------
ID Patient ID 1 3 Char
VISIT Date of Visit 4 6 MMDDYY
DX Diagnosis 10 2 Char (see list)
HR Heart Rate 12 3 Numeric
SBP Systolic Blood Pressure 15 3 Numeric
DBP Diastolic Blood Pressure 18 3 Numeric
RX_1 Treatment 1 21 1 Char (see list)
RX_2 Treatment 2 22 1 Char (see list)

DX Codes: RX Codes:

01 Cold 1 Immunization
02 Flu 2 Casting
03 Fracture 3 Beta Blocker
04 Routine Physical 4 ACE Inhibitor
05 Heart Problem 5 Antihistamine
06 Lung Disorder 6 Ibuprofen
07 Abdominal Pain 7 Aspirin
08 Laceration 8 Antibiotic
09 URI
10 Lyme Disease
11 Ear Ache

Listing of Raw Data from CLIN_X.DTA

1 2 3
123456789012345678901234567890
------------------------------
7770901941108809806468
04505189404086180092
033010594090841280788
0451021950508821012034
033090995020681200706
033101095070721220748
0451031950508418010234
009123195030881340882
033051894010881280786
17706069504058120074
045122195060781500808
00909099507078128082
009070195080581280844
 Appendix 209

1770818951008213008086
4450919950308814408826
77709129507086100062
0091015950106612208257
0781101950509019011073
033101095100581200708

CLINICAL.DTA Data File Description

Starting
Variable Description Column Length Format Codes
----------------------------------------------------------------------
ID Subject ID 1 3 Char
GENDER Gender 4 1 Char M,F
DOB Date of Birth 5 6 MMDDYY
VISIT Visit Date 11 6 MMDDYY
PRIM_DX Primary DX Code 17 2 Char (see list)
SEC_DX Secondary DX Code 19 2 Char "
HR Heart Rate 21 3 Numeric
SBP Systolic Blood Pres. 24 3 Numeric
DBP Diastolic Blood Pres. 27 3 Numeric
VITAMINS Pt. Taking Vitamins? 30 1 Char 1=Yes, 0=No
PREGNANT Is PT Pregnant? 31 1 Char 1=Yes, 0=No

DX Codes:

01 Cold 07 Abdominal Pain

02 Flu 08 Laceration
03 Fracture 09 Immunization
04 Routine Physical 10 Lyme Disease
05 Heart Problem 11 Ear Ache
06 Lung Disorder
210 The SAS Workbook

Listing of Raw Data from CLINICAL.DTA

1 2 3 4
1234567890123456789012345678901234567890
----------------------------------------
123M102146051595010506213607610
278F11015505159504 06810406400
444F07086505169501 07812806211
756M122562061195070906615009600
811F030475060195050608816607400
193F10105908159504 07011206811
M10159502159604 08009805410
978M062894070295040908809806210
586M120409051895050608616209610
919F09069306209511 09205805010
529M07135008099510 05811407600
324F01011075119506 08620010010
012F100908760195030809018409800
812M10116710109508 06613408210
338M07289007129511 08210607010
959F121565111195070908817609810
007F04196402069504 07410606411
291M080975060395011106813208600
984F12237905159502 07409606201
669M02287707079509 06214009000
999F01019502019401 08018810200
229F06188009189509 07410406010
885M091085101595110908811408400
178M030491071695010908810206800
F010190010696050607012008000
966F040591081695010909205403610
782M060710051895050608822018000
374M05068807229504 08312007000
285F06077009119511 08416009600
884M07121609119505 07819014000
258M08098211189504 07214409800
733F090978030495070207412007011
449M10103110189501 10
941F101070101995030408819011001
 Appendix 211

Program to Create the SAS Data Set

CLINICAL

*--------------------------------------------------------------------*
| Program Name: CLINICAL.SAS in subdirectory C:\WORKBOOK |
| Purpose: To Create a Permanent SAS data set called CLINICAL from |
| the raw data file CLINICAL.DTA along with permanent |
| formats. Both the SAS data set and permanent formats are |
| to be stored in the subdirectory C:\WORKBOOK |
*--------------------------------------------------------------------* ;

LIBNAME WORKBOOK 'C:\WORKBOOK';

OPTIONS PAGENO=1 NOCENTER FMTSEARCH=(WORKBOOK);

FILENAME CLIN 'C:\WORKBOOK\CLINICAL.DTA';

PROC FORMAT LIBRARY=WORKBOOK;

VALUE $GENDER 'M' = 'Male'
'F' = 'Female';
VALUE $DXCODES '01' = 'Cold'
'02' = 'Flu'
'03' = 'Fracture'
'04' = 'Routine Physical'
'05' = 'Heart Problem'
'06' = 'Lung Disorder'
'07' = 'Abdominal Pain'
'08' = 'Laceration'
'09' = 'Immunization'
'10' = 'Lyme disease'
'11' = 'Ear Ache';
VALUE $YESNO '0' = 'No'
'1' = 'Yes';
RUN;
212 The SAS Workbook

DATA WORKBOOK.CLINICAL;
INFILE CLIN;
INPUT @1 ID $3.
@4 GENDER $1.
@5 DOB MMDDYY6.
@11 VISIT MMDDYY6.
@17 PRIM_DX $2.
@19 SEC_DX $2.
@21 HR 3.
@24 SBP 3.
@27 DBP 3.
@30 VITAMINS $1.
@31 PREGNANT $1.;

LABEL ID = 'Pt. Number'

GENDER = 'Gender'
DOB = 'Date of Birth'
VISIT = 'Visit Date'
PRIM_DX = 'Primary DX'
SEC_DX = 'Secondary DX'
HR = 'Heart Rate'
SBP = 'Systolic Blood Pressure'
DBP = 'Diastolic Blood Pressure'
VITAMINS = 'Pt. Taking Vitamins?'
PREGNANT = 'Is Pt. Pregnant?';

FORMAT PRIM_DX SEC_DX $DXCODES.

DOB VISIT MMDDYY8.
GENDER $GENDER.
VITAMINS PREGNANT $YESNO.;

RUN;

CPR.DTA Data File Description

Starting
Variable Description Column Length Format Codes
---------------------------------------------------------------------------
SUBJECT Subject number 1 3 Numeric
V_FIB Ventricular Fibrillation 4 1 Numeric 1=Yes, 0=No
RESP Pt on Respirator? 5 1 Numeric 1=Yes, 0=No
AGEGROUP Age greater than 70? 6 1 Numeric 1=Yes, 0=No
SURVIVE Did patient Survive? 7 1 Numeric 1=Yes,
0=No
 Appendix 213

Listing of Raw Data from File CPR.DTA

1
1234567890
----------
0011001
0021111
0031110
0040010
0050010
0060110
0070011
0081001
0090000
0101011
0110000
0120101
0130100
0141111
0150101
0161001
0170011
0180010
0190110
0201011
0210110
0220010
0231110
0240001
0251011
0261110
0270110
0280110
0291100
0300100
0310001
0320010
0330001
0340100
0350010
0360001
214 The SAS Workbook

DEMOG1.DTA and DEMOG2.DTA Data

File Descriptions

Starting
Variable Description Column Length Type Codes
----------------------------------------------------------------------------
ID Subject ID 1 3 Char
DOB Date of Birth 4 6 MMDDYY
GENDER Gender 10 1 Char M=Male, F=Female
STATE State of Residence 11 2 Char
EMPLOYED Employed? 13 1 Char 1=Yes,
0=No

Listing of Raw Data from DEMOG1.DTA

1 2 3 4 5 6
123456789012345678901234567890123456789012345678901234567890
------------------------------------------------------------
178102146MNJ1
982110166FNY0
723030452MNJ1
258101074FNY1
139122550FNJ0
145010444MNY1

Listing of Raw Data from DEMOG2.DTA

1 2 3 4 5 6
123456789012345678901234567890123456789012345678901234567890
------------------------------------------------------------
559110168FNY0
714010244MNH0
934052890FNJ0
569112850FNY1
 Appendix 215

DIALYSIS.DTA Data File Description

Starting
Variable Description Column Length Type Codes
----------------------------------------------------------------------------
ID Subject ID 1 3 Char
GENDER Gender 4 1 Char M=Male, F=Female
DOB Date of Birth 5 8 MM/DD/YY
VISIT Visit Number 13 1 Numeric
HR Heart Rate 14 3 Numeric
SBP Systolic Blood Pressure 17 3 Numeric
DBP Diastolic Blood Pressure 20 3 Numeric

Listing of Raw Data from DIALYSIS.DTA

1 2 3
123456789012345678901234567890
------------------------------
001M10/21/461080140080
001 2082142084
001 3078138078
002F11/17/221066120070
003F04/04/181084150090
003 2088152102
004M12/21/101072120074
004 2070122076
004 3078128078
005F08/02/311092180110
006 1076180112
006 2080178090
216