Table of Contents
List of Programs ix
Preface xvii
Acknowledgments xix
The RETAIN Statement
Introduction 1
Demonstrating a DATA Step with and without a RETAIN Statement 1
Generating Sequential SUBJECT Numbers Using a Retained Variable 7
Using a SUM Statement to Create SUBJECT Numbers 9
Demonstrating That Variables Read with a SET Statement
Are Retained 10
A Caution When Using a RETAIN Statement 11
The LAG and DIF Functions
Introduction 13
Using the LAG Function to Compute Differences 13
Demonstrating Some Related Functions: LAG2, LAG3, and So Forth 16
Demonstrating the DIF Function 17
FIRST. and LAST. Temporary Variables
Introduction 19
How to Create FIRST. and LAST. Temporary Variables 19
Using More Than One BY Variable 22
A Simple Application Using FIRST. and LAST. Variables 24

Flags and Counters
Introduction 27
Using a Flag Variable to Determine If a Particular Event Ever
Occurred in Any One of Several Observations for Each Subject 27
Counting the Number of Positive Outcomes for Each Patient 29
Summarizing Data Using PROC MEANS and
PROC FREQ
Introduction 33
Using PROC MEANS to Output Means to a Data Set 34
Comparing CLASS and BY Statements with PROC MEANS 37
Computing Other Descriptive Statistics 38
Automatically Naming the Variables in the Output Data Set 40
Demonstrating an Alternative Way to Select Specific Descriptive
Statistics for Selected Variables 41
Adding Additional Variables to the Summary Data Set Using an ID
Statement 42
Specifying More Than One CLASS Variable 44
Selecting Multi-Way Breakdowns Using the TYPES Statement 47
Using the PROC MEANS CHARTYPE Option to Simplify the _TYPE_
Interpretation 49
Comparing PROC MEANS and PROC FREQ for Creating an Output
Data Set Containing Counts 50
Counting Frequencies for a Two-Way Table 52
Using PROC SQL with Longitudinal Data
Introduction 55
Creating a Demonstration Data Set 55
A Simple SQL Query 57
Using PROC SQL to Count Observations within a BY Group 58
Demonstrating a HAVING Clause 59
Using PROC SQL to Create a Macro Variable 60
Using a Summary Function to Compute Group Means 62

Restructuring SAS Data Sets Using Arrays
Introduction 65
Creating a New Data Set with Several Observations per Subject from
a Data Set with One Observation per Subject 66
Another Example of Creating Multiple Observations from a Single
Observation 69
Going from One Observation per Subject to Many Observations per
Subject Using Multidimensional Arrays 72
Demonstrating the Use of a Multidimensional Array 74
An Alternative Program 77
Another Example of a Multidimensional Array 78

Restructuring SAS Data Sets Using PROC TRANSPOSE
Introduction 81
Going from One Observation to Several Observations 81
Another Example of Creating Multiple Observations from a Single
Observation 84
Going from One Observation per Subject to Many Observations
per Subject 86
Creating a Data Set with One Observation per Subject from a Data Set
with Multiple Observations per Subject 88
Study One: Operations on a Clinical Database
Introduction 94
Description of the Clinical Data Set 94
Selecting the First or Last Visit for Each Patient 95
Computing Differences between the First and Last Visits 97
Another Method of Computing Differences between the First and
Last Visits 99
Computing Differences between Every Visit 100
Counting the Number of Visits for Each Patient (DATA Step Approach) 101
Counting the Number of Visits for Each Patient (PROC FREQ) 103
Counting the Number of Visits for Each Patient (PROC MEANS) 103
Counting the Number of Visits for Each Patient (PROC SQL) 104
Selecting All Patients with n Visits (DATA Step Approach) 105
Selecting All Patients with n Visits (PROC FREQ Approach) 106
Selecting All Patients with Two Visits (Using PROC SQL) 107
Selecting All Patients with Two Visits (Using SQL in One Step) 107
Using PROC SQL to Create a Macro Variable 108
Computing Summary Statistics for Each Patient (Using PROC MEANS) 109

Computing Summary Statistics for Each Patient (Using PROC SQL) 110
Adding a Value from the First Visit to Each Subsequent Visit 111
Looking Ahead: Making a Decision about the Current Observation
Based on Information in the Next Observation 114
Using Flags to Ascertain Vitamin Use 117
Using PROC FREQ to Ascertain Vitamin Use 118
Counting the Number of Routine Visits for Each Patient 119
Study Two: Operations on Daily Weather Data and
Ozone Levels
Introduction 121
The OZONE Data Set 121
Computing Weekly Averages 122
Using the MOD Function to Group Data Values 125
Computing a Moving Average for a Single Variable 127
Study Three: Producing Summary Reports on a Library
Data Set
Introduction 129
Computing the Number of Books per Patron Visit and by Library 130
Computing the Number of Patrons by Day of Week and Library 134
Generating a Table of LC Categories by Age Group and Overall 135

Useful Macros
Introduction 141
Listing All or Part of a Data Set 141
Computing Differences between Successive Observations 143
Computing Differences between the First and Last Observations
per Subject 145
Computing a Moving Average 147
Computing Cell Means and Counts 149
Counting the Number of Observations per Subject 151
List of List of Data Files and SAS Data Sets
The TEST_SCORES Data Set 153
The CLINICAL Data Set 154
The CLIN_FIRST Data Set 156
The OZONE Data Set 157
The LIBRARY Data Set 160
165