Multiple Imputation of Missing Data Using SAS

Patricia Berglund, Steven G. Heeringa

Description

Find guidance on using SAS for multiple imputation and solving common missing data issues.
Multiple Imputation of Missing Data Using SAS provides both theoretical background and constructive solutions for those working with incomplete data sets in an engaging example-driven format. It offers practical instruction on the use of SAS for multiple imputation and provides numerous examples that use a variety of public release data sets with applications to survey data.
Written for users with an intermediate background in SAS programming and statistics, this book is an excellent resource for anyone seeking guidance on multiple imputation. The authors cover the MI and MIANALYZE procedures in detail, along with other procedures used for analysis of complete data sets. They guide analysts through the multiple imputation process, including evaluation of missing data patterns, choice of an imputation method, execution of the process, and interpretation of results.
Topics discussed include how to deal with missing data problems in a statistically appropriate manner, how to intelligently select an imputation method, how to incorporate the uncertainty introduced by the imputation process, and how to incorporate the complex sample design (if appropriate) through use of the SAS SURVEY procedures.
Discover the theoretical background and see extensive applications of the multiple imputation process in action.
This book is part of the SAS Press program.

Sujets

• COMPUTERS
• Enterprise software
• Data quality
• Data management
• General officer
• Mathematics
• General

Informations

• Publié par: SAS Institute
• Date de parution: 01 juillet 2014
• Nombre de lectures: 0
• EAN13: 9781629592039
• Langue: English
• Poids de l'ouvrage: 3 Mo

Extrait

Multiple Imputation of Missing Data Using SAS
Patricia Berglund and Steven Heeringa
support.sas.com/bookstore
The correct bibliographic citation for this manual is as follows: Berglund, Patricia and Heeringa, Steven, 2014. Multiple Imputation of Missing Data Using SAS . Cary, NC: SAS Institute Inc.
Multiple Imputation of Missing Data Using SAS
Copyright 2014, SAS Institute Inc., Cary, NC, USA
ISBN 978-1-62959-203-9
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.
July 2014
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 support.sas.com/bookstore or call 1-800-727-3228.
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.
Contents
About This Book
About The Authors
Acknowledgements
Chapter 1: Introduction to Missing Data and Methods for Analyzing Data with Missing Values
1.1 Introduction
1.2 Sources and Patterns of Item Missing Data
1.3 Item Missing Data Mechanisms
1.4 Review of Strategies to Address Item Missing Data
1.4.1 Complete Case Analysis
1.4.2 Complete Case Analysis with Weighting Adjustments
1.4.3 Full Information Maximum Likelihood
1.4.4 Expectation-Maximization Algorithm
1.4.5 Single Imputation of Missing Values
1.4.6 Multiple Imputation
1.5 Outline of Book Chapters
1.6 Overview of Analysis Examples
Chapter 2: Introduction to Multiple Imputation Theory and Methods
2.1 The Origins and Properties of Multiple Imputation Methods for Missing Data
2.1.1 A Short History of Imputation Methods
2.1.2 Why the Multiple Imputation Method?
2.1.3 Overview of Multiple Imputation Steps
2.2 Step 1-Defining the Imputation Model
2.2.1 Choosing the Variables to Include in the Imputation Model
2.2.2 Distributional Assumptions for the Imputation Model
2.3 Algorithms for the Multiple Imputation of Missing Values
2.3.1 General Theory for Multiple Imputation Algorithms
2.3.2 Methods for Monotone Missing Data Patterns
2.3.3 Methods for Arbitrary Missing Data Patterns
2.4 Step 2-Analysis of the MI Completed Data Sets
2.5 Step 3-Estimation and Inference for Multiply Imputed Data Sets
2.5.1 Multiple Imputation-Estimators and Variances for Descriptive Statistics and Model Parameters
2.5.2 Multiple Imputation-Confidence Intervals
2.6 MI Procedures for Multivariate Inference
2.6.1 Multiple Parameter Hypothesis Tests
2.6.2 Tests of Linear Hypotheses
2.7 How Many Multiple Imputation Repetitions Are Needed?
2.8 Summary
Chapter 3: Preparation for Multiple Imputation
3.1 Planning the Imputation Session
3.2 Choosing the Variables to Include in a Multiple Imputation
3.3 Amount and Pattern of Missing Data
3.4 Types of Variables to Be Imputed
3.5 Imputation Methods
3.6 Number of Imputations (MI Repetitions)
3.7 Overview of Multiple Imputation Procedures
3.8 Multiple Imputation Example
3.9 Summary
Chapter 4: Multiple Imputation for the Analyzsis of Complex Sample Survey Data
4.1 Multiple Imputation and Informative Data Collection Designs
4.2 Complex Sample Surveys
4.3 Incorporating the Complex Sample Design in the MI Imputation Step
4.4 Incorporating the Complex Sample Design in the MI Analysis and Inference Steps
4.5 MI Imputation and Analysis for Subpopulations of Complex Sample Design Data Sets
4.6 Summary
Chapter 5: Multiple Imputation of Continuous Variables
5.1 Introduction to Multiple Imputation of Continuous Variables
5.2 Imputation of Continuous Variables with Arbitrary Missing Data
5.3 Imputation of Continuous Variables with Mixed Covariates and a Monotone Missing Data Pattern Using the Regression and Predictive Mean Matching Methods
5.3.1 Imputation of Continuous Variables with Mixed Covariates and a Monotone Missing Data Pattern Using the Regression Method
5.3.2 Imputation of Continuous Variables with Mixed Covariates and a Monotone Missing Data Pattern Using the Predictive Mean Matching Method
5.4 Imputation of Continuous Variables with an Arbitrary Missing Data Pattern and Mixed Covariates Using the FCS Method
5.4.1 Imputation of Continuous Variables with an Arbitrary Missing Data Pattern and Mixed Covariates Using the FCS Method
5.5 Summary
Chapter 6: Multiple Imputation of Classification Variables
6.1 Introduction to Multiple Imputation of Classification Variables
6.2 Imputation of a Classification Variable with a Monotone Missing Data Pattern Using the Logistic Method
6.3 Imputation of Classification Variables with an Arbitrary Missing Data Pattern and Mixed Covariates Using the FCS Discriminant Function and the FCS Logistic Regression Method
6.4 Imputation of Classification Variables with an Arbitrary Missing Data Pattern and Mixed Covariates: A Comparison of the FCS and MCMC/Monotone Methods
6.4.1 Imputation of Classification Variables with Mixed Covariates and an Arbitrary Missing Data Pattern Using the FCS Method
6.4.2 Imputation of Classification Variables with Mixed Covariates and an Arbitrary Missing Data Pattern Using the MCMC/Monotone and Monotone Logistic Methods with a Multistep Approach
6.5 Summary
Chapter 7: Multiple Imputation Case Studies
7.1 Multiple Imputation Case Studies
7.2 Comparative Analysis of HRS 2006 Data Using Complete Case Analysis and Multiple Imputation of Missing Data
7.2.1 Exploration of Missing Data
7.2.2 Complete Case Analysis Using PROC SURVEYLOGISTIC
7.2.3 Multiple Imputation of Missing Data with an Arbitrary Missing Data Pattern Using the FCS Method with Diagnostic Trace Plots
7.2.4 Logistic Regression Analysis of Imputed Data Sets Using PROC SURVEYLOGISTIC
7.2.5 Use of PROC MIANALYZE with Logistic Regression Output
7.2.6 Comparison of Complete Case Analysis and Multiply Imputed Analysis
7.3 Imputation and Analysis of Longitudinal Seizure Data
7.3.1 Introduction to the Seizure Data
7.3.2 Exploratory Analysis of Seizure Data
7.3.3 Conversion of Multiple-Record to Single-Record Data
7.3.4 Multiple Imputation of Missing Data
7.3.5 Conversion Back to Multiple Record Data for Analysis of Imputed Data Sets
7.3.6 Regression Analysis of Imputed Data Sets
7.4 Summary
Chapter 8: Preparation of Data Sets for PROC MIANALYZE
8.1 Preparation of Data Sets for Use in PROC MIANALYZE
8.2 Imputation of Major League Baseball Players Salaries
8.3.1 PROC GLM Output Data Set for Use in PROC MIANALYZE
8.3.2 PROC MIXED Output Data Set for Use in PROC MIANALYZE
8.4 Imputation of NCS-R Data
8.5 PROC SURVEYPHREG Output Data Set for Use in PROC MIANALYZE
8.6 Summary
References
Index
About This Book
Purpose
Multiple Imputation of Missing Data Using SAS provides both theoretical background and constructive solutions for those working with incomplete data sets in an engaging example-driven format. It offers practical instruction on the use of SAS for multiple imputation and provides numerous examples using a variety of public release data sets.
Is This Book for You?
Written for users with an intermediate background in SAS programming and statistics, this book is an excellent resource for anyone seeking guidance on multiple imputation. The authors cover PROC MI and PROC MIANALYZE in detail along with other procedures used for analysis of complete data sets. They guide analysts through the multiple imputation process, including evaluation of missing data patterns, choice of an imputation method, execution of the p