Trivellore Raghunathan – författare

The Survey of Income and Program Participation (SIPP) is one of the U.S. Census Bureau''s major surveys with features making it a uniquely valuable resource for researchers and policy analysts. However, the Census Bureau faces the challenge of protecting the confidentiality of survey respondents which has become increasingly difficult because numerous databases exist with personal identifying information that collectively contain data on household finances, home values, purchasing behavior, and other SIPP-relevant characteristics.

A Roadmap for Disclosure Avoidance in the Survey of Income and Program Participation addresses these issues and how to make data from SIPP available to researchers and policymakers while protecting the confidentiality of survey respondents. The report considers factors such as evolving privacy risks, development of new methods for protecting privacy, the nature of the data collected through SIPP, the practice of linking SIPP data with administrative data, the types of data products produced, and the desire to provide timely access to SIPP data. The report seeks to balance minimizing the risk of disclosure against allowing researchers and policymakers to have timely access to data that support valid inferences.

Multiple Imputation in Practice: With Examples Using IVEware provides practical guidance on multiple imputation analysis, from simple to complex problems using real and simulated data sets. Data sets from cross-sectional, retrospective, prospective and longitudinal studies, randomized clinical trials, complex sample surveys are used to illustrate both simple, and complex analyses.

Version 0.3 of IVEware, the software developed by the University of Michigan, is used to illustrate analyses. IVEware can multiply impute missing values, analyze multiply imputed data sets, incorporate complex sample design features, and be used for other statistical analyses framed as missing data problems. IVEware can be used under Windows, Linux, and Mac, and with software packages like SAS, SPSS, Stata, and R, or as a stand-alone tool.

This book will be helpful to researchers looking for guidance on the use of multiple imputation to address missing data problems, along with examples of correct analysis techniques.