ST14
Managing Missing Data with Multiple Imputation using PROC MI in SAS(R)
Contributed
Dr. Hossein N. Yarandi
University of Florida
Abstract:
In the data analysis phase of scientific research, missing values present a challenge to investigators. In particular, for the analysis of survey data, accommodation of the incomplete data is critical to making valid inferences. Common approaches for addre
ssing missing data generally include complete-case analysis, available-case analysis, and various single-value imputation methods. These methods have been the subject of increasing criticism with respect to their tendency to underestimate standard errors,
overstate statistical significance, and it introduces bias. Many existing methods cannot be viewed as an adequate approach for addressing the degree and complex patterns of missingness. An alternative approach for managing incomplete data is multiple imput
ation. Through this method, we can replace each missing value with a set of plausible values that represent the uncertainty about the right value to impute. The multiply imputed data sets are then analyzed by using standard procedures for complete data and
combining the results from these analyses. Multiple imputation does not attempt to estimate each missing value through simulated values but rather to represent a random sample of the missing values. These process results in valid statistical inferences th
at properly reflect the uncertainty caused by missing values. PROC MI in SAS creates multiply imputed data sets for incomplete multivariate data. It uses methods that incorporate appropriate variability across the imputations. The method of choice depends
on the patterns of missingness. The purpose of this study is to apply multiple imputation to missing data in a Medical Expenditure Panel Survey using PROC MI.
Biography:
Hossein N. Yarandi, PhD, is an Associate professor in the College of Nursing and Biostatistics Unit at the University of Florida. He teaches research, statistical methods, and design at the baccalaureate, masters, and doctorate levels; supervises theses an
d dissertations, and consults with the faculty members about research grants and projects. His research interests are on research methodology, health care economics and financing health care.