Kernel Density Estimation with Missing Data and Auxiliary Variables
Dr. Suzanne Dubnicka, Department of Statistics, Kansas State University

Monday, April 2, 2007
4:10 p.m.
223 Weber


The concept of density function is central to many statistical methods, either directly or indirectly.  Often summary measures regarding this density, such as the mean or variance, are of primary concern.  However, there are situations for which the shape of the underlying density function is of primary interest.  Unfortunately, in many situations such as large clinical trials, data are incomplete.  We will present a weighted kernel approach to estimating the density when some responses are missing but auxiliary variables are available. The weights are based on inverse propensity scores which will be estimated under the assumption that the responses are missing at random.  The proposed method will be illustrated on data from the AIDS Clinical Trial Group protocol 175 and will be evaluated in a simulation study.  As with all kernel density estimators, bandwidth selection is an important issue and will be discussed.  Uses for the proposed kernel density estimator will also be explored.



  home for studentspeoplecoursesdistance degreeresearchconsulting seminars resources
PRIMESSTARMAPemail us equal opportunitydisclaimerweb master • last modified: August 11 2003