Washington Statistical Society on Meetup

2010 Hansen Lecture

Carl-Erik Särndal, Statistics Sweden
"Dealing with Survey Nonresponse: In Data Collection, in Estimation"

Discussants: J. Michael Brick, Westat
Roger Tourangeau, University of Michigan & Joint Program in Survey Methodology, University of Maryland

Abstract: In dealing with survey nonresponse, statisticians need to consider (a) measures to be taken at the data collection stage, and (b) measures to be taken at the estimation stage. One may practice a form of responsive design: In the later stages of the data collection, one tries to realize an ultimate response set that is better balanced or more representative than if no special effort is made. When the data collection is terminated, one still faces, at the estimation stage, the question of how to achieve the best possible reduction of nonresponse bias in the estimates. This effort is aided by a bias indicator, constructed as a function of selected powerful auxiliary variables. The concept of balanced response set extends the well known idea of balanced sample. It has a bearing on both aspect (a) and aspect (b). The statistical properties of a proposed measure of lack of balance are explored and illustrated.