Accuracy in Estimation with Nonresponse: A Function of Degree of Imbalance and Degree of Explanation

• Activity Number: 630
• Type: Invited
• Date/Time: Thursday, August 13, 2015 : 8:30 AM to 10:20 AM
• Sponsor: Journal of Survey Statistics and Methodology
• Author(s): Carl-Erik Sarndal*
• Companies: Statistics Sweden
• Keywords: Auxiliary variables ; Fieldwork planning ; Intervention in fieldwork ; Responsive design

Abstract

Responsive design is concerned with managing data collection, through planning and intervention, to obtain a well-balanced set of respondents. The ultimate goal, though, is accurate estimation even when there is nonresponse. Auxiliary variables are central here and at the estimation stage. For accuracy, there are two factors, i.e., (i)in the data collection the level of imbalance achieved with the auxiliary information, and (ii)in the estimation the degree to which the auxiliaries explain the study variable. Since reduced imbalance does not by itself guarantee low bias, we ask whether this costly effort in data collection is justified. Could one have done as well by using the auxiliary information at the estimation stage? We outline theory for a two factor explanation of accuracy and apply it to two surveys at Statistics Sweden. The factors, degree of imbalance and degree of explanation, are systematically varied, and their joint effect on the accuracy of the estimates is evaluated empirically. One conclusion is that the calibration adjusted estimate realizes some improvement in accuracy by the earlier reduction in imbalance.