Journal of Official Statistics, Vol.14, No.3, 1998. pp. 277295

Current Issue
Personal Reference Library (PRL)
Personal Page

Design and Analysis of Experiments Embedded in Sample Surveys

This article focuses on the design and analysis of large-scale field experiments embedded in ongoing surveys. Historically, experimental designs and sampling theory have formed two separated domains in applied statistics. However, in the design and analysis of embedded experiments, statistical methods from experimental designs and sampling theory can be combined in order to improve the quality and efficiency of these experiments. We will discuss how parallels between randomized experiments and random survey samples can support the design of field experiments embedded in ongoing surveys. This is illustrated with a series of embedded experiments conducted at Statistics Netherlands. This article focuses on a design-based approach for the analysis of such experiments. Generally, the objective of embedded field experiments is to draw inferences on finite population parameters. To this end the analysis has to include the probability structure imposed by the applied sampling design of the survey as well as the randomization of the applied experimental design used to assign the experimental units to the different treatments. We worked out such a design-based method for the analysis of the two-sample problem.

Experimental design; sampling design; Horvitz-Thompson estimator; generalized regression estimator; two-sample problem.

Copyright Statistics Sweden 1996-2017.  Open Access
ISSN 0282-423X
Created and Maintained by OKS Group