Abstract:
|
Unit-nonresponse is usually compensated for by weighting adjustment, and calibration techniques are frequently used for benchmarking purposes and for improving survey estimates. Traditionally, nonresponse adjustment and calibration are performed in two separate steps, and this weighting procedure is called the two-step procedure. Lately, there has been considerable discussion on whether the calibration step can take care of the nonresponse adjustment as well in a single step (called the one-step procedure). The answer to this question is not clear cut, and in this paper we want to address this issue for various situations arising in practice in terms of relationship between auxiliary information and survey variables and response mechanisms. This is an empirical investigation built upon the work by Haziza and Lesage (2015).
|