Comments on: Deville and Särndal’s calibration: revisiting a 25 years old successful optimization problem
- 27 Downloads
It is a pleasure to comment on this interesting article by Devaud and Tillé, whose paper gives us the opportunity to reflect on the important impact produced by this method of sampling parameter estimation. Since the seminal work by Deville and Särndal, calibration has been one of the most useful tools available with which to incorporate auxiliary information in survey sampling. This technique ensures that the estimates obtained are coherent with those already in the public domain, while simultaneously reducing non-coverage, non-response and selection biases (Arcos et al. 2014). Although other important estimation methods that also use auxiliary information have been proposed (e.g. the empirical likelihood method; Chen and Qin 1993 or that of model-based estimators; Valliant et al. 2000), in practice, the vast majority of national statistical agencies use calibration as a reweighting technique and have developed software to compute calibrated weights in accordance with...
This study was partially supported by Ministerio de Economía y Competitividad (Grant MTM2015-63609-R).
- Ferri R, Castro L, Rueda M (2019) Superpopulation models using machine learning methods for estimation in online surveys. In: Proceeding of mathematical and computational modelling, approximation and simulation. MACMAS 2019Google Scholar
- Harms T, Duchesne P (2006) On calibration estimation for quantiles. Surv Methodol 32:37–52Google Scholar
- Montanari GE, Ranalli MG (2009) Multiple and ridge model calibration. In: Proceedings of workshop on calibration and estimation in surveys. Statistics CanadaGoogle Scholar
- Plikusas A (2006) Non-linear calibration. In: Proceedings, workshop on survey sampling. Central Statistical Bureau of Latvia, Ventspils, Latvia, RigaGoogle Scholar
- Valliant R (2019) Comparing alternatives for estimation from nonprobability samples. Technical report. https://doi.org/10.13140/RG.2.2.14618.90566