×

Robust inference in two-phase sampling designs with application to unit nonresponse. (English) Zbl 1373.62045

Summary: Influential units occur frequently in surveys, especially in business surveys that collect economic variables whose distributions are highly skewed. A unit is said to be influential when its inclusion or exclusion from the sample has an important impact on the sampling error of estimates. We extend the concept of conditional bias attached to a unit and propose a robust version of the double expansion estimator, which depends on a tuning constant. We determine the tuning constant that minimizes the maximum estimated conditional bias. Our results can be naturally extended to the case of unit nonresponse, the set of respondents often being viewed as a second-phase sample. A robust version of calibration estimators, based on auxiliary information available at both phases, is also constructed.

MSC:

62D05 Sampling theory, sample surveys
62F35 Robustness and adaptive procedures (parametric inference)

Software:

robustbase
PDFBibTeX XMLCite
Full Text: DOI

References:

[1] Beaumont, A unified approach to robust estimation in finite population sampling, Biometrika 100 pp 555– (2013) · Zbl 1382.62002 · doi:10.1093/biomet/ast010
[2] Chambers, Proceedings of the Conference on Statistical Science Honouring the Bicentennial of Stefano Franscini’s Birth pp 125– (1997) · doi:10.1007/978-3-0348-8930-8_10
[3] Chambers , R. L. Kokic , P. Smith , P. Cruddas , M. 2000 Winsorization for identifying and treating outliers in business surveys Proceedings of the Second International Conference on Establishment Surveys, Amer. Statist. Assoc. 717 726 Alexandria, Virginia
[4] Elliott, Little model-based alternatives to trimming survey weights, Journal of Official Statistics 16 pp 191– (2000)
[5] Estevao, Ten cases of auxiliary information for calibration in two-phase sampling, Journal of Official Statistics 18 pp 233– (2002)
[6] Esteavo, Survey estimates by calibration on complex auxiliary information, International Statistical Review 74 pp 127– (2006) · doi:10.1111/j.1751-5823.2006.tb00165.x
[7] Favre-Martinoz, A method of determining the winsorization threshold, with an application to domain estimation, Survey Methodology 41 pp 57– (2015)
[8] Hulliger, Outlier robust Horvitz-Thompson estimators, Survey Methodology 21 pp 79– (1995)
[9] Hulliger , B 1999 Simple and robust estimators for sampling Proceedings of the Section on Survey Research Methods 54 63 American Statistical Association Virginia, Alexandria
[10] Kim, Nonresponse weighting adjustment using estimated response probability, The Canadian Journal of Statistics 35 pp 501– (2007) · Zbl 1143.62008 · doi:10.1002/cjs.5550350403
[11] Kokic, Optimal Winsorizing cutoffs for a stratified finite population estimator, Journal of Official Statistics 10 pp 419– (1994)
[12] Maronna, Robust statistics: Theory and methods (2006) · Zbl 1094.62040 · doi:10.1002/0470010940
[13] Moreno-Rebollo, Influence diagnostics in survey sampling: Conditional bias, Biometrika 86 pp 923– (1999) · Zbl 0949.62010 · doi:10.1093/biomet/86.4.923
[14] Moreno-Rebollo, Influence diagnostics in survey sampling: Estimating the conditional bias, Metrika 55 pp 209– (2002) · Zbl 1320.62024 · doi:10.1007/s001840100142
[15] Rivest, On Searls Winsorized means for skewed populations, Survey Methodology 21 pp 119– (1995)
[16] Särndal, Model assisted survey sampling (1992) · doi:10.1007/978-1-4612-4378-6
[17] Zaslavsky, Downweighting influential clusters in surveys: Application to the 1990 post enumeration survey, Journal of the American Statistical Association 96 pp 858– (2001) · doi:10.1198/016214501753208889
This reference list is based on information provided by the publisher or from digital mathematics libraries. Its items are heuristically matched to zbMATH identifiers and may contain data conversion errors. In some cases that data have been complemented/enhanced by data from zbMATH Open. This attempts to reflect the references listed in the original paper as accurately as possible without claiming completeness or a perfect matching.