Unit level small area estimation for business surveys: comparing transformation-based and robust models

Bocci, Chiara and Smith, Paul A. (2023) Unit level small area estimation for business surveys: comparing transformation-based and robust models. 64th ISI World Statistics Congress, , Ottawa, Canada. 16 - 20 Jul 2023. 6 pp .

Record type: Conference or Workshop Item (Paper)

Abstract

Small area estimation methods are generally based on models which have assumptions of normal errors, but many types of data do not have a normal distribution. Several approaches have been suggested to deal with skewed data, and here we investigate transformations (with and without bias correction) and compare them with previous work with robust models which are less affected by the tails of the distributions. We investigate the properties of these models with a real data set of Italian retail businesses which mimics a structural business survey. Transformation based approaches improve small area estimates, but are not as effective as the best robust approaches. The assessment of which robust approaches are best is qualitatively the same as in previous work, and
corroborates the earlier findings with a different data set.

This record has no associated files available for download.