Publication:
On the maxbias curve of residual admissible robust regression estimates

Thumbnail Image
Identifiers
URI: http://hdl.handle.net/10016/10349
Files
ws9562.pdf (377.73 KB)
Publication date
1995-12
Defense date
Authors
Advisors
Tutors
Journal Title
Journal ISSN
Volume Title
Publisher
Impact
Google Scholar
Export
Research Projects
Organizational Units
Journal Issue
Abstract
The robustness properties of a regression estimate are throughly described by its maxbias curve. However, this function is difficult to compute, especially when the regressors are not elliptically distributed. In this paper, we propose a general method for computing maxbias curves, valid for a large number of robust regression estimates, namely, those estimates defined by residual admissible functionals. Our results are also useful to compute maxbias curves when the regressors are not elliptically distributed. \Ve provide several examples of application of the method which include S-, T-, and signed R-estimates. MM-estimates are also studied under a related, although slightly different, approach.
Description
Keywords
Robust regression, Bias robustness, Maxbias curve, S-estimates, Tau-estimates, R-estimates, MM-estimates
Bibliographic citation
Collections
DES - Working Papers. Statistics and Econometrics. WS