Measure of location-based sestimators in simple linear regression
2013 (English)Report (Other academic)
In this paper we consider certain measure of location-based estimators (MLBEs)for the slope parameter in a linear regression model with a single stochastic regressor. Themedian-unbiased MLBEs are interesting as they can be robust to heavy-tailed samples and,hence, preferable to the ordinary least squares estimator (LSE). Two dierent cases are consideredas we investigate the statistical properties of the MLBEs. In the rst case, the regressorand error are assumed to follow a symmetric stable distribution. In the second, other typesof regressions, with potentially contaminated errors, are considered. For both cases the consistencyand exact nite-sample distributions of the MLBEs are established. Some results for thecorresponding limiting distributions are also provided. In addition, we illustrate how our resultscan be extended to include certain heteroscedastic regressions. Finite-sample properties of theMLBEs in comparison to the LSE are investigated in a simulation study.
Place, publisher, year, edition, pages
Department of Statistics, Uppsala University , 2013. , 18 p.
Working paper / Department of Statistics, Uppsala University, 2013-2
IdentifiersURN: urn:nbn:se:uu:diva-199580OAI: oai:DiVA.org:uu-199580DiVA: diva2:620148