|You are working with the text-only light edition of "H.Lohninger: Teach/Me Data Analysis, Springer-Verlag, Berlin-New York-Tokyo, 1999. ISBN 3-540-14743-8". Click here for further information.|
|See also: robust regression|
A statistical procedure is considered to be robust if it performs well even when required assumptions are not met, or if the procedure performs well for a large number of distributions. A "statistical procedure" could be any item, from an estimate to a statistical test, or from a modeling technique to cluster analysis. Robustness is a big issue in applied data analysis, since practical problems tend to create outliers.
A simple example of a robust statistic is the median
in comparison with the mean. The median is less affected by outliers than
Last Update: 2004-Jul-03