A Note on Contaminated Samples of Size Three

Abstract

Estimation of the mean a nd s ta nd ard d e viati on us in g th e c lo sest two of three observations in a sample fro m a normal popul ati on with conta min ati on by s lippage of the mean is investi gated by a sampling s tud y. Li ebl e in 's res ult s , whi ch indicated th a t the use of th ese s tatistics is not advisable for noncont a mina ted sa mples, a re born e out b y thi s stud y for cont a min a t.e d sa mples a s we ll. Key Word s : Outli e rs, cont a min a te d samples, robus t estimates, best two of three.