ROBUST INFERENCE FOR LOCATION PARAMETERS : ONE- AND TWO-SAMPLE PROBLEMS

ISBN-13: 9783639008524 / Angielski / Miękka / 2009 / 124 str.

Bei Feng

Opis

Estimating measures of location is a fundamentalstatistical problem. The sample mean is not always a good choice to estimate location because it isnot resistant to the influence of outliers. To treatthis problem in a precise manner when nonnormality ispresent, we may use robust location estimates. Inthis work, we present a number of robust locationfunctionals and deter- mine their breakdown point andinfluence function. We study robust locationestimates such as the sample trimmed mean, the sampleWinsorized mean and estimates based on symmetricquantiles. Confidence interval estimation andhypothesis testing are examined from a robustperspective. Both the one-sample case and thetwo-sample case are considered, the latter under twosituations : independence and dependence. A fewpractical examples illustrate the study. Some Rprogrammes are presented in this book.

O autorze

BEI FENG graduated in mathematics (Wuhan Normal Institute,Hankou
Branch, China, 1985),and obtained a Master Science in Statistics
(Laval University, Canada, 2003). She works for
Research unit on children''s psychosocial maladjustment, Laval
University, Canada. As a statistician, she has many research
interests especially in robust statistics.