A ONE-SIDED TRIMMED MEAN ESTIMATOR FOR THE MEAN OF POSITIVELY SKEWED DISTRIBUTIONS |
| |
Authors: | Ling Chen |
| |
Institution: | Dept of Statistics, Florida International University, University Park, Miami, FL 33199, USA ch.enltgrace.liu.edu. |
| |
Abstract: | This article proposes estimating the mean of positively skewed distributions by a one-sided trimmed mean estimator, called the upper-mean, and uses it to assess soil contact exposures at sites with toxic contamination. An optimal upper-mean is found by maximising the probability of the estimator falling in a target range. Monte Carlo studies are conducted for several positively skewed distributions and for a distribution obtained from real data. The simulation results show that the upper-mean is a better estimator than the upper 95% confidence limit estimator currently used, because it is more probable that the estimator is covered by the target range. |
| |
Keywords: | Order statistic over-estimation skewness under-estimation |
|
|