Improved penalized mean for estimating the mean concentration of contaminants

Chen and Fernigan proposed a non-parametric, conservative method that involved using a penalized mean to estimate the average concentration of contaminants in soils. The method assumes a random sample obtained from a whole site involved in the US Superfund program. However, in some cases, about 10% of known data are collected from the 'hot spots'. In this paper, two procedures are proposed to use the information from hot spots data or an extreme value to estimate the mean concentration of contaminants. These procedures are evaluated using a data set of chromium concentrations from one of the Environmental Protection Agency's toxic waste sites. The simulation results show that these new procedures are cost-effective.