WEIGHTING INDIVIDUAL DATUM FOR NONPARAMETRIC ANALYSIS 1

Many mining permits require a comparison of a stratified reference area to the reclaimed area for bond release. While this is easily accomplished by performing a test that uses stratified sample estimates of means and variances, such as the parametric t-test, nonparametric tests are often required by regulators if the data are not normally distributed or have small sample sizes. We propose three methods of weighting data that weight individual data points such that they can be used for nonparametric comparisons. These methods allow weights to be generated based on the pre-mining distribution of reference areas, the number of samples taken in the reference areas, and a combination of both of the aforementioned weights. These weights maintain comparability of the reference data to the reclaim data by confining the mean of the weights to one. This method provides a solution for nonparametric analysis where permits require comparison of a stratified reference to the reclaimed area.