Correct interpretation of experimental and observational data requires not only the specification of the physical situation under observation (that is, some sort of model), but also the realization that measurements are unlikely to conform exactly to that specification. The observed and expected deviations from some specified model are the grist of statistical analysis. One of our strengths is evaluating adequate specifications for physical situations (any such specification is a model; and practically all models are only approximations), taking account of the deviations using statistical analyses, and making correct statements about what can, and cannot, be predicted on the basis of the statistics obtained. Correspondingly, we are expert at peer-reviewing what others have done when attempting this same approach.
We have applied statistical analyses in numerous evaluations of data from contaminated sites, landfills, tests of exhaust stacks from various facilities, and investigations of air quality. We have also pioneered methods for the statistical analysis of data generated in toxicologic and epidemiologic studies.
In a recent court case, we were asked to evaluate the likely time-course of water contamination due to diffusion of perchloroethylene from an asphaltic tank liner in a pumped water system, given a limited set of measurements of perchloroethylene in the water. Here the relevant physics of contaminant diffusion from thin films shows the expected functional form of the time course of diffusion of the perchloroethylene from the asphalt layer. The design of the pumped water system, and the fluctuating demand-load on the water system, led to an unpredictable short-term variation of concentration with time and sampling. The functional form for (long-term) diffusion, combined with a statistical model for the unpredictable short-term fluctuations, allowed consistent and coherent analysis of the available observations, and allowed a convincing presentation of the probable variation of concentration over a period of 15 years.