Summer 2007 | EPA in Arlington, VA

During the summer of 2007, I interned at the Environmental Protection Agency (EPA) in Washington, DC in the Office of Pesticide Programs (Health Effects Division, Chemistry and Exposure Branch). During this time I worked on numerous projects with my main project focused on validating the risk evaluation program CARES. In short, this program estimates the pesticide residue exposure of the population, which is important for setting pesticide use limits. CARES uses a Monte Carlo simulation for each run, computing hypothetical residue exposures for a collection of individuals. For this project I had the opportunity to fly to North Carolina for a two-day trip to receive training in the program, write documents outlining my approaches, and give two presentations to discuss my findings. Adjustments and corrections to CARES are ongoing and the new version of CARES is expected to be released in early 2008.

Other projects I spent time on during my internship included analysis of controlled experiments and using R to compute the parameters in a exponential power model (also referenced as a low-dose shoulder model within EPA). The controlled experiments were generally hierarchical (nested) setups. Although I did . In the computation of parameters of the exponential power model, code originally used by Woody Setzer was adjusted for the given data. This model permits a "low-dose shoulder" (ie, it is like an exponential model but also permits there to be a significant immediate drop or a delayed drop at the lower values, which looks much like a shoulder, in the exponential model). In addition, this powered exponential code was made easier for general use by non-statisticians.

Some photos: