ISBIS4 Abstract

Contact Author's Name: Eric Smith
Title of Abstract: Improving evaluation of water quality using random effect models and power priors
Author(s): Eric Smith, Keying Ye, Raina Duan, Zhengrong Li
Affiliation: Department of Statistics, Virginia Tech, Blacksburg, VA 24061

Evaluation of water quality is based on selection of a standard then using a decision rule. Standards may be based on magnitude, duration or frequency of potential violation. The decision rule is often based on a test such as a binomial test or a tolerance limit. In many environmental monitoring programs to evaluate standards, the approach is to sample many locations using monthly or quarterly sampling. This often results in small sample sizes for making reliable decisions. An alternative is to make use of other information from prior samples or from neighboring sites. This information may be combined using a model to connect the information from different sites. Two approaches are considered, a random effects model and a Bayesian approach using power priors. The methods are illustrated using data from lakes and streams in the state of Virginia.