A Comparison of Shoreline Assessment Study Designs Used for the Exxon Valdez Oil Spill

Shoreline assessment studies may be used after an oil spill to determine whether an injury has occurred and estimate the extent of injuries. Design of these studies generally cannot meet the randomization and replication requirements of classical experimental designs. For example, oil spills cannot be replicated, nor can oiling be applied randomly to experimental units (sites). Furthermore, species counts and other measures of abundance generally do not meet the assumptions of normal theory models. This leaves the statistician and principal investigators the task of finding design and analysis strategies that take into account these departures from standard statistical practices. This paper examines the study design, analysis methods and statistical power (i.e. the ability to detect oil spill effects) of three shoreline assessment programs conducted after the Exxon Valdez oil spill (Exxon's Shoreline Ecology Program, the Exxon Valdez Oil Spill Council's Coastal Habitat Injury Assessment study, and NOAA's Biological Monitoring Survey) and shows that the types of techniques used are influenced by study objectives.