A Statistical Model to Predict the Incidence of Pathogenic Protozoa Using Surrogate Variables

A description of bacteriological studies in three geographic areas (Narragansett Bay, Rhode Island; the New York Bight apex; and the Philadelphia-Camden Dumpsite) is presented and the results of some basic statistical analyses evaluated. A method is then described whereby surrogate variables may be used in the SAS "MAXR" program to generate prediction tables and formulae which enable an environmental manager to predict the incidence of pathogenic protozoa in the sediments using these techniques. Variables used in the modeling procedure are divided into tactical and strategic categories based upon cost per sample analysis and models are presented which demonstrate that the less expensive data produce models nearly as valid as those relying upon the more labor intensive methods. Mention is made that sufficient information exists in computerized marine pollution data archives to permit such model building for numerous environmental pollutants.