You are here:
A CONDITIONAL PROBABILITY APPROACH FOR ANALYZING SURVEY DATA TO ESTIMATE PROBABILITY OF IMPAIRMENT
Citation:
Paul, J F., G Pesch, W B. Galloway, AND C J. Strobel. A CONDITIONAL PROBABILITY APPROACH FOR ANALYZING SURVEY DATA TO ESTIMATE PROBABILITY OF IMPAIRMENT. Presented at New England Association of Environmental Biologists, Newport, RI, March 13-15, 2002.
Description:
A question that arises is how can survey data, collected with a random design, provide an initial screening for identifying unsampled areas that are likely to have biological impairment? A random sampling design provides estimates of relative fraction of the population of interest that are above or below a given value of an indicator, for example, fraction of first order stream miles in a state that have a fish IBI (index of biotic integrity) less than 3 (on a 1 to 5 scale). The estimate of fraction of the population is the probability of occurrence (e.g., the probability of observing impaired first order stream miles). Physical and chemical data are typically collected in conjunction with biological data, plus extensive data sources exist for land cover in catchments of the sampling sites. These data allow one to calculate conditional probabilities for impairment; e.g., probability of impairment given that greater than ten percent of the catchment area is urban land. These conditional probabilities provide a screening tool for identifying areas most likely or least likely to be impaired. Data from the Maryland Biological Stream Survey are used to illustrate the approach.