![[logo] US EPA](http://www.epa.gov/epafiles/images/logo_epaseal.gif){kind=logo}
Probability Sampling
description | simple example | MAIA example | how it works
Description: The hallmark of EPA's Environmental Monitoring and Assessment Program (EMAP) is the selection of water bodies based on a probability sampling design. Stream sites within a region are randomly selected for sampling and the results from the random sample can then be applied to the entire region with a known level of uncertainty.
Simple example: You are sampling a stream site and have defined 10 transects across the stream for taking channel measurements. You also want to sample benthic invertebrates but do not want to sample from every transect. To randomly select 1 transect, you might number the transects from 1 to 10, then pick numbers out of a hat or use a random numbers table.
MAIA example: Klemm et al. (in review) used samples from 562 stream sites to evaluate the condition of the invertebrate assemblage in the MAIA region. Because the probability sample set did not necessarily include the extremes of the distribution of sites, reference (least impaired) and test (degraded) sites were also sampled.
Figure
Figure: Probability sampling sites (yellow), reference sites (blue and green) and test (sites) for benthic invertebrates. Probability sites were randomly located within the region.
How the method works: EMAP's probability sampling design is based on a method called 2-stage cluster sampling. This method is used when the sampling population is too large to sample every unit, for example the population of all water bodies in the U.S. For the first stage, a grid of evenly spaced hexagons was draped over a map of the region. Every 16th hexagon was selected and all the streams within the hexagon were enumerated. For the second stage of sampling, a random number of stream sites were selected from within each hexagon (Herlihy et al. [in review]).