Double sampling is taking a second set of samples in a one-stage survey because the retrospective power of the test did not meet design objectives. At the same time, double sampling causes the Type I error rate to exceed the rate specified for the one-stage survey.
Before the initial round of sampling takes place, Data Quality Objectives (DQOs) should mention any allowances for double sampling and be approved by the appropriate regulators. During the DQO process, double sampling could be considered as an option to setting the Type I error rates.
For more information on the MARSSIM process, visit MARSSIM on EPA.gov