The statistical routines in Blossom include the multi-response permutation procedure family of statistics developed by Dr. Paul Mielke, Jr.; the regression quantile estimates for linear models developed by Dr. Roger Koenker; and permutation procedures for linear models developed by Brian Cade and colleagues. These procedures have the common feature of considering quantiles of data distributions as the basic unit of analysis and use permutation theory for inferences via hypothesis tests or confidence intervals. The use of permutation theory for inferences requires fewer assumptions about the data and allows the use of test statistics that have an unknown or intractable sampling distribution. The test statistics and estimates based on quantiles (or distances between quantiles) allow a variety of questions about ecological phenomena to be better answered than would be possible by focusing exclusively on means (expected values) and variances, as is done in most conventional statistical procedures.
U.S. Department of the Interior | U.S. Geological Survey
URL: http://www.fort.usgs.gov/resources/spotlight/blossom/blossom_statprocedures.asp
Page Contact Information: AskFORT@usgs.gov
Page Last Modified: 11:34:02 AM
