Adaptive Statistical Methods

Adaptive methods use the data to determine what statistical procedure should be used and then use the data again to carry out the procedure. At first glance, it may appear that it is cheating to use the data to determine what test should be used. However, if the adaptive method is properly constructed an adaptive method can maintain its size and can be a more powerful test.

I have developed two approaches to adaptive testing.

The most recent approach, which uses permuations of residuals to compute the p-value, is developed in the book "Adaptive Tests of Significance Using Permutations of Residuals with R and SAS" by T. W. O'Gorman, which has been published by John Wiley & Sons in 2012. To obtain more information about this approach click here.

For more information about this book go to the Wiley web site.

An earlier approach, which used permutations of independent variables and a different adaptive weighting method, was described in the book "Applied Adaptive Statistical Method," by T. W. O'Gorman, which was published by SIAM in 2004. To obtain more information about this approach click here.

If you are interested in adaptive testing please do not hesitate to contact me at my e-mail address (ogorman@math.niu.edu) .