The regression to the mean adjustment can be seen as a limited form of hierarchical/multilevel models with a fixed population mean, so any one score gets shrunk toward the population mean.
(I was reading about them because apparently the pooling eliminates multiple comparison problems, and Gelman is a big fan of them.)
The regression to the mean adjustment can be seen as a limited form of hierarchical/multilevel models with a fixed population mean, so any one score gets shrunk toward the population mean.
(I was reading about them because apparently the pooling eliminates multiple comparison problems, and Gelman is a big fan of them.)