A default conjugate prior for variance components in generalized linear mixed models (comment on article by Browne and Draper)
For a scalar random-effect variance, Browne and Draper (2005) have found that the\nuniform prior works well. It would be valuable to know more about the vector case, in\nwhich a...