To incorporate empirical Bayes shrinkage of LFCs, we postulate a zero-centered normal prior for the coefficients βir of model (2) that represent LFCs (i.e., typically, all coefficients except for the intercept βi0): (10)\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document} $$ \beta_{ir} \sim N\!\left(0, {\sigma_{r}^{2}} \right). $$ \end{document}βir∼N0,σr2.
