Skip to main content
. 2015 Sep 21;17(2):205–220. doi: 10.1093/biostatistics/kxv038

Fig. 1.

Fig. 1.

True models under homogeneous and heterogeneous structures. Each block contains five covariates, and the omitted blocks do not harbor any important covariates. In the first structure, a covariate needs to be active (or inactive) in all of the studies, whereas in the second structure, a covariate can be partly active among the studies.