The LMIXED Procedure
PROC LMIXED Compared to Other SAS Procedures
The LMIXED procedure provides linear mixed modeling functionality that is most comparable to that of the MIXED, HPMIXED, and HPLMIXED procedures in SAS/STAT software. The GLIMMIX procedure and the NLMIXED procedure can also be used to fit linear mixed models, but these procedures are primarily designed to fit generalized linear mixed models (GLMMs) and nonlinear mixed models, respectively.
The LMIXED procedure, like PROC HPMIXED, can fit linear mixed models by using sparse matrix storage and sparse matrix computations. These sparse matrix techniques are particularly suitable for problems in which the crossproducts matrix is large and sparse. Unlike PROC HPMIXED, the LMIXED procedure does not yet support the CONTRAST, ESTIMATE, LSMEANS, REPEATED, and TEST statements.
The LMIXED procedure, like PROC HPLMIXED, can fit linear mixed models by using specialized efficient algothrims. This is particularly suitable for models that have thousands of levels in a common subject effect. Unlike PROC HPLMIXED, the LMIXED procedure does not yet support the REPEATED statement.
The LMIXED procedure is designed to fit the following large linear mixed models:
models that have thousands of levels in the common subject effect. For information about the common subject effect, see Common Subject Effect.
models that have a large and sparse design matrix.