The LMIXED Procedure

Syntax: LMIXED Procedure

The following statements are available in the LMIXED procedure:

PROC LMIXED <options>;
CLASS variable <(options )> <variable <(options)>> </ global-options>;
BY variables;
DISPLAY <table-list> </ options>;
DISPLAYOUT table-spec-list </ options>;
EFFECT name=effect-type (variables </ options>);
MODEL dependent = <fixed-effects> </ options>;
RANDOM random-effects </ options>;
PARMS <(value-list) …> </ options>;
WEIGHT variable;
OUTPUT OUT=CAS-libref.data-table<COPYVARS=(variables)><keyword <=name>>…<keyword <=name>>;
BLUP OUT=CAS-libref.data-table <options>;
OPTIMIZATION <options>;

Items within angle brackets ( < > ) are optional.

The PROC LMIXED and MODEL statements are required, and the MODEL statement must appear after the CLASS statement if a CLASS statement is included. The RANDOM statement must follow the MODEL statement. The RANDOM statement can appear multiple times. Other statements can appear only once.

The BY, MODEL, OUTPUT, RANDOM, PARMS, BLUP, OPTIMIZATION, and WEIGHT statements are described in full after the PROC LMIXED statement in alphabetical order. The CLASS and EFFECT statements are described by summarizing their functionality and syntax; for more information about these statements, see Chapter 3: Shared Concepts.

Table 10.1 summarizes the basic functions and important options and statements that are supported by the LMIXED procedure.

Table 10.1: Summary of PROC LMIXED Statements

Statement	Description	Important Options
PROC LMIXED	Invokes the procedure	The DATA= option specifies the input data table; the METHOD= option specifies the estimation method.
CLASS	Declares qualitative variables that create indicator variables in and matrices	None
MODEL	Specifies the dependent variable and the fixed effects, setting up	The S option requests a solution for fixed-effects parameters.
RANDOM	Specifies random effects, setting up and	The SUBJECT= option creates block-diagonality; the TYPE= option specifies the covariance structure; the S option requests a solution for the random effects.
OUTPUT	Creates a data table that contains observationwise statistics	The ALLSTATS option requests that all statistics be computed.
PARMS	Specifies a grid of initial values for the covariance parameters	The HOLD= and NOITER options hold the covariance parameters or their ratios constant; the PARMSDATA= option reads the initial values from a data table.
OPTIMIZATION	Controls aspects of the nonlinear optimizations	The TECHNIQUE= option specifies the optimization technique.
BLUP	Creates a data table that contains the estimation of fixed and random effects	The OUT= option specifies the output data table; the SOLVER= option specifies the mixed model equations solver.

Last updated: December 21, 2018