The NLMIXED Procedure
-
Overview
-
Getting Started
-
Syntax
-
DetailsModeling Assumptions and NotationIntegral ApproximationsBuilt-in Log-Likelihood FunctionsHierarchical Model SpecificationOptimization AlgorithmsFinite-Difference Approximations of DerivativesHessian ScalingActive Set MethodsLine-Search MethodsRestricting the Step LengthComputational ProblemsCovariance MatrixPredictionComputational ResourcesDisplayed OutputODS Table Names
-
Examples
- References
The ESTIMATE statement enables you to compute an additional estimate that is a function of the parameter values. You must provide a quoted string to identify the estimate and then a valid SAS expression. Multiple ESTIMATE statements are permitted, and results from all statements are listed in a common table. PROC NLMIXED computes approximate standard errors for the estimates using the delta method (Billingsley, 1986). It uses these standard errors to compute corresponding t statistics, p-values, and confidence limits.
The ECOV option in the PROC NLMIXED statement produces a table containing the approximate covariance matrix of all the additional estimates you specify. The ECORR option produces the corresponding correlation matrix. The EDER option produces a table of the derivatives of the additional estimates with respect to each of the model parameters.
The following options are available in the ESTIMATE statement:
-
ALPHA=
-
specifies the alpha level to be used in computing confidence limits. The default value corresponds to the ALPHA= option in the PROC NLMIXED statement.
- DF=d
-
specifies the degrees of freedom to be used in computing p-values and confidence limits. The default value corresponds to the DF= option in the PROC NLMIXED statement.