Predicted Means
The predicted means are the expected value for the generalized linear model. When covariates are present in the model, predicted means are computed from the value of those covariates as set in the Covariate values group box. By default, covariates are set to their respective means.
The computation of the predicted means for each effect in the model is constructed by first generating the L matrix for the given effect. The L matrix is generated exactly in the same manner as in the General Linear Models (GLM) module when computing least square means which is also equivalent to how SAS computes the L matrix. The vector of means on the logit scale for a given effect is then computed as
.
For a given level of an effect, the corresponding row vector l from the matrix L is used to compute the predicted mean as given by:
