RE: posthoc step

From: "Bachman, William (MYD)" bachmanw@iconus.com Subject: RE: [NMusers] posthoc step Date: Tue, December 7, 2004 3:25 pm On a much more simplistic level, one can consider the posthoc step in the following manner: 1. the objective function value is a function of individual predictions, observations, fixed and random effect parameters and distributions of the parameters (which we minimize by "optimizing" the parameters). 3. we have the prior parameters and distributions of these parameters (and the objective function value) from the first order analysis. 4. Bayes theorem allows us to "back calculate", the individual parameters from the known individual observations, population parameters and distributions (and objective function value). In other words, its like a system of known and unknowns where we now rearrange the objective function equation and solve for the individual parameters and predictions. No apologies to the statisticians! :) Bill