Multiple experiments with a single individual

[Eric Hack]

Frederic,

 

I have a question about how to define the levels in the simulation input file to implement a hierarchical population model.  I have data on several individuals, and multiple experiments were performed per individual.  Suppose I have Experiments A through Z.  Then my file might look like the following.

 

Level {

Distributions for population means

Distributions for population variances

Likelihood definition

            Level {

            Distributions at the subject level

                        Experiment {      A

                        Individual 1 experiment

                        }

                        Experiment {      B

                        Individual 2 experiment

                        }

                        …

                        Experiment {      Z

Individual N experiment

}

            }

}

 

MCSim “clones” the subject level distributions, and each experiment would receive a different parameter vector.  Suppose experiments B and C were performed on the same individual.  How do I tell MCSim to use the same parameters for experiments B and C?  Should I enclose them in a third level statement, such as the following?

 

Level {

Distributions for population means

Distributions for population variances

Likelihood definition

            Level {

            Distributions at the subject level

                        Experiment {      A

                        Individual 1 experiment

                        }

                        Level { #Open level for individual in experiments B and C

                                    Experiment {      B

                                    Individual 2 experiment number 1

                                    }

                                    Experiment {      C

                                    Individual 2 experiment number 2

                                    }

                        } #Close level for individual in experiments B and C

                        …

                        Experiment {      Z

Individual N experiment

}

            }

}

 

 

Thank you,

Eric Hack