Optimal sample plans for multiresponse and multisubject experiments

Abstract

[EN] In designed experiments quite often several variables are observed, which complicates the covariance structure of the data, especially when the objective is to observe the evolution of different characteristics along the time. Initial studies have been made assuming that the samples were taken on the same experimental unit, and thus some characteristics were assumed to keep constant for all the process. In the present work, a multisubject experiment is considered, which allows a variety of modifications of the previous paradigm: from considering different characteristics in the individuals, to equal or distinct models for the responses in each subject and the possibility to use several evolution models. Analytical results have been obtained, and an example were different bacteria are observed along the time has been analyzed and discussed.