Combinatorial problems in repeated measurements designs (Q2638697)

In the recent past a number of papers on design of experiments have considered the structure of designs with certain desirable properties. In a repeated measurements designs (RMD) there are t treatments, n experimental units and the experiment lasts for p periods. Each experimental unit receives one treatment during each period. Thus the design may be represented as a \(p\times n\) array containing entries from \(\{\) 1,2,...,t\(\}\). The term ``repeated measurements design'' (RMD) is also used to describe experiments in which at most one treatment is applied to an experimental unit and successive readings are taken over time. Any \(p\times n\) array containing entries from \(\{\) 1,2,...,t\(\}\) can be used as a design. However, some arrays are better than others and the arrays that are `best' depend on the model that is being proposed to analyze the results of the experiment and the terms in that model that are involved in estimating. Here two linear models have been proposed for analyzing results from a repeated measurements experiment; d(k,u) represents the treatment applied in design d to unit u during period k. In this survey paper, many results about the structure of optimal RMDs, over classes of competing designs, two linear models and the construction methods available for these designs are summarized. But the structure of optimal designs when the treatments to be applied have a factorial structure are not considered. Besides giving the construction for these RMDs this survey paper is concluded with a summary of the open problems.