Data Structures and Algorithms for an Open System to Design and Analyse Generally Balanced Designs
Roger William Payne, Michael F. Franklin · COMPSTAT · 1994
The design of experiments is now an important facility in many statistical programs and packages. Algorithms are available for constructing partial replicates and designs containing effects confounded with blocks. Alternatively, programs may offer a repertoire of pre-selected designs. Less attention, however, seems to be given to the question of how these confounded designs will eventually be analysed, and to ways of avoiding the constraints on choice of design that arise from menu-based systems. In this paper we focus on the important class of generally balanced designs, and describe the information and associated data structures required to form any particular design, and to specify the analysis. We also discuss the tools and algorithms available for deriving these details, and explain how they can be put together to provide an open system in which users can investigate designs and then add them to an available repertoire.