ABSTRACT
The greatest body of knowledge about the design of experiments is for linear steady-state relations between a response and treatments (or inputs, influences). Although statistical considerations strongly encourage random experimental order in both time and space, other considerations may require that experiments be run in a particular order. In factorial experiments a sufficient number of levels of each factor are examined to obtain an approximate evaluation of its effects and its interactions with other variables. Although a completely randomized order through the trial sequence might be preferred to minimize the impact of correlation with uncontrolled conditions, one might somewhat override complete randomization to start with the most important trials relative to how a model could be developed. Factorial designs, screening trials, are easy to understand and visualize.
