Abstract: The problem of constructing trend-resistant factorial designs is discussed. Suppose a factorial experiment is to be run in a time sequence with one observation taken at a time. Then the experimenter has to decide in which order to observe the treatment combinations. A common practice is to randomize. However, sometimes randomization may lead to an undesirable ordering, and a systematic run order may be preferred. Attention is focused on the construction of systematic run orders of factorial designs in which the estimates of important factorial effects are orthogonal to some polynomial trends. Some recent work on this subject is unified and extended.
Key words and phrases: Design key, generalized foldover method, polynomial trend, run order.