A method of constrained randomization for pâ‚nâ‚x pâ‚‚nâ‚‚ x... p/k n/k factorials
Richard L. Schwaller · Montana State University ScholarWorks (Montana State University) · 1971
This paper presents a method of constrained randomization for (Formula not captured by OCR) factorial experiments in blocked designs when the requirement (imposed by the nature of the process under investigation) restricts the number of factor levels which may be changed from treatment combination to adjacent treatment combination.The method of constrained randomization requires the construction of a sequence of treatment combinations in which only one level of one factor is changed for each adjacent pair of treatment combinations and also for the first and last treatment combination, considering them as an adjacent pair.The existence and construction of such sequences is considered in Chapter II.Chapter III shows how these sequences are used to generate random sequences having the same properties, This method of constrained randomization provides a basis for statistical analysis utilizing the randomization model.In Chapter IV, unbiased estimates for factorial effects are obtained using the method of constrained randomization.The average variance of an estimated effect or interaction is found and analysis of variance tables for both the randomization model and the infinite model are given.