Design and analysis of serial experiments
Lynn Weidman · 1975
Several cases of the following situation- are considered. Each experimental unit receives a prescribed series of treatment applications, and a particular characteristic of interest is measured in association with each application. The experimenter is not only interested in the effect a treatment has on the observation associated with its application, called direct effect, but also in the effects it has on succeeding observations, called residual effects, and its combined effect with following applications on their associa.ted observations, called interaction effects. Comparing the treatment effects to select the most effective overall treatment is the purpose of these experiments. The design problem is to select the series in such a manner that variances for all the effects in a particular class of effect, e.g., direct effects, are as equal as possible and covariances between any two