Inference for a one-way factorial experiment

We develop estimating equations for Factor Level means in a completely randomized one way factorial experiment. This development closely parallels the development for cluster means in two-stage sampling when all clusters are selected at the first stage (see c00ed52.doc). We assume the factor has H levels that correspond to H distinct treatments, and refer to a distinct factor level as a treatment. Each treatment is potentially assigned to each of M=Hm subjects in a population. The experiment consists of randomly assigning m subjects to each treatment (with each subject receiving only one treatment), and then observing a response. Thus, the response can potentially be observed on any of the treatment-subject combinations. The solution given is general, and there are no examples or applications..

The expected value and variance of a superpopulation for a one way factorial experiement

We develop expressions for the expected value and variance in a one way factorial model. These expressions are derived for the superpopulation vector in the context of a one factor study. The factor has H levels that correspond to H distinct treatments. Each treatment is potentially assigned to each of M=Hm subjects in a population. The experiment consists of randomly assigning m subjects to each treatment (with each subject receiving only one treatment), and then observing a response.