ABSTRACT
This chapter considers the analysis of experimental data consisting of multiple measurements taken at various locations in an animal. A traditional mixed linear model is proposed and fitted to the data in order to draw statistical inferences about tumor hemodynamics. Assuming normality of the data, frequentist and Bayesian approaches are compared, the former employing the method of residual/restricted maximum likelihood, and the latter a straightforward simulation from the joint posterior density function formed from standard noninformative priors. Although more computationally intensive, the Bayesian approach provides more information and makes inference more readily attainable than traditional frequentist statistical methods.
