The importance of finding appropriate parameterizations for nonlinear statistical models is highlighted. The purpose of this paper is to explore the principles of reparameterization, using an example from real data. It is shown that stable parameterizations allow likelihood-based confidence...