Optimal design for linear regression models in the presence of heteroscedasticity caused by random coefficients

摘要

Random coefficients may result in heteroscedasticity of observations. For particular situations, where only one observation is available per individual, we derive optimal designs based on the geometry of the design locus.