Experiments for Non-Linear Functions

The paper reviews the work that has been done on the planning of experiments with response functions non-linear in at least one of the parameters. Apart from older work on the design of dilution series experiments and of quantal bioassays, this field is relatively recent, the mathematical and computing aspects being more complex than for linear responses. In particular, an efficient design requires good advance estimates of the unknown parameters. Research has concentrated on i optimum estimation of the parameters, under an asymptotic criterion, for both sequential and non-sequential approaches, ii tests of the adequacy of the model, discrimination between models and model-building.