by Schwabe, R., Graßhoff, U., Großmann, H., Holling, H..
Series: 2005-06, Preprints
In this paper we show that product type designs
are optimal in partially heteroscedastic
multi-factor linear models.
This result is applied to obtain locally $D$-optimal
designs in multi-factor generalized linear models
by means of a canonical transformation.
As a consequence we can construct optimal designs
for logistic response in paired comparison experiments
which are close to the concept of utility balance.
optimal design, paired comparison, Bradley-Terry model, logistic model,