We consider the maximum variance optimality criterion of Elfving (1959
) in the context of (nonlinear) response models, Some practical guidel
ines for the construction of minimax optimal designs are given. In som
e cases this criterion yields one point designs as a consequence of di
fferent scales of the elements in the Fisher information matrix. As an
alternative a ''standardized'' maximum variance criterion is introduc
ed and applied to the calculation of efficient designs. The results ar
e illustrated for binary response models and it is demonstrated that i
n these models standardized minimax optimality should be preferred to
ordinary minimax optimality.