Single-index models generalize linear regression. They have applicatio
ns to a variety of fields, such as discrete choice analysis in econome
trics and dose response models in biometrics, where high-dimensional r
egression models are often employed. Single-index models are similar t
o the first step of projection pursuit regression, a dimension-reducti
on method. In both cases the orientation vector can be estimated root-
n consistently, even if the unknown univariate function (or nonparamet
ric link function) is assumed to come from a large smoothness class. H
owever, as we show in the present paper, the similarities end there. I
n particular, the amount of smoothing necessary for root-n consistent
orientation estimation is very different in the two cases. We suggest
a simple, empirical rule for selecting the bandwidth appropriate to si
ngle-index models. This rule is studied in a small simulation study an
d an application in binary response models.