In a nonparametric regression model with a multivariate explanatory variabl
e we consider the problem of testing the hypothesis that specific interacti
ons in a canonical decomposition of the model vanish. A simple consistent t
est is developed which is bared on the difference between the corresponding
sum of squares of the ordinary ANOVA for the multi-factor case and a nonpa
rametric variance estimator. This presents an analogue of the classical ANO
VA in the nonparametric regression setup. Asymptotic normality of the intro
duced test statistic is derived under the null hypothesis and under fixed a
lternatives. The finite sample behaviour of the proposed procedures is illu
strated by a small simulation study and a data example. (C) 2001 Academic P
ress.