Building models of nonlinear relationships are inherently more difficu
lt than linear ones. There are more possibilities, many more parameter
s and thus more mistakes can be made. It is suggested that a strategy
be applied when attempting such modelling involving testing for linear
ity, considering just a few model types of parsimonious form and then
performing post-sample evaluation of the resulting models compared to
a linear one. The strategy proposed is a 'simple-to-general' one and t
he application of a heteroskedasticity correction is not recommended.