ABSTRACT
Some nonlinear models can, by taking a suitable transformation, be made linear; the logistic regression model described in Chapter 7 is an example. (The process of transformation is not necessarily straightforward, however, since both the expectation function and the disturbance term are transformed, possibly invalidating the assumption of constant variance and normality required for the usual linear regression approach. Linearization should only be used when the transformed data are adequately described by a model with an additive normal error. See Kolkiewicz, 2005, for more details.) In this chapter, however, we shall concentrate, in the main, on models that cannot be linearized, the so-called
intrinsically nonlinear models.
