A Nonlinear Dynamic Model Applied to Data with Two Times of Measurement

Abstract

A good model of developmental phenomena should not only “explain” the data in the sense that the model is not falsified by the data but it should also be built on the basic theoretical assumptions about the process under study. This is often not the case when standard statistical models are applied to developmental data. An alternative is then to apply nonlinear dynamical system modeling. This approach is followed in the article and it is illustrated by a provisional model of the development of boys’ problem behavior between the ages 10 and 13. The underlying interactionistic theory of development was mirrored in a nonlinear model that predicted outcomes marginally better than a standard regression model. It is pointed out in the article that it might be possible to use nonlinear dynamic system modeling also in contexts where data are available only from a few measurement occasions.