ABSTRACT
Developmental models should allow their users to predict the course and form of developmental change, provided the relevant parameters and conditions are known. It is almost trivial that, if all relevant conditions for a particular prediction are alike, the resulting predictions should be similar. It should not be the case, for instance, that a prediction depends on who exactly is making that prediction, or on which moment it is made. To put it differently, a model should entail a deductive procedure that takes a certain number of conditions (parameters, values, etc.) as its input, and produces a specific prediction as its output. The problem with most of our developmental models, however, is that they hardly ever contain such a deductive procedure. The relationship between a model and its predictions is often loose and only intuitively established. Given the same sets of nontrivial conditions, different users will come up with different predictions. Often it is unclear what the relevant inputs to the procedure should be, which steps should be taken to arrive at a prediction, or how precise those predictions should be.
