>
> Funny, I don't see anything about looking to see if perhaps the model is
> indeed Bravo Sierra. I can get a good fit with any regression model by
> simply exhausting the degrees of freedom. That is, if I keep putting in
> enough variables (especially lags of the dependent variable) I can get a
> great fit to the data. However, there is nothing to ensure the
> predicitive power of the model.
>
Ya' know, to a certain degree, you're absolutely right -- you can pretty
much get any regression model to fit any data set. On the other hand,
perhaps we should assume that the scientists and modellers are using a
little bit of Ockham's razor in coming up with the best possible -- though
simplified -- explanation of the data. I don't know what kind of skeptic
you want to be here, but it seems naive to argue that just because there
is "some doubt" in the model, that we ought to go the length to
say that it is "shot through with doubt" and discard it. I think it far
more conservative to presume that the folks running these regressions were
smart enough not to be too lax with the degrees of freedom, and that since
entire boatloads of people seem to think this model a feasible explanation
of the data, that perhaps we ought to heed some of its predictions.
Ben
|