I think you've misunderstood me. I did not mean emotional motivation in the sense that, "well, he's doing X because it's rational, but he's doing Y because he is having a regression into the time when his mother caught him.. enjoying himself with the sheep."
I mean, emotional motivation everywhere. Perhaps hidden under levels of rationalization, but that's the guy pulling the strings. "He did X because he feels threatened by Y due to his experience at time Z..." and so forth.

To make a substantiative claim: you cannot really model a politician, or any professional or non-professional person without making a psychological analysis of his entire history, including his childhood. Emotional responses are patterned by experience, and without knowing enough of this person's experiences throughout his life, again, including his childhood, you will always be forced to use ad-hoc explainations, wihtout being able to use it for predictions.

"Feels threatened" has many practical uses. If someone feels threatened by someone else, he will act in certain ways. Those ways can be tested for in labs, I would think. That way, maybe you can use small-scale experimentation in the way of making great predictions. (In the sense that you can use the small experiments in order to help you with the modeling of the "cells," and using methods of statistical mechanics to make predictions concerning "bodies." But this is psychohistory again.)

Of course, you can raise any kind of sociological theories without meddling with psychology - but I doubt that they will or do have anything in relation with the actual social structures. It must be remembered, that a society is just a group of individuals. There is only so much you can do with the classical ideal gas model. Eventually, you have to answer for the individual atoms. (Though you may get good approximations - it's just that it's better to know how the more involved picture looks, just so that you know "why" your approximation works.)

By the way, I'm afraid that I can't quite see the point in theories that do not allow prediction. But maybe you can explain it to me.

I don't like being accused of rape, even in a metaphorical manner.

In any case, since you are not even intending to create a model that gives predictions, my whole argument is pretty pointless. The whole discussion becomes one of "which theory is prettier," which is, of course, of its own importance, but which seems to me rather pointless.

If social sciences cannot make predictions, then why are they called sciences, and why do people invest in them so?

I meant "prettier" in a metaphorical sense. Since you can't check those theories, a selection of one over the other is _pure_ aesthetics. Of course, Science has aesthetical considerations when choosing theories, but at least the theories that are not adequate are screened out by experimentation.

How was it found to be wrong? How do you check whether a social theory is at odds with reality?

Well, there you have it - you do use a sort of "experimentation" in social sciences. Why, then, are social theories there not tested under such criteria? For instance, Theory A is falsifiable by a professional survey stating that the number of murders per day has decreased by less than 10%. (It predicts, of course, that the number of murders should decrease by 10% or more, for some reason known only to its originator, and some illuminaries.)

(Ask Konqueror, KDE's web browser.)