. . . and you can claim that everything you do is "just a model," and therefore can't be expected to conform to reality.

Having come from one of the soft disciplines myself, Anthropology, I kinda agree that most of what PoliSci and Econ put out is pretty vapid.  ESPECIALLY when they use rational choice models, which Econ seems always to do (thus the "it's only a model" claim) and which is increasingly common in PoliSci.

The hard sciences really are different.  They have math.  You can talk about things in math, and the meaning of the equations always stay the same.  The equation might be wrong, or it might be right, but it means the same thing to everybody.  Furthermore, some things can be proven unequivocally wrong.  That really helps.  

The soft sciences can't communicate much about gender or kinship or discrimination in math, andh thus have to use words to describe it.  The meanings of words and sentences shift from person to person, reader to reader.  Therefore no two people ever read the same article or book, and therefore any discipline which has to communicate through words has to wade through mud to do much of anything because it's so gosh darn hard to agree on terms and meanings.  Furthermore, you can't really prove anything definitively wrong, because the basis for which you judge something to be wrong or right is just as subjective as the words you used to describe the thing under discussion to begin with.

And to the poster who mentioned Heisenberg . . . really.  Until you can do the math necessary to understand what exactly Heisenberg meant with the uncertainty principle (and given that it deals with quantum phenomenon, you simply CANNOT understand any way other than mathematically - it is too inhuman), don't go extrapolating him to the whole world.  It just looks bad.