Election predictions: Reply

Abstract

Contrary to Aubert's claim, my paper on election predictions does not seek to draw empirical conclusions from mathematical premisses alone. The empirical premiss, approximated by the continuity assumption, is that sufficiently small changes in the predicted vote will cause only small changes in the actual vote. The technical criticisms by Ofsti and Osterberg of the reaction function are answered by specifying the function's domain. Other criticisms are also answered, and the reply concludes by placing the election prediction theorem in the context of other theorizing about human expectations and outguessing phenomena.