Much empirical work on Congressional elections implicitly assumes that cand
idates are vote-maximizers; this may be a fair assumption for challengers,
but it is not a good description of incumbent behavior. I present a general
intertemporal utility maximizing model of candidate behavior, which includ
es vote-maximization as a special case. I then demonstrate that these model
s have important consequences for both the design and interpretation of emp
irical work.