To a substantial extent, political scientists continue to be in a state of shock, or even denial, over the results of the November 1994 mid-term elections. This degree of surprise has been largely attributed to the failure of the 1994 elections to conform to numerous models that have successfully predicted past election results on the basis of some combination of presidential popularity and economic performance. In a broader sense, it has become fashionable to argue alternately that the voters weren't really saying anything, or that they did not know what they were saying, or that they were saying things not fit for polite company—angry, mean, and even racist things. Such an interpretation does a disservice both to our profession and to our friends and neighbors, the real people who cast those votes.

It will be some time before we are able to untangle the factors that led to the Republican sweep and to see whether it was a short-term aberration or the beginning of a long-term trend in congressional elections. Nevertheless, I would like to offer a two-part explanation for the results and the consequent academic confusion: the election models that we use as both a tool and a crutch are at the same time not precise enough and too precise.

They are not precise enough because they do not capture all that needs to be taken into account even on the dimensions of the economy and presidential popularity.