Why gain in the Senate but midterm loss in the House? Evidence from a natural experiment

In this note we use the Senate's six-year election cycle to explain why the "iron law of midterm loss" that applies so consistently to the House works with less certainty in the Senate; in fact, since 1946 there have been three instances (1962, 1970, and 1982) where the Senate has experienced no midterm loss. To explain the differing nature of midterm seat change in the Senate, we employ a natural experiment in which Senate midterm elections (1946-1994) are categorized in the following way: (1)The same party controlled the White House two and six years prior to the midterm; or (2) a different party held the presidency six years as compared to two years before the midterm. We hypothesize that, in the first situation, midterm loss forces are mutually reinforcing; thus, the Senate experiences large and unidirectional seat changes against the party that holds the White House. In the later situation, however, the electoral cycle effects (t-2 and t-6) run counter to one another and, therefore, seat change is not unidirectional, midterm loss is lessened, and there is even the potential for midterm gain. In fact, all of the midterm gains in the Senate in the 20(th) century occur in this situation.