Electoral Security of Members of the U.S. House, 1900–2006

Previous studies have documented that the increase in the incumbency advantage in the 1960s did not decrease the probability of defeat of incumbents in the U.S. House. I define a method for establishing bounds on the probability of incumbent defeat and find that it decreases significantly in the 1950s, before the rise of the incumbency advantage. Incumbency advantage does not have a direct relationship with incumbent defeat rates, raising questions about the use of the incumbency advantage as a means for making inferences about the electoral security of incumbents.