Gary Jacobson (1987a, 1987b) has suggested that increased vote proportions for incumbent House candidates during the 1960s and 1970s have not resulted in greater electoral safety, primarily because of greater observed volatility in interelection vote swings occurring during the same era. Using data on House elections from 1824 to 1978, we find that both "marginal" and "safe" incumbents are much more vulnerable to electoral defeat during the 19th century than during the 20th century. This pattern seems to be explained by greater homogeneity of interelection vote swings during the 20th century. Overall, our results lend strong support to Jacobson's assertion that the meaning of electoral marginality varies substantially over time.

It is a rare thing to receive excellent and truly constructive criticisms of one's research. Gary Jacobson's response to our paper (Garand, Wink, and Vincent 1993) certainly falls into that category. His careful analysis of our arguments provides an excellent statement of one viewpoint as to how the relationship between incumbent vote margins and subsequent incumbent victory should be studied.Professor Jacobson's primary concern with our work pertains to the issue of how uncontested seats are treated in our analysis. In his previous research, Jacobson (1987) utilized only contested seats to explore how incumbent vote margins in one election were translated into patterns of incumbent victory in subsequent elections. Our research is somewhat of a departure from that approach, insofar as we include all incumbents (including those running in uncontested seats) in our analysis. Jacobson suggests that this strategy is problematic for several reasons.

No abstract