Disclaimer. Don't rely on these old notes in lieu of reading the literature, but they can jog your memory. As a grad student long ago, my peers and I collaborated to write and exchange summaries of political science research. I posted them to a wiki-style website. "Wikisum" is now dead but archived here. I cannot vouch for these notes' accuracy, nor can I say who wrote them.
Jacobson. 1987. The marginals never vanished: Incumbency and competition in elections to the US House of Representa. AJPS 31:126-141.
Mayhew (1974) identified the "case of the vanishing marginals," spawning a literature on the incumbency advantage: why had it increased? Jacobson argues that the incumbency advantage has not increased.
Note that Jacobson eliminates only the argument that incumbents are more secure than they used to be. He does not solve the new puzzle: Why do incumbents get such a larger vote margin than they used to get (but he does suggest some answers)?
Jacobson notes that members of Congress (MCs) do indeed win by larger margins now than they used to, a change that occurred rapidly in the 1960s (see table 1); however, this increase in margins has not made incumbents safer.
First, incumbents still lose just as frequently as they used to (Table 2).
Instead (and second), the definition of "marginal" has changed. MCs winning 60-65% of the vote in the 1970s were as likely to lose as MCs who won 55-60% in the 1950s (Table 3, Figure 1). Thus, the "marginal range" has expanded.
And third, the "swing ratio" has "declined little, if at all" (p 131). The swing ratio of the "percentage of seats changing hands for a given aggregate shift in the national partisan vote" (p 127); in other words, if the Democrats (nationwide) win 2% more votes (in Congressional races) in 1978 than in 1976, how many seats do they gain in Congress? This ratio has declined slightly, but insignificantly, over time (p 132).
Why is it that incumbents are no more likely to win even though their margins are, on average, much larger?
The "interelection vote swing" has increased. If we predict the Democratic vote share in each (aggregate Congressional) election based on the previous election's share, the standard deviation increases steadily over time (Table 4). In other words, the previous election is less related to the current election than it used to be.
Thus, freshmen MCs are more likely to win (given their margin) today than they used to be (Figure 2) and non-freshman MCs are less likely to win than they used to be (Figure 3). The earlier seniority effects are basically gone.
This might reflect the declining importance of partisanship in voting behavior. "But a less partisan, candidate-oriented electorate is also evidently more fickle" (p 136). "Not only are House incumbents objectively no safer now than they were 30 years ago, they also bear much more personal responsibility for whatever level of safety they do enjoy" (p 138).
More generally, this is how Jacobson sees Congresional elections evolving over time (p 138-9): During the 1960s, House incumbents responsded to increasingly candidate-centered campaigns and new sources of campaign funds to temporarily shore up their incumbency advantage. But within a few electoral cycles, they "faced new institutional players": PACs, polling firms, direct mail firms, "renascent national party campaign committees." These new players were able to fight back, eliminating the short-lived incumbency advantage of the 1960s.
Though Mann (1978) may have slightly overstated his case, then, it does appear that MCs are "unsafe at any margin."
Research by the same authors
Research on similar subjects