¹ Erikson, Robert, ‘Malapportionment, Gerrymandering, and Party Fortunes in Congressional Elections’, American Political Science Review, 66 (1972), 1234–45CrossRefGoogle Scholar; Tufte, Edward, ‘The Relationship between Seats and Votes in Two-party Systems’, American Political Science Review, 67 (1973), 540–7CrossRefGoogle Scholar; Mayhew, David, ‘Congressional Elections: The Case of the Vanishing Marginals’, Polity, 6(1974), 295–317.CrossRefGoogle Scholar

² For detailed reviews see Beth, Richard, ‘“Incumbency Advantage” and Incumbency Resources: Recent Articles’, Congress and the Presidency, 9 (1981–1982), 119–36CrossRefGoogle Scholar and ‘Recent Research on “Incumbency Advantage” in House Elections’, Congress and the Presidency, 11 (1984), 211–24.CrossRefGoogle Scholar

³ ‘Congressional’ elections generally referred to ‘House’ elections until very recently. Although Kostroski showed that post-war Senate developments tracked House developments until the mid-1970s, after that time the two arenas diverged, with Senate elections arguably becoming the most (contd) competitive arena in the American system. House–Senate differences are generally attributed to the greater visibility of Senate elections – fewer and more prestigious, they attract superior challengers who are able to raise adequate campaign funds – and to the greater importance of national issues in Senate campaigns. See Kostroski, Warren, ‘Party and Incumbency in Post-War Senate Elections: Trends, Patterns, and Models’, American Political Science Review, 68 (1973), 1213–34CrossRefGoogle Scholar; and Alford, John and Hibbing, John, ‘The Disparate Electoral Security of House and Senate Incumbents’ (paper presented at the Annual Meeting of the American Political Science Association, Atlanta, 1989).Google Scholar

⁴ Collie, Melissa and Brady, David, ‘The Decline of Party Voting Coalitions in the House of Representatives,’ in Dodd, Lawrence and Oppenheimer, Bruce, Congress Reconsidered, 3rd edn (Washington, DC: CQ Press, 1985), pp. 272–87.Google Scholar

⁵ Edwards, George, Presidential Influence in Congress (San Francisco: Freeman, 1980), pp. 70–8.Google Scholar

⁶ Mann, Thomas, ‘Is the House of Representatives Unresponsive to Political Change?’ in Reichley, A. James, ed., Elections American Style (Washington, DC: Brookings Institution, 1987), pp. 266–8.Google Scholar

⁷ Collie, and Brady, , ‘Decline of Partisan Voting Coalitions’Google Scholar; King, Anthony, ‘The American Polity in the Late 1970s: Building Coalitions in the Sand’, in King, Anthony, ed., The New American Political System (Washington, DC: American Enterprise Institute, 1978), pp. 371–95Google Scholar; Cain, Bruce, Ferejohn, John and Fiorina, Morris, The Personal Vote (Cambridge, Mass.: Harvard University Press, 1987), chaps 8–9.CrossRefGoogle Scholar

⁸ Jacobson, Gary C., ‘The Marginals Never Vanished: Incumbency and Competition in Elections to the US House of Representatives, 1952–82’, American Journal of Political Science, 31 (1987), 126–41.CrossRefGoogle Scholar

⁹ To be precise, the data show a greater than 4 per cent increase in average vote percentages accompanied by a 1 per cent drop in the rate of defeat. Obviously, it is not logically necessary that an increase in average margin produce an equivalent increase in re-election rates, especially when the latter are already at the 93–4 per cent level. But Jacobson is correct in arguing that increased vote margins have real importance only to the extent that they correspond to some decrease in the likelihood of defeat. Ironically, in the elections that have taken place since publication of Jacobson's article, re-election rates have soared into the high 90s.

¹⁰ Mann, Thomas, Unsafe at Any Margin: Interpreting American Elections (Washington, DC: American Enterprise Institute for Public Policy Research, 1978).Google Scholar

¹¹ Jacobson, , ‘The Marginals Never Vanished,’ p. 130.Google Scholar

¹² This counter-trend is evident in Jacobson's Figure 1, and he remarks that ‘House members with very close elections were more vulnerable in the 1950s, but vulnerability diminished sharply as the vote margin increased: the frequency of defeats drops below 5 percent beyond the 57 percent margin’ (p. 130). The significance of this exception does not figure in later arguments, however.

¹³ Even a gross comparison of the 50–60 per cent categories supports this statement. In the 1950s the overall rate of defeat in this category was 13.9 per cent. In the 1970s it dropped to 11.8 per cent.

¹⁴ We emphasize that we are referring to rates of defeat, but as Mayhew documented long ago, there are also changes in the numbers of districts in various categories. There are now fewer districts in the range traditionally identified as marginal, and correspondingly many more in the ranges traditionally categorized as safe. Slightly higher defeat rates in the latter, more numerous, category balance the significantly lower rate of defeat in the former, less numerous, category.

¹⁵ Jacobson, , ‘The Marginals Never Vanished’, p. 133.Google Scholar

¹⁶ One methodological point deserves mention. Jacobson's Table 4 reports the inter-election swing in the Democratic candidate's vote in incumbent-contested districts, whereas our Table 2 contains the swing in the incumbent's vote in those districts. The two series are not identical because where a Republican wins in the first election of a pair, the incumbent swing will be the negative of the Democratic swing. Both series tell largely the same story – increased volatility, but the temporal increase in incumbent swing is smaller (2.8 per cent between the 1950s and 1970s, versus 3.7 per cent for the Democratic swing), and the mid-1960s jump is more pronounced (incumbent swings are larger in years of national tides than are Democratic swings). Given the focus on incumbent re-election rates in this discussion, the inter-election vote swing for incumbents would seem to be the more appropriate measure.

¹⁷ Cover, Albert and Mayhew, David, ‘Congressional Dynamics and the Decline of Competitive Congressional Elections’, in Dodd, Lawrence and Oppenheimer, Bruce, eds, Congress Reconsidered (New York: Praeger, 1977).Google Scholar

¹⁸ Tables 1–3 calculated separately for freshmen and non-freshmen incumbents are available on request.

¹⁹ Cain, , Ferejohn, and Fiorina, , Personal Vote, chap. 3.Google Scholar

²⁰ Mayhew, , ‘Case of the Vanishing Marginals’, p. 295.Google Scholar

²¹ Fiorina, Morris, Congress – Keystone of the Washington Establishment (New Haven, Conn.: Yale, 1977), pp. 12–13.Google Scholar

²² Mann, , ‘Is the House of Representatives Unresponsive?’, pp. 266–8.Google Scholar

²³ Jacobson remarks that ‘The pattern for the 1950s indicates that new members were often swept into office by a partisan tide, then swept out again once it had receded’ (‘The Marginals Never Vanished’, p. 136).

²⁴ Mann offers a scenario for a Republican capture of the House, but he offers it ‘not as a likely occurrence, but as a plausible one’ (see ‘Is the House Unresponsive to Political Change?’ p. 278).

²⁵ Jacobson, , ‘The Marginals Never Vanished’, p. 128.Google Scholar

²⁶ Niemi, Richard and Fett, Patrick, ‘The Swing Ratio: An Explanation and Assessment’, Legislative Studies Quarterly, 11 (1986), 75–90CrossRefGoogle Scholar. The Butler method involves tracing out a hypothetical seats-votes curve by calculating how many actual outcomes would have been changed by uniform positive and negative swings of given magnitudes, often the twenty numbers between – 10 per cent and + 10 per cent inclusive.

²⁷ Brady, David and Grofman, Bernard, ‘Sectional Differences in Partisan Bias and Electoral Responsiveness in US House Elections 1850–1980’, British Journal of Political Science, 21 (1991), 247–56.CrossRefGoogle Scholar

²⁸ Gudgin, Graham and Taylor, Peter, Seats, Votes and the Spatial Organization of Elections (London: Pion Press, 1979), pp. 56–9, passim.Google Scholar

²⁹ The theoretical concept and statistical estimation of the swing ratio are treated at length in our paper, ‘And Still More on the Concept of Swing Ratios’ (Research Paper No. 990, Graduate School of Business, Stanford University, 1991).

³⁰ This extension of the time series is less innocent than it might appear at first glance, since there were relatively large seat swings in the late 1940s elections, and relatively small ones in the elections of the late 1980s. One of the referees has observed that if Jacobson had examined the longer period his argument might have been different. Nevertheless, even if we estimate swing ratios over the shorter period, all the differences reported below still appear, except with lower significance levels: the different measure of party vote is more consequential than extension of the data series.

³¹ Erikson, ‘Malapportionment’; Fiorina, Congress; Born, Richard, ‘Generational Replacement and the Growth of Incumbent Reelection Margins in the US House’, American Political Science Review, 73 (1979), 811–17CrossRefGoogle Scholar; Alford, John and Brady, David, ‘Personal and Partisan Advantage in US House Elections, 1846–1986’, in Dodd, Lawrence and Oppenheimer, Bruce, eds, Congress Reconsidered, 4th edn (Washington, DC: CQ Press, 1989), pp. 153–69.Google Scholar

³² Separate coefficients for the two periods can be estimated within a single equation by using interaction terms. Define a dummy variable, D, that equals 0 between 1946 and 1964, and 1 between 1966 and 1990. Then estimate the following equation:

Dem Seat% = a₁ + a₂ D + b₁ (Dem vote%) b₂ D (Dem vote%)

The 1946–64 prediction is (a₁ + b₁ (Dem vote%)) and the 1966–90 prediction is ((a₁ + a₂) + (b₁ + b₂) (Dem vote%)). One can utilize a t-test to determine whether the swing ratio (b₁, b₁ + b₂) differs between the two periods, and more generally an F-test will indicate whether the dual restrictions that a₂ = 0 and b₂ = 0 (the two periods are identical) result in a significant loss in explanatory power.

³³ King, Gary and Gelman, Andrew, ‘Systemic Consequences of Incumbency Advantage in Congressional Elections’, American Journal of Political Science, 34 (1990), 1142–64.Google Scholar

³⁴ Jacobson, , The Electoral Origins of Divided Government (Boulder, Colo.: Westview Press, 1990), pp. 82–93.Google Scholar

³⁵ Shepsle, Kenneth A., Models of Multiparty Electoral Competition (Chur, Switzerland: Harwood Academic, 1991), chaps. 4–7.Google Scholar

³⁶ Gelman and King adopt such a strategy in seeking to measure the incumbency advantage. See Gelman, Andrew and King, Gary, ‘Estimating Incumbency Advantage without Bias’, American Journal of Political Science, 34 (1990), 1142–64.CrossRefGoogle Scholar

³⁷ These estimates are based on Table 6. If the regressions in Table 5 are used, Republican House majorities are also predicted in 1980 and 1990. The regressions in Table 6 have superior fits, so we utilize them here.

³⁸ These tests are one-tailed tests of the hypothesis that the swing ratio is lower after 1964 than before. For the shorter time series utilized by Jacobson, the significance levels are 0.10 for pre-1964 against post-1964 incumbents, 0.10 for pre-1964 against post-1964 open seats and 0.01 for post-1964 incumbents against post-1964 open seats. The difference between pre-1964 incumbents and open seats remains non-significant.

³⁹ After some recent elections partisan commentators have complained that redistricting and unequal turnout have deprived Republican candidates of their fair share of House seats. Possibly this is true, but there is a prior problem that is usually overlooked; namely, that Republicans do not get enough votes. They have not beaten the Democrats in the aggregate vote since 1952.

⁴⁰ In a personal communication Thomas Mann has reminded us that one has to go hunting where the ducks are. Open seats constitute such a small proportion of all races that it is hard to see how a majority can be built by winning them. We agree, but we also cannot see how under present conditions a majority strategy can be based on defeating incumbents. The bottom line would seem to be that the Republicans have a long row to hoe.

⁴¹ Collie, Melissa, ‘Incumbency, Electoral Safety, and Turnover in the House of Representatives, 1952–1976’, American Political Science Review, 75 (1981), 119–31.CrossRefGoogle Scholar

⁴² Weisberg, R., ‘Have House Elections Become a Meaningless Ritual?’ in Luttbeg, Norman, ed., Public Opinion and Public Policy, 3rd edn (Itasca, III.: F. E. Peacock, 1981).Google Scholar