¹ At least this is the case in the San Francisco Bay area. See Prewitt, Kenneth, “Political Ambitions, Volunteerism, and Electoral Accountability,” American Political Science Review, 64 (March, 1970), 5–17, at p. 9 CrossRefGoogle Scholar.

² For example, in Connecticut during the period 1946–1958, 982 out of 1201 incumbents won re-election to the state legislature, for a success rate of 81.8 per cent ( Barber, James David, The Lawmakers [New Haven: Yale University Press, 1965], on p. 8 Google Scholar); and David Leuthold reports an incumbent success rate of 73 per cent in Missouri state races during the period 1960–1964 ( Electioneering in a Democracy [New York: Wiley, 1968], p. 127 Google Scholar). See also Hyneman, Charles S., “Tenure and Turnover of Legislative Personnel,” Annals of the American Academy of Political and Social Science, 195 (1938), 21–31 CrossRefGoogle Scholar, for data covering ten state legislatures over the decade 1925–1935. Finally, Wisconsin data collected by three of my students, Gary Londergan, Patrick Maguire, and Timothy Smock, show an even higher state-level success rate than those found elsewhere. Of the 50 state senators running for re-election in the period 1958–64, 46, or 92.0 per cent, won. Of the 315 state assemblymen running for re-election in the period 1956–62, 271, or 86.0 per cent, won.

³ Computed from figures supplied in Jones, Charles O., Every Second Year (Washington, D.C.: The Brookings Institution, 1967), p. 68 Google Scholar.

⁴ Computed from figures supplied in Turrett, J. Stephen, “The Vulnerability of American Governors, 1900–1969,” Midwest Journal of Political Science, 15 (February, 1971), 108–32, at p. 118CrossRefGoogle Scholar.

⁵ Cummings, Milton C. Jr., Congressmen and the Electorate (New York: The Free Press, 1966), chapter 3Google Scholar.

⁶ A fair inference, it would seem, from the data presented in Cummings, pp. 78–79.

⁷ Computed from figures supplied in Cummings, p. 60.

⁸ Jones, Charles O., “The Role of the Campaign in Congressional Politics,” in The Electoral Process, ed. Jennings, M. Kent and Zeigler, L. Harmon (Englewood Cliffs, N.J.: Prentice-Hall, Inc., 1966), pp. 21–41 Google Scholar.

⁹ Erikson, Robert S., “The Advantage of Incumbency in Congressional Elections,” Polity (Spring, 1971), 395–405 Google Scholar.

¹⁰ Erikson, Robert S., “Malapportionment, Gerrymandering, and Party Fortunes in Congressional Elections,” American Political Science Review, 66 (December, 1972,) 1234–45, at pp. 1240 ff.CrossRefGoogle Scholar

¹¹ Matthews, Donald R., U. S. Senators and Their World (New York: Vintage Books, n.d. First published 1960). See especially pp. 239–42Google Scholar.

¹² Hinckley, Barbara, “Incumbency and the Presidential Vote in Senate Elections,” American Political Science Review, 64 (September, 1970), 836–42CrossRefGoogle Scholar.

¹³ Hinckley, p. 838.

¹⁴ Weber, Ronald E. and Munger, Frank, “Party Identification and the Classification of State Party Systems,” paper delivered at the 1968 Annual Meeting of the American Political Science Association, Washington, D.C Google Scholar. Their basic approach is to use what is known about the relationship between party identification and the demographic and socioeconomic characteristics of voters to construct a number of “voter-types.” Determining the proportion of a state's electorate fitting each such “type” permits an estimate of each party's identifiers in that state. The approach is similar to that employed in the Simulatics project reported by Pool, Ithiel de Sola, Abelson, Robert P., and Popkin, Samuel in Candidates, Issues, and Strategies, rev. ed. (Cambridge: MIT Press, 1965)Google Scholar.

¹⁵ Arguments and evidence for this general position can be found, among other places, in Wonnacott, Ronald J. and Wonnacott, Thomas H., Econometrics (New York: John Wiley & Sons, Inc., 1970), especially pp. 124–25Google Scholar; Tufte, Edward R., “Improving Data Analysis in Political Science,” in The Quantitative Analysis of Social Problems, ed. Tufte, Edward R. (Reading, Mass.: Addison-Wesley, 1970)Google Scholar; and, for the coup de grâce to the goodness-of-fit criterion in theory construction, Feller, William, An Introduction to Probability Theory and Its Applications, Volume II (New York: John Wiley & Sons, Inc., 1966), p. 52 Google Scholar. The relevant portions of Feller's point are summarized and quoted in Przeworski, Adam and Soares, Glaucio A. D., “Theories in Search of a Curve,” American Political Science Review, 65 (March, 1971), 51–68 CrossRefGoogle Scholar, in fn. 34 on p. 67.

¹⁶ Miller, Clem, Member of the House (New York: Charles Scribner's Sons, 1962), quoted at p. 93 Google Scholar.

¹⁷ Incumbent is defined here, and throughout the study, as a senator who has achieved the office through election, even a special election. Rates of running and winning always refer only to general election contests. A person holding the office at the time of the general election by virtue of appointment was not considered an incumbent. The reason for this distinction is presented in the final section.

Although I do not deal with primary contests here, I might note for the record that only about 7 per cent of the postwar senators who sought renomination were denied it, a higher figure than the House's 1.6 per cent (from data in footnote 3), to be sure, but still small enough to pose no real problem to the present analysis.

¹⁸ Computed from figures in Jones, , Every Second Year, p. 68 Google Scholar.

¹⁹ Jones, p. 68.

²⁰ Clapp, Charles, The Congressman (Garden City, N.Y.: Anchor Books, 1963), p. 374 Google Scholar.

²¹ Jones, , Every Second Year, p. 68 Google Scholar.

²² Jones, p. 68. Note, however, that this trend refers to the safety of districts for the party. Although we assume the general pattern is similar for the safety of incumbents, an empirical demonstration of that assumption's validity is clearly beyond the scope of the present paper.

²³ Naturally, trend lines such as these are highly sensitive to extreme values at the end points of the time series. For example, adding 1946 to the present pair of series brings the two trend lines a good deal closer together due to the high success (100 per cent) of Republican incumbents and the low success (66.7 per cent) of Democrats in that year. The Democratic trend is then described by the equation, Ŷ = 1.12 T + 81.1, while the Republican estimate becomes Ŷ = 1.54 T + 69.8, where T = 0 in 1946 in both instances. And yet despite such extreme readings, the difference between the parties does still remain, and still points to the same question: why the different rate of change for each party over time?

²⁴ These within-state patterns are based on data from 1946–1970.

²⁵ Maryland retained in office only two of the six senators seeking reelection. The three states returning an even half of their incumbents are Kentucky, Pennsylvania, and Wyoming.

²⁶ Ranney, Austin, “Parties in State Politics,” in Politics and the American States, ed. Jacob, Herbert and Vines, Kenneth N. (Boston: Little, Brown and Company, 1965), pp. 61–99 Google Scholar; see especially pp. 63–70.

²⁷ Campbell, Angus, Converse, Philip, Miller, Warren, and Stokes, Donald, The American Voter (New York: John Wiley and Sons, 1960), p. 142 Google Scholar.

²⁸ The following figures are taken from Campbell, Angus and Miller, Warren E., “The Motivational Basis of Straight and Split Ticket Voting,” American Political Science Review, 51 (June, 1957), 293–312 CrossRefGoogle Scholar.

²⁹ Campbell and Miller; see their Table VI, at p. 306.

³⁰ A brief review of the electoral advantages of congressional incumbency can be found in Bailey, Stephen K., Congress in the Seventies (New York: St. Martin's Press, 1970), pp. 9–11 Google Scholar. See also Robert Erikson, “The Advantage of Incumbency in Congressional Elections.”

³¹ Stokes, Donald E. and Miller, Warren E., “Party Government and the Saliency of Congress,” chapter 11 in Elections and the Political Order, by Campbell, Angus, Converse, Philip, Miller, Warren, and Stokes, Donald (New York: John Wiley and Sons, 1966), pp. 194–211, at p. 205 Google Scholar.

³² Stokes, and Miller, , in Campbell, et al., Elections and the Political Order, p. 205 Google Scholar.

³³ See Leuthold, Electioneering in a Democracy, for an admirable discussion of congressional campaigns largely in terms of resource acquisition. His observations and findings undoubtedly apply, mutatis mutandis, to senatorial contests as well.

³⁴ Such an inference can easily be drawn from Leuthold, Table 1, p. 67, at least insofar as labor unions are concerned. In addition, chapter 6 deals with the whole range of fund-raising questions. A comment regarding incumbents is telling: “Nonincumbents may have needed more money but they found it harder to raise, and ended up with less money than incumbents” (p. 77).

³⁵ The close relationship between national and state electoral results is documented in Bean, Louis, How to Predict Elections (New York: Alfred A. Knopf, 1948)Google Scholar and in Key, V. O. Jr., American State Politics (New York: Alfred A. Knopf, 1956), especially in chapter 2Google Scholar.

³⁶ Miller, Warren E., “One Party Politics and the Voter,” American Political Science Review, 50 (September, 1956), 707–25CrossRefGoogle Scholar.

³⁷ For a stimulating analysis of left voting which emphasizes the importance of the form of the relationships among phenomena, as well as a more detailed development of the “contextual” argument, see Przeworski and Soares, “Theories in Search of a Curve.” On the problem of forms of relationships as it relates to an American electoral situation, see Burnham, Walter Dean and Sprague, John D., “Additive and Multiplicative Models of the Voting Universe: The Case of Pennsylvania, 1960–68,” American Political Science Review, 64 (June, 1970), 471–90CrossRefGoogle Scholar.

³⁸ The two alternative forms mentioned here, as well as several others, are developed further and the parameters estimated in my paper, “Form and Substance in Theory Construction: Electoral Behavior in Postwar Senate Elections,” presented at the Annual Meeting of the American Political Science Association, Washington, D.C., September 5–9, 1972 Google Scholar. In addition, the contextual argument is elaborated in that paper several steps beyond the sketch given here.

³⁹ The source relied upon for all election data was the Statistical Abstract of the United States (Washington, D.C.: Government Printing Office, various dates). Errors noted were corrected (e.g., on two occasions special Senate elections were listed in the general election category) but otherwise the figures were used as given. Incumbency information was culled from the sources listed in Table 1 of the present paper.

⁴⁰ At the end points of the election series, the following modifications prevailed: the “base party vote” for 1948 was selected from the period 1946–1952; that for 1968 from 1964–1970; and that for 1970 from 1966–1970.

⁴¹ The concept is developed and explicated by Converse, Philip E. in “The Concept of a Normal Vote,” in Elections and the Political Order, Campbell, et al., pp. 9–39 Google Scholar.

⁴² An extreme case is that of Nevada, where the “base party vote” for the Democrats exhibited an exceptionally strong and consistent increase during most of the postwar period. Whether this is due to a “true” growth of Democratic loyalists or to the fact that there is only one Representative who, having been the incumbent since 1956, has been able to steadily tighten his hold on the electorate over the last decade and a half, is a moot question. The potential for distortion in such a situation is clear.

⁴³ Kabaker, Harvey M., “Estimating the Normal Vote in Congressional Elections,” Midwest Journal of Political Science, 13 (February, 1969), 58–83 CrossRefGoogle Scholar.

⁴⁴ Uncontested Senate elections were omitted from the analysis. Cases where the Representatives from a state were unopposed eliminated that state from the analysis for that year (i.e., the “tides” variable was considered to have a missing value for that year, which required deletion of the entire state). Such an occurrence was quite infrequent. All statistical computations were carried out on Washington University's IBM 360/50 via the Statistical Package for the Social Sciences (SPSS) routines.

⁴⁵ Yet this should not lead us to conclude that the parties are in any important sense “the same,” for the similarity may very well be due mainly to the nature of the variables employed. The Republican S-values are almost always complements of the Democratic S-values. Although the P-values for the two parties are certainly not linear transformations of each other, there is nonetheless enough of an inverse relationship between them (and between the incumbency variables for the two parties) to make the similarity in behavior of the separate party equations as much of a technical expectation as of a substantive finding. I am grateful to an anonymous referee for this observation.

⁴⁶ See, e.g., Pomper, Gerald M., Foster, Badi G., Jacob, Charles E., McWilliams, Wilson C., Rosenthal, Alan, and Sigler, Jay A., The Performance of American Government (New York: The Free Press, 1972), pp. 81–86 Google Scholar.

⁴⁷ The most notable exponent of such an interpretation is Walter Dean Burnham. See his “The Changing Shape of the American Political Universe,” American Political Science Review, 59 (March, 1965), 7–28 CrossRefGoogle Scholar, and his more recent and comprehensive Critical Elections and the Mainsprings of American Politics (New York: W. W. Norton, 1970)Google Scholar. The concept of a “critical election” was first given academic sanction in Key, V. O. Jr., “A Theory of Critical Elections,” Journal of Politics, 17 (February, 1955), 3–18 CrossRefGoogle Scholar. Also pertinent are Schattschneider, E. E., The Semi-Sovereign People (New York: Holt, Rinehart, and Winston, 1960)Google Scholar; Pomper, Gerald M., Elections in America (New York: Dodd, Mead and Company, 1968)Google Scholar; The American Party Systems, ed. Chambers, William Nisbet and Burnham, Walter Dean (New York: Oxford University Press, 1967)Google Scholar; among others.

⁴⁸ See, e.g., Burnham, Critical Elections. Erikson also sees the mid-1960s as significant. See his “Malapportionment, Gerrymandering, and Party Fortunes in Congressional Elections,” at pp. 1240 ff.