¹ Attempts to delineate this phenomenon more precisely have been made: cf. Stokes, Donald E. and Iverson, Gudmund R., “On the Existence of Forces Restoring Party Competition,” Public Opinion Quarterly, 26 (Summer, 1962), 159–171CrossRefGoogle Scholar; and Sellers, Charles, “The Equilibrium Cycle in Two-Party Politics,” Public Opinion Quarterly, 29 (Spring, 1965), 16–38CrossRefGoogle Scholar.

² Bean, Louis, How to Predict Elections (New York: Alfred A. Knopf, 1948)Google Scholar. In an earlier work, Ballot Behavior: A Study of Presidential Elections (Washington, D.C.: American Council on Public Affairs, 1940), Bean discussed how state trends differ from the national trend (chapters 2–5), but he was interested in predicting elections and not measuring inter-party competition. Bean's ideas were presented in a more sophisticated way by Gosnell, Harold P., Grass Roots Politics: National Voting Behavior of Typical States (Washington, D.C.: American Council on Public Affairs, 1942)Google Scholar.

³ How to Predict Elections, Table 8 (pp. 186–87) and chapter ten, especially pp. 120–21.

⁴ Bean's data extend from 1916 to 1944.

⁵ For example, cf. Schattschneider, E. E., The Semisovereign People (New York: Holt, Rinehart and Winston, 1960), pp. 89–90Google Scholar.

⁶ Zeller, Belle (ed.), American State Legislatures: Report of the Committee on American Legislatures, American Political Science Association (New York: Thomas Y. Crowell Company, 1954), chapter 12Google Scholar.

⁷ Essentially this criticism holds true for the inter-party competition index based on percent-age of elections won applied to Iowa counties by Gold, David and Schmidhauser, John R., “Urbanization and Party Competition: The Case of Iowa,” Midwest Journal of Political Science, 4 (1960), 62–75CrossRefGoogle Scholar, and to the rather crude inter-party competition index used for purposes of illustration by Key, V. O. Jr., Politics, Parties, and Pressure Groups (fifth edition; New York: Thomas Y. Crowell Company, 1964), chapter 11Google Scholar.

⁸ Golembiewski, Robert T., “A Taxonomie Approach to State Political Party Strength,” Western Political Quarterly, 11 (1958), 494–513CrossRefGoogle Scholar.

⁹ Jewell, Malcolm E., The State Legislature: Politics and Practice (New York: Random House, 1962)Google Scholar.

¹⁰ Ranney, Austin and Kendall, Willmoore, “The American Party Systems,” this Review, 48 (1954) 477–485Google Scholar; it will also be found in their work Democracy and the American Party System (New York: Haroourt, Brace and Company, 1956), pp. 161–64. The same technique with variations was used extensively by others: Key, V. O. Jr., American State Politics: An Introduction (New York: Alfred A. Knopf, 1956)Google Scholar, used gubernatorial elections in 21 non-Southern states over a varying span of time; Ransone, Coleman B. Jr., The Office of Governor in the United States (University, Alabama: University of Alabama Press, 1956)Google Scholar, added two needed categories but restricted himself to gubernatorial elections; Eulau, Heinz, “The Ecological Basis of Party Systems: The Case of Ohio,” Midwest Journal of Political Science, 1 (1957), 125–135CrossRefGoogle Scholar, limited himself to the Ohio House of Representative elections and reduced the categories to three; Standing, William H. and Robinson, James A., “Inter-Party Competition and Primary Contesting: The Case of Indiana,” this Review, 52 (1958), 1066–1077Google Scholar, objected to the Ranney-Kendall index but applied the same technique to three state-wide races; Leiserson, Avery, Parties and Politics: An Institutional and Behavioral Approach (New York: Alfred A. Knopf, 1958)Google Scholar, changed the terminology but used the technique; Lockard, Duane, New England State Politics (Princeton: Princeton University Press, 1959)CrossRefGoogle Scholar, and The Politics of State and Local Government (New York: The Macmillan Company, 1963), shortened the base period from 40 years to 16 and considered only Presidential and gubernatorial elections; Cox, Edward F., “The Measurement of Party Strength,” Western Political Quarterly, 13 (1960), 1022–1042CrossRefGoogle Scholar, added some statistical sophistication to the Ranney-Kendall index; Cutright, Phillips, “Ubanization [sic] and Competitive Party Politics,” Journal of Politics, 25 (1963), 552–564CrossRefGoogle Scholar, used the Ranney-Kendall technique to verify the substantive results of Eulau, op. cit., while rejecting those of Gold and Schmidhauser, op. cit.; Wirt, Frederick M., “The Political Sociology of American Suburbia: A Reinterpretation,” Journal of Politics, 27 (08, 1965), 647–666CrossRefGoogle Scholar, also used the Ranney-Kendall technique while acknowledging its limitations; all of the studies cited below are also indebetd to Ranney and Kendall.

¹¹ Ranney, Austin, “Parties in State Politics,” Politics in the American States: A Comparative Analysis, Jacob, Herbert and Vines, Kenneth N. (eds.), (Boston: Little, Brown and Company, 1965), chapter 3Google Scholar.

¹² Schlesinger, Joseph A., “A Two-Dimensional Scheme for Classifying the States According to Degree of Inter-Party Competition,” this Review, 49 (1955), 1120–1128Google Scholar, and “The Structure of Competition for Office in the American States,” Behavioral Science, 5 (I960), 197–210. Avery Leiserson, Parties and Politics: An Institutional and Behavioral Approach, op. cit., used the same terminology as Schlesinger but not his technique using instead percentage of vote; Casstevens, Thomas W. and Press, Charles, “The Context of Democratic Competition in American State Politics,” American Journal of Sociology, 68 (03, 1963), 536–543CrossRefGoogle Scholar used the results from Schlesinger's 1955 and 1960 articles, from Ranney and Kendall, and from Key's American State Politics to discuss the ecological basis of American party competition (cf. also a comment upon Casstevens and Press' article by Outright, Phillips, American Journal of Sociology, 69 (11, 1963), 291–292CrossRefGoogle Scholar, and a rejoinder by the authors in ibid., 69 (November, 1963), 292–294, centering on the economic indicators used and on the inter-party competition index); Dawson, Richard E. and Robinson, James A., “Inter-Party Competition, Economic Variables, and Welfare Policies in the American States,” Journal of Politics, 25 (1963), 265–289CrossRefGoogle Scholar, combined the technique of Ranney and Kendall with that of Schlesinger and found a high correlation between them when ranking the states according to inter-party competition.

¹³ Hofferbert, Richard I., “Classification of American State Party Systems,” Journal of Politics, 26 (1964), 550–567CrossRefGoogle Scholar.

¹⁴ Its analysis is put off in order to criticize all the major attempts with their suggested techniques at one time. Hofferbert has used his index in his artiole “The Relation between Public Policy and Some Structural and Environmental Variables in the American States,” this Review, 50 (March, 1966), 73–82.

¹⁵ Such as not indicating trends, not revealing much about the Southern states, and not always being able to measure alternation after reappor-tionment of the U. S. House, all of which Schlesinger mentioned. One criticism which he implicitly noted on page 208 of his 1960 article, but the import of whioh he did not seem to realize, is that a measure of alternation in the office of U. S. Representative measures Congressional district competition, not state-wide competition. To compare a district measure with state-wide measures certainly does produce “a special effect upon the result” as he said on page 200 of the same article. Furthermore, as he presents it, there is no party identification connected with the rate of alternation. This identification can easily be added and probably would have produced different configurations in his diagrams (pages 202–205 of his 1960 article). Finally, because there are two offices with the same title (U. S. Senator) Schlesinger treated them as if they were one office being held by two persons at the same time. Measuring alternation in office for the two offices of U. S. Senator produces no more problems than measuring it for a four-year-term governor and two-year-term U. S. Representative.

¹⁶ Hofferbert has the same trobule as Schlesinger in handling the office of U. S. Senator; he includes only three offices; he considers his base period (1932–1962) as one which “adequately reflects the contemporary alignment between the parties” (page 558) but nowhere substantiates his contention or even tries to do so and in fact has to shorten the period to 1946–1962 for Minnesota and Wisconsin; his ranking system fails to distinguish between parties; his method requires even more computations than Schlesinger's; his example to show the efficacy of ranking turned out to be a poor one (a Rho of only+0.58 was an unfortunate result although the term “significant correlation” is open to debate); and gremlins were very active with his Table I (pp. 562–563) because Kansas was left out entirely, Georgia's average rank is an obvious typographical error, Vermont's is also incorrect, and New Mexico seems to have been wrongly ranked.

¹⁷ An attempt, not very successful, was made to obtain groupings (two-party, weak two-party, modified one-party, and one-party) from Hoffer-bert's ranking. Actually the rank versus group quibbling is silly because we still do not know exactly what we want to do with these data. Both methods must be used and results compared. More important, however, is that changing an interval scale into an ordinal scale (as Hofferbert had done) unnecessarily limits applicable procedures.

¹⁸ Cf. Cox, op. cit., pages 1022–28.

¹⁹ Loc. cit. and Schlesinger (1960), op. cit., page 198.

²⁰ Ranks, on the other hand, are expressed on an ordinal scale suitable only for calculating some nonparametric correlation coefficients.

²¹ It is hard to escape the contention that all of these decisions must be arbitrary. Perhaps they must remain so for the present. An awareness of the problems involved and of the limitations are the only real bench mark at this time. Cf. Golembiewski, op. cit., pp. 494–499; Jewell, op. cit., p. 13; Dawson and Robinson, op. cit., p. 273; Hofferbert, op. cit., pp. 554–555, although on page 558 he appears to have forgotten. In any event, the appearance of arbitrariness will disappear when satisfactory results are obtained.

²² In this instance the following problems were encountered at this point. In New York the votes cast in one party's column for a candidate of another party were counted for the candidate's party. In the three state-wide elections in California (during the time selected) in which there was only one major party candidate due to crossfiling, his votes were attributed to his party. Although not a satisfactory solution, it can be defended (perhaps) on the ground that one party was not strong enough to field a candidate who could win over the opposition party's candidate in the first party's primary. In Wisconsin the Progressive Party was counted as a third party—cf. Epstein, Leon D., Politics in Wisconsin (Madison: University of Wisconsin Press, 1958)Google Scholar, chapter III. One-party Southern states were counted as oneparty although there is strong opposition to doing so—cf. Sindler, Allan P., “Bifactional Rivalry as an Alternative to Two-Party Competition in Louisiana,” this Review, 49 (1955), 641–642Google Scholar. As noted in the text below, a composite vote for U. S. Representative was obtained for use by adding the votes in each district to get a state-wide total for each party. (In Arkansas the vote in unopposed races is not tabulated so when there were opposition candidates in only a few of the U. S. Representative districts from 1958 to 1964, it was not possible to compute the percentages accurately. In these cases the elections were not counted. If there were no opposition candidates at all, then the candidates' party received 100% of the vote.) Whenever conflicts appeared in vote totals, the governmental source was taken as authoritative. The source for the data was primarily the Statistical Abstract of the United States (1940–1964) supplemented where necessary by Scammon, Richard M. (ed.), America Votes 5 (Pittsburgh: University of Pittsburg Press, 1964)Google Scholar, the World Almanac, for various years, the CQ Weekly Report, The Political Almana, 1952 (New York: B. C. Forbes & Sons Publishing Co., Inc., 1952)Google Scholar, and The New York Red Book, 1944 (Albany: Williams Press, Inc., 1944). As soon as the Inter-University Consortium for Political Research has completed its data bank, it will be the best source of election data.

²³ For the years 1940–1958, Σg = 48×4 = 192; for the years since 1960, Σg = 50×4 = 200.

²⁴ An index applicable to multiparty systems is now under development as is one to be used for primary elections.

²⁵ On this problem cf. Stokes, Donald E., “Spatial Models of Party Competition,” this Review, 57 (06, 1963), 368–377Google Scholar.