¹ “Oscillations in Politics,” Annals of the American Academy of Political and Social Science, Vol. 12, pp. 69–97 (07, 1898)CrossRefGoogle Scholar.

² See particularly Key, V. O. Jr., Politics, Parties and Pressure Groups (New York, 1950), pp. 595–603Google Scholar.

³ Chapin, F. S., “Variability of the Popular Vote at Presidential Elections,” American Journal of Sociology, Vol. 18, pp. 222–240 (09, 1912)CrossRefGoogle Scholar.

⁴ Millspaugh, A. C., “Irregular Voting in the United States, Political Science Quarterly, Vol. 33, pp. 230–254 (06, 1918)CrossRefGoogle Scholar.

⁵ Ogburn, W. F. and Jaffe, Abe J., “Independent Voting in Presidential Elections,” American Journal of Sociology, Vol. 42, pp. 186–201 (09, 1936)CrossRefGoogle Scholar.

⁶ Fletcher, R. and Fletcher, M., “Consistency in Party Voting from 1896 to 1932,” Social Forces, Vol. 15, pp. 281–285 (12, 1936)Google Scholar.

⁷ Gosnell, H. F. and Coleman, W. G., “Political Trends in Industrial America,” Public Opinion Quarterly, Vol. 4, pp. 473–486 (Sept., 1940)CrossRefGoogle Scholar. See page 482 for a unique computation of independent voting. The authors arrive at the independent vote by taking the low Democratic and Republican percentages over any given series of elections, subtracting the sum of these minimal percentages from 100, and calling the remainder “the independent vote.”

⁸ See his How to Predict Elections (New York, 1948), especially pp. 74 ff.Google Scholar, where the author attempts a measure of “political flexibility.”

⁹ “Ballot Laws and Their Workings,” Political Science Quarterly, Vol. 21, pp. 38–58 (March, 1906)CrossRefGoogle Scholar.

¹⁰ Op. cit.

¹¹ Gosnell, Harold F. and Gill, Norman N., “An Analysis of the 1932 Presidential Vote in Chicago,” American Political Science Review, Vol. 29, pp. 967–984 (Dec., 1935)CrossRefGoogle Scholar.

¹² Detroit Voters and Recent Elections (Detroit, 1938)Google Scholar.

¹³ Lazarsfeld, Paul F., Berelson, Bernard, and Gaudet, Hazel, The People's Choice (New York, 1948)Google Scholar.

¹⁴ Survey Research Center, Institute for Social Research, A Study of the Presidential Vote: November, 1948 (Ann Arbor, 1949)Google Scholar.

¹⁵ These variations in findings are summarized by Key, op. cit. (above, n. 2), pp. 599–601.

¹⁶ Op. cit. (above, n. 3), p. 240.

¹⁷ Herring, Pendleton, The Politics of Democracy (New York, 1940), p. 248Google Scholar.

¹⁸ Toward a More Responsible Two-Party System (New York, 1951), p. 90Google Scholar.

¹⁹ Schattachneider, E. E., The Struggle for Party Government (College Park, Maryland, 1948), p. 6Google Scholar.

²⁰ The Survey Research Center Study of 1948 was the product of two nationwide surveys conducted during October and November, 1948. The sample was selected by the method of probability and area sampling. The data used here are taken from the survey made after the November election. The author is indebted to Angus Campbell, Director of the Center, for permission to use these data.

The Washtenaw County Study was based on the same sampling method but was not so rigorous in methodology. Certain rural areas of the county were not included in the study; the sample was fulfilled more completely in some areas (particularly Ypsilanti and Ann Arbor) than others, and the interviewing was not as professional. As a consequence, only in a few instances are the data relied on here, and then only for suggestive purposes.

²¹ Self-classified “partisans” were considered “independents” if they admitted to “often splitting” their ballot, quite frequently voting for “the opposite party,” or splitting their ticket in 1948. On the other hand, self-classified “independents” were considered “partisan” if they demonstrated none of these tendencies.

²² The latter figure is found in the Gallup poll released June 25, 1948.

²³ See, for example, “Quizzing Taft,” U. S. News and World Report, 03 14, 1952, p. 56Google Scholar.

²⁴ “Definite Republicans” and “Definite Democrats” were respondents who voted consistently for one party in 1948, 1944, and “previously.” “Independents” were those who split their ballots and who indicated no consistent pattern of adherence to one of the two major parties. The political predisposition of “independents” was determined by analyzing the frequency and direction of transference of party allegiance, and their preference if nonvoters in 1948, as well as from the pattern of respondents' split-voting, if ascertainable. It was too difficult to determine the political predisposition of the independents who never voted—hence eighteen instead of twenty categories. Even so, the data on voting frequency are subject to certain reservations. Although the responses to several questions were used, no doubt there still was an over-reporting of “always voting.”

The number of cases in each of the “infrequent” and “never” (except Democratic “never”) categories was small, by necessity. The allocation of the “nevers” was most difficult, greatest reliance being placed on voting preference in 1948 “if you had voted.” Demographic data were not helpful here.

²⁵ Eldersveld, Samuel J., “The Influence of Metropolitan Party Pluralities in Presidential Elections Since 1920,” American Political Science Review, Vol. 43, pp. 1189–1206 (12, 1949)CrossRefGoogle Scholar.

²⁶ A sample of Michigan counties was used for this study since an examination of data for all counties would have required more time than was available. The 35 counties include 10 from the Upper Peninsula, 13 from the upper half of the Lower Peninsula, and 12 from the lower half of the Lower Peninsula, including Wayne and almost all of the Detroit precincts.

²⁷ An analysis was made of precincts which were high and low in split-voting in 1950 in comparison to their split-voting percentages in 1940. The five highest precincts in 1950 were also high in 1940, above the city average of 27.3 per cent. But the correlation of all first ward precincts in 1950 with 1940 was not significant (.286). The author is indebted to Professor Donald S. Hecock of Wayne University, Detroit, for data on split-voting in previous elections in Detroit.

²⁸ The Washtenaw Study revealed the following interesting pattern of split-voting in 1948:

Split for national and statewide candidates 33%

Split for President 10%

Split for Congressman 11

Split for Governor 10

Split—not ascertained 2

Split for local candidates 28%

No one office was primarily responsible for split-voting.

²⁹ Key, p. 602.

³⁰ Data on Detroit split-voting by precincts for 1936 through 1946 were secured from Professor Donald S. Hecock, Wayne University, Detroit, who has done a prodigious job in collecting these figures.