¹ Diamond, William, “Urban and Rural Voting in 1896”, American Historical Review, Vol. 46 (1941), pp. 281–305CrossRefGoogle Scholar.

² Holcombe, Arthur N., The Political Parties of Today (1924), p. 116Google Scholar.

³ Quantitative Methods in Politics, pp. 166–175.

⁴ The New Party Politics, p. 128.

⁵ The Middle Classes in American Politics (1940), p. 219Google Scholar.

⁶ See Key, V. O., Politics, Parlies, and Pressure Groups (1948), pp. 169–173Google Scholar, and Robinson, Edgar E., The Presidential Vote, 1896–1932 (1934), p. 31Google Scholar.

⁷ Elmo Roper's summary of his study of new voting trends in 1948 was published in the New York Herald Tribune, June 19, 1949, p. 1Google Scholar. His data indicate, he concludes, “significant long-term implications,” even, perhaps, “a one-party set-up in America.”

⁸ The 1940 census lists 14 cities with a population over 500,000. Of these, Buffalo is excluded here, as is also Washington, D. C., for which, of course, no such study as this is possible. The terms “urban” and “metropolitan” are used interchangeably here, although according to sociological criteria the latter is perhaps preferable. It is not intended to imply that the voting population outside of these 12 cities is non-urban or even non-metropolitan. Rather, the objective is to contrast the r61e of the metropolitan city with that of the rest of the state.

⁹ Projection of the analysis into the pre-1920 era is difficult, due to the unavailability or unreliability of urban election statistics. Official city-wide presidential election returns are in some states more poorly reported than are returns for counties and other elections. The sources used here for the state and national vote were: state manuals; E. E. Robinson, op. cit.; and the statement of the vote prepared after each election by the clerk of the House of Representatives. The 1948 figures were secured from statements of the vote supplied by state officials, or from the Associated Press tabulations of national and state votes, as reported in the New York Times, Dec. 11, 1948, and the Congressional Record, Jan. 18, 1949, Appendix, A 252. The following sources were used for computing the urban statistics used in this study: state manuals; Robinson, op. cit.; reports of city and county boards or departments of elections or canvassers (Detroit, Cleveland, Chicago, Milwaukee, St. Louis, Pittsburgh, and Boston); correspondence with city and state officials; city yearbooks; and newspaper tabulations. The New York Times, Dec. 11, 1948, reports returns for 13 cities for 1944 and 1948 which, however, are not entirely correct. The Chicago Daily News Almanac (1937) was helpful in unearthing the returns in earlier elections for Chicago. Los Angeles county figures were used throughout, since strictly urban returns were not available, except for recent elections. Inasmuch as Los Angeles constitutes over 50 per cent of the county population, and since 85 per cent of the county population is urban, it was felt that for our purposes it was defensible to use county data. The New York state and city Democratic figures for 1936–1948 include the votes of the American Labor and Liberal parties when these parties supported the Democratic candidates.

¹⁰ Thus, if the city's plurality is over 50 per cent of the state plurality, it can be considered a basic factor in swinging the state to either of the two major parties; if it is close to or over 100 per cent of the state plurality, it is of far greater influence—no doubt the dominant factor in that election. Table 2 should be interpreted from this standpoint. It presents the percentage which the city plurality was of the state plurality.

¹¹ This problem can be simply posed for presidential elections in the following terms: Does the metropolitan party plurality control the state's electoral decision and, therefore, its Electoral College votes (a) when metropolitan and non-metropolitan voters support different parties, and (b) when these voters support the same party?

¹² In 1920, 15.8 per cent; 1928, 18.3 per cent; 1932, 18.3 per cent; 1940, 20.7 per cent; 1948, 21.3 per cent.

¹³ Urban trends before 1920 and for cities of smaller size, in various population intervals, should be analyzed. Further, coördinated studies of the urban rôle in state and local elections, upon which the writer has begun research, are needed before definitive conclusions are permissible. Other studies of urban electoral influence which have explored some of these possibilities are: Pollock, James K. and Eldersveld, Samuel J., Michigan Politics in Transition (1942)Google Scholar; Gosnell, Harold F., Grass Roots Politics (1942)Google Scholar; Eldersveld, Samuel J. “A Study of Urban Electoral Trends in Michigan, 1920–1940”, MS (doctoral dissertation), University of Michigan, 1946Google Scholar.

¹⁴ In 1924, only Philadelphia, Detroit, and Los Angeles increased the actual size of their Republican pluralities, and only Los Angeles, St. Louis, and Baltimore increased the proportionate size of their pluralities.

¹⁵ Comparing 1928 with 1920, only Los Angeles increased the size and proportion of its Republican plurality.

¹⁶ In 1920, nine cities produced their largest (or only) Republican pluralities; in 1924, two (Philadelphia and Detroit); in 1928, one (Los Angeles). Similarly, in 1920, nine cities contributed their largest proportion of the state Republican plurality; in 1924, three. It is true that the highest mean percentage contribution to state pluralities was in 1924 (31.4 per cent), although in 1920 the highest mean actual plurality was cast. The Republicans secured more Electoral College votes by virtue of urban pluralities in 1924 (39) than in 1920 (8) or 1928 (13). Furthermore, it is not meant here to convey the impression that the actual Republican vote decreased from 1920 to 1928. As a matter of fact, it did decrease in New York City, Boston, San Francisco, and St. Louis. This is a study, however, of changes in terms of plurality contributions, not in fluctuations of the actual vote. And for most of these cities 1920 represented the high point of their “relative,” or plurality, Republican strength.

¹⁷ Op. cit., p. 175. Professor Rice's hypothesis was that “urban communities, admittedly more dynamic and more sensitive to currents of opinion than rural communities, tend to swing more strongly in any direction which public opinion in the nation as a whole is taking.”

¹⁸ Professor Key has, however, already indicated how slight is the verification for the cycle hypothesis. See his summary analysis of such studies, op. cit., pp. 605–608.

¹⁹ The number of states won by the Democrats in 1932–1948 due to urban pluralities was as follows: 1932, seven; 1936, six; 1940, nine; 1944, eight; 1948, five. The 149,000 Democratic plurality in Detroit in 1940 was not quite sufficient to offset the non-metropolitan Willkie vote.

²⁰ Percentage-wise, the number of cities in each election contributing their highest proportion of the state Democratic plurality was as follows: 1928, one (Boston); 1940, three (New York, Los Angeles, and Milwaukee); 1944, five; 1948, three (Chicago, Cleveland, and San Francisco). It must be remembered that in 1944 two of these states, Ohio and Wisconsin, went Republican; in 1948, four did so.

²¹ Detroit in 1944 and Boston in 1948.

²² See Table 3. The mean size of metropolitan pluralities was 292,000 in 1936, as compared to 120,000 in 1948. Six cities cast their smallest Democratic plurality in 1948. These were New York, Philadelphia, Baltimore, Milwaukee, San Francisco, and Los Angeles. As a consequence, New York, Pennsylvania, and Maryland went Republican, and California barely eked out an 18,000 Democratic plurality. Detroit in 1948 cast its lowest Democratic plurality since 1932, which accounts largely for Republican success in that state. Of the six states which went Democratic in 1948, three (California, Illinois, and Ohio) had only slight majorities, for which metropolitan voters were largely responsible.

²³ The “gross” percentage used in Table 5 is misleading for some election years, since it is based on the sum of 12 city pluralities for the party winning the presidency. National pluralities, on the other hand, are generally arrived at after a summation of total Republican and Democratic votes (not pluralities), as reported by the states (not cities and counties). (By laborious summation of state pluralities, other national plurality figures can be computed. For example, the national plurality for 1920 would then be R7,936 (not R7,005), and for 1948 would be D2,774 (not D2,135).) This being so, it is necessary to relate the metropolitan pluralities to the state vote, on the basis of which national pluralities are computed. The “net” percentage in Table 5 attempts to do this. It is a composite figure, summing up metropolitan pluralities, or portions thereof, which enter into the state plurality figure. It, however, is also misleading because it leaves out of consideration important metropolitan Democratic pluralities from 1932 on, because their states went Republican. The third percentage used (“party total per cent”) represents the relationship between the total Republican or Democratic metropolitan vote and the total national vote for the winning major party. The trend is observable by reference to all three indices, although as a true measure of amplitude none is perfect.

²⁴ Elmo Roper's post-election analysis of Truman's victor y in 1948 (New York Herald Tribune, June 19, 1949) poses some interesting questions. Comparing 1948 with 1944, he concludes that one reason for the Truman success was the larger turnout and greater Democratic strength of lower-income wards in 1948. Although his study is suggestive of a trend, it does not explain the Truman victory in these states. Three of Roper's seven cities actually decreased the size of their Democratic pluralities. Increased Democratic votes in some wards were not a major reason for Truman's success.

Another conclusion of Roper is that the “long-run implication” of the 1948 election “is that the big city vote will go increasingly Democratic.” The above study of metropolitan pluralities suggests that the swing of the big city vote to the Democrats was a greater and more consistent phenomenon from 1928 to 1944 than from 1944 to 1948. The pollsters' observations may help to explain their mispredictions in 1948, but the trend changes in urban voting which they discovered after the 1948 election, if true, actually were more in evidence before 1948.

²⁵ Roosevelt and Truman would also have lost these elections if one subtracts from their total Electoral College vote in those years other significant blocks of sectional, and “non-metropolitan,” votes. Although it is not so easy to perform such subtraction in 1940 and 1944, it is easier to do so for 1948. The 59 Electoral College votes of the Border, for instance, which Truman won, and not by virtue of metropolitan pluralities, were vital to his victory. See table 7, footnote 37.

²⁶ Op. cit., p. 47.

²⁷ These were: New Hampshire in 1916, Massachusetts and Rhode Island in 1928. In 1912, seven went for Wilson, Vermont for Taft, and Pennsylvania for Roosevelt.

²⁸ This section possessed 140 Electoral College votes in 1932 and 138 in 1944. In each election, the distribution was as follows: 1932, 56 Republican, 84 Democratic (64 due to metropolitan pluralities); 1936, 8R, 132D (64 metropolitan); 1940, 8R, 132D (100 metropolitan); 1944, 8R, 130D (98 metropolitan); 1948, 118R, 20D (16 metropolitan). The Republicans won five and seven states in 1932 and 1948, respectively; the Democrats won seven states in each election in 1936–1944.

²⁹ Of the 498 Electoral College votes won by the Democrats in the Northeast in 1932–1948, 342 were because of metropolitan pluralities.

³⁰ In this analysis, the following states are included in “the Mid-West”: Ohio, Indiana, Illinois, Michigan, Wisconsin, Minnesota, and Iowa. This conforms to Professor Ewing's classification, but varies from Professor Holcombe's categories, inasmuch as he includes the last three states in the “West.”

³¹ Op. cit., p. 86.

³² In 1916, Ohio went Democratic; in 1924, Wisconsin gave LaFollette its Electoral College votes. In 1932, five of these states went Democratic, Michigan and Minnesota supporting Roosevelt.

³³ In 1940, only Indiana, Michigan, and Iowa went Republican; in 1944, Indiana, Iowa, Wisconsin, and Ohio; in 1948, only Indiana and Michigan.

³⁴ The Mid-West possessed 122 Electoral College votes in 1932 and 117 in 1944. The distribution of these votes in each election was as follows: 1932, 122 Democratic (74 due to metropolitan pluralities); 1936, 122D (48 metropolitan); 1940, 44R, 78D (67 metropolitan); 1944, 59R, 58D (47 metropolitan); 1948, 31R, 86D (65 metropolitan). Of the 466 Electoral College votes won by the Democrats in the Mid-West 1932–1948, 301 were because of metropolitan pluralities.

³⁵ This is particularly true for 1932 and 1936. The only victories won in these states by the Democrats after 1936 which were not directly related to metropolitan pluralities were in Minnesota, 1940–1948, and in Iowa, 1948.

³⁶ No detailed analysis can be made here of the significance of Baltimore and St. Louis pluralities for Border politics, or of San Francisco and Los Angeles pluralities in the West. Both Border states fluctuated cyclically between the two major parties after the Civil War. California, however, was Democratic only three times before 1932. The consistent Democratic strength in these states from 1932 on, not unusual for the Border states, was, however, largely due to metropolitan pluralities. This was not so for California and Missouri, for the period 1896–1928, with four exceptions: St. Louis did swing the state to the Republicans in 1908 and 1924; California went Progressive in 1912 and Republican in 1928 due to metropolitan pluralities. Baltimore, on the other hand, was a major factor in Republican successes in Maryland in 1896, 1900, 1908, 1920, and 1924, and contributed an important Democratic plurality to Wilson in 1916. Except for Baltimore, these cities have been consistently more instrumental in major party success from 1932 on.

³⁷ The following breakdown by elections summarizes the data on total sectional Electoral College votes won by the Democrats in 1932–1948, as well as indicating the number of such votes controlled by metropolitan pluralities. It will be seen that several formulae are possible.

³⁸ The New Party Politics, p. vii.

³⁹ See Gras, N. S. B., An Introduction to Economic History (1922)Google Scholar; McKenzie, R. D., The Metropolitan Community (1933)Google Scholar; and D. J. Bogue, The Structure of the Metropolitan Community; A Study of Dominance and Subdominance (1949).

⁴⁰ Much research is necessary, for example, before the “urban middle class” aspect of the Holcombe hypothesis is verified. Coördinated factorial analyses of the composition of the urban vote are needed. These should describe urban voting behavior in terms of turnout, consistency of voting, third-party support, and political attitudes, in local, state, and national elections, and for contrasting types of cities.

⁴¹ This study was completed under a grant from the Faculty Research Fund of the Horace H. Rackham School of Graduate Studies, University of Michigan.