Published online by Cambridge University Press: 02 September 2013
This study aims to analyze some features of voting behavior in a particular state party system and to illustrate new methods of analysis of election data. In this paper we use the gross election statistics for Michigan counties and cities to relate recent voting trends to the ecological characteristics of counties and to investigate the effect of certain variables on voting in cities.
A closer scrutiny of Michigan election results is warranted at this time because major interest groups in this state no longer place a high premium on “keeping out of politics.” Traditionally, interest groups in American state and national politics are supposed to avoid any open and continuing commitments to particular party organizations or factions. Labor and business groups in Michigan, however, skirting the Taft-Hartley and corrupt practices acts (which purport to limit union and corporate political activities), have either modified or abandoned their past policies in this respect. The Michigan CIO fully accepts a partisan role and since 1948 a liberal-CIO coalition has controlled the Democratic party organization.
2 Stephen, B. and Sarasohn, Vera H., Political Party Patterns in Michigan (Detroit: Wayne State University Press, 1957)Google Scholar.
3 V. O. Key throughout his analysis of party politics in American states classifies Michigan as a “sure” Republican state. This classification, however, ia based on long term data and Key does note that Michigan politics are undergoing a change. American State Politics: An Introduction (New York, 1956)Google Scholar.
4 The 1955 and 1957 Republican state conventions have been marked by floor fights over the state party chairmanship in which General Motors and Ford managements were reported to have backed opposing candidates. Detroit News, February 20, 1955 and February 9, 1957.
6 U. S. Bureau of the Census, U. S. Census of Population: 1950, Vol. II, Characteristics of the Population, Part 22, Michigan, Chapter B (Washington, 1952)Google Scholar.
7 The counties falling in each type are listed as follows: Urban (13): Calhoun, Genesee, Ingham, Jackson, Kalamazoo, Kent, Macomb, Midland, Muskegon, Oakland, Saginaw, Washtenaw, Wayne; Farm-Urban (7): Bay, Berrien, Lenawee, Monroe, Ottawa, St. Clair, St. Joseph; Farm (21): Allegan, Barry, Branch, Cass, Clinton, Eaton, Gratiot, Hillsdale, Huron, Ionia, Isabella, Lapeer, Livingston, Mason, Montcalm, Newaygo, Oceana, Sanilac, Shiawassee, Tuscola, Van Buren; Mineral (5): Gogebic, Houghton, Iron, Keweenaw, Marquette; Forest Lower (27): Alcona, Alpena, Antrim, Arenac, Benzie, Charlevoix, Cheboygan, Clare, Crawford, Emmet, Gladwin, Grand Traverse, Iosco, Kalkaska, Lake, Leelanau, Manistee, Mecosta, Missaukee, Montmorency, Ogemaw, Osceola, Oscoda, Otsego, Presque Isle, Roscommon, Wexford; Forest Upper (10): Alger, Baraga, Chippewa, Delta, Dickinson, Luce, Mackinac, Menominee, Ontonagon, Schoolcraft.
8 Except for 1958 returns, all election statistics were obtained from the Michigan Manual (Lansing: Secretary of State, biennially)Google Scholar. Returns for the 1958 election were obtained directly from the Elections Division of the Secretary of State's Office.
9 The F values for the respective elections are: 1948, 13.41; 1950, 20.26; 1952, 24.43; 1954, 17.88; 1956, 16.70; 1958, 16.74 where Fα.01 (n 1=5, n 2 = 80) =3.25. The importance of the analysis of variance test can hardly be over-estimated in this instance since it indicates that the county type classifications are more than mere artifacts. We also calculated the coefficients of variation for both parties by county type in each election; but the results of this latter analysis were inconclusive.
11 The figures for 1950 are used in order to get consistency with the remainder of this analysis and correspondence with 1950 census data. A breakdown identical to Table II was performed for the 1956 gubernatorial vote and revealed similarly wide variations within the size categories.
12 Occupational and labor force data were obtained from U. S. Department of Commerce, Bureau of the Census, U. S. Census of Population: 1950, Vol. II, Characteristics of the Population, Part 22, Michigan, Chapter B (Washington, 1952)Google Scholar.
13 The equation giving the expected percentage of the Democratic vote based on the percentage of laborers in the total labor force was Y = 13.67 + 1.16x, where Y is the expected percentage of the two-party vote received by the Democratic candidate for governor in 1950 and x is the percentage of the total labor force constituted by laborers in 1950. The equation Y = 29.87 +2.58x describes the regression line relating the Republican vote to the managerial classification, where Y equals the expected percentage of the two-party vote received by the Republican candidate for governor and x is the percentage of the total labor force composed of managers in the same year cited above. Substituting the known x in both equations for each of the cities in our study, it was possible to calculate the expected percentage of the vote (Y) received by each of the candidates and to compare it with the percentage actually received.