Electoral Change in the American South, 1948-1976: The Influence of Population Size

Southeastern Geographer Vol. XIX, No. 2, November 1979, pp. 80-90 ELECTORAL CHANGE IN THE AMERICAN SOUTH, 1948-1976: THE INFLUENCE OF POPULATION SIZE Gerald L. Ingalls and Stanley D. Brunn The last time the Republicans were in Atlanta was 100 years ago. They burned it down. Jimmy Carter Democratic Governor of Georgia 1971(2) Students of electoral behavior have maintained a long-standing interest in the processes that lead to changes in the character of electoral response. Implicit in much of this research is the contention that urban environments foster electoral change. (2) Correspondingly students of politics have focused a great deal of attention upon the American South, especially upon the link between large urban environments and the political changes represented by improved Republican Party fortunes. The South's political tradition is strongly Democratic, particularly in rural areas which even today remain bastions of one party politics. Retween 1948 and 1976 the "solidly" Democratic South was breached in a number of places by Republican electoral gains (Fig. 1). Several studies have provided evidence that electoral support given G.O.P. presidential candidates in the South has been greater in large than in small cities. (3) Recause these studies were of presidential elections , however, we are uncertain whether the generally positive relationship between size-of-place and level of Republican electoral support holds true at other levels of competition. Specifically, are G.O.P. inroads as significant in senate races? Is the support given to G.O.P. candidates consistent through time or do victories in one election erode in the next? Finally, has support for G.O.P. candidates remained an "urban" phenomenon or has it spread to rural areas? This paper explores the nature of electoral change in the American South from 1948 to 1976 by focusing Dr. Ingalls is Associate Professor of Geography at the University of North Carolina at Charlotte, in Charlotte, NC 28223. Dr. Brunn is Professor of Geography at Michigan State University in East Lansing, Ml 48824. Vol. XIX, No. 2 81 on the direction and strength of the relationship between electoral support for G.O.P. candidates and sizes of county population. THE APPROACH. The study area consists of a sample of 424 contiguous counties drawn from the 1,139 counties or county equivalents in the 11 original states of the Confederacy (Fig. 1). This sample represents about 40 percent of all "southern" counties. It includes counties from three Rim or Rorder South states and four Deep South states, which provide considerable diversity in population size, cultural and economic characteristics (Rlack ReIt agricultural areas to industrial areas), and cities of various sizes. Changes in patterns of electoral response were measured by use of the Republican vote. The data consisted of 14 electoral variables and three population variables. (4) The electoral variables are the percentage of the total vote going to Republican candidates in the eight presidential elections from 1948 to 1976 and the senatorial elections of 1948, 1954, 1960, 1966, 1972, and 1976. The population data are the total population for each county in the sample area for 1950, 1960, and 1970. All variables were measured in county units. Recause the data deviated from a normal distribution, nonparametric tests and graphic associations were employed to test for the expected relationships between county population size and the level of Republican vote. Statistical relationships were derived by computing Spearman 's rho coefficients for each of the 14 possible rank-order relationships . Population data for 1950 were used with the 1948 and 1952 presidential elections and the 1948 and 1954 senatorial elections; population data for 1960 were used for the presidential elections in 1956, 1960, and 1964, and for senatorial elections in 1960. Population data for 1970 were used for the presidential races in 1968, 1972, and 1976, and for senatorial races in 1966, 1972, and 1976. In testing the relationships, null hypotheses were postulated. These assumed the following form: H0 no significant relationship exists between the level of support for the Republican party in the 1948 (1952, etc.) presidential (or senatorial ) election and the level of population size (total population) in 1950, 1960 or 1970. The tests for statistical relationships permit discussion of the general...