Five demographic characteristics-education, income, sex ratio, age, and percentage of children-were compared with the percentage of vote pro-integration. Higher income and education were found to correlate positively at a very significant level with a higher vote percentage. Significant correlations with sex ratio, age, and percentage of children were found only in the moiety of precincts highest in percentage of vote pro-integration. A TOPIC of major interest at the present time is that of liberal-conservative differences. Analyses in this field have outlined a number of consistent variations in the attitudes of different groupings and classes, making understandable some of the political behavior of these groups which is not directly tied to interest. For example, the lower classes tend to be liberal in their attitudes on economic issues, but less liberal concerning race relations.' This paper is concerned with some of the nonethnic demographic correlates of pro-integration voting by white precincts in a metropolitan southern county as a means of determining the differing attitudes of different types of precincts. The method of analysis used, that of correlation of demographic and voting data, is a departure from the usual practice of obtaining information from a sample survey such as that which underlies Melvin M. Tumin's study of attitudes toward desegregation in Guilford County (GreensboroHigh Point), North Carolina,2 but it has a major advantage over the survey method. Surveys measure what people say they have done or what they will do rather than what actually occurs, and at least one study has demonstrated that the tendency is to give a socially acceptable answer.3 In the case of an issue as controversial as integration is in the South, one would have to accept any survey data with extreme caution, but casting a vote is an anonymous and concrete action. Unfortunately, the use of voting data does have some limitations. The major one is that aggregate data cannot be linked to specific individuals except in rare cases.4 However, through the use of demographic data for the voting units, it can still be used for comparisons among precincts with different characteristics as is done in this paper. One other objection to the use of voting data is that elections are rarely so clear-cut as to the issues involved that one can determine which questions, if any, are being resolved. But, through the careful selection of elections for analysis, some important questions can be isolated. The election under study in this paper is of this type. * I am grateful to Dr. Richard L. Simpson of the University of North Carolina for the suggestions he made concerning this paper. I S. M. Lipset and R. Bendix, "Political Sociology," Current Sociology, VI (1957), 90. 2 Desegregation: Resistance and Readiness (Princeton, N. J.: Princeton University Press, 1958). 3H. J. Parry and H. M. Crossley, "Validity of Responses to Survey Questions," Public Opinion Quarterly, XIV (1950), 61-80. 4W. S. Robinson, "Ecological Correlations and the Behavior of Individuals," American Sociological Review, XV (1950), 351-357.