¹ The role of the city manager in three of the cities, Alpha, Beta, and Gamma, was reported on in Adrian, Charles R., “Leadership and Decision-Making in Manager Cities: A Study of Three Communities,” Public Administration Review, Vol. 18 (1958), 208–213CrossRefGoogle Scholar.

² Historically, two other claims of significance were made on behalf of nonpartisan elections: (1) that they raised the caliber of candidates for elective office; and (2) that they provided for greater integrity in the political process. These considerations are not included in this article.

³ Some attention has been given to the city which is only formally nonpartisan. See Adrian, Charles R., “A Typology for Nonpartisan Elections,” Western Political Quarterly, Vol. 12 (1959), pp. 449–458CrossRefGoogle Scholar; and Freeman, J. Lieper, “Local Party Systems: Theoretical Considerations and a Case Analysis,” American Journal of Sociology, Vol. 44 (1958), pp. 282–289CrossRefGoogle Scholar.

⁴ Most of the data used in this article are public information and if our findings were to stop here, there would be no useful purpose in masking the identity of the cities. However, much confidential information has been gathered for additional studies involving these cities.

⁵ The term “prevailing slate” will be used in referring to the “citizens' committee” endorsees in each of the Alpha and Beta elections and to the groups of councilmen who banded together on two occasions in Gamma. The less successful blocs in Alpha will be called “opposition slates.” The Delta Union slate does not lend itself to the following analysis because of the ward elections in that city.

⁶ The fact that the slopes are all below one indicates that the distribution of precincts is in every case dispersed over a wider range when measured by percentage of party support than when measured by percentage of slate support. This calculation may be influenced by the fact that we are comparing an individual election on one side to the average of a group of elections on the other.

⁷ In the previous analysis a simple coefficient of correlations was computed because of the necessity of determining the slope. In addition, the hypothesis implied a straight linear relationship. In this case the easier-to-compute rank order correlation, with its curvilinear assumptions, is adequate to indicate higher and lower orders of relationship.

⁸ These calculations are based upon precinct data. Conceivably it is the Republican voters in predominantly Democratic precincts who constitute a large percentage of the non-voters, but this is statistically unlikely. Our analysis is subject to a second qualification. The correlation figures could be affected by “plunking,” i.e., the practice of voting for less than the maximum allowable number of candidates. If plunking were more characteristic of Democratic precincts, the correlations would be reduced. But plunking is a technique of minorities, or shows a voter's lack of knowledge about candidates. In either case it is a functional equivalent of non-voting. It shows a disassociation from local politics by large numbers of Democratic voters.

⁹ A precinct boundary map was unavailable for the earlier elections, forcing a restriction of this analysis to the two years. The coefficients of correlation are significant at the one per cent level.

¹⁰ See Pearson, G. W., “Prediction in a Non-Partisan Election,” Public Opinion Quarterly, Vol. 12 (Spring, 1948), p. 115CrossRefGoogle Scholar, in which it was reported that in a nonpartisan election poll, Denver voters were not influenced in their choices by knowledge of the candidates' political affiliation.