The effects of source valence, including its components source credibility, attraction, and homophily, on voter preference were examined in this study. A systematic sample with a random start of 350 registered Democratic voters in Tallahassee, Florida, was used to test the effects of eight dimensions of source valence on voter preference. The differential valence of the two candidates was the best predictor of voter preference. The eight-variable model accounted for 61% of the variance in voter preference, while attitude homophily alone accounted for 58% of that variance.The relationship between voter evaluations of political candidates and voting preferences has been the subject of considerable investigation. These evaluations have been called candidate images, voter attitudes, voter perceptions of the candidate and a variety of other labels. One important goal of this research was to ascertain the relationship between these voter evaluations or images of the candidates and voter preferences in electoral contests.The literature reveals several problems that have prevented a thorough test of this relationship. First, many studies on candidate image have suffered from the operationalization of image as a general evaluative construct rather than an empirically based, multidimensional construct. Second, studies which have utilized multidimensional measures of image were conducted principally in presidential elections, preventing generalization to state and local elections. Finally, most studies have been conducted in general elections where the impact of Perer A. Andersen (Ph.D., Florida State University, 1975) is assistant professor of speech communication at West Virginia University, Morgantown, WV 26506.party affiliation is more powerful than the voters' evaluation of the candidates. Assessing the impact of voters' evaluations in primary elections would eliminate this party-identification effect and would allow the specific effect of voters' evaluation of the candidates to be studied. The purpose of this study was to assess the relationship between voter evaluations of candidates and voting preference.Previous candidate evaluation or image research has demonstrated that evaluations of candidates have a substantial impact on voting. Stokes (1966) maintains that fluctuations in electoral attitudes have focused primarily on the candidates themselves. This study, however, did not carefully measure the various dimensions of candidate image. Leuthold ( 1968) found that in congressional elections nearly two-thirds of the voters thought that the candidate was their most important reason for voting, but no actual measures of image were collected. Instead, the conclusion is based on reports of the voter's reason for hedhis voting decision. Likewise, Conway (1968) reports that in low stimulus elections, personal appeals and campaign-
IN MEMORIAMWord of the sudden death of Dr. Robert J. Kibler was received while this article was in press. His passing is a severe loss to all who have known and loved him, as well as to ICA, wh...