A study was designed to investigate the consequences of various coding strategies on the semantic differential as a function of responses to an ideal source in the prediction of overall credibility and homophily. The study was also intended to provide additional evidence for the disconfirmation of the underlying extremity assumption of the semantic differential. One week prior to the 1976 presidential election, 390 undergraduates completed two questionnaires containing credibility and homophily items: the first questionnaire in response to the ideal male candidate, and the second in response to the current candidate, Gerald Ford. The second set was augmented by a request to indicate how nearly ideal Ford was perceived to be. Data were processed by a single sample t-test, strength-of-association test, and multiple regression analysis. Additional evidence was found for the disconfirmation of the extremity assumption of the semantic differential. Neither recording strategy provided a better predictor for credibility or homophily than did the standard coding strategy. (Data tables are included.) (Author/DF)