This thesis proposes a mathematical model which postulates a precise relationship between message and sender. The experiment was designed so that the experimenter could note variations in the evaluations of identical statements as they were perceived to have been emitted by various sources. To test the appropriateness of the model, the statements were chosen by the experimenter and the sources selected by the subjects in a way that a two-dimensional structure--a competence dimension and a like dimension--should appear from the analysis. From the analysis of the pooled data, two distinct but highly correlated dimensions appeared. These dimensions could be clearly identified as a "like" dimension and a "competence" dimension. Statements and sources relevant to competence clustered around the competence dimension; statements and sources relevant to liking clustered around the dimension identified as the like dimension. This analysis accounted for nearly 99 percent of the variance.