Transformed Facial Similarity as a Political Cue: A Preliminary Investigation

Experimental subjects evaluated a candidate for local office whose face was digitally altered to absorb the subjects’ facial structure. For half of the subjects, the photograph of the candidate was morphed such that the image presented was a blend composed of 60% of the unfamiliar Caucasian male and 40% of the subject. For the other half the photograph was unaltered. Given previous research on implicit familiarity (Zajonc, 1968, 1980), we predicted that the morph would advantage the candidate. The results demonstrated no main effect of the similarity manipulation. However, there was a robust similarity by gender interaction. Male subjects evaluated the morphed candidate more favorably than the unaltered photograph, while female subjects rated him more negatively. We discuss potential explanations for this interaction effect, and explore the possible implications of facial similarity as a political cue in both highand low-information elections. Facial Similarity as a Political Cue 3 TRANSFORMED FACIAL SIMILARITY AS A POLITICAL CUE: A PRELIMINARY INVESTIGATION Does a candidate’s appearance matter to voters? Although visual imagery is an essential element of media-based campaigns, with the exception of a few scattered studies on candidates’ attractiveness (e.g. Efran & Patterson, 1974; Redlawsk & Lau, 2003; Rosenberg & McCafferty, 1987), political scientists have tended to discount the significance of nonverbal cues. In fact, political communication researchers generally take for granted that candidates’ visual attributes are secondary to message-based considerations including their policy positions, character traits (such as competence and integrity), performance credentials, and, most notably, partisan affiliation. This explicitly cognitive or substantive calculus is thought to be so dominant that even in campaigns where candidates’ non-verbal and physical attributes, most notably their race and gender, do influence vote preference, researchers attribute the connection to voters’ tendency to infer particular policy positions from these attributes. Gender, for instance, is taken as a liberal “signal” because most voters believe that women are generally more liberal and inclined to oppose “male” positions such as increased military spending or harsh remedies for crime (McDermott, 1988; Iyengar et al., 1997). Similarly, increased support for African-American candidates among African-American voters is considered symptomatic of group solidarity or policy agreement (Bobo & Gilliam, 1990). In effect, a candidate’s race or gender is thought to trigger gender or race-based stereotypes, which include links about the candidate’s preferences on