Does the Medium Matter?: Cuing Race and Ethnicity in Survey Experiments

• In Event: Identity and Measurement in Racial Politics

Abstract

How do different experimental designs that study candidates influence how respondents process information about a candidate's race and ethnicity? Traditionally, studies looking at how a candidate’s race and ethnicity shapes vote choice have relied on vignette experiments, where respondents receive a paragraph candidate biography and researchers vary relevant characteristics. However, there has been a growing shift towards using conjoint experiments to study preferences towards political candidates. In conjoint experiments, information is presented line-by-line in a table format rather than lumped in a paragraph, potentially changing how the information is processed by the respondent. Using two survey experiments, we test how differences in the treatment mode (vignette or table) used to present candidates’ characteristics influence how respondents use information about a candidate’s race and ethnicity. Counter to our initial expectations, our first experiment demonstrates that respondents expressed higher levels of support for Latino candidates in a table experimental design (compared to a vignette experimental design). This is true even as we found hypothesis-congruent results for Black candidates, particularly among respondents high in racial resentment. Building on these findings in a second survey experiment, we test the effect of treatment mode on support for Latino vs white candidates, looking at how the potential mechanisms of information salience, cognitive load, and stereotype control vary across the two experimental set-ups.

Authors

• Sara Morell, The College of New Jersey

• Angela Ximena Ocampo, University of Texas at Austin