Language-of-Interview Effects and Latino Mass Opinion

This paper extends the well-studied phenomenon of interviewer effects (race-, gender-, and ethnicity-of-interviewer) to the yet unexamined domain of language-of-interview. Using data from the 1989-1990 Latino National Politics Survey, the paper finds that the answers that survey respondents give differ, sometimes markedly, depending on whether an interview is conducted in Spanish or English. The paper then considers several explanations for why language-of-interview should matter. Methodological answers – that such effects result from sample characteristics or measurement error – fail to fully explain the effects. The influence of social desirability also does not fully explain language-of-interview effects. In addition to these factors, the findings suggest that interviews in different languages enable and evoke a distinct set of meanings, experiences, and attitudes about what it means to be Latino in the United States. Viewed thus, language-of-interview effects are more than just corrective for how we conduct surveys but also a valuable tool for examining how language informs and influences Latino mass opinion.