How Interview Language Shapes Bilingual Latinos' Political Opinions

Why Language in Surveys Matters?

Public opinion researchers increasingly find that respondents' answers vary by the language used in interviews, but theoretical explanation is limited. Efren Perez asks why interview language changes responses among bilinguals and why some political concepts are more vulnerable to this effect. The question matters for measuring public opinion accurately, for comparing groups across languages, and for understanding political expression among bilingual communities.

Theory: Language and Mental Accessibility

Perez proposes that language affects survey responses by making some political concepts more mentally accessible. The core claim is that certain constructs become associated with a particular language during learning and socialization; when interview language matches the language tied to a construct, recalling that construct from memory becomes easier and opinions about it intensify.

How the Study Tests the Idea

The hypothesis is tested with an experimental manipulation of interview language in two U.S. surveys of English–Spanish bilingual Latino adults. Key elements of the design include:

• Random assignment to English or Spanish interview conditions.
• Outcome measures of opinion levels, item nonresponse (refusal rates), and response times.
• A range of constructs that vary in stability and crystallization (for example, American identity, anti-Obama affect, and partisanship).
• Tests for emotional mediation, measuring anxiety, anger, pride, and efficacy to rule out affective explanations.

Key Findings

• Interview language systematically influences the accessibility of some political concepts: respondents report stronger opinions on concepts tied to the interview language.
• Example: respondents interviewed in English express higher levels of American identity than those interviewed in Spanish.
• English interviewees are less likely to refuse factual knowledge items about U.S. politics and tend to answer them more quickly, consistent with easier retrieval when language and construct align.
• Constructs at opposite ends of the crystallization spectrum—highly labile attitudes (e.g., anti-Obama affect) and very crystallized identities (e.g., partisanship)—do not show these language-linked accessibility patterns.
• Analyses rule out that the observed language effects are primarily driven by changes in self-reported anxiety, anger, pride, or efficacy when bilingual respondents are interviewed in one language versus the other.

Implications for Measurement and Representation

These results imply that interview language can shape both whether respondents retrieve particular political concepts and how strongly they express opinions, with consequences for survey design, interpretation of cross-language comparisons, and the study of Latino public opinion. Survey practitioners should consider language-linked accessibility when translating items, choosing interview language protocols, and interpreting differences across linguistic groups.