Context effects in national health surveys: effects of preceding questions on reporting serious difficulty seeing and legal blindness.

The first task that a respondent faces in a survey interview is to arrive at the intended interpretation of the survey question. Although the wording of the question may seem unambiguous, its pragmatic interpretation may nevertheless be ambiguous. People are concerned with this pragmatic meaning of questions rather than with their literal meaning (see Clark and Schober 1991; Grice 1975; Schwarz 1994, 1996, 1999; Suchman and Jordan 1990; Todorov 1997). To arrive at the intended interpretation of a specific question, respondents may rely on different cues in the context. Among the most potent cues in a survey context are the preceding questions (Bishop, Oldendick, and Tuchfarber 1984; Martin, DeMaio, and Campanelli 1990; Strack 1992; Strack, Schwarz, and Wänke 1991; Tourangeau and Rasinski 1988; Tourangeau et al. 1989). The context effects of preceding questions are stronger for general than for specific questions (Schuman and Presser 1981). Presumably, the intended referent of general questions is more susceptible to alternative interpretations and, hence, to context effects. For instance, “Do you have serious difficulty seeing, even when wearing glasses or contact lenses?” can have different interpretations if preceding questions covered medical conditions related to vision than if they covered conditions unrelated to vision. Respondents may be more likely to infer that serious difficulty seeing involves a medical vision condition in the former than in the latter context.