The answers given by respondents to survey questions arise out of complex cognitive processes which involve more than strategies of recall or the interpretation of specific vocabulary used in questions. Respondents also employ a wide-ranging pattern of social knowledge surrounding the specific content area on which the survey touches. As part of the Census Bureau's on- going attempt to improve coverage in the census and in surveys, a small-scale cognitive study, using thirteen vignettes, was designed to investigate respondents' concepts about residence. This paper describes the use of vignettes in cognitive interviews, as a means of examining respondents' contextual knowledge. The preparation of the vignettes, selected questioning strategies employed during the interview, and analysis of respondents hidden social assumptions are discussed. The vignettes were deliberately ambiguous, ethnographically based, and avoided specific residence terminology. Manipulating hypothetical situations and encouraging respondents to ask for information were productive questioning strategies in revealing the social assumptions which influence responses. These hidden assumptions, about kinship, personal intentions, character, and other aspects of social relationships, often directly mediate respondents' judgments about residence. Such factors are likely to affect actual decisions about which persons to include on survey rosters.
