Conferences

Proxy respondents often are used in epidemiologic and health survey research when the subjects of interest (i.e., index or direct respondents) can not give information by themselves. Including proxy respondents increases sample size and study power. These benefits must be weighed against possible sources of measurement errors which can undermine the validity of survey questionnaire. The accuracy of proxy-derived versus self-derived data has received attention in the literature; less studied has been possible differences between proxy and self respondent response process.

The objective of this study was to assess the proxy respondent information for a wide range of variables common in disability research and to explore the response strategies of both types of informants. We compared data gathered from proxy respondents with self-respondents by means of cognitive interviewing. In this sense, we analysed sources of error in the response process for questions of a national disability survey. The questionnaire includes questions related to different limitations about daily life activities. 13 direct informants and 13 proxies answered to 11 “target” questions about the same content for both types of informants. Data were analysed by means AQUAD (Huber, 2008). Results will be discussed in order to the level of agreement between both type of informant, interviewer errors identifying potential disabled people and deviations to expected interpretation of key concepts for each target questions.