Exploring Proxy Response Bias in a Large-Scale Survey of People with Disabilities

Surveys that collect health-related data from people with disabilities often allow proxy respondents to answer questions for sample members who, because of their disabilities, cannot respond for themselves. The rationale for using proxy respondents is to minimize the potential for bias associated with nonresponse. But bias could still be present if the responses provided by the proxies are systematically different from those provided by self-respondents with disabilities.

In this study, we will examine this bias, called proxy response bias, in a large-scale survey of people with disabilities—the National Beneficiary Survey (NBS). Our study has three objectives: (1) to determine if the use of proxy respondents in the NBS is related to the demographic characteristics of the sample members; (2) to examine the size and direction of the differences between proxy and self-reported responses to questions on health-related questions; and (3) to assess whether the proxy-sample member relationship affects the differences between proxy and self-reported responses.