Journal of Official Statistics, Vol.18, No.1, 2002. pp. 13–23
Are Nonrespondents to Health Surveys Less Healthy Than Respondents?
G. Cohen and J.C. Duffy
Abstract:In health surveys of the elderly the use of a sampling frame linked to the national registration system, enables comparison of the subsequent mortality of respondents and nonrespondents. To the extent that mortality is an indicator of health at the time of the survey such comparison can inform estimates of the nonresponse bias on self-reported health variables.
We describe a 1993 postal survey of an age-stratified sample of 4,869 people aged 55 and over drawn from the Community Health Index in Lothian Region, Scotland, and a follow-up survey about three years later. Mortality up to mid-1998 was significantly higher among nonrespondents to the first survey. Nonresponse to the follow-up was associated with worse self-reported health three years earlier as well as with indicators of low socioeconomic status. After controlling for self-reported health in 1993, respondents and nonrespondents to the 1996 follow-up differed significantly in mortality over the 18 months after the follow-up survey.
Keywords:Nonresponse bias; mortality; self-reported health; short form 36; limiting long-term illness.
Copyright © Statistics Sweden 1996-2018. Open AccessISSN 0282-423XCreated and Maintained by OKS Group