Assessing non-response bias: a case study from the 1985 Welsh Heart Health Survey

A number of techniques are used to assess nonresponse bias. This paper presents a case study of one rarely used method, which involved interviewing a sample of non-responder s to the mail-return questionnaire used in the 1985 Welsh Heart Health Survey. The aim of the interviews was to obtain information about a number of key health behaviours and risk factors for heart disease, so that comparisons could be made with responders to the survey. The method proved feasible and showed that the one important difference between the two groups was that more of the non-responders smoked. With the exception of smoking, this study gives greater confidence about the accuracy of data from the 1985 Welsh Heart Health Survey in providing a reliable indication of the prevalence of health behaviours and risk factors in the Welsh population. The findings have more general implications for the use of mail-return surveys in population studies, and the definition of 'acceptable' response rates.