• postal questionnaires
• general practice
• non-response bias

Postal surveys have been used for many years by researchers to obtain information about the attitudes, knowledge and self reported behaviour of general practitioners. It is widely believed that poor response rates undermine the validity of survey research,1 2 but only a few studies have tackled the methodological problem of whether a low response rate matters.3

Non-responders to general practitioner postal surveys have been characterised, as being older, isolated and less well qualified,4 however these results relate to individual questionnaires and univariable analyses. To the best of our knowledge no studies have examined the characteristics of GPs who routinely do not return postal questionnaires. These GPs could form an identifiable and sizable proportion of general practice and may be an important source of non-response bias.