Abstract
The bias resulting from nonresponse causes problems for all epidemiological research. This study investigated the characteristics of those who refused to participate in a community survey of female psychiatric disorder. The demographic characteristics of refusers were similar to those previously reported. When compared to responders, those who refused tended to be older, never married and not widowed, of lower socioeconomic status and urban dwellers. No differences in physical illness requiring hospitalization or inpatient and outpatient psychiatric illness, as assessed through hospital records, were discerned between refusers and responders. Reasons for refusal are discussed. The bias problem may get worse as changing community attitudes to privacy and patients' rights lead to greater refusal rates.