Fate of incorrectly addressed mailed questionnaires.

Mail surveys are a widely used and often inexpensive method for conducting epidemiologic studies, but may suffer from poor response rates. Some of this nonresponse may be due to letters that never reach study subjects rather than nonparticipation. We mailed letters to invalid addresses and to fictitious occupants at valid addresses in order to determine the proportion of letters that were not returned to us by the postal service as undeliverable. All letters mailed to invalid addresses were returned, whereas 13.3% (95% Cl 2.0-24.7%) of letters sent to fictitious occupants were not. Because some undeliverable mail is not returned, investigators may erroneously assume nonresponse for some subjects who never have the opportunity to participate.