Nonresponse Bias in Mail Surveys: The Case of the Department of Defense Post-Service Survey

The bias caused by nonresponse to mailed questionnaires is difficult to handle. The paper examines several approaches to identifying and correcting survey data thus biased. It would appear that neither the direction nor the magnitude of the bias can be known with confidence unless the investigator already has considerable information about the sample population or it is possible to find and interview an adequate sample of those who did not respond. The paper uses questionnaire data collected from former service men to demonstrate the alternative procedures for identifying and removing non-response bias purely by statistical methods.