|
Abstract:
|
Many surveys use administrative records to obtain demographic information and then request permission to obtain detailed information on behaviors from the respondent. A complete response results when both the demographic information and the detailed survey data are obtained. Non-responses arise from non-contact and from refusal to participate. If non-response rate is high and complete responders differ from non-responders, statistics that use only data from complete responders may be severely biased. If complete responders underreport undesirable behaviors in surveys and thus produce false responses, measurement error bias would be resulted. Through the combined asse of paradata, survey data, and biomarker data, this paper describes the extent of the possible biases, factors associated, and implications of adjustment. Data from the Arrestee Drug Abuse Monitoring System are used.
|
