| | Cohort studies often conduct periodic surveys to determine disease outcomes and update exposure and other information. Ignoring individuals who do not repond to one or more surveys in analyses can lead to biased results if responders and non-responders differ according to important characteristics associated with outcomes of interest. We describe in this paper a simple weighting method that can be used to adjust for potential non-response bias in studies that have baseline information available on everyone.
|
U.S. Radiologic Technologists Study
