Wells, ElizabethHarrop, Erin Nicole2014-10-132014-10-132014-10-132014Harrop_washington_0250O_13330.pdfhttp://hdl.handle.net/1773/26249Thesis (Master's)--University of Washington, 2014BACKGROUND: Participant attrition threatens validity in longitudinal research. Non-random attrition creates "attrition bias," that could compromise result generalizability. This study explores the impact of attrition on a longitudinal substance use study. METHODS: Easy-to-contact (ETC) participants, difficult-to-contact (DTC) participants, and study non-completers were compared on demographic characteristics. Following this, outcome variables for ETC and DTC participants were compared, using DTC participants as theoretical proxies for non-completers. RESULTS: DTC participants and non-completers differed on few demographic characteristics. However, DTC participants were more likely to have used a substance during the follow-up period compared to ETC participants. DISCUSSION: These results suggest that attrition bias may result in lower reports of relapse. This bias may not be a major threat to validity in etiological studies with at least 65% retention. However, additional testing for attrition bias in efficacy studies is important, as differential attrition by condition (particularly among substance users) could threaten conclusion validity.application/pdfen-USCopyright is held by the individual authors.attrition; attrition bias; Follow-up; Follow-up difficulty; retention; substance useSocial workPsychologyQuantitative psychology and psychometricssocial work - seattleThesis