Walters, Scott T.2020-09-022020-09-022019-05https://hdl.handle.net/20.500.12503/29855This three-paper model dissertation investigates issues related to self-reported substance use. Self-report is a less invasive and expensive method of collecting substance use behavior when compared to a toxicological test, but the self-report method has been shown to be unreliable in some populations. We found that self-report missed some use captured by a saliva toxicological test administered to a subsidized housing population enrolled in a technology-assisted health coaching program. Concordance was highest among marijuana users and increased over time. Higher rates of concordance were found when the recall window was expanded from a restricted biological recall window to match the toxicological test to the full 90 day window of the Timeline Follow-Back. Participants who reported using substances more frequently reported having more problems related to their substance use. We also found that both substance use problems and the frequency of consumption of a combined Other category of substances, including cocaine, amphetamine, methamphetamine, opiates, prescription pills, or phencyclidine were predictive of lower quality of life. This dissertation validates previous literature indicating that self-report is a fair to moderately good measure of actual substance use behavior in vulnerable populations that may intentionally or unintentionally misreport their substance use. Programs limited to self-reported measures may consider widening their recall windows to increase accuracy of self-report.application/pdfensubstance useTimeline Follow-backself-reportsubsidized housingSubstance-Related DisordersVulnerable PopulationsHealth PromotionThesis