|
Citation
|
Sokolovsky AW, Janssen T, Barnett NP, Colby SM, Bernstein MH, Hayes KL, Jackson KM. Drug Alcohol Depend. 2018; 193: 83-90. |
|
Affiliation
|
Center for Alcohol and Addiction Studies, Brown University, 121 Main St., Providence, RI, 002903, USA. |
|
Abstract
|
BACKGROUND: Recanting - denying previous reports of lifetime substance use - occurs frequently in longitudinal investigations of adolescent substance use. While research has focused on how individual differences contribute to recanting, intra-individual factors associated with recanting over time remain understudied.
METHODS: Adolescents (n = 1023) were assessed six times between 2009 - 2015. The sample included participants who reported ever-sipping alcohol in at least one assessment who maintained or recanted ever-sipping at the subsequent assessment (n = 543, 53.1% of full sample; 54.5% female; 84.9% white; 89.5% non-Hispanic). The majority (58.6%) of the sample recanted ever-sipping. We fit linear mixed models to investigate whether prospective changes in perceived peer drinking, peer approval, alcohol expectancies, and fear of reprisal predicted recanting. To explore whether mechanisms of recanting differed for delayed (i.e., two assessments or later) recanting, we refit the models in a subset of data excluding immediate (i.e., subsequent assessment) recanters.
RESULTS: Prospective increases in perceived peer drinking (OR = 0.65), peer approval of drinking (OR = 0.82), and positive and negative alcohol expectancies (OR = 0.96; 0.98, respectively) predicted lower odds of recanting. Similar effects were observed among only delayed recanters.
CONCLUSION: Time-varying, intra-individual factors uniquely predicted recanting over time. Although most recanting occurs immediately following the initial report of ever-sipping, the observed effects were consistent between delayed recanters and the sample as a whole. Considering the systematic patterns evident in recanting, researchers should consider using computer-assisted or other research methods that minimize or verify recanting when it occurs while also informing missing data models.
Copyright © 2018 Elsevier B.V. All rights reserved.
Language: en |