A joint project of the Graduate School, Peabody College, and the Jean & Alexander Heard Library

Title page for ETD etd-03222017-134449


Type of Document Dissertation
Author Hennessy, Emily Alden
Author's Email Address emily.a.hennessy@vanderbilt.edu
URN etd-03222017-134449
Title Adolescent recovery capital and application of exploratory methods
Degree PhD
Department Community Research and Action
Advisory Committee
Advisor Name Title
Emily E. Tanner-Smith Committee Chair
Andrew J. Finch Committee Member
Craig Anne Heflinger Committee Member
Kevin J. Grimm Committee Member
Keywords
  • adolescent
  • addiction
  • data mining
  • recovery
  • recovery capital
  • substance use
Date of Defense 2017-03-17
Availability unrestricted
Abstract
Research suggests that adolescent recovery from substance use disorders is a complex and dynamic process requiring multiple resources at intersecting ecological levels. The recovery capital framework is one model that allows for the modeling of these different resources, but has only been studied among adult populations. Thus, the present dissertation explores the relevance of recovery capital for adolescents and also incorporates a demonstration of exploratory methods for social scientists studying similar complex issues and populations. The first paper presents a latent class analysis to distinguish whether adolescents who are in need of treatment have different patterns of recovery capital. The results suggest that there are five qualitatively distinct classes of recovery capital among this adolescent population and that demographic characteristics are predictive of the type of recovery capital class to which an adolescent belongs. The second paper uses data from an ongoing observational study to address whether recovery capital resources predict attendance at a recovery high school (RHS), one form of community recovery capital, using four different quantitative approaches: logistic regressions, SEARCH, classification trees, and random forests. The results of this study indicate that predictors of RHS attendance are diverse, represent factors in multiple recovery capital domains, and are not necessarily linked to higher levels of recovery capital. Additionally, the different exploratory approaches highlight potential important variable interactions for future research to explore. The final empirical paper uses the dataset in paper two to demonstrate the utility of data mining approaches as compared to traditional logistic regression approaches for covariate selection prior to propensity score estimation. The results suggest that logistic regressions produce the best balance on included covariates, yet the random forest method retains the largest sample and identifies key interactions that are important to include in propensity score estimation. Together, these three studies highlight the applicability of the recovery capital model as an ecological framework specific to addiction and recovery for understanding adolescent recovery processes, while also identifying gaps in the current recovery capital model. In addition, these studies demonstrate both the utility and potential challenges of utilizing exploratory quantitative methods to study complex social science research questions.
Files
  Filename       Size       Approximate Download Time (Hours:Minutes:Seconds) 
 
 28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access 
  Hennessy.pdf 3.80 Mb 00:17:35 00:09:02 00:07:54 00:03:57 00:00:20

Browse All Available ETDs by ( Author | Department )

If you have more questions or technical problems, please Contact LITS.