Existing studies have important methodological limitations that have hindered our ability to develop a useful guideline for effective applications of alternative measurements for daily patterns of drug use and related risk behaviors in a variety of research and clinical settings. The long-term goal of this project is to extract crucial information from daily process data of health risk behaviors to inform prevention and intervention. The objective in this particular application is to develop cost-effective measurement and cutting-edge methodology to collect and analyze daily process data. To accomplish this objective, three specific aims will be pursued:1) Develop cutting-edge statistical methodology and free software to address the missing data issue of prospective measurements including the interactive voice response (IVR) system and text-messaging (TM) and to examine reactivity and validity based on daily patterns of health risk behaviors (drug/alcohol use, violence, and HIV risk behaviors);2) Conduct an experimental study using a high risk sample with a novel design to evaluate compliance, reactivity and validity under different measurements and assessment schedules;and 3) Analyze the 9 waves of data collected through the proposed project and a prior study to build developmental models for at risk youth in order to inform prevention and intervention. The study design is innovative because the methodology development (Aim 1), experimental study (Aim 2), and statistical modeling (Aim 3) are all built upon prospective data from a high risk sample during the critical developmental period for drug use and related risk behaviors (61% African-American, 100% drug users, 83% involving in multiple HIV risk behaviors). The statistical methodology is innovative because we build a bridge between the statistical science and substance abuse field by introducing and extending new statistical concepts and techniques to evaluate psychometric properties of measurements for daily patterns of drug use and related risk behaviors. The proposed research is significant because this is the first study that compares the IVR, TM and the most commonly used retrospective measurement, the timeline follow back (TLFB), for measuring daily patterns of multiple risk behaviors in terms of compliance, reactivity, and validity. The finding of this study can inform prevention scientists and clinicians about effective applications of alternative measurements to assess risk for long-term health problems, develop screening protocols, design intervention facilitating behavioral changes, and evaluate intervention effects longitudinally.
This study will develop cost-effective measurement and cutting-edge methodology with free software that can be used to collect and analyze daily report of drug use and related health risk behaviors including alcohol use, violence, and HIV risk sexual behavior. The proposed work will have a substantial positive influence on public health because we will provide prevention scientists and clinicians with powerful tools to identify at ris youth, target problematic behavior patterns in intervention, and monitor progress of behavior changes over time.
|Yang, Hanyu; Li, Runze; Zucker, Robert A et al. (2016) Two-stage model for time varying effects of zero-inï¬‚ated count longitudinal covariates with applications in health behaviour research. J R Stat Soc Ser C Appl Stat 65:431-444|
|Yang, James J; Li, Jia; Williams, L Keoki et al. (2016) An efficient genome-wide association test for multivariate phenotypes based on the Fisher combination function. BMC Bioinformatics 17:19|
|Yang, Guangren; Yu, Ye; Li, Runze et al. (2016) Feature Screening in Ultrahigh Dimensional Cox's Model. Stat Sin 26:881-901|
|Yang, Songshan; Cranford, James A; Li, Runze et al. (2015) A time-varying effect model for studying gender differences in health behavior. Stat Methods Med Res :|
|Yang, Hanyu; Cranford, James A; Li, Runze et al. (2015) Two-stage model for time-varying effects of discrete longitudinal covariates with applications in analysis of daily process data. Stat Med 34:571-81|
|Dziak, John J; Li, Runze; Zimmerman, Marc A et al. (2014) Time-varying effect models for ordinal responses with applications in substance abuse research. Stat Med 33:5126-37|