Research

Research from my laboratory largely focuses on further elucidating the mechanisms that contribute to substance use disorders (SUDs), primarily alcohol use disorders (AUDs), across the lifespan. Research from my lab often has a particular focus on individual difference factors in decision-making, as indexed by studies assessing various measures of impulsivity facets using a multi-methodology approach. My research seeks to employ innovative statistical techniques to more precisely model the dynamic relations between AUDs and relevant covariates. Given the important health implications of emerging substance-use technologies (i.e., electronic cigarettes) and policy changes relevant to cannabis, research from my lab has also focused on electronic cigarette use as well as cannabis use among emerging adults. To enhance the clinical impact of my earlier work focusing on dynamic relations between personality and alcohol involvement across emerging adulthood, I extended my work to clinical populations to determine the potential clinical utility of assessing personality features across treatment programs commonly used to mitigate substance-related harms.

Central to this program of research is a focus on the measurement of core constructs in the field of substance use and abuse in order to better understand the functional relations among constructs at various levels of analysis. This has involved expanding my program of research to use multitrait, multimethod frameworks for assessment. This has included multimethod assessments of substance use as well as various impulsivity-like facets within substance-involved samples. In collaboration with my students, I have also engaged in detailed psychometric evaluations of commonly employed measures of impulsigenic traits, including work from my lab focused on the factor structure and measurement invariance of key scales.

I have also expanded my research to determine the extent to which modern analytic approaches, such as machine learning, show marked improvements in classification compared to more traditional approaches (e.g., general linear modeling). I was awarded a grant from the Department of Defense to apply machine learning methodology to enhance the classification of suicidal behaviors by developing risk algorithms. This work suggested that 1) prior work showing marked superiority of machine learning approaches over traditional approaches were based on a flawed validation approach and 2) general linear modeling performed as well as more advanced approaches (e.g., random forest) in classifying suicidal behaviors and outcomes in a large dataset. I am currently in collaboration with large treatment sites for SUDs to further evaluate the value added of machine learning approaches to classification of key outcomes relevant to addiction treatment, including treatment outcomes and treatment completion.