The Methods research area develops and applies innovative qualitative and quantitative methods for public mental health research, with a focus on statistical methods and economic models. These methods, applied across other program areas, are crucial for generating accurate answers to research questions. Faculty in the methods area address complications regarding missing data and non-adherence in randomized trials, as well as teach students how to analyze complex data such as DNA or complex longitudinal data, how to measure and model variables that are not directly observable and how to model the cost and benefit trade-offs of preventive interventions. There are strong links between the methods research area and other groups in the Department, such as the substance use research group, the Center for Prevention and Early Intervention and the Center for the Prevention of Youth Violence.
There are three particular research areas within the Methods area: causal inference, latent variables and measurement, and mixed methods.
- The causal inference area, led by Dr. Elizabeth Stuart, focuses on the development of statistical methods for estimating the effects of exposures, programs, or policies. This includes methods for non-experimental studies, such as estimating the long-term consequences of adolescent drug use or studying state opioid policies, as well as methods for designing and analyzing randomized experiments, such as of school-based preventive interventions, including work on mediation analysis within trials, and assessing the generalizability of trial results to target populations.
- A second area of focus, led by Dr. Rashelle Musci, surrounds measurement, including latent variables and measure harmonization. This includes the development and application of novel latent variable methods and data harmonization tools. With increasing availability of high quality extant data, methodology surrounding data harmonization is becoming increasingly important. The Department of Mental Health is on the front line of this research in regards to mental health with the harmonization of a variety of dataset types (ie., surveillance data, randomized controlled trial data, electronic medical records) for use in answering important questions related to mental health.
- A third area, led by Dr. Joseph Gallo, focuses on how to combine qualitative and quantitative methods, known as mixed methods. Mixed methods research is defined as the collection, analysis, and integration of both quantitative (e.g., RCT outcome) data and qualitative (e.g., observations, interviews) data to provide a more comprehensive understanding of a research problem than might be obtained through quantitative or qualitative research alone. Typical applications of mixed methods in the health sciences involve adding qualitative interviews to follow up on the outcomes of intervention trials, gathering both quantitative and qualitative data to assess patient reactions to a program implemented in a community health setting, or using qualitative data to explain the mechanism of a study correlating behavioral and social factors to specific health outcomes.
The Methods research area also has strong links with other departments and centers in the school. This includes joint appointments with the Department of Biostatistics, as well as links to methods-related groups such as the causal inference and health economics working groups. Student involvement in the Methods area consists of research assistance opportunities, as well as advising by faculty members in statistical and economic methods. Relevant coursework includes term-long and summer institute courses in the Department of Mental Health, such as the Methods seminar, courses in the design of cluster-randomized trials, and a two-term sequence on statistics for psychosocial research. Courses in the Biostatistics department are also relevant, including a causal inference course taught by Dr. Stuart. Many students interested in this program area also do a concurrent MHS in Biostatistics.
Our Work in Action
Faculty work on statistical methods for program and policy evaluation, helping assess the effectiveness of public health interventions. Specific methods questions include how well results from randomized trials carry over to target populations of policy interest, how to estimate the effects of exposures or interventions that can’t be randomized (e.g., childhood environmental exposures), and how to use longitudinal data to estimate the effects of state policies (e.g., opioid prescribing cap laws).
A group of faculty, staff, and students have been working on ways to measure mental health during the COVID-19 pandemic, including partnering with large-scale national and international surveys. Combining established measures of mental health with large-scale data collection activities, including the Understanding America Study out of the University of Southern California, and a Facebook-platform based survey administered through Carnegie Mellon University and the University of Maryland. Those collaborations and data collection efforts have enabled examination of patterns of mental distress during the pandemic in the US and internationally, and how mental distress relates to individual and area characteristics and policies.
Rashelle Musci, PhD, MS, is a methodologist and child mental health expert studying latent variable methods, intergenerational transmission, and impact of prevention programs.
Trang Quynh Nguyen, PhD '14, MHS '14, MS, uses causal inference methods to contribute to sound and effective research on physical and mental health and social justice.
Elizabeth Stuart, PhD, uses statistical methods to help learn about the effects of public health programs and policies, often with a focus on mental health and substance use.
Training and Funding Opportunities
- Data Analytics for Behavioral Health
- Substance Use Epidemiology
- Mental Health Services and Systems
- Psychiatric Epidemiology
- Mixed Methods Research Training Program: This funding opportunity is for early to mid career (PhDs). Funded by NIH. This is not for incoming doctoral students.