Causal Mediation Analysis
Master causal mediation analysis in a rigorous three-day boot camp. Engage in seminars and hands-on sessions to explore mediating mechanisms.
Modules/Weeks
Weekly Effort
Discipline
School
Format
Cost
Course Description
The Causal Mediation Analysis Training is a three-day intensive boot camp of seminars and hands-on analytical sessions to provide an overview of concepts and data analysis methods used to investigate mediating mechanisms.
- Identify limitations of traditional mediation methods: Understand scenarios where standard approaches fail to capture mediation effects accurately.
- Analyze mediation under the counterfactual framework: Articulate concepts and assumptions necessary for identifying mediating mechanisms.
- Apply regression techniques for single and multiple mediators: Formulate and implement regression-based approaches to assess mediation effects effectively.
- Utilize software for mediation analysis: Gain proficiency in using statistical software for mediation analysis and interpreting results for practical applications.
To contact support for this course, please email [email protected].
Course Prerequisites
Investigators from any institution and from all career stages are welcome to attend, and we particularly encourage trainees and early-stage investigators to participate. There are four requirements to attend this training:
- Each participant must be familiar with linear and logistic regression.
- Each participant must have experience with programming in R.
- Although the instructors will provide an overview of the fundamentals of causal inference (potential outcomes, directed acyclic graphs, and marginal structural models), we invite the participants to read chapters 1-7, 11, and 12 of Hernán MA, Robins JM (2020). Causal Inference: What If. Boca Raton: Chapman & Hall/CRC (free).
- Each participant is required to have a personal laptop/computer and a free, basic RStudio Cloud account. All lab sessions will be done using RStudio Cloud (now known as Posit).
What You Will Learn
Mediation analysis is an emerging field in causal inference relevant for comparative effectiveness research, evaluating and improving policy recommendations, and explaining biological mechanisms. Training in the potential outcomes framework for causal inference is important to understand the assumptions required for valid mediation analyses. This course will equip participants with foundational concepts and cutting edge statistical tools to investigate mediating mechanisms.
This three-day intensive course will cover some of the recent developments in causal mediation analysis and provide practical tools to implement these techniques and assess the mechanisms and pathways by which causal effects operate. Led by a team of experts in causal mediation techniques at Columbia University, this course will integrate lectures and discussion with hands-on computer lab sessions using R. The course will cover the relationship between traditional methods for mediation in environmental health, epidemiology, and the social sciences and new methods in causal inference using a wide variety of examples to illustrate the techniques and approaches. We will discuss 1) when the standard approaches to mediation analysis are valid for dichotomous, and continuous, outcomes, 2) alternative mediation analysis techniques when the standard approaches will not work, using ideas from causal inference and natural direct and indirect effects 3) the no-unmeasured confounding assumptions needed to identify these effects, and 4) how regression approaches for mediation analysis can be extended in the presence of multiple mediators.
By the end of the workshop, participants will be able to:
- Understand when traditional methods for mediation fail.
- Articulate concepts about mediation under the counterfactual framework and assumptions for identification.
- Formulate and apply regression approaches for mediation for single and multiple mediators.
- Develop facility with the use of software for mediation and interpretation of software output.
Instructors
Linda Valeri is an Assistant Professor of Biostatistics at Columbia University Mailman School of Public Health. Dr. Valeri is an expert in causal inference with a focus on statistical methods and computational tools for causal mediation analysis, measurement error, and missing data. She is currently working on methods for mediation analysis with high-dimensional exposures, as well as intensive longitudinal and time-to-event data on mediators in the presence of competing risks. She is interested in translating statistical methods in public health and precision medicine to improve our understanding of mental health, environmental determinants of health, and health disparities. Dr. Valeri is also a passionate teacher. In the past ten years she has been teaching full semester as well as short courses on causal mediation analysis at premier academic institutions such as Columbia University, Harvard University, University of Michigan, Erasmus Universiteit Rotterdam (Netherlands), Universite’ de Bordeaux (France), and University of Milan (Italy).