Ever pondered whether to include a variable in an analysis – is the variable necessary, will its inclusion/exclusion introduce bias in the analysis?  Why is the variable important anyways – is it a mediator, a moderator and a confounder some combination of these???

This course introduces how to incorporate DAGs (Directed Acyclic Graphs) in analysis planning to help think through what variables to collect and use (or refrain from using) in an analysis.

Topics include:

• Introducing the structure of DAGs and the rules for using them to guide variable selection
• A deep dive into the types of variable relations captured in DAGs with emphasis on mediation, moderation, effect modification, confounding and colliders and their potential to introduce bias in analysis
• Hands on example of how to create and evaluate DAGs in R using the package daggity

Use of DAGs may help inform data collection and analytical plans in ways that enhance reproducibility.

Course prerequisites: None. Knowledge of regression analysis and R is helpful but not required.

Registration and fee (refundable for UNCCH students) are required for Zoom link access.