mindmap
root(Course in Causal Inference)
{{BLOCK 3:<br/>Quasi experimentation}}
{{BLOCK 2:<br/>Experimentation}}
{{BLOCK 4:<br/>Observational studies}}
{{BLOCK 1:<br/>Introduction to DAGs}}
Incorporating Causal Inference in Statistics Courses
1. The team
- G. Alexi Rodríguez-Arelis (UBC)
- Philippe Boileau (McGill University)
2. Why is the topic important?
“The goal of causal inference is to estimate the effect in a population of intervening on one variable, the treatment, on another variable, the outcome.” (Cummiskey et al. 2020)
- In introductory courses, we must go beyond the usual warning “correlation does not imply causation”
- What is the underlying process that is generating the data?
2. Why is the topic important?
- Carver et al. (2016) indicate that “students should understand that statistics is a problem-solving and decision making process that is fundamental to scientific inquiry and essential for making sound decisions”
- Most importantly, when drawing causal conclusions on observational data, students should be skeptical (Cummiskey et al. 2020)
The Directed Acyclic Graph (DAG)
- DAGs are handy tools to depict causality
- Statisticians and subject-matter experts can work together to build them!
DAGs from Cummiskey et al. (2020).
4. Potential overall structure of course content
- There are four blocks that start with an introduction to DAGs
- Then, we go from full control to no control on the study treatments
- A stronger emphasis on data science-related applications such as A/B testing
A mind map going clockwise!
5. Challenges ahead
- Is there room within the course to include Bayesian thinking?
- How smooth would it be to introduce the fundamentals of causality via DAGs and Bayesian networks? (Lu, Zheng, and and 2023)
5. Challenges ahead
- What are the pros and cons of having regression analysis as a prerequisite?
- How theoretical should we conceived the course? Lübke et al. (2020) provide a fair simulation-based starting via generative modelling and regression analysis
5. Challenges ahead
- Using different textbooks to develop course material such as:
- “Causal Inference: What If” by Hernán (2024), a review on causal inference with (i.e., parametric) or without (i.e., non-parametric) models
- “A First Course in Causal Inference” by Ding (2024), a review on randomized and observational studies
- “Causal Inference: The Mixtape” by Cunningham (2021), which provides insights on regression discontinuity for quasi-experimentation
Questions?
