Schedule

In general, on Mondays we will have brief lectures and demonstrations in R, while on Wednesdays, we will have student-led paper discussions and/or replication exercises of published papers. In the latter half of the course, we will shift to work-shopping our own projects and discussing additional issues and considerations with applied causal inference. In this way, we will use a partially flipped classroom and strive to create a collaborative and inclusive classroom environment to discuss, ask questions, collaborate, and get feedback on your analyses.

The tentative schedule of content, subject to change based on student interests and availability of guest speakers, is as follows:

Monday Wednesday Goals/Topics
Week 1 (1/13) Course Intro Intro to causal inference and counterfactual causality • Intro to course and goals
• Understand students’ goals for the course
• Set norms and expectations
• Intro to concepts in causal inference and motivation for applying it to ecology and evolution
• Clarify causal versus other research questions and aims
• Brief Introduction to two main frameworks for counterfactual causality (potential outcomes and graphical causal modeling frameworks)
Week 2 (1/20) No class - MLK Day

Intro to the main frameworks for counterfactual causal inference

Guest speaker: Dr. Suchinta Arif

• Structural Causal Models and Directed Acyclic Graphs
Week 3 (1/27) Intro to the main frameworks for counterfactual causal inference Randomized Controlled Experiments (or RCTs) and experimental design • Potential outcomes framework
• Application of potential outcomes framework to RCTs
• Review key assumptions of RCTs
• Critique experimental designs in ecology and identify solutions
• Dissect experimental design with respect to assumptions required for causal inference
Week 4 (2/3) Observational data and counterfactuals; Introduction to quasi-experimental methods

Creating and Analyzing DAGs

(Guest speaker: [Dr. Zach Laubach](https://laubach.github.io))

Due: first draft of research question, or lit review proposal, and DAG analysis on 2/7

• Articulate how and why observational data deviates from assumptions of RCTs
• Understand challenges of applying causal inference to observational data, including confounding and selection bias
• DAGs as a tool from the graphical causal modeling framework
• Principles of covariate and confounder selection

Project/Workshopping:

• Students create and workshop DAGs for their own research/study systems

Week 5 (2/10) Pre-regression matching Paper discussion and replication exercise • Learn pre-regression matching as a method
• Demo application in R
• See & critique how it is applied in the literature
Week 6 (2/17) Difference in Difference (DiD) Paper discussion and replication exercise • Learn DiD designs, including identification assumptions and interpretation
• Demo application in R
• Compare DiD to matching and experiments
• See & critique how it is applied in the literature
Week 7 (2/24) Panel methods continued: Two way fixed effects and extensions

Paper discussion and replication exercise - Class on zoom

Due Draft of DAG OR literature review proposal (2 page max.)

• Learn within estimators (two-way fixed effects) including identification assumptions and interpretation
• Compare panel designs with conditioning on observables designs and with random effect/mixed effect models in R and understand the differences in assumptions
• Compare applications in literature and the assumptions required for the conclusions drawn
Week 8 (3/3) Synthetic Control (Asia Kaiser)

Extensions of diff-in-diff and fixed effects

(Dr. Jarrett Brynes)

On zoom

• Learn synthetic control including identification assumptions and interpretation
• Compare it to other types of designs
Week 9 (3/10) Instrumental Variables (IV) IV Paper discussion • Learn IV as a method
• Demo application in R
• See & critique how it is applied in the literature
Week 10 (3/17) One-on-one consultations on projects on zoom RegressionDiscontinuity Designs (RDD)

• Make progress on project and get feedback on proposed research design and challenges so far from Laura and your classmates

• Learn RDD as a method

Week 11 (3/24) Spring break (no class) Spring break (no class)
Week 12 (3/31) RDD Paper discussion DUE: In class - brief presentation on project

• Demo application in R
• See & critique how it is applied in the literature

• Make progress on project and get feedback on proposed research design and challenges so far from Laura and your classmates

Week 13 (4/7) Project presentations continued; Intro to robustness checks Comparison of study designs cont’d; Sensitivity tests • Articulate, compare, and parse the assumptions, estimands, and generalizability of different designs
• Intro to Sensitivity Tests
Week 14 (4/14) Mediation Analysis Power, Inference, and Robust and Clustered Standard Errors • Introduction to mediation analysis

Articulate additional assumptions for mediation •
• Standard errors, power, and inference with quasi-experimental methods
Week 15 (4/21) Special topic: Generalizability (Guest speaker: Dr. Becks Spake) Project presentations


Introduction to challenges with generalizability


Crititically consider Generalizability, transportability, and approaches to increase them
• Clearly communicate the application of causal inference methods to a research question in ecology OR present a topic we did not cover as a class

Week 16 (4/28) Project presentations May 4th: Final Write up due by email • Clearly communicate the application of causal inference methods to a research question in ecology OR present a topic we did not cover as a class

Potential additional topics/units to be voted on by the class:

Generalizability in experimental and observational studies

  1. Korell et al. 2019
  2. Spake et al. 2022
  3. Spake et al. 2021

Heterogeneous treatment effects and conditional average effects

e.g. Causal Forests

Deeper dive into mechanisms and mediation analysis

Deeper dive into Sensitivity Tests

Replication/Reproducibility/Pre-registration

e.g. Kimmel et al. 2023 Empirical evidence of widespread exaggeration bias and selective reporting in ecology