Schedule

Check here to see what you should be doing before, during, and after each class day.

Overview

Week	Monday	Wednesday	Friday	Announcements
1 (9/2 - 9/6)		First day of class! Foundational ideas	Causal identification	Work on Assignment 1 due Wed 9/11 at 5pm
2 (9/9 - 9/13)	Causal graph fundamentals	Simulating data using causal graphs Assignment 1 due today at 5pm	Simulating data using causal graphs (continued)	Work on Assignment 2 due Fri 9/27 at 5pm
3 (9/16 - 9/20)	Identifying causal effects with causal graphs	Identifying causal effects with causal graphs (continued)	Synthesis day: time to work on Assignment 2	Work on Assignment 2 due Fri 9/27 at 5pm
4 (9/23 - 9/27)	Randomized experiments	Target trial framework	Matching (Part 1) Assignment 2 due today at 5pm
5 (9/30 - 10/4)	Matching (Part 2)	Weighting	Weighting (continued)	Work on Assignment 3 and addressing feedback on Assignment 2. Both due Wed, 10/9.
6 (10/7 - 10/11)	Regression discontinuity designs	Event studies / interrupted time series	Leslie won't be in class today but start thinking about project topics and finding data if pursuing the data analysis option	Work on Assignment 3 and addressing feedback on Assignment 2. Both due Wed, 10/9.
7 (10/14 - 10/18)	Interrupted time series and synthetic control	Synthetic control	Fall Break!
8 (10/21 - 10/25)	Synthesis day	Synthesis day	Fixed effects models	Work on Assignment 4 due Wed 10/30 at 5pm
9 (10/28 - 11/1)	Difference-in-differences (DiD)	DiD continued	Sensitivity analyses for unmeasured confounding	Assignment 4 due Wed 10/30 at 5pm. Work on Assignment 3 revisions due Mon 11/4 at 5pm. Work on Project Milestone 1 due Fri 11/8 at 5pm.
10 (11/4 - 11/8)	Instrumental variables	Project work day	Project work day	Assignment 3 revisions due Mon 11/4 at 5pm. Project Milestone 1 due Fri 11/8 at 5pm.
11 (11/11 - 11/15)	Project work day	Mini-lesson: mediation analysis Slides	Mini-lesson: time-varying treatments Slides	Project Milestone 1 due Fri 11/22 at 5pm.
12 (11/18 - 11/22)	Mini-lesson: causal discovery Slides	Mini-lesson: doubly robust estimation Slides	Mini-lesson: generalizability/transportability Slides	Assignment 5 due Wed 11/20 at 5pm. Project Milestone 2 due Fri 11/22 at 5pm.
13 (11/25 - 11/29)	How do we make the world a better place? How can we make Macalester a better place? How can we make our home departments better communities?	🦃 Thanksgiving Break	🦃 Thanksgiving Break
14 (12/2 - 12/6)
15 (12/9 - 12/13)	Project presentations (day 1 of 3)	Project presentations (day 2 of 3) (last day of classes)
16 (12/16 - 12/20)	Tuesday 12/17 Project presentations (day 3 of 3)			Final project paper due Wed 12/18 at 5pm

Week 1: Foundations

Day 1: Welcome! (9/4)

Before class:

• Get acquainted with our course by reading the syllabus and touring our course website and Moodle page.

During class: Introductions and foundations

After class:

Required

The following chapters from The Effect lay the foundation for asking good questions. They’re written in a fun, conversational style and have some nice humor interspersed throughout.

• The Effect Chapter 1: Designing Research (~10 minutes)
  • Video alternative: Designing Research (~8 minutes)
• The Effect Chapter 2: Research Questions (~30 minutes)
  • Video alternative: Research Questions (~11 minutes)

Optional

If you would like to have an additional reference throughout the course that leans more technical and economics-leaning, I recommend starting Causal Inference: The Mixtape by reading Chapter 1: Introduction (~35 minutes).

• Note: Scott uses the term endogenous in Section 1.3 without defining it. This is an economics term that essentially parallels the term confounding in statistics. That is, an endogenous variable is (often) one that is a confounder.

Day 2: Causal identification (9/6)

Before class:

Required

• The Effect Chapter 5: Identification (~50 minutes)
  • Video alternatives:
    • Identification: Data Generating Processes (~8 minutes)
    • Identification: Alternative Explanations (~9 minutes)
    • Identification: Alcohol and Mortality (~9 minutes)

Week 2: Causal graphs

Day 1: Causal graph fundamentals (9/9)

Before class:

Required

• The Effect Chapter 6: Causal Diagrams (~40 min)
  • Video alternatives:
    • Causal Diagrams: Causality (~9 minutes)
    • Causal Diagrams (~12 minutes)
• The Effect Chapter 7: Drawing Causal Diagrams (~30 min)
  • Video alternatives:
    • Drawing Causal Diagrams (~13 minutes)
    • Simplifying Causal Diagrams (~12 minutes)

Formulate a research question that relates to an area that you’re interested in. In class, we will be using the principles in Chapters 6 and 7 to draw a causal diagram that addresses this research question.

Optional

• Mixtape Chapter 3: Directed Acyclic Graphs but skip 3.1.3 Backdoor Criterion (~40 minutes)

Day 2: Simulating data using causal graphs (9/11)

Before class:

Required

• Video: Key structures in causal graphs (~12 min) (slides)
  • Note 1: In this video, I refer to a concept called “exchangeability”. This is a concept that I included in the last offering of this course. Exchangeability is a causal identification assumption. When the exchangeability assumption is satisfied, we can identify causal effects. When it is not satisfied, I call this “a lack of exchangeability”. In the language we have used, this happens when there are alternate (noncausal) explanations for a relationship between two variables. So in the video, whenever you hear “creates a lack of exchangeability”, replace that with “leads to the presence of alternate (noncausal) explanations for the relationship between two variables”. When you hear “achieve conditional exchangeability”, replace that with “we are able to address this alternate explanation when analyzing the data”.
  • Note 2: I use the formal probability ideas of marginal and conditional dependence and independence. You can get the essential ideas from this video by replacing terms in the list below. (If you want to learn or review these probability ideas (not required!), watch my Probability Essentials video.)
    • “A and B are marginally [independent/dependent]” = “A and B are [unrelated/related] in a general population”
    • “A and B are conditionally independent” = “A and B are [unrelated/related] in subgroups (in each group defined by one or more variables)”
• 2 readings from Andrew Heiss’s Program Evaluation course:
  • Generating random numbers (~35 min)
  • The ultimate guide to generating synthetic data for causal inference (Stop when you get to the subsection “Visualizing one variable”) (~25 min)

During class: Simulating data using causal graphs

Day 3: Simulating data using causal graphs (9/13)

• Continuation of last class

Week 3: Causal graph wrap-up

Day 1: Identifying causal effects with causal graphs (9/16)

Before class:

Required

• Video: Causal and Noncausal Paths (~5 min)
• Video: D-Separation (~8 min)
• Reading: The Effect Chapter 8: Causal Paths and Closing Back Doors (~35 minutes)
  • Video alternatives:
    • Causal Pathways (~9 minutes)
    • Closing Causal Pathways, and Collider Variables (~11 minutes)
    • Testing Causal Diagrams, and Placebo Tests (~8 minutes)

Optional

• Mixtape Section 3.1.3: Backdoor criterion (~5 minutes)

During class: Identifying causal effects with causal graphs

Day 2: Identifying causal effects with causal graphs (9/18)

• Continuation of previous class

During class: Identifying causal effects with causal graphs and Testing causal graphs

Day 3: Synthesis day (9/20)

• Pause day to work on Assignment 2.
• Please create a rough outline of your slides before class.
• We will have a chance to get feedback on the outline, work on the slides, and then get additional feedback on the slide content.

Week 4: Randomized experiments, target trials, matching

Day 1: Randomized experiments (9/23)

Before class: No required reading or videos for today.

Optional

• Reading: Treatment Allocation and Randomization (~20 minutes) from Penn State’s online notes for the course Design and Analysis of Clinical Trials

During class: Randomized experiments

Day 2: Target trial framework (9/25)

Required

• Reading: Hernán, M. A., Sauer, B. C., Hernández-Díaz, S., Platt, R., Shrier, I. (2016). Specifying a target trial prevents immortal time bias and other self-inflicted injuries in observational analyses. Journal of Clinical Epidemiology, 79, 70–75. (~35 minutes)

During class: Target trial framework

Day 3: Matching (9/27)

Before class:

Required

• The Effect Chapter 14: Matching (~160 minutes)
  • Video alternatives:
    • Matching (~11 minutes)
    • Five Questions About Matching (~13 minutes)
    • Distance Matching (~9 minutes)
    • Propensity Score Matching (~10 minutes)
    • Coarsened Exact Matching and Entropy Balancing (~8 minutes)
    • When Matching Goes Wrong (~14 minutes)
    • Treatment Effect Estimation with Matching (~7 minutes)

Optional

• Mixtape Chapter 5: Matching and Subclassification (~140 minutes)
• Journal article: Matching Methods for Causal Inference: A Review and a Look Forward (~60 minutes)

During class: Matching (Part 1)

Week 5: Matching and weighting

Day 1: Matching (9/30)

• Continuation of previous class

During class: Matching (Part 2)

Day 2: Weighting (10/2)

Before class:

Required

• Video: Inverse Probability Weighting (Time: 10:24) (Slides)

During class: Weighting

Day 3: Weighting (10/4)

• Continuation of previous class

During class: Weighting

Week 6: Regression discontinuity and longitudinal/panel data

Day 1: Regression discontinuity (10/7)

Before class:

Required

• The Effect Chapter 20: Regression Discontinuity (~130 minutes)
  • Video alternatives:
    • Regression Discontinuity (~10 minutes)
    • Estimating Regression Discontinuity (~11 minutes)
    • Adjustments to Regression Discontinuity (~8 minutes)
    • Checking Regression Discontinuity Assumptions (~11 minutes)

Optional

• Mixtape Chapter 6: Regression Discontinuity (~170 minutes)

During class: Regression discontinuity designs

Day 2: Event studies / interrupted times series (10/9)

Before class:

Required

• The Effect Chapter 17: Event Studies (~70 minutes)
  • Video alternatives:
    • Event Studies (~12 minutes)
    • Interrupted Time Series (~8 minutes)
    • Event Studies in Finance (~11 minutes)

During class: Event studies / interrupted time series

Day 3: Flex Day (10/11)

No class today because I will be attending a workshop.

On your own: Start exploring options for the course project.

Week 7: Synthetic control

Day 1: Synthetic control (10/14)

Before class:

Required

• The Effect Section 21.2.1: Synthetic Control (~10 minutes)
  • Video alternatives:
    • Other Standard Research Designs (~5 minutes) (Just the part on the synthetic control method)

Optional

• Mixtape Chapter 10: Synthetic Control (~70 minutes)

During class: Interrupted time series and synthetic control

Day 2: Synthetic control (10/16)

During class: Interrupted time series and synthetic control

---

🍁 Fall Break: Thursday, October 17 - Sunday, October 20 🍁

🍁 No class on Friday, October 18 🍁

Week 8: Fixed effects models

Day 1: Synthesis day (10/21)

During class: Working on Assignment 4

Day 2: Synthesis day (10/23)

During class: Working on Assignment 4

Day 3: Fixed effects models (10/25)

Before class:

Required

• The Effect Chapter 16: Fixed Effects (~70 minutes)
  • Video alternatives:
    • Fixed Effects (~12 minutes)
    • Estimating Fixed Effects (~8 minutes)
    • Random Effects (~6 minutes)

During class: Fixed effects models

Week 9: Diff-in-diff, sensitivity analyses

Day 1: Difference-in-differences designs (10/28)

Before class:

Required

• The Effect Chapter 18: Difference-in-Differences (Section 18.1: How Does it Work? and Section 18.2: How Is It Performed?) (~60 minutes)
  • Video alternatives:
    • Difference in Differences (~11 minutes)
    • Parallel Trends (~9 minutes)
    • Estimating Difference in Differences (~10 minutes)
    • Supporting Parallel Trends (~9 minutes)

Optional

• Mixtape Chapter 9: Difference-in-Differences (~230 minutes)

During class: Difference-in-Differences

Day 2: DiD continued (10/30)

Before class:

Required

• The Effect Chapter 18: Difference-in-Differences (Section 18.3: How the Pros Do It) (~20 minutes)
  • Video alternatives:
    • Dynamic Difference-in-Differences (~10 minutes)
    • Staggered Treatment in Difference-in-Differences (~9 minutes)

During class: Difference-in-Differences

Day 3: Sensitivity analyses (11/1)

Before class: No required reading or videos for today.

During class: Sensitivity Analyses

Week 10: Instrumental variables and project work

Day 1: Instrumental variables (11/4)

Before class:

Required

• The Effect Chapter 19: Instrumental Variables (~85 minutes)
  • Video alternatives:
    • Instrumental Variables (~10 minutes)
    • Instrumental Variable Validity (~10 minutes)
    • Estimating Instrumental Variables (~12 minutes)
    • Testing Instrumental Variables Assumptions (~11 minutes)

Optional

• Mixtape Chapter 7: Instrumental Variables (~150 minutes)

During class: Instrumental Variables

Weeks 11-14: Assorted topics, project work time

Potential topics

• Doubly robust estimation
• Causal discovery
• Interference
• Generalizability/transportability
• Machine learning and causal inference
• Time-varying treatments framework
• Dynamic treatment regimes
• Mediation analysis

🦃 Thanksgiving Break: Wednesday, November 27 - Sunday, December 1 🦃

Week 15: Project presentations

Day 1: TBD

Before class:

Required

• (~xx minutes)

Optional

• (~xx)

Day 2: Last day of class!

Before class:

Required

• (~xx minutes)

Optional

• (~xx)

During class: