The Effect: An Introduction to Research Design and Causality Book Club
Introduction
Book club meetings
Pace
Icebreaker
git and GitHub
Book Design
Levels of Understanding
1
Designing Research
Main Goal
1.1
I Have a Question
1.2
Empirical Research
1.3
Why Research Needs a Design
1.4
In This Book
Meeting Videos
Cohort 1
2
Research Questions
2.1
What is a Research Question?
2.2
Why Start with a Question?
2.3
Where Do Research Questions Come From?
2.4
How Do You Know if You’ve Got a Good One?
Actionable
Meeting Videos
Cohort 1
3
Describing Variables
3.1
Descriptions of Variables
School
Scorecard
3.2
Types of Variables
3.3
The Distribution
3.3.1
Categorical Variables
3.3.2
Histograms
3.3.3
Density Plot
3.4
Summarizing the Distribution
3.4.1
Centrality
3.4.2
Percentiles
Variation
Standard Deviation
Interquartile Range
Skew
3.5
Theoretical Distributions
NHST
Null Hypothesis Significance Testing
Meeting Videos
Cohort 1
4
Describing Relationships
4.1
Relationships
4.2
Conditional distributions
4.3
Conditional means
4.4
Line-fitting/regression
4.5
Conditional Conditional Means (not a typo) AKA using controls
Meeting Videos
Cohort 1
5
Identification
5.1
The Data Generating Process
Introduction
Two Parts of DGPs
Hair Color Example
Two core ideas
5.2
Where’s Your Variation?
DGP and Isolating Variation
Avocado DGP
5.3
Identification
Example of Family Dog, Rex, Escaping House
Identification Process
Identification in Practice
5.4
Alcohol and Mortality
Overview
Study Findings
Book Exercises
Example Answers
Issues with Causal Explanation
Controls in Study to Address Alternate Explanations
Lingering Issues
5.5
Context and Omniscience
Meeting Videos
Cohort 1
6
Causal Diagrams
6.1
Causality
6.2
Causal Diagrams
Coin-Cake Example
Unobserved Variables
6.3
The Real World
6.4
Research Questions
6.5
Moderators
Meeting Videos
Cohort 1
7
Drawing Causal Diagrams
7.1
Our Idea of the World
7.2
Thinking through the DGP
First Draft DAG
7.3
Simplify
Revised DAG
7.4
Avoiding Cycles
7.5
Assumptions
Meeting Videos
Cohort 1
8
Causal Paths and Closing Back Doors
8.1
I Walk the Line
8.2
Any Way You Like It
Wine and Lifespan Example
8.3
Path types
8.4
Closing paths
8.5
Colliders
8.6
Using Paths to Test the DAG
Placebo Test Example
Interpretation of Placebo Tests
Meeting Videos
Cohort 1
9
Finding Front Doors
9.1
Looking Ahead
9.2
Trying to Push a String
Charter School Example
9.3
What the World Can Do For Us
9.4
Lottery-Bankruptcy Example
Air Pollution Example
Medicare Example
9.5
Riding a Shooting Star
Meeting Videos
Cohort 1
10
Treatment Effects
10.1
For Whom the Effect Holds
10.2
Different Averages
10.3
I Just Want an ATE
10.4
Who Cares?
Meeting Videos
Cohort 1
11
Causality with Less Modeling
11.1
Confidence
11.2
Wide Open Spaces
Control Groups
11.3
But Am I That Wrong?
Meeting Videos
Cohort 1
12
Opening the Toolbox
12.1
Methods that we’ll be checking out
12.2
Structure of upcoming chapters
Meeting Videos
Cohort 1
13
Regression
13.1
Basics
13.2
Error terms
13.3
Regression assumptions
13.4
Sampling variation
13.5
Hypothesis testing in OLS
13.6
Mantras about hypothesis testing
13.7
Regression tables
13.7.1
Interpretation
13.7.2
Controls
13.8
Subscripts in regression equations
13.9
DAG to Regression
13.10
Getting fancier
13.11
Binary/discrete variables
13.12
Polynomials
13.13
Variable transformation
13.13.1
Options
13.13.2
Interpretation of log
13.14
Interaction terms
13.15
Nonlinear regressions
13.15.1
good link functions
13.15.2
Interpretation
13.16
Standard errors
13.16.1
Assumptions
13.16.2
Fixes (mostly sandwich estimators)
13.16.3
Bootstrapping
13.17
Sample Weights
13.18
Collinearity
13.19
Measurement error
13.20
Penalized regression
Meeting Videos
Cohort 1
14
Matching
SLIDE 1
Meeting Videos
Cohort 1
15
Simulation
SLIDE 1
Meeting Videos
Cohort 1
16
Fixed Effects
SLIDE 1
Meeting Videos
Cohort 1
17
Event Studies
17.1
How does it work?
17.2
Prediction and Deviation
17.3
Terminology
17.4
How it’s used in Finance
17.5
How it’s used with regressions
17.5.1
Example: Improved ambulance care
17.6
How it’s used when taking time series seriously
17.7
Forecasting with Time Series Models
17.8
Joint tests
Meeting Videos
Cohort 1
18
Difference-in-Differences
SLIDE 1
Meeting Videos
Cohort 1
19
Instrumental Variables
19.1
How does it work?
19.2
Assumptions
19.3
Canonical designs
19.4
Instrumental Variables estimator
19.5
Example: Insurance takeup
19.5.1
2SLS
19.5.2
GMM
19.6
IV and treatment effects
19.7
Checking IV assumptions
19.8
How the Pros do it
19.9
Don’t just test for weakness, fix it
19.10
Way past LATE
19.11
Nonlinear IV
19.12
Validity violation
Meeting Videos
Cohort 1
20
Regression Discontinuity
SLIDE 1
Meeting Videos
Cohort 1
21
A Gallery of Rogues: Other Methods
SLIDE 1
Meeting Videos
Cohort 1
22
Under the Rug
SLIDE 1
Meeting Videos
Cohort 1
Published with bookdown
The Effect: An Introduction to Research Design and Causality Book Club
11.1
Confidence
Draw a causal diagram (DAG) based on data generating process (DGP)
List out all paths from treatment to outcome
Close bad paths
But what if our causal diagram doesn’t represent the data?