Economics 104 Winter Quarter 2019
Instructor: Carlos Dobkin
Office Hours: Monday 2:00-3:00 (E2 Room 441)
This class is focused on how to use regression to perform causal inference and to present and describe these results clearly.
We will cover the analysis of Randomized Control Trial (RCT) which is the gold standard for causal inference. We will then examine four
other approaches: regression, instrumental variables, regression discontinuity, and difference in differences. Class attendance is
required. I will cold call people in class and have a short quiz at the end of most classes. The grade is based on participation (5 percent), in class quizzes (15 percent),
write ups and homework (25 percent), and two papers (55 percent). You may use any statistical package you want to conduct the analysis for the homework
and papers but only Stata will be supported in section. You can purchase Stata (DO NOT GET SMALL STATA) at
Stata Grad Plan.
The sections will be focused on teaching you the statistical programming.
You will use R or Stata to do the statistical analysis for the Homework. Please install it on your laptop before the first section.
You can download R for free by going to the this website and clicking download R. Once it is installed
you should install RStudio.
The sections will be taught focusing on Stata
The textbook which you can purchase on Amazon is "Mastering 'Metrics: The Path from Cause to Effect"
Frequently Asked Questions
- What do you want for the write up of the journal articles? I would like 2-3 paragraphs that convey the central message of the paper.
- Why can't I download the papers from the reading list below? You need to be on a campus IP address to access the papers as they are from sites UCSC pays for.
- What is the face sheet I need to turn in with a revised paper? It is one page of bullet points describing all the major and minor fixes you made to the paper in
the revision. You should fix and note how you fixed both the issues flagged in the comments on your original paper and anything else that you noticed needed work
such as narrative flow, grammar or details of your argument.
- What version of Stata should I get Either Stata IC or SE will be sufficient for the class. Small Stata will not be sufficient as it only handles 1,200 observations.
Teaching Assistants Office Hours
- Dan Oliver (firstname.lastname@example.org) - Monday 4:30-6:30 E2 403G
- Luka Kocic (email@example.com) - Tuesday 12-2 E2 403B
- Andrew Barber (firstname.lastname@example.org) - Wednesday 3:30-5:30 in 403B
- Homework 1: Evaluating the effectiveness of a turn out the vote effort Homework 1
- Homework 2: Does Regression Get us the "Right" Answer? Homework 2
- Homework 3: Does the Minimum Legal Drinking Age Reduce Alcohol Consumption Homework 3
- Homework 4: Does the Minimum Legal Drinking Age Reduce Mortality Homework 4
Data & Useful Links
Schedule of Lectures, Readings and Assignments (This is subject to change depending on how rapidly we proceed)
- Lecture 1 (Tuesday, January 8): Overview and Review
- Overview of causal inference and the research designs we will examine (Randomized Controlled Trial, Regression Correction on Observables, Instrumental Variables, Regression Discontinuity and Difference in Differences).
- Review of how Ordinary Least Squares (OLS) is estimated.
- Lecture 2 (Thursday, January 10): Review of Inference in Ordinary Least Squares
- Central Limit Theorem
- Significance testing
- Lecture 3 (Tuesday, January 15): Introduction to the Potential Outcomes Framework (Mastering 'Metrics Chapter 1.1)
- Potential outcomes framework
- Average Treatment Effect
- Regression on a Dummy Variable
- Write up of article due #1: Holland, Paul "Statistics and Causal Inference." Journal of the American
Statistical Association, 1986, 81, 945-960.
- Lecture 4 (Thursday, January 17): Randomised Controlled Trial (Mastering 'Metrics Chapter 1.2)
- Why settings where people opt into treatment are problematic
- Randomized Controlled Trials can generate unbiased estimates
- Homework 1 Due:
- Lecture 5 (Tuesday, January 22): Variance of Estimates (Mastering 'Metrics Chapter 1 Appendix)
- Lecture 6 (Thursday, January 24): Contrasting Estimates from an RCT with Regression Adjustment (Mastering 'Metrics Chapter 2)
- Lecture 7 (Tuesday, January 29): Instrumental Variables - Introduction (Mastering 'Metrics Chapter 3)
- Lecture 8 (Thursday, January 31): Instrumental Variables - Assumptions and Intuition (Mastering 'Metrics Chapter 3)
- First draft of first paper due:
- Lecture 9 (Tuesday, February 5): Instrumental Variables - Local Average Treatment Effect, Two Stage Least Squares and Weak Instruments (Mastering 'Metrics Chapter 3)
- Lecture 10 (Thursday, February 7): Regression Discontinuity: Overview and examples (Mastering 'Metrics Chapter 4)
- Second draft of first paper due:
- Lecture 11 (Tuesday, February 12): Regression Discontinuity: Making Figures (Mastering 'Metrics Chapter 4)
- Lecture 12 (Thursday, February 14): Regression Discontinuity: Choosing Regression Specification (Mastering 'Metrics Chapter 4)
- Final draft of first paper due:
- Lecture 13 (Tuesday, February 19): Regression Discontinuity - Assessing Validity, IV Interpretation and Estimating Standard Errors
- Lecture 14 (Thursday, February 21): Introduction to Difference in Differences (Mastering 'Metrics Chapter 5)
- Lecture 15 (Tuesday, February 26): Difference in Differences - Examples and Triple Differences (Mastering 'Metrics Chapter 5)
- Lecture 16 (Thursday, February 28): Difference in Differences - Panel with Fixed Effects
- Lecture 17 (Tuesday, March 5): Overview of How to Evaluate an Empirical Study
- Lecture 18 (Thursday, March 7): More tools: Inference in Small Samples, Measurement Error, Robust Errors and Power Analysis
- First draft of second paper due:
- Lecture 19 (Tuesday, March 12): Overview of Course
- Lecture 20 (Thursday, March 14): Presenting Study Results to an Audience
- Second draft of second paper due:
- Final revision of second paper: Due by 5:00 PM Wednesday March 20th