Final Project

The final project is an opportunity to engage with a published research paper that asks a causal question. You will choose one paper from a set that we have prepared. For that paper, you will carry out two tasks.

1. Summarize what the authors have done

Choose one causal estimand from the paper. For that estimand,

• Define the estimand using potential outcomes
• Present the identification assumptions that link the quantity to observable data
• Discuss the estimator that the authors use to estimate the target quantity

In many papers, the three steps above are stated in English and may involve some ambiguity. You may not be certain what quantity the authors wanted to estimate, or exactly what they assumed. In these settings, your task is to choose one interpretation of what the authors have done and make your interpretation precise.

2. Propose a new quantity to estimate

In the second part, you will propose a new causal estimand. If you were to collect new data or conduct new analysis to estimate a new quantity, what would that quantity be? For this part of the assignment, you can write as though you have unlimited resources to collect any amount and kind of data you would want.

• Define a new causal estimand
• Present the identification assumptions that link the quantity to observable data
• Discuss the estimator that you would use to estimate the target quantity

There is one restriction: for part 2, your proposed analysis cannot involve a randomized treatment. While experiments are terrific ways of conducting research, the identification and estimation assumptions in randomized experiments can be trivial. In order to practice the skills learned in this class, we want you to focus on a non-randomized setting.

Format of the final project

Working in groups

We anticipate that most students will carry out the final project in groups of 4–5 students who are all in the same discussion section. Near the middle of the semester, we will circulate a form for you to tell us if there are people you’d like to work with, or if you’d like us to randomize you to a group so you can meet new people. If you prefer to work alone, just come talk to us, and we can discuss how that could work.

Writeup

Writeup due Nov 21 at 5pm

Your group will submit a writeup that is between 1,500 and 2,000 words, typeset using RMarkdown.

Presentation

Presentations Nov 29 during discussion

Each group will make a 10-minute presentation, using slides. Your presentation should involve both parts (1) and (2).

Grading

We will post rubrics on Canvas when we assign final project groups.

Published papers for the project

Your group will study one of the following papers:

1. Stuart, E. A., & Green, K. M. (2008). Using full matching to estimate causal effects in nonexperimental studies: Examining the relationship between adolescent marijuana use and adult outcomes. Developmental Psychology, 44(2), 395.
   • This paper is an example of matching
2. Halloran, M. E., & Hudgens, M. G. (2018). Estimating population effects of vaccination using large, routinely collected data. Statistics in Medicine, 37(2), 294-301.
   • This paper is an example of addressing interference
3. Eggers, A. C., & Hainmueller, J. (2009). MPs for sale? Returns to office in postwar British politics. American Political Science Review, 103(4), 513-533.
   • This paper is an example of regression discontinuity
4. Acharya, A., Blackwell, M., & Sen, M. (2016). The political legacy of American slavery. The Journal of Politics, 78(3), 621-641.
   • This paper is an example of instrumental variables