Applied Econometric Analysis Project

Data Selection

Try to find data you’re interested in but satisfies all the requirements. It should have plenty of columns to explore, different kind of variables, and be large enough to do some interesting analysis (sufficient number of rows).

Analysis

  • Figure out what each column is. What does it mean? Explain it clearly.
  • Are there problems? Do you need to clean the data? What about missing data? How will you handle that? Is that a problem? We think about this as “sample selection”.
  • Think about the relationships between all the variables and the outcomes. Make interesting figures, think about why it’s true, think about outliers. This is still in the “descriptive” stage.
  • Then, think about your regression model and follow the steps in the project requirements.
  • Try to make your code and figures as organized and clear as possible.
  • Don’t show everything you do, just the most important parts. You can make an appendix if you need to (or want to) show more details.

Suggested Writing Process

Here is a writing process I think is helpful:

  1. No judgment word dump. Just write down your thoughts. Don’t worry about grammar, spelling, or structure. Just get your ideas down. Put everything you are possibly thinking in any order. No judging allowed just stick it on the paper. Don’t even think about touching the delete button just keep going. Even ask questions in the middle.
  2. Organize your thoughts and put them into bullet points, and see if there is a natural flow that emerges.
  3. Think about how it all makes sense together, and write some sentences in your own words to connect the ideas.
  4. Read it several times and see if it makes sense, sounds convincing, and is clear.

AI Usage

I spend a lot of time with my students. I listen carefully, stay long after class and office hours, and tutor students every year. I do this so I can hear their voices, and cater my content and materials to their needs. I read every single note from the professor, the labs, the materials, and the book, so I know what is at the students’ disposal when I am creating my slides, questions, etc. I also have a handful of students I know do not use AI. They don’t like it or are not good at it, or simply don’t need it. I deeply care about my students, so I think about the feedback they give me and the stories they tell–how they are struggling, why certain professors connect or don’t connect with them, bad experiences with other TAs, etc.

What’s my point? The point is that I know what AI sounds like and what an undergrad’s voice sounds like. I also know that most students don’t use words like “idiosyncratic shocks”, or “mean reversion” when they don’t really know what those mean rigorously. I would honestly prefer students just say what they would actually say casually about the material instead of trying to sound fancy. I think it would be hilarious and wonderful to see a student say something like:

“This relationship is lowkenuinely nonlinear af when you look at the resids.”

Also, keep in mind that ability to use AI is correlated with other abilities. So poor prompting and lazy formatting from the lowest performing students on an AI slop project is pretty obvious.

Try to stand out! Write something genuine. Write something relatable that shows me you personally thought carefully about the data. This is an opportunity to utilize one of the most important skills you can have in your career: communication. As time goes on it will be more and more important to differentiate yourself from the slop, so I urge to to take the opportunity now.

I am a scholar and a teacher and a coach, and I absolutely love my job. I love giving students feedback and helping them grow and succeed!

WarningI have no interest in being the AI police.

If I think your project is slop, then your grade will reflect that. However, if it has distinct hallmarks of AI, and especially if it has undeniable artifacts of AI, I will report it to the professor and document it rigorously.

I know students are under immense pressure to perform and cut corners, but taking the time to work through my writing process above or another process that works for you is the best way to execute this project.

How can you use AI ethically?

  1. Use AI as a partner to bounce ideas off of, but always start with your own hypotheses and questions.
  2. As it to read your work and give high level feedback, or a few suggestions on how to reword. This prevents your voice from getting taken over.
  3. Tell it to read your work critically and point out flaws in the logic. Are there counterexamples that don’t work?
  4. Ask it where you can read more on the topic, then go back and forth about a concept until you feel clear that you have understood it.

Feedback

I will be grading approximately 70 projects. If you are interested in getting serious feedback on your project, put a note in your introductory paragraph of the project that indicating why and how you want extra feedback on your project, being very specific. This project can be just another assignment, or it can be something where you can really showcase your skills, and practice communicating your project ideas and innovations–a common interview question for industry or research positions.