Econometrics

You should expect to spend roughly eight hours per week studying for ECON 3300 outside of class. I have designed the course with that expectation in mind. I recommend allocating those eight hours roughly as follows:

1. Preview each upcoming lecture notebook. The goal is not to understand everything in advance, but to familiarize yourself with the main ideas so that you have a framework for organizing what you learn in class. (1 hour/week)
2. Study each lecture notebook on your own before attempting the problem set. Work through the material carefully, take written notes, and keep track of anything you do not understand. (2 hours/week)
3. Talk through anything you are struggling with. Discuss difficult concepts with your classmates, peer mentors, or me during office hours. (1 hour/week)
4. Complete the problem set. You are encouraged to work with your classmates, but you must submit your own handwritten solutions and should be able to explain the reasoning behind everything you write. (3 hours/week)
5. Study for the upcoming quiz after completing the problem set. Use the quiz as an opportunity to make sure you can work with the material independently. (1 hour/week)

If you follow this routine consistently, you will be doing your part to get the most out of the course. I will do my best to make sure that this effort is rewarded with a deep understanding of the material and that your grade is a fair reflection of what you have learned.

The use of AI is prohibited in this course.

I prohibit AI use in this course for three main reasons.

1. If you use econometric tools, you should understand them. Econometrics is used to inform decisions with real consequences for people, from evaluating public policies to guiding business decisions. A regression result is of little value if you cannot interpret it or understand how it was produced. If we are going to use econometric evidence to make recommendations, I believe we have an obligation to understand the methods and assumptions underlying those recommendations.
2. Learning requires struggling with the material. AI can give you an answer without requiring you to work through the problem yourself. Dr. Leo shared with me a useful analogy comparing this to using GPS. You may follow and understand every GPS instruction and reach your destination, but that does not mean you have learned the route. If instead you make the trip several times on your own, take a few wrong turns, and have to figure out where you are, you will eventually understand the route much better. The same is true in econometrics. Making mistakes and working through confusion are important parts of learning.
3. AI can be wrong in subtle ways. AI often produces answers that sound convincing even when the reasoning is incorrect. Its mistakes can involve subtle issues of interpretation or statistical reasoning that are difficult to recognize unless you already understand the material. My goal is for you to develop enough independent knowledge that, when you use AI in the future, you can evaluate its answers rather than simply trust them.

AI is absolutely becoming an integral part of professional work. But the most common way people interact with AI today, through chatbots such as ChatGPT, is already quite easy to use, and it will only become easier. There are more advanced ways of using AI to write statistical code or develop software, but even these can be learned relatively quickly.

As AI becomes more ubiquitous, the ability to use it will become less scarce. As we know from ECON 1050, when the supply of a skill increases, the price employers are willing to pay for that skill tends to fall! By the time you graduate, employers may simply expect you to know how to use AI. What will remain more valuable is the knowledge and judgment that allow you to use AI effectively.

Econometrics will remain valuable because it teaches you how to interpret data, evaluate empirical claims, and recognize when an analysis is misleading. Developing this knowledge is difficult, and it is difficult to outsource. My goal in this course is therefore to help you develop the econometric understanding that will make AI more useful to you later, rather than use AI as a substitute for developing that understanding now.

Absolutely not. Although I typically grade on a curve, I have structured the course so that your grades reward a deep understanding of the material rather than your ability to produce work that you do not understand.

While AI use is prohibited, I cannot effectively enforce that rule on problem sets. For that reason, problem sets are graded primarily for effort, and you are rewarded for completing them successfully only to the extent that you also perform well on the corresponding in-class quiz. See the syllabus for the formal grading scheme.

The class project is also designed so that your grade reflects how well you understand your own work. A large share of the project grade comes from your presentation, where I will ask you questions about your analysis and your understanding of the project. The remainder of the course grade comes from the in-class midterm and final exam.

In short, not using AI will not put you at a disadvantage. If anything, using AI in place of working through the material yourself is likely to hurt your grade by weakening the understanding that the course is designed to assess.

AI can be detrimental to learning, but it can also be very useful for completing tasks when you already have the knowledge needed to critically evaluate its output. I use AI regularly to help me write code for interactive plots, simulate data, proofread lecture notes, produce early drafts of course materials, and think through econometric questions. That use extends to the materials for this course.

My promise to you is that I will never give you AI-generated material that I have not carefully reviewed and revised myself. I read every word to ensure that it is accurate, appropriate for the course, and consistent with what we discuss in class. The first drafts of many lecture notebooks on this website were developed with assistance from AI, but I then spent several hours revising each one so that the final version reflects exactly what I believe you should learn in this course.

I will never use AI to grade or evaluate your work.

You will receive feedback throughout the course in several ways.

Most Tuesdays, you will complete a 15-minute quiz covering material from the previous week. I will return your graded quiz in Thursday’s class and review the solutions with the class, giving you an opportunity to ask questions about anything you found confusing.

I will not return your problem sets, but I will post complete solutions on both this website and on Brightspace. Please review these carefully, especially for any questions you found confusing. For the class project, I will provide each group with verbal feedback following the presentation and a detailed written assessment on Brightspace.

I will post solutions to the midterm and final exam on Brightspace. You are welcome to review your graded midterm with me during office hours.

You are also always welcome to discuss your progress in the course with me during office hours or after class.

Specifically, by the end of this course, you will be able to

1. Describe, compare, and contrast various econometric methods.
2. Mathematically derive simple econometric and statistical estimators.
3. Apply econometric methods to analyze data using statistical software and basic coding techniques.
4. Interpret regression results using appropriate economic language and effectively communicate them to a broader audience.
5. Identify and think critically about the underlying assumptions, limitations, and strengths of various econometric estimators.
6. Formulate a research question, prepare and analyze data in Stata using appropriate econometric methods, and effectively explain and evaluate your findings.