Project 1 👶

Project 1 👶#

Randomized Experiments

In this project, we will conduct our first causal study start to finish by examining the Infant Health and Development Program (IHDP) dataset, which is a classic dataset in the causal inference literature.

Projects in this course are split into two notebooks to make it easier to manage the code:

  • Part 1 is focused on implementing the functions and interactive plots and will be primarily autograded (outside of the interpretation questions).

  • Part 2 is focused on using the functions you implemented in Part 1 to analyze the IHDP dataset.

Tip

I recommend that you test your functions in Part 1 both with your own assertions and with the autograder before moving on to Part 2. That way you’ll have more certainty that your functions are correct for the analysis you’ll do in Part 2.

Learning Objectives#

  • Practice following the causal roadmap with a real-world dataset.

  • Explore the benefits of stratification in a randomized experiment.

  • Translate mathematical quantities into code by implementing two difference-in-means estimators.

  • Exercise your data manipulation and visualization skills with real and simulated data.

Note

This project is due Thursday 10/9 at 11:59pm.

Grading guidelines

The course projects offer an opportunity to practice the full causal inference workflow, from building estimators and formulating questions to conducting analyses and communicating your findings effectively. Here are some guidelines on how to approach the project:

  • Like the worksheets, a portion of points will be autograded – feel free to submit as many times as you want to check your code’s correctness!

  • For visualizations:

    • Help your reader understand your findings through visually clear figures

    • Label your axes, provide legends when appropriate, and add figure titles to provide context

  • For written responses:

    • Support your ideas with specific evidence from your analysis or prior knowledge

    • Write concisely but clearly – one-word/one-phrase answers usually don’t give enough space to show what you’ve learned

If you’re uncertain about any portion of the project, please do come to my office hours, Bhargavi’s TA hours, or reach out on Ed!

Contents