AI-ENHANCED LEARNING

Choose your
AI tutor

Bring your questions.
Find your way to understanding.

Every student learns differently. Our AI tutors are built as customized Google Gemini Gems — each with a distinct pedagogical approach, trained on the course content. "Learn by coding" is also a key feature of these tutors. You can use the Google Colab scratchpad to easily engage with the code the tutors provide.

Find your learning approach
6 chapters5 causal methodsPython code · Real datasets
A QUESTION TO START WITHIllustrative exchange

Think it through.
One step at a time.

STUDENT

Why randomly assign health insurance in the RAND experiment?

LEARNING COACH

People who choose different insurance plans may differ in health, too. Random assignment helps make the groups comparable before treatment.

What would you compare at the start to check whether the groups look similar?

Start with a question. Build the reasoning.
An example of guided reasoning. Actual tutor responses vary.Meet the five tutors ↗

FIVE WAYS TO LEARN

Find your thinking partner.

Choose the approach that fits
the question in front of you.

🎓

Learning Coach

Recommended for first-time learners

Your step-by-step guide through causal inference. Starts with intuition before introducing formulas, breaks concepts into manageable pieces, and provides worked examples using chapter datasets. Checks your understanding at each stage and guides you to answers with positive reinforcement. Covers all 6 chapters: RCTs, Regression, IV, RD, DD, and the Wages of Schooling.

Step-by-Step Worked Examples Comprehension Checks Homework Help Beginner Friendly

Try: "Walk me through the RAND Health Insurance Experiment from Chapter 1 step by step"

Best for: Building intuition chapter by chapter, getting unstuck on homework problems

Start Learning →
🏛

Case-Study Explainer

For understanding the stories behind the data

The domain expert who brings each natural experiment to life. Uses a Context-Design-Identification-Policy framework to explain the historical circumstances, institutional details, and policy features that make each study's causal strategy work. Covers 11 case studies including the RAND HIE, Oregon Health Lottery, MDVE, KIPP schools, MLDA mortality, and Great Depression banking.

Historical Context Policy Analysis 11 Case Studies Identification Strategy

Try: "Explain the institutional context behind the Oregon Health Plan lottery from Chapter 1"

Best for: Understanding why each causal method works in its real-world setting, preparing policy discussions

Start Learning →
📝

Exam Coach

For midterm and final preparation

Your test-preparation specialist. Starts with diagnostic questions to find your weak spots, then builds targeted practice sets with increasing difficulty. Drills the distinctions between IV, RD, and DD, provides flashcard-style concept summaries, and flags common exam pitfalls. Question formats include multiple choice, short answer, Wald estimator calculations, and scenario-based problems.

Practice Questions Weak Spot Diagnosis Method Distinctions Flashcard Summaries Increasing Difficulty

Try: "Quiz me on the difference between IV and RD for Chapters 3 and 4"

Best for: Drilling key concepts before exams, identifying gaps in your understanding of the 5 causal methods

Start Learning →
🤔

Socratic Challenger

For students with some foundation

Teaches through probing questions, never direct answers. Challenges your assumptions about correlation and causation, pushes you to identify validity threats, and asks the hard questions: ‘Why can’t we just compare?’, ‘What assumption does the heavy lifting?’, ‘Who are the compliers?’ Gives direct help after 3 rounds if you’re stuck, or immediately for Python syntax.

Questions Only Assumption Challenges Validity Threats Deep Thinking Critical Reasoning

Try: "I think comparing average outcomes between treated and control groups always gives a causal effect"

Best for: Developing critical thinking about identification, preparing to defend research design choices

Start Learning →
💻

Code-First Experimenter

For hands-on, learn-by-doing coders

Starts with Python code and real data, then builds up to theory. Shows you the result first and explains why it works. Encourages ‘What if’ experiments: change the bandwidth, drop control variables, compare OLS to IV. Uses pandas and pyfixest with all course datasets streamed from GitHub.

Python Code Real Datasets What-If Experiments pyfixest Google Colab Ready

Try: "Show me IV estimation of returns to schooling using quarter-of-birth data from Chapter 6"

Best for: Learning causal methods through code, extending chapter notebooks with your own experiments

Start Learning →

How It Works

1

Choose a Tutor

Pick the learning style that matches your needs and goals.

2

Ask Anything

Ask about any chapter, concept, or dataset. The tutor knows the full curriculum.

3

Learn by Doing

Follow the tutor's guidance, run code in Colab, and build your causal inference skills.