Course Materials
# Module 01
Introduction
Learning objectives: Distinguish statistical modeling, machine learning, and artificial intelligence; compare prediction-focused and explanation-focused analysis; explain how analytical goals, assumptions, data, and evaluation criteria shape a quantitative research workflow.
Before class
During class
# Module 02
Machine Learning for Classification
Learning objectives: Distinguish classification from regression and supervised from unsupervised learning; explain how decision trees use entropy and information gain to select splits; describe bagging and random forests; evaluate binary classifiers using a confusion matrix, accuracy, precision, recall, and F1.
Before class
- ISLP, Chapter 1: “Introduction”Required
- ISLP, Chapter 4: “Classification”Required
During class
# Module 03
Logistic Function
Learning objectives: Convert among probabilities, odds, and log-odds; explain how the logistic function maps a linear predictor to a probability; fit binary logistic regression and interpret coefficients, odds ratios, and predicted probabilities; explain maximum likelihood estimation and evaluate model predictions.
Before class
- Odds and Log (Odds)Required
- Logistic RegressionRequired
During class
- Logistic Regression in ROptional
# Module 04
Ordinal Logistic Regression
Learning objectives: Identify when an ordinal outcome calls for ordinal logistic regression; explain cumulative logits and the proportional-odds assumption; fit an ordinal logistic regression model and interpret its coefficients, odds ratios, and predicted probabilities; assess model assumptions and communicate results for ordered outcomes.
Before class
# Module 05
Workshop 1: Proposal/Pitch
Learning objectives: Clarify research questions, assess project feasibility, and offer constructive feedback to prioritize proposal revisions.
# Module 06
Multinomial and Count Models
Learning objectives: Distinguish multinomial logistic regression from binary and ordinal models; fit a multinomial logistic regression model using an appropriate reference category; interpret coefficients, relative risk ratios, and predicted probabilities for nominal outcomes.
Before class
- “Count Data”Optional
# Module 07
Introduction to NLP
Learning objectives: Distinguish binary, multiclass, and multilabel classification; compare task-specific and embedding-based classifiers; evaluate predictions using suitable metrics and error analysis.
Before class
During class
- Draft a sentiment codebook, label a small set of reviews, and compare disagreements.
- Compare task-specific and embedding-based classifiers using a confusion matrix, precision, recall, and F1.
After class
- Write a short results memo comparing the two approaches and discussing misclassified examples.Required
# Module 08
NLP Applications: Zero-Shot and Topic Modeling
Learning objectives: Apply zero-shot classification using candidate labels; distinguish classification from clustering; explain how BERTopic uses embeddings, clustering, and topic representations to identify themes.
Before class
# Module 09
Synthetic Controls
Learning objectives: Explain how synthetic controls construct a comparison for a treated unit; assess donor-pool selection, pretreatment fit, and data requirements; interpret policy effects and evaluate the credibility of synthetic control case studies.
Before class
After class
# Module 10
Machine Learning for Regression
Learning objectives: Explain the bias–variance tradeoff and recognize underfitting and overfitting; evaluate regression predictions on held-out data; interpret machine learning results in an applied study of residential water use.
Before class
After class
# Module 11
Staggered Diff-in-Diff
Learning objectives: Explain identification assumptions for difference-in-differences with multiple time periods and staggered treatment adoption; interpret treatment effects across groups and time; connect modern DiD estimators to applied policy studies and replication code.
Before class
- Baker, Andrew, Brantly Callaway, Scott Cunningham, Andrew Goodman-Bacon, and Pedro H. C. Sant’Anna (2025), “Difference-in-Differences Designs: A Practitioner’s Guide.” Working paper, June 18 version. (opens in a new tab) Focus on Section 5.2. The same example progresses from basic DiD to multiple periods and staggered adoption; mortality estimates serve as pedagogical illustrations.Required
- Kakani, Pragya, Jessica Lu, Dima Mazen Qato, Sean Nicholson, and William L. Schpero (2026), “Medicaid Expansion and Retail Pharmacy Availability.” JAMA Health Forum 7(2): e256940. (opens in a new tab) Supplementary online content: methods, tables, and expansion timing figure (opens in a new tab).Required
# Module 12
Factor Analysis and Structural Equation Modeling
Learning objectives: To be added.