In social science research, we often have panel data (multiple units with repeated observations for each unit). We often wonder which panel data models to use. Most economists prefer fixed effect models. Sometimes people use multi-level models. In this talk, I'd like to talk about multi-level models: what are they, when should we use them, and why.

• Audience: Harvard Faculty, Students, and Staff are all welcome, with priority given to HBS faculty, doctoral students, and RAs
• Prerequisites: Familiarity with fundamental statistical concepts is recommended
• Software requirements: None

Part 1

The seminar introduces principles of causal inference for randomized experiments and observational studies. We discuss the potential outcomes framework and, through its lens, give an overview of modern quasi-experimental methods, including regression and matching. We end with a brief literature review.

• Audience: Harvard Faculty, Students, and Staff are all welcome, with priority given to HBS faculty, doctoral students, and RAs
• Prerequisites: Familiarity with fundamental statistical concepts, including probability, distribution, and linear regression is recommended.

Part 2

This is an applied part of a series of workshops on principles of causal inference (attendance of the previous parts is not required, but recommended). During this workshop we will go into details of the matching method, one of the most popular quasi-experimental methods in causal inference.

• Audience: Harvard Faculty, Students, and Staff are all welcome, with priority given to HBS faculty, doctoral students, and RAs
• Prerequisites: Familiarity with fundamental statistical concepts, including probability, distribution, and linear regression is recommended. Basic knowledge of R is required.

Part 3

This is an applied part of a series of workshops on principles of causal inference and matching methods (attendance of the previous parts is not required, but recommended). During this workshop we will practice implementing matching methods in R using various packages, including MatchIt, cem, RItools, and cobalt.

• Audience: Harvard Faculty, Students, and Staff are all welcome, with priority given to HBS faculty, doctoral students, and RAs
• Prerequisites: Familiarity with fundamental statistical concepts, including probability, distribution, and linear regression is recommended. Basic knowledge of R is required.
• Software requirements: Please have R and RStudio installed and bring a laptop if you would like to follow the hands-on part.

Moving beyond basics, this seminar will introduce new skills to improve and accelerate your research: importing/exporting data, altering your database structure, "views" for common queries, and optimizing queries with indexes & other features.

• Audience: HBS and FAS affiliates, with priority given to HBS faculty, doctoral students, and staff
• Pre-Requisites: Basic knowledge of mySQL