A 2-Day Seminar Taught by Felix Elwert, Ph.D.
This course offers an in-depth survey of modern instrumental variables (IV) analysis. IV analysis is an important quasi-experimental technique with numerous applications in economics, the social and biomedical sciences, business, marketing, and education.
IVs allow us to get unbiased estimates of causal effects even when there is selection bias, unobserved confounding, or imperfect compliance. The technique applies equally to randomized trials and observational studies. IV analysis is a very powerful tool—as long as the underlying assumptions are met.
This seminar will take students from the basic Wald estimator up to powerful recent developments, including non-parametric tests of the exclusion assumption. Students will get extensive hands-on experience by analyzing real worked examples from economics, sociology, and the health sciences. We will carefully dissect key technical and substantive assumptions to empower students to recognize, understand, and empirically test these assumptions in practice.
This seminar puts a premium on a rigorous and practical understanding. We will capitalize on three complmentary perspectives: modern potential-outcomes notation, visually intuitive directed acyclic graphs (DAGs), and the traditional algebraic approach. This will enable students to recognize IVs in their own studies, understand assumptions thoroughly, and read the specialist literatures in different fields.
Topics include single instruments, instrumental-sets, weak instruments, first-stage diagnostics, over-identification tests, exclusion tests, two-stage least squares (2SLS), natural experiments, encouragement trials, Mendelian randomization, split-sample IV, continuous and categorical outcomes, compliance classes, local average treatment effects (LATE), and Balke-Pearl bounds.
We will use the latest commands in Stata and learn theoretical and practical insights that transfer across software packages.
Who should attend?
If you want to understand how and when you use instrumental variables analysis in practice, this course is for you. The material is equally applicable to experimental and non-experimental data. Participants should have a good working knowledge of multiple regression and basic knowledge of Stata (point-and-click graphical user interface or basic command line operation). Students need no prior programming experience.
This seminar will use Stata for the examples. Although not required, participants are welcome to bring their own laptop computers with version 13 of Stata installed. Power outlets will be available at each seat.
Schedule and materials
The class will meet from 9 am to 4 pm each day with a 1-hour lunch break from 12 pm to 1 pm.
Participants receive a bound manual containing detailed lecture notes (with equations and graphics), examples of computer printout, and many other useful features. This book frees participants from the distracting task of note taking.
Registration and Lodging
The fee of $895 includes all course materials.
Lodging Reservation Instructions
A block of rooms has been reserved at the Club Quarters Hotel, 1628 Chestnut St., Philadelphia, PA at a nightly rate of $142 for a Standard room. This hotel is about a 5-minute walk from the seminar location. To make a reservation, you must call 203-905-2100 during business hours and identify yourself by giving the group code STA601. For guaranteed rate and availability, you must make your reservation by May 1, 2014.
1. Causal effects
a. ATE: Average treatment effects
b. LATE: Local average treatment effects
c. Identification in randomized trials and observational studies
d. Problems: Confounding, selection, and attrition bias
2. Approaches to IV analysis
a. Potential outcomes
b. Directed acyclic graphs
3. Types of IV analysis
a. Single instruments
b. Instrumental sets
4. Uses of IV analysis
a. Parametric estimation of treatment effects
b. Nonparametric testing of a null hypothesis
a. Wald estimator
b. Split-sample IV
c. Two-stage least squares
d. Indirect least squares
e. Linear and nonlinear outcome models
f. Standard errors
g. Why one should never control for an IV in OLS
6. Understanding assumptions and their consequences
a. Weak instruments
c. Exclusion restriction
7. Testing assumptions
b. Over-identification tests
c. Balke-Pearl bounds
d. Sensitivity analysis
8. Running examples
a. Economics (wage determination)
b. Medical (drug side effects)
c. Social science (contagion in social networks)