Field
Experiments
in Economics
From causal inference fundamentals to running your own randomized controlled trial — a concentrated crash course for doctoral researchers. You will leave with the conceptual toolkit and hands-on R skills to design, execute, and analyze a field experiment.
↓ Download Handout (PDF)
Workshop Syllabus
Participants apply each concept to their own research in a short structured worksheet. By session 6, each person has a sketch of a real experimental design — question, randomization strategy, outcome measure, and sample size estimate — and delivers a 2-minute pitch.
- Economic methodology: theory → observation → experiment
- Fundamental problem of causal inference
- Observational methods & their core assumptions
- Regression, RDD, IV — what each buys you (and costs you)
- Why random assignment solves the counterfactual problem
- Limitations & ethical considerations
Handout §1 · 30 min lecture · 15 min discussion: "Where does your research face a causal identification problem?"
- Strategic, descriptive, process & impact questions
- When is impact evaluation needed — and when not?
- Needs assessment methods
- Prioritization criteria: influence, scalability, new knowledge
- Cost-effectiveness framing
Handout §2 · 30 min lecture · 15 min worksheet: "What is your research question — and is it answerable with an RCT?"
- Opportunities to randomize: new programs, oversubscription, phase-in
- Individual vs. cluster-level randomization
- Spillovers, attrition & compliance — choosing the right level
- Simple, stratified, and matched randomization
- Best practice checklist
- R demo: simple randomization in 10 lines
Handout §3 · 35 min lecture + live demo · 10 min worksheet: "What would you randomize and at what level?"
- Outcome → indicator → instrument → variable chain
- Administrative vs. survey data; respondent & enumerator choices
- Validity, reliability, and field-testing measures
- Non-survey instruments: mystery clients, list randomization, biomarkers
- Baseline surveys — when and why
Handout §4 · 30 min lecture · Case: Bertrand & Mullainathan resume audit · 10 min worksheet: "What is your primary outcome and how do you measure it?"
- Sampling variation, CLT, confidence intervals
- Type I / Type II errors; significance vs. power
- Determinants of power: N, effect size, variance, ICC
- Individual vs. cluster-randomized formulas
- R demo:
pwrpackage — one live worked example
Handout §5 · 35 min lecture + live demo · 10 min worksheet: "What sample size does your design need?"
- ITT estimation & adding covariates
- Key threats in brief: attrition, non-compliance, spillovers
- Pre-analysis plans — why they matter
- Generalizability & from findings to policy
- 2-minute design pitches by participants
Handout §§6–8 · 25 min lecture · 20 min pitches: each participant presents their worksheet sketch — question, design, outcome, sample size, one key threat
What You Will Take Away
pwr package to determine required sample sizes for individual- and cluster-randomized designs.Before You Arrive
□ Download Before the Workshop
- Handout PDF — print or load on your device: igorasanov.com
- Worksheet — one page you will fill in across sessions 2–6 (link shared before workshop)
- R demos are instructor-led — no installation needed
□ Assumed Background
- Basic statistics (hypothesis testing, regression)
- Introductory econometrics (OLS) is helpful but not required
- No prior R experience required — we start from basics
- Curiosity about causal questions in your research area
Instructor
Igor Asanov is an experimental economist specializing in causal inference and causal machine learning, with research spanning education, mental health, and labor markets. He has designed and analyzed large-scale randomized controlled trials across Ecuador, Vietnam, Europe, and multiple other countries, with findings published in Nature Human Behavior (forthcoming), PNAS, World Development, JEBO, and Social Science & Medicine. He teaches PhD-level courses in field experiments, econometrics, and causal machine learning and AI at the University of Kassel.
www.igorasanov.com →