OPCO 2100 - Methods for Research and Policy Evaluation

Course taught in English. The objective of this course is to introduce quantitative methods for causal inference, relying on mathematical formalization and applying them to concrete research situations. Students will learn how to handle data - assembling a dataset, cleaning it, and preparing it for quantitative analysis. The methods covered will include a review of linear models (simple and multiple regressions) and nonlinear models (probit, logit), before moving on to the core of the course: causal inference, namely DiD, RDD, and IV (2SLS, 2SRI). Each session will be divided equally between the presentation of methods and their mathematical formulation, and in-class exercises designed to help students master the associated coding techniques. The final goal is to replicate a scientific article that applies one or more of these methods, enabling students to develop technical skills useful for research or for the quantitative evaluation of public policies. Another objective is to cultivate critical thinking regarding the strengths and limitations of such research methods. It is therefore particularly well suited for students who wish to develop their skills in computer science, mathematics, and quantitative research methods.

Basic coding skills (beginners welcome) - Stata, R, or Python are all acceptable. Students should know how to use an IDE of their choice, how to import data, and have some introductory knowledge of a programming language.

Basic statistical concepts - e.g. students should know what a mean, a median, a standard deviation, a statistical sample and statistical inference are.

The course is structured with a 1-hour lecture (CM) followed by a 1-hour practical session (TP). Students are required to have a computer with RStudio or Stata installed (Stata licenses are available through Sciences Po). Each session begins with an introduction to the methods to be used, during which their mathematical formalization is covered. The second part of the session focuses on applying these methods to real-world data using RStudio or Stata. Classes are held every two weeks for a total of 2 hours and are conducted in English.

The course will be assessed using a simplified pass/fail system. Evaluation will be based on a short report (five pages) replicating a scientific article or a report that uses a causal inference method, provided by the lecturers at the beginning of the semester. The writing of this report will be supported incrementally throughout the semester, with guidance provided during each class, so that students can progress step by step in the making of the paper.