Quantitative Public Policy Analysis II

Prof. Cautrès is very passionate about teaching. He created an extremely meaningful learning experience, relating and applying methods to practical issues. I have never learned statistics this well before. — participant from Japan

Whether you think they should or they should not, numbers, data, and quantitative methods matter to today's public policy and policy analysis. Policymakers and administrators alike use numbers to support their (normative) arguments on what policies should be implemented, whether governments should or should not provide certain services, or whether it is the right time to engage in policy reform. At the same time, policy analysts use data and wide variety of quantitative methods to predict and evaluate the success or failure of new policies and to engage in evidence-based research on the impact of past policy interventions.

This course is the second part in a two-course sequence (cf. Quantitative Public Policy I). It is designed to provide participants with advanced statistical skills that allow them not only to engage with quantitative policy reports and publications, but to actively use advanced quantitative methods to analyze public policies as part of their academic, public sector consulting, or civil service careers.

Dates

This one-week, 17.5-hour course runs Monday-Friday, July 10-14, 2023. The course is scheduled for 9:00 am-12:30 pm.

Classroom Location

Faculty of Arts and Social Science, AS1 03-03

Image preview

Instructor

Bruno Cautrès, Sciences Po Paris

Detailed Description

Building on the material covered by the first course in the two-course Quantitative Public Policy Analysis sequence (cf. Quantitative Public Policy I), this course moves from basic statistical concepts and regression-based reasoning for policy evaluation to more advanced tools and provides a survey of quantitative methods for empirical studies and public policy research.

The main question addressed in this course is that of causal inference: how can one answer the question of whether X (a public policy, treatment, or decision) can be qualified as a 'cause' with respect to Y? This question, which may initially appear trivial, requires – in order for it to be answered – the use of advanced methodologies. All these methods, each with their own characteristics, try to control or even neutralize endogeneity.

The focus of the course is therefore on statistical methods for causal inference, i.e., methods designed to address questions that concern the causal link and the impact of causes on outcomes, where causes can be a policy intervention or a change in political institutions and economic conditions, and an outcome might be (changes in) household income, public support, election results, crime rates, etc.

The course starts out with a discussion of the strengths and limitations of multiple regression analysis and the relation between regression and causal modeling before covering a variety of quasi-experimental designs for causal inference. Participants will learn about a number of extensions and alternatives to the standard linear regression model that are commonly employed by advanced public policy analysts today, among them:

Panel data methods, incl. fixed and random effects, difference-in-differences
Instrumental variable estimation
Regression discontinuity designs
Limited dependent variables techniques, e.g., binary logit, ordinal logit analysis
Quantile regression analysis (time permitting)

In addition to introducing participants to these advanced methods, the course also analyzes their strengths and weaknesses and provides participants with hands-on experience on how to apply statistical techniques to real-world data. Part of the learning-by-practice approach are regular replications of the results of published research papers. Applications are drawn from public policy as well as various social science disciplines, such as economics, political science, and sociology.

Prerequisites

We strongly encourage participants to combine this course with the introductory Quantitative Public Policy Analysis I or the complementary Experimental Methods course. Alternatively, participants should be familiar with the material covered by Quantitative Public Policy Analysis I to guarantee that they get the most out of this course. Experience with the statistical software Stata is helpful, but not required.

Requirements

Participants are expected to have access to an internet-connected computer. Access to data, temporary licenses for the course software, and installation support will be provided by the Methods School.

Core Readings

Will be provided.