Design, Causality and Modelling

This course introduces students to the fundamental principles of experimental design and data analysis techniques that aim to uncover causal mechanisms. The central objective is to understand how causal questions can be answered using data, particularly in situations where controlled experiments are impractical or impossible. Students learn how to formulate meaningful research questions, distinguish causation from association, and identify the information required to draw credible causal conclusions. The course adopts structural causal models and causal diagrams (Directed Acyclic Graphs, DAGs) as a unifying framework for representing assumptions about the data-generating process and for reasoning about confounding, mediation, and identification. Students learn how causal paths arise in complex systems and how back-door and front-door criteria can be used to identify causal effects.

Building on these foundations, the course introduces treatment effect estimation, including average and conditional treatment effects, and discusses how different causal estimands address different scientific and policy questions. Particular attention is given to the role of research design in isolating causal variation and reducing bias. A broad range of practical methods for causal analysis is covered. These include randomized experiments and ANOVA, multivariate linear regression, non-linear regression models, matching methods, inverse probability weighting, difference-in-differences designs, and instrumental variables. Students learn both the statistical implementation of these methods and the assumptions required for valid causal interpretation.

• Formulate and evaluate causal research questions and distinguish between prediction, explanation, and intervention-based analyses.
• Describe and reason about data-generating processes using structural causal models and understand the distinction between correlation and causation.
• Construct and interpret causal diagrams (Directed Acyclic Graphs) to represent causal assumptions and identify sources of bias.
• Apply identification strategies based on back-door and front-door criteria to determine when causal effects can be estimated from data.
• Define and interpret treatment effects, including the Average Treatment Effect (ATE), Average Treatment Effect on the Treated (ATT), and related causal estimands.
• Design and analyze empirical studies using randomized experiments, ANOVA, linear and non-linear regression models, and other causal modelling approaches.
• Estimate causal effects from observational data using matching methods, difference-in-differences, and instrumental variable techniques.
• Critically assess the assumptions, strengths, and limitations of causal inference methods in applications from economics, public policy, social sciences, and the life sciences.

• Quality education
• Industry, innovation and infrastructure
• Peace, justice and strong institutions

In presence

The course combines lectures with interactive tutorials and practical exercises. Students will engage with both conceptual material (e.g., causal reasoning using diagrams) and applied work involving real or simulated data to estimate treatment effects. Short in-class experiments and guided replication of published studies will be used to reinforce key concepts. Emphasis is placed on active participation, critical reflection on assumptions, and collaborative problem-solving.

The course is evaluated by means of a midterm exam and a final exam that make up 40% and 60% of the final grade, respectively.