• Articulate the fundamental problem of causal inference.
• Explain how various research designs seek to draw causal inference.
• Present a substantive discussion of how the assumptions of a research design can be violated in a specific setting.
• Use statistical software to conduct an independent empirical analysis of a problem using the research designs and methods introduced in the course.
• Interpret and analyse the results of analyses appropriately in relation to a given research problem.
• Critically evaluate the strengths and weaknesses of the research designs and methods introduced in the course in their application to a given research problem.

In business, public policy and the non-governmental sector, the demand for “evidence” is stronger than ever. Managers and policy-makers place an increasing premium on knowing “what works”, when they decide on which direction to take the company or the country. Additionally, recent decades have seen an explosion of data availability. Combined with advances in causal inference, this puts social scientists in a better position to answer the demand for evidence than they ever have been.

This course provides a case-based introduction to techniques that allow us not only to measure which policies “work”, but also to quantify how much they work. That is, drawing causal inferences. We will cover four broader areas:

1) The so-called potential outcomes framework for thinking about causal inference.

2) Research designs that allow for drawing causal inferences.

3) The statistical implementation in the software R.

4) Substantive interpretation of the results of a statistical analysis.

The course will be taught as a mix of lectures in larger groups and exercise classes in smaller groups. The teaching format is “particular general particular”. We will introduce a particular question, then discuss how such a question is handled in general by reviewing core concepts from the literature, and we then return to the particular application by focusing implementation and extensions.

Throughout the course, we will follow an applied, hands-on approach, emphasizing the implementation and interpretation of statistical analyses. Hence, students will spend a substantial amount of time working with software. The exercise classes will be based on group work prior to class. Solutions and implementations will then be discussed in class. The course software will be R. All activities will be based on data from actual research projects.

In relation to Nordic Nine

The course supports the Nordic Nine (NN) capabilities by training analytic skills and critical thinking through hands-on work with data. In the lectures, students will learn fundamental concepts from statistics and causal inference. In the exercises, they will apply these concepts by criticizing a published piece research (NN6), discussing constructively how to improve its research design (NN6), and honing data analytic skills by replicating the results with real-world data (NN2). This will form the basis of class discussions, where we will deliberate about how a particular piece of research informs us about society (NN1), and helps resolve the challenges facing us (NN3).