Multilevel and Longitudinal Modelling

Full course description

In this comprehensive exploration of advanced statistical techniques, we will delve into a variety of powerful tools aimed at enhancing your ability to conduct sophisticated analyses and draw robust conclusions in the realm of social sciences.

Our journey begins with an in-depth exploration of Factor Analysis, a method designed to unveil latent variables and underlying structures within observed variables. Through hands-on applications, you will master the art of reducing complex data sets into more manageable components, allowing for a deeper understanding of intricate relationships.

Moving beyond Factor Analysis, we will navigate the intricate landscape of Structural Equation Modelling. SEM provides a framework to examine complex relationships by simultaneously estimating multiple interdependent equations. By the end of this section, you will be equipped to model and assess intricate causal networks.

As we progress, we will tackle Multilevel Regression models, an essential tool when dealing with hierarchically structured data. Whether exploring individual and group-level effects or understanding variance at different levels, multilevel regression will empower you to analyze data with nested structures effectively.

The course will then shift focus to Longitudinal Models, indispensable for studying changes over time. We will explore handling time-dependent data, allowing you to uncover trends, trajectories, and causality in longitudinal studies.

Finally, we will approach the issue of causality in greater detail. To do so, we will start from what is considered the gold standard to identify causal effects: randomized controlled trials. This will allow us to discuss different experimental approaches in the social sciences (with particular reference to survey experiments). We will then apply the logic of causal inference to studies in which researchers can only employ observational data. We will, therefore, reflect back about longitudinal studies looking in particular at panel structures and fixed effects; and we will discuss a particular situation that researcher have in the past used to detect causality: event studies. We will then discuss additional methods that have been developed to detect causality in observational studies. Namely, we will discuss instrumental variables, regression discontinuity, and difference-in-differences. These methods can be particularly useful for assessing causal effects in situations where randomized controlled trials are impractical or impossible. We will, therefore, discuss their intuitions, advantages, and limitations. This last section of the course will focus on theory, with the objective of giving you the main intuition behind complex methods that have been developed in the social sciences. After the classes, you should be able to delve deeper into the topics we have discussed, and to approach replication files to employ these methods in your research.

While underlying theory will be discussed, the greatest emphasis will be on application and interpretation of models and results (but for the case of causal inference methods). The emphasis of the course is on understanding advanced statistical tools and developing the competence to apply these tools to analyze data in order to be able to examine particular substantive research problems. Summarizing, you will learn how to integrate the theory (from your substantive course) and the methods, and put them together in a coherent paper. As the final exam consists of writing an integrated research paper (in parallel with your substantive specialization), you will have the opportunity to have one-to-one meetings about your own research projects. The ability of selecting the right technical tools and techniques in order to research an original research question is a crucial aspect in order to facilitate substantive, theoretical and methodological innovation.