Applied Panel Data Analysis in Social Sciences

14-15 November 2024

Elisa Volpi, UNIGE

Professor Macarena Ares, University of Barcelona

This panel data course is tailored for PhD students and researchers in the social sciences, focusing on the practical application of micro-level (individual-level) panel data analysis. The course offers and introduction to longitudinal data analysis on an applied level using Stata. The course equips participants with the skills needed to leverage panel data for addressing complex social science research questions. With a strong emphasis on hands-on learning, it covers key concepts, techniques, and challenges associated with longitudinal analyses.

The course begins with an exploration of the theoretical foundations of modeling panel data, emphasizing how panel data can enhance our ability to identify causal effects, as framed in the potential outcomes framework. It also delves into the assumptions underpinning causal effect identification in these analyses.

On a practical level, students will begin by learning how to prepare panel data sets, understand the data structure, and identify potential challenges in working with such data.

The core of the course revolves around linear regression techniques, encompassing fixed-effects regression, random-effects regression, and hybrid models. Students gain a comprehensive understanding of these estimation strategies, enabling them to choose the most suitable approach for their research questions.

Practical application is a central feature of the course, with participants working extensively in a computer lab environment. They apply their knowledge to real-world panel data sets, such as the widely-used British Household Panel Survey/ Understanding Society or the Swiss Household Panel. Through these hands-on exercises, students develop the skills needed to analyze and interpret panel data effectively, and to face some of the very practical challenges of, e.g., studying short- vs. long-term change, or modelling the effects of transitions across discrete statuses.

By the end of this course, participants will have a solid grasp of panel data analysis techniques, from theory to practical implementation. They will be equipped to tackle a wide range of social science research questions. The course will encourage participants to work with their own data and to share the challenges they face with fellow participants.

Tentative schedule:

November 14^th

Theory: 

1. Modelling panel data

2. Longitudinal analyses and causal inference

3. Fixed-effects, First-difference and Random-effects models

Practical session:

1. The structure of panel data

2. Dataset preparation and coding decisions

3. Pooled OLS

November 15^th

Theory:

1. Challenges in FE and RE models

2. Modelling transition analyses

3. Hybrid models

Practical session:

1. Fixed-effects regression

2. Random-effects regression

3. Hybrid models

UNIBE

The workshop will take place at the University of Bern, Sitzungszimmer 217 (14.11) and 204 (15.11), Hauptgebäude.

15