This reading course provides a hands-on and paper-based approach to understanding and analyzing data. For many projects, collection of new data or experimental designs are the only way to answer a research question or to provide the decisive complementary evidence. Different ways to collect data can have important implications for model estimation and evaluation, parameter inference, and policy conclusions. Standard econometric methods start from assumptions about the sampling procedure and try to cope with the limitations of a given dataset. Instead, we start at the design stage and examine the interplay between sampling and experimental methods, statistical inference and estimation of causal effects. We will use the German Business Panel as point in case and implement cutting-edge methods to gain insights into the causal mechanisms behind reported outcomes. In each session, one of the participants will present a research paper, which we will discuss in light of concrete implementation at trial scale. Participants are encouraged to present research that is valuable for their own thesis or may be assigned to present a topic.

In addition to presenting a paper and participating in the discussion, students are expected to write a short technical report that summarizes the methods and implications in a way useful for peers who want to use the newly collected data or learn about experimental results.

• As the programming part of the course starts with an introduction to Python, it is not required to have previous knowledge in Python. At the same time, programming experience will be helpful for successfully completing the course.
• As the methods covered in this course can be applied to many different settings, the course explicitly targets students/researchers from different areas, including accounting, economics, finance, marketing, and management.
• Note that the course focusses on quantitative approaches. Thus, it might not be the best fit for students who exclusively work qualitatively with text data.

The goal of this lab exercises is to guide students through the typical steps of a scientific data-science project from problem formulation to data acquisition, selection of methods, analysis and presentation / documentation. The focus of this lab will be on analyzing textual data, for example large scale news or social media datasets, using techniques and methods from the domain of natural language processing. The students will present their results and write a paper about their research.

This reading course provides a hands-on and paper-based approach to understanding and analyzing data. For many projects, collection of new data or experimental designs are the only way to answer a research question or to provide the decisive complementary evidence. Different ways to collect data can have important implications for model estimation and evaluation, parameter inference, and policy conclusions. Standard econometric methods start from assumptions about the sampling procedure and try to cope with the limitations of a given dataset. Instead, we start at the design stage and examine the interplay between sampling and experimental methods, statistical inference and estimation of causal effects. We will use the German Business Panel as point in case and implement cutting-edge methods to gain insights into the causal mechanisms behind reported outcomes. In each session, one of the participants will present a research paper, which we will discuss in light of concrete implementation at trial scale. Participants are encouraged to present research that is valuable for their own thesis or may be assigned to present a topic.

In addition to presenting a paper and participating in the discussion, students are expected to write a short technical report that summarizes the methods and implications in a way useful for peers who want to use the newly collected data or learn about experimental results.