The Digital Causality Lab Meets AI

• Course type: Lecture and exercise
• Institution: Faculty of Business Administration
• Funding period: 01.09.2024 to 28.02.2025
• Short title: DCL Vol. 3

Alignment of the Digital Causality Lab

In a world characterised by data, it is becoming increasingly important to assess cause-and-effect relationships appropriately. Methods and tools from the field of causal inference help to analyse empirical relationships with regard to their causality. In view of the increasing amount of (false) information, it is essential to teach causal data skills: With the right knowledge, studies can be categorised and viewed critically. Students should be able to recognise systematic biases, e.g. due to background variables ("confounders") or sample selection, and assess their implications for data analysis and interpretation.

The assessment of causal relationships is an essential step in making appropriate and targeted decisions - not only in a private or professional context, but also with regard to social expertise. As part of this second funding phase of the project, the communication of causal knowledge was opened up to a broad audience.

^{UHH / Bach}

Review and results

With the help of the funding, a modern, interesting and varied course could be developed and implemented. As part of the project, a new low-threshold introductory course on methods and tools of causal inference was created. New teaching materials were created for this, based on current books, data examples and supplementary content. The focus was on teaching causal modelling intuitively. Formal, mathematical content was taught based on this. In addition, high-quality instructional videos were recorded and produced, which were embedded in an online course. The newly developed course replaces a previous subject-specific introductory course in business administration on the topic of causal inference. The didactic concepts and interactive learning apps developed in the first project phase were integrated into the new interdisciplinary course. The teaching course will remain open to students of all subjects in the future as part of the free elective area.

Students were able to give free rein to their creativity in their own project work and apply the skills they had learnt in the lecture independently. The data products developed reflect the diverse interests of the students. They cover topics on statistical methods of causal inference, statistical software, as well as newer topics such as causality and artificial intelligence. The source code and the various phases of development are available via publicly accessible GitHub repositories. A gallery of student projects can be found on the Digital Causality Lab website.

Tips from lecturers for lecturers

Hybrid teaching was an important factor in the success of the event, particularly through the conception and production of learning videos and their embedding as part of a high-quality online course. Many valuable experiences were also gained in the practical implementation and combination with a regular classroom session. The course builds on the DCL project in the second funding round, in which new didactic approaches were taken on the basis of research-based learning. As a result, the implementation of research-based learning in teaching was also improved in this funding phase, which will benefit this and other courses in the long term. It was essential not only to compare the goals set and actually achieved in teaching, but also to exchange ideas with the students. The participants themselves were very interested and also wanted to make a contribution to improving teaching.

Persons involved

Dr. Philipp Bach

Postdoc at the Chair of Statistics with application in business administration

Main focus:

• Estimation of causal effects and inference in high dimensionsal settings with machine learning methods
• Machine Learning and Deep Learning
• Applications in labor, health and financial economics

Kontakt

Prof. Dr. Martin Spindler

Professorship for Statistics with Application in Business Administration

Main focus:

• Statistics, Econometrics
• Causal Inference
• Machine Learning / AI, in particular Causal Machine Learning
• Fields of Application: Health Economics, Insurance

Kontakt