«Statisticians are second-class mathematicians, third-rate scientists and fourth-rate thinkers. They are the hyenas, jackals and vultures of the scientific ecology: picking over the bones and carcasses of the game that the big cats, the biologists, the physicists and the chemists, have brought down.»
- Stephen Senn, in: Dicing with Death (2003)
I was on the trajectory of becoming a big cat but then fully embraced the life of a hyena. I am a research fellow at the Institute of Mathematics of the University of Zurich in the Applied Statistics group of Professor Reinhard Furrer. There, I try to develop a consistent framework for good statistical practice in exploratory animal research in order to improve the reproducibility of animal studies.
Research focus
I work at the interface of applied statistics and biomedical research, with a special focus on the statistical design and analysis of exploratory research. Beyond that, I am interested in the prerequisites that need to be fulfilled for scientific knowledge to be reliable and in the societal and political role that scientific knowledge plays.
PhD thesis
Experimental design and analysis necessitate the combination of scientific and mathematical models in a consistent fashion. Every experiment usually begins with a hypothesis about the world, stated in quite general terms and of broad scope: the Primary Model. To inquire about the hypothesis scientifically, it has to be transformed into a hypothesis that is empirically testable with current methods and resources: the Experimental Model. Hence, a suitable experimental context with clearly defined parameters and outcomes must be specified. Both levels are usually stated in the scientific language of the discipline in which the experiment takes place, and only the next two levels are in the realm of statistics. First, the data generation process must be described by a Data Generation Model, followed by the specification of the experimental process in the form of the Experimental Design Model which is then combined with a particular Data Analysis Model for analysis. that specifies the statistical measures and tests that are used to analyze the data.
Publications
«The presumption that all we need is an agreement on numbers - never mind if they’re measuring different things - leads to pandemonium.»
- Deborah Mayo, in: Statistical Inference as Severe Testing (2018), xi
For a list of publications, see here. For popular science and non-scientific articles, see here.
Lectures
M-9 «Communicating Animal Research» (Talk & Exercises), organized multiple times a year in English and German by the Institute of Laboratory Animal Science of the University of Zurich. With Joel Lüthi (English), Florian Dehmelt (German), and Maike Heimann (English & German).
M-13 «Basic Statistics for Animal Experimentation» (Talk & Exercises), organized two to three times a year by the Institute of Laboratory Animal Science of the University of Zurich. With Reinhard Furrer and Bernadetta Tarigan.
M-22 «Reproducibility in Animal Research», organized once to twice a year by the Institute of Laboratory Animal Science of the University of Zurich. With Philippe Bugnon and Maike Heimann.
BME321 Design of Experiments (Exercises), organized once a year by the Institute of Mathematics of the University of Zurich. With Reinhard Furrer, Bernadetta Tarigan, Hanno Würbel, and Paulin Jirkof.
Crash Course in Statistics, organized once a year by the Neuroscience Center Zurich. With Christoph Luchsinger.
Talks
«Random allocation is not random sampling: Logical and statistical pitfalls when making inferences from randomized experiments» (Talk), 5th of February 2025, Perspectives on Scientific Error Workshop.
«Als Wissenschaftler*in in der Politik» (Vortrag), 12. Oktober 2023, MD-PhD-Retreat
Statistical Design: Why even bother? (Invited Talk), 8th of September 2022, organized by the Institute of Laboratory Animal Science.
«Sample Size Estimation for Exploratory Animal Research: A critical assessment» (invited talk), 07. June 2022, organised by the Max Planck Lab Animal Science Lectures
«Statistical Design in Animal Research: Important Principles for Experimental Researchers» (invited talk), 07. September 2021, organised by AgroVet-Strickhof.
Wissenschaft in den Medien (guest lecture in German), 8th of December 2020, «Wissenschaftskommunikation im Internet: Kommunikatoren, Inhalte und Effekte», organised by the Department of Communication and Media Research, University of Zurich.
Stupid Statistics!? (Lightning Talk), 12th of November 2019, Swiss Statistics Meeting 2019.
Writing about science - a crash course (guest lecture), 4th of November 2019, BIO 376, University of Zurich.
Natural Language Processing for automated case classification (in German), 7th of March 2019, Swiss National Insurance Fund Suva, Lucerne.
It’s the philosophy, stupid! (guest lecture, in German), 30th of October 2018, «Wissenschaftskommunikation im Internet: Kommunikatoren, Inhalte und Effekte», organised by the Department of Communication and Media Research, University of Zurich.
Writing about animal research: It’s the philosophy, stupid!, 14th of February 2018, Basel Declaration Society Conference in San Francisco.
Click on the title of the talks to see the slides. For a complete list of talks and presentations, see here.
Resources
Resources that might be useful to you if you are working in the area of animal research, study design or reproducibility and good statistical practice
Guidance document for animal researchers (preliminary version - currently being overhauled): What to consider when planning and describing your study in a license application for animal experiments. IMPORTANT NOTE: This is a living document and will be extended and adapted if deemed necessary. Also, note that it is geared toward the license application process in the Canton of Zurich but that it is not an official document issued or endorsed by the cantonal authorities or the Animal Research Commission of the Canton of Zurich.
Repositories
You can find my Github repositories here. Note that I am currently in the process of merging the different Github I created over the years, thereby overhauling the repositories worth keeping and clearing out others. As of now, only my current Master thesis project and a small pet project is visible, other repositories will become visible over time. In the meantime, share my delight for this CV I just created using pagedown!
Weight of Statistical Evidence: Detection and Correction of Publication Bias. My Master thesis at EPFL.
Stupid Statistics: A (growing) collection of statistics done wrong (or rather: statistics done in a way that could be improved). Did you stumble across a good example for bad stats? Then send me an e-mail or send a pull request with quick explanation of why you believe it is bad statistics.
Commission Work
I am a member of the Animal Research Commission of the Canton of Zurich: The commission assesses the scientific and ethical validity of animal research proposals of all animal experiments conducted in the canton of Zurich, predominantly from ETH Zürich and the University of Zurich. In my commission work, I put special emphasis on experimental design and good statistical practice along the lines that are outlined in this guidance document here. Due to the confidentiality rules that apply, I can in general not publicly comment on issues regarding individual applications. For more details, see this text here (in German).
Reviews
I reviewed scientific papers for the following journals, primarily focusing on questions of animal research and biostatistics:
Laboratory Animals, Preclinical pilot studies: five common pitfalls and how to avoid them, February 2024
Laboratory Animals, Preclinical pilot studies: five common pitfalls and how to avoid them, December 2023
Springer Nature Scientific Reports, Over 110 Million Mice and Rats Are Bred and Used in United States Laboratories, November 2020
Springer Nature Scientific Reports, Over 110 Million Mice and Rats Are Bred and Used in United States Laboratories, October 2020
Springer Nature Scientific Reports, The “Completely Randomised” and the “Randomised Block” Are the Only Common Experimental Designs Which Can Avoid Bias and Irreproducibility in Pre-clinical Research, July 2020
Springer Nature Scientific Reports, The “Completely Randomised” and the “Randomised Block” Are the Only Common Experimental Designs Which Can Avoid Bias and Irreproducibility in Pre-clinical Research, March 2020
Boasting section
In Switzerland, showcasing your achievements too openly is usually regarded as rather impolite. It’s OK to take pride in personal achievements but please don’t rub it under people’s nose! That means that I’m a bit in a pickle as the academic system demands the exact opposite behaviour, namely: Look, fellow scientists, I received this scholarship, got that grant and won those prizes - can I get tenure now?
So, as a compromise, I provide you with some successful (yay!) and unsuccessful (nay!) attempts at competing in the scientific game for funds and fame:
(Yay!) November 2024: Funding proposal «Designing Experiments for Exploration: Statistical Design in the Context of Small Sample Statistics with Applications to Exploratory Animal Research» accepted by the UZH Doc.Mobility Programme (CHF 36’150)
(Nay!) May 2023: Funding proposal «3R Practices: An interdisciplinary study of practical, epistemic, and statistical trade-offs to foster the application of the 3R principles» rejected (NRP 79 «Advancing 3R»)
(Yay!) November 2022 Funding proposal accepted for the project «Swiss Young Network for Science Policy and Diplomacy (SYNESPOD)» by the Swiss Young Academy.
(Yay!) May 2022: Funding proposal accepted for the project «How Small Is Big Enough? Tackling Challenges in Fusing Big Numbers of Small Data Sets» by the Emeritus Foundation (CHF 57’545)
(Nay!) October 2021: Funding proposal rejected for a DACH-project to assess the congruence between animal research applications and subsequent publications with regard to study design and statistical analysis.
(Yay!) June 2021: Funding proposal accepted for the «Forschungskredit Candoc» by the University of Zurich for the project «Hierarchical Fusion Modeling From Heterogeneous Data With Applications To Statistical Inference In Preclinical Research» (CHF 27’911)
(Yay!) November 2020: Funding proposal accepted for the project «What can we learn from COVID-19 fake news about the spread of scientific misinformation in general?» by the Swiss Young Academy
(Yay!) November 2020: Funding proposal accepted for the project «Who Gets Heard? Selecting Scientific Experts in Swiss Legislative Processes» by the Swiss Young Academy
(Nay!) October 2020: An application for project funding that I wrote together with my PhD advisor got rejected by the Swiss National Science Foundation.
(Yay!) May 2020: I was selected as a member of the Swiss Young Academy.
(Yay!) 2016-2019: I received the Werner Siemens Fellowship from the Swiss Study Foundation and the Werner Siemens Foundation.