Bernhard Knapp
Research Subjects
Bernhard Knapp has a total of 56 co-authors. Here are the top co-authors by number of shared publications:
- Anastasia Meshcheryakova
Department of Pathophysiology and Allergy Research, Center of Pathophysiology, Infectiology and Immunology, Medical University of Vienna, 1090 Vienna, Austria
14 shared publications - Andrea Bileck
Department of Analytical Chemistry, Faculty of Chemistry, University of Vienna, Vienna, Austria
14 shared publications - Christine Pirker
Center for Cancer Research and Comprehensive Cancer Center, Medical University of Vienna, Borschkegasse 8a, 1090 Vienna, Austria
14 shared publications - Christopher Gerner
Institute of Analytical Chemistry, Faculty of Chemistry, University of Vienna, Austria
14 shared publications Dagmar Födinger
14 shared publicationsElisabeth Hofstätter
14 shared publications
Latest Publications
- Open Access
- 19 January 2026
- Published by Center for Open Science in SocArXiv
On the Inappropriateness of the Hodges–Lehmann Estimator for Ordinal Grading Data
The Hodges–Lehmann estimator (HL) is widely described (in statistical terms) as a nonparametric, robust, consistent and efficient estimator of central location. While this characterization is correct under standard assumptions—continuous, unimodal distributions on a meaningful metric scale—its application to ordinal grading data such as university marks is problematic. Using explicit counterexamples, we show that the HL estimator can exhibit severe jump discontinuities: minimal changes in the sample leads to large, discrete changes in the HL estimate. This behavior persists under sample size inflation by replication. This shows that the HL should not be used to represent a central location in ordinal grading data as frequently present at universities.
- 19 January 2026
- Published by Center for Open Science in SocArXiv
Open AccessOn the Inappropriateness of the Hodges–Lehmann Estimator for Ordinal Grading Data
The Hodges–Lehmann estimator (HL) is widely described (in statistical terms) as a nonparametric, robust, consistent and efficient estimator of central location. While this characterization is correct under standard assumptions—continuous, unimodal distributions on a meaningful metric scale—its application to ordinal grading data such as university marks is problematic. Using explicit counterexamples, we show that the HL estimator can exhibit severe jump discontinuities: minimal changes in the sample leads to large, discrete changes in the HL estimate. This behavior persists under sample size inflation by replication. This shows that the HL should not be used to represent a central location in ordinal grading data as frequently present at universities.
- Open Access
- 16 December 2025
- Published by Center for Open Science in SocArXiv
Is a single question sufficient? - The practical unidimensionality of course evaluations
This study analyzes the results of the student feedback questionnairesfrom the University of Applied Sciences Technikum Wien from 2010 to 2023(6, 451 courses). In these questionnaires attitudes towards eight items (modalities,structure, explanations, outcomes, practicality, integration, respectfulness,and overall satisfaction) were elicited on a Likert scale. The aggregatedata shows that responses to different items are highly correlated andcollinear, with Pearson correlation coefficients of up to 0.9. The analysis ofthe variance inflation factors and the principal components confirms this. Wesuggest that extensive questionnaires may be inefficient and that a reductionto a single question may be warranted.
- 16 December 2025
- Published by Center for Open Science in SocArXiv
Open AccessIs a single question sufficient? - The practical unidimensionality of course evaluations
This study analyzes the results of the student feedback questionnairesfrom the University of Applied Sciences Technikum Wien from 2010 to 2023(6, 451 courses). In these questionnaires attitudes towards eight items (modalities,structure, explanations, outcomes, practicality, integration, respectfulness,and overall satisfaction) were elicited on a Likert scale. The aggregatedata shows that responses to different items are highly correlated andcollinear, with Pearson correlation coefficients of up to 0.9. The analysis ofthe variance inflation factors and the principal components confirms this. Wesuggest that extensive questionnaires may be inefficient and that a reductionto a single question may be warranted.
- Open Access
ChatGPT as a medical doctor? A diagnostic accuracy study on common and rare diseases
Seeking medical advice online has become popular in the recent past. Therefore a growing number of people might ask the recently hyped ChatGPT for medical information regarding their conditions, symptoms and differential diagnosis. In this paper we tested ChatGPT for its diagnostic accuracy on a total of 50 clinical case vignettes including 10 rare case presentations. We found that ChatGPT 4 solves all common cases within 2 suggested diagnoses. For rare disease conditions ChatGPT 4 needs 8 or more suggestions to solve 90% of all cases. The performance of ChatGPT 3.5 is consistently lower than the performance of ChatGPT 4. We also compared the performance between ChatGPT and human medical doctors. We conclude that ChatGPT might be a good tool to assist human medical doctors in diagnosing difficult cases, but despite the good diagnostic accuracy, ChatGPT should be used with caution by non-professionals.
Open AccessChatGPT as a medical doctor? A diagnostic accuracy study on common and rare diseases
Seeking medical advice online has become popular in the recent past. Therefore a growing number of people might ask the recently hyped ChatGPT for medical information regarding their conditions, symptoms and differential diagnosis. In this paper we tested ChatGPT for its diagnostic accuracy on a total of 50 clinical case vignettes including 10 rare case presentations. We found that ChatGPT 4 solves all common cases within 2 suggested diagnoses. For rare disease conditions ChatGPT 4 needs 8 or more suggestions to solve 90% of all cases. The performance of ChatGPT 3.5 is consistently lower than the performance of ChatGPT 4. We also compared the performance between ChatGPT and human medical doctors. We conclude that ChatGPT might be a good tool to assist human medical doctors in diagnosing difficult cases, but despite the good diagnostic accuracy, ChatGPT should be used with caution by non-professionals.