Biostatistics; Epidemiology; Regression Analysis
My main research interest lies in statistical modeling, especially with small samples or rare events. In this situation, penalized regression techniques such as Firth's penalization or ridge regression usually yield more reliable estimates than maximum likelihood estimation. For instance, one interesting question is, which penalized logistic regression models yield acceptable effect estimates as well as acceptable predicted probabilities with small samples. Another important aspect is the assessment of the model performance by resampling techniques.
On the other hand, I am interested in the use of large administrative data bases, for example health claims data bases. A thorough understanding of the origin and structure of the data combined with a careful modeling approach is necessary to prevent wrong conclusions.
- Puhr, R. et al., 2017. Firth’s logistic regression with rare events: accurate effect estimates and predictions? Statistics in Medicine. Available at: http://dx.doi.org/10.1002/sim.7273.
- Mansournia, M.A. et al., 2017. Separation in Logistic Regression: Causes, Consequences, and Control. American Journal of Epidemiology, 187(4), pp.864–870. Available at: http://dx.doi.org/10.1093/aje/kwx299.
- Geroldinger, A. et al., 2018. Mortality and continuity of care – Definitions matter! A cohort study in diabetics A. Gruneir, ed. PLOS ONE, 13(1), p.e0191386. Available at: http://dx.doi.org/10.1371/journal.pone.0191386.