Daten zum Projekt

One and the same data, different results - how much do the results of epidemiological studies depend on who does the analysis?

Initiative: "Experiment!" (beendet)
Ausschreibung: Explorative Phase
Bewilligung: 02.03.2021
Laufzeit: 1 Jahr 6 Monate


Epidemiological study results are usually highly scattered and metaanalyses are often performed to determine a pooled estimator. It remains open why study results vary so strongly: Is it mainly because the data sets are different, which may be due to different populations, but also due to different measurement methods and errors in data collection, or is it because the data are analyzed differently? In this project, one potential source of scattering of results - the diversity of data - will be systematically eliminated. Several epidemiologists and biometricians will be given two sharply defined research questions and identical data sets. They will be asked to work independently on these research questions, and finally their results and methods of analysis will be compared.


  • Priv.-Doz. Dr. Bernd Kowall

    Universität Duisburg-Essen
    Medizinische Fakultät
    Institut für Medizinische Informatik, Biometrie
    und Epidemiologie
    Abteilung Klinische Epidemiologie