Errors in research practice frequently stem from insufficient capabilities in applying the
fundamentals of scientific reasoning. In the biomedical sciences, such mistakes are significant
contributors to the increasing numbers of article retractions and hence also exacerbate the
public’s mistrust in the scientific enterprise. Particularly in times of a global pandemic, these
tendencies can be detrimental for science and society. The role of big data in many fields of
science is continuously on the rise, hence, we need more practitioners who are not only capable
to of solving statistical problem sets on paper but able to transfer those skills into research
practice. The goal of this proposal is to produce and initially evaluate educational materials
that can help mitigate this situation. In a pilot study, we will produce the “R3easoning”
module, a guided case study approach that builds on the three R’s of good scientific
practice: Rigor, Reproducibility and Responsibility. The module showcases common
errors in the data science fields with the help of expert interviews. Experienced practitioners
from the JHSPH departments of Molecular Microbiology and Immunology, Epidemiology and
Biostatistics, as well as data management experts from the Johns Hopkins Welch Medical Library,
will provide insights into what they learned conceptually from pitfalls in scientific reasoning during
their careers in science. Students apply these concepts to their disciplinary context, formulate
recommendations for improvement, and critique each other’s rationales.
The R3easoning module is designed as an all-online, staged case study approach on basic error
analysis in data science practice. Due to the module’s subdivision into thematic units, either the
entire R3easoning module or portions can be flexibly integrated into a variety of data science
programs, depending on curricular space. The R3easoning module will be piloted and tested in a
large enrollment, graduate level, online course on statistical reasoning at the JHSPH. The course
serves graduate students across a variety of biomedical and public health sciences. This setting
provides a unique chance for course participants to broaden their research skills, communicate,
and collaborate across disciplinary boundaries.
The R3easoning module, which will be made freely available after revision and initial validation,
could be used by educators and research practitioners at several levels of their development at
other institutions to test whether differences in understanding and practice skills are measurable.
Results from the work proposed here could serve as a basis for future, long-term and larger-scale
follow-up studies across institutions and learner populations.