Member Research and Reports

Member Research and Reports

WashU: Shift the Research Culture to Boost Reproducibility in Public Health Research

Recent increases in research retractions have suggested that new strategies are needed to maintain or improve research quality in public health and other fields. While public health researchers use some promising research practices, there is room for improvement, according to new research from the Brown School at Washington University in St. Louis.

Replication, or conducting an entire study again to verify results, is the gold standard for ensuring research quality. When replication is not feasible, reproducing study results given the data source is one alternative. Brown School researchers surveyed 247 members of the American Public Health Association Applied Public Health Statistics section in late 2017. They found that the participants included many of the methodological details necessary for reproducibility in their publications. However, few participants shared statistical code, data, or both. Likewise, many reported that their data and code would be difficult for colleagues to find if they left their institution. Sharing data and code are important for facilitating reproducibility.

The most-cited barriers to sharing were data privacy and lack of time. Participants suggested training researchers on reproducible research practices and requirements by journals and funders could increase the use of reproducible research practices. The Brown School team developed a reproducibility toolkit for researchers to address some of the barriers identified by participants.

“Solving this problem will require a major cultural shift by the public health community,” concluded the lead author, Dr. Jenine Harris, associate professor at the Brown School. “Funders and journals rewarding or requiring reproducible research could encourage changes in current practice, as could incorporating reproducible research into public health degree programs.”

The paper was published September 12 in PLOS One.