Conflicting Data: Dealing with the Reproducibility Issue

Conflicting Data: Dealing with the Reproducibility Issue

28 October 2017
1:30 pm – 2:45 pm
  • Marriott Marquis: Nob Hill Room

F1) Conflicting Data: Dealing with the Reproducibility Issue

There is nothing worse than getting a story wrong. But what happens when the science itself is wrong? More than half of scientists today think that there is a substantial problem with reproducibility in research, according to a 2016 Nature survey. And the most newsworthy papers—the most novel ones from the hottest research fields and with the most controversial findings—are the most likely to turn out to be anomalies and flukes.

Studies may be difficult to reproduce for a variety of reasons—conflicting results may arise from relatively benign variations in technical setups or misinterpreted p-values. At the same time, well-publicized instances of outright malpractice and fraud threaten the public’s trust in science. This reproducibility issue poses a conundrum for science reporters, editors, and PIOs tasked with reliably reporting on new developments in science. It is no service to our audiences to report scientific findings in a way that increases the chance that much of what we report may turn out to be wrong.

This session will invite both scientists and journalists who have studied and reported on the reproducibility problem to discuss how these issues arise and how to look for red flags in scientific publications. We will also explore different responses to the issue for journalists, including a few possible approaches for improving science news coverage.