Dr. Moore Examines Scientific Reliability on 270-Researcher Team

By Christopher Sarachilli | November 18, 2015

Assistant Professor of Psychology Dr. Katherine Moore is one of 270 researchers published in a Science Magazine article for her work with the Reproducibility Project: Psychology (RPP), which replicates previously published studies and compares the results to the original findings. The study was featured in The Atlantic, The New York Times, and Wired.com, among other outlets.

The project found that fewer than half of the 100 replicated psychology studies produced the same findings as the original. The authors proposed several reasons for the discrepancy. For instance, publication bias is a possibility: a particular experiment may show significant results if attempted enough times, simply due to statistical chance. Even if multiple studies find nothing conclusive, the one study with significant results often is published.

As a result, “what’s out there, what’s published, might be an inflated view of what’s going on in the world,” said Dr. Moore, who began researching with the project before starting at Arcadia this summer.

The issue stems beyond psychology, she said, noting that other sciences, such as biology, will likely produce similar percentages of replication as psychology. RPP aims to raise awareness across all sciences and introduce changes that can increase the amount of reproduction studies published and prevent anomalous results from gaining too much legitimacy.

“The project’s goals are to develop and implement better scientific practices, including those that reduce the publication and sensationalizing of statistically shaky evidence,” said Dr. Moore. Studies with null findings and replication studies, while less glamorous than those with exciting new developments, also deserve attention and have their own benefit to scientific understanding.

Dr. Moore, with student Ryan Donohue of Elmhurst College, replicated a study on visual working memory and examined participants’ ability to determine if an item in a display has changed on repeated viewings. Their results “largely matched the findings from the initial investigation,” according to the study, which can be found on Open Science Framework.

For more on her findings and RPP, contact Katherine Moore at moorek@arcadia.edu.