Faculty members and graduate students at the University of Minnesota have formed a workshop to hold discussions about reproducibility in research studies.
The discussions come during a national movement to replicate research in social science fields, such as psychology. The movement has shown many previous studies are not reliable. After discussions last spring regarding ways the University can address these research practices, the Minnesota Center for Philosophy of Science designed workshops for faculty and students to discuss ways to develop replicable research methods.
“Any scientific discipline will depend upon reproducible findings, that’s how you build a science,” said Matt McGue, a professor in the Department of Psychology.
The Reproducibility Working Group meets biweekly this semester to discuss the issue of reproducibility in psychological research and focus on topics such as measurement.
Alan Love, director of the Minnesota Center for Philosophy of Science, said the purpose of holding these conversations across campus is for researchers across all disciplines to be actively thinking about complex issues within their own methods.
McGue said issues of reproducible research arise when researchers shift their methods or measurements to get the results they want.
In this process of shifting methods, researchers may lose the likelihood of their study being replicable, McGue said.
“The significance of this, now, is that researchers have determined that certain research practices that we thought were appropriate actually increase the likelihood that we produce results that are not reproducible,” McGue said.
Faculty members and graduate students from the philosophy, psychology and statistics departments have been attending the workshops. McGue said all members offer a unique perspective to the discussion of reproducibility as there are intersections across all three areas.
“I think it is very beneficial to have the different perspectives in a room discussing this common problem,” he said.
The movement for producing replicable research is stronger among younger researchers because they are part of these ongoing conversations, he said.
Keven Joyal-Desmarais, a Ph.D. candidate in social psychology, said he is taking what he learns from the workshops and implementing it in his own research methods.
“The [workshop] helps me further develop my thinking of reproducibility as well as better equip me to have discussions with others,” Joyal-Desmarais said.
In an effort to make research methods more sound, McGue is implementing a program for student researchers that allows them to preregister their studies so they are publicly available and minimizes the chance of data being skewed. As a wave of replication studies show that many previous studies are not replicable, McGue said that work such as these workshops cannot completely eliminate the studies that were not replicable.
“It would be conservative to demand everything be reproduced,” McGue said. “We want to address limitations in research … but we want to take risks.”