Interdisciplinary research is often hailed as essential for solving society’s biggest problems. However, conventional wisdom says that interdisciplinary work is penalized in peer review, making it more difficult to publish and get funded.

In a new paper, University of Michigan School of Information PhD candidate Sidney Xiang, along with UMSI associate professor Daniel Romero and assistant professor Misha Teplitskiy, found that this conventional wisdom is only partially true. It depends on how one defines “interdisciplinarity.” 

They find that work rooted in interdisciplinary references was associated with higher acceptance rates. In contrast, work with titles and abstracts that cross disciplinary boundaries was less likely to be accepted. However, this boundary-crossing penalty disappears when the paper includes diverse references or is submitted to an explicitly interdisciplinary journal, showing the importance of alignment between interdisciplinary topics, knowledge base and audience.

“The negative association between interdisciplinary topics and evaluations may have to do with the difficulty of pleasing multiple audiences, but interdisciplinary references may reflect the benefits of drawing on a larger pool of non redundant information,” Teplitskiy says.

“Evaluating Interdisciplinary Research: Disparate Outcomes for Topic and Knowledge-Base” was published in the April issue of the Proceedings of the National Academy of Sciences. It analyzes peer review data from more than 128,000 submissions across 62 STEM journals from the Institute of Physics Publishing (IOP Publishing).

The significance of this research comes from the large number of papers the researchers analyzed and the way interdisciplinarity was measured not just by topic, but also by the diversity of sources each paper drew from, allowing for a clearer understanding of how different types of interdisciplinarity affect peer review outcomes.

“Our paper highlights the importance of thinking about the multiple ways interdisciplinarity is involved in the scientific process and the alignment between these dimensions, such as methods, teams, topics, references and audience,” Romero says. “Our findings suggest that when these align, the conventional interdisciplinarity penalty disappears.”

One key contribution of the paper is the unique dataset the researchers drew from, which includes both accepted and rejected papers. 

“That’s important because most studies in this space can only look at successfully published papers, which is a biased sample of all papers being reviewed,” Xiang says. “To fully understand the evaluation dynamics, it was important to have a sample including unsuccessful cases as well.”

Xiang, a third year PhD candidate at UMSI and lead author of the paper, joined in 2022. With a background in applied mathematics and physics, Xiang enjoys studying the science of science and looking at how computational methods like data mining, simulations, machine learning and network analysis impact social phenomena and human behavior.

Xiang says she learned a lot about how to tackle big projects through this research. 

“My advice for aspiring PhD students, especially those coming straight out of undergrad, is that one of the key things you learn in a PhD program is how to tackle high-uncertainty projects with no clear end date,” she says. “These long-horizon projects are a big part of the experience, and it’s something you have to adapt to during the PhD.”

“Evaluating Interdisciplinary Research: Disparate Outcomes for Topic and Knowledge-Base” was published in the April issue of the Proceedings of the National Academy of Sciences. Read the abstract below: 

Interdisciplinary research is essential for addressing complex global challenges, but there are concerns that scientific institutions like journals select against it. Prior work has focused largely on how interdisciplinarity relates to outcomes for published papers, but which papers get accepted for publication in the first place is unclear. Furthermore, journals may evaluate two key dimensions of interdisciplinarity,—topic and knowledge base,—differently. Topic interdisciplinarity (measured through title and abstract) may incur evaluation penalties by cutting across disciplinary evaluation standards and threatening symbolic boundaries, while knowledge-base interdisciplinarity (measured through references) may incur benefits by combining a large pool of nonredundant information. Evaluations may also depend on how well these dimensions align with each other and the intended audience. We test these arguments using data on 128,950 submissions to 62 journals across STEM disciplines, including both accepted and rejected manuscripts. We find that a 1SD increase in knowledge-base interdisciplinarity is associated with a 0.9 percentage-point higher acceptance probability, while a 1SD increase in topic interdisciplinarity corresponds to a 1.2 percentage-point lower acceptance probability. However, the penalty for high topic-interdisciplinarity diminishes when knowledge-base interdisciplinarity is also high, and when submitted to journals designated as “interdisciplinary.” These findings challenge the narrative of a uniform bias against interdisciplinary research and highlight the importance of distinguishing between its dimensions, as well as their alignment with each other and the intended audience.

RELATED

Learn more about Sidney Xiang, Daniel Romero and Misha Teplitskiy by visiting their UMSI profiles. 

Check out UMSI’s PhD in Information program and apply today!