UVA-Led Global Research Team Aims to Understand What Makes Research Findings Credible

University of Virginia psychologist Brian Nosek is helping lead a global effort to understand how we can know when research findings can be trusted. The answer, according to a sweeping new set of studies, is far from simple.

Nosek is a key contributor to the Systematizing Confidence in Open Research and Evidence (SCORE) program, an international collaboration involving hundreds of researchers and thousands of studies in a broad range of disciplines. The project examined how well scientific findings hold up under scrutiny and whether humans or machines can predict which results will stand the test of time.

The findings, published in the journal Nature, reveal a nuanced picture of scientific credibility. About half of the findings tested could be replicated successfully, and while human experts were fairly good at predicting which results could be replicated, machine-learning approaches struggled to do so reliably.

According to Nosek, AI was not nearly as effective as the team had expected, but it does show signs of promise.

Additionally, the findings suggest that no single measure can determine whether a study is trustworthy.

“Credibility is not reducible to one single thing,” Nosek said. “There’s no single number that tells you whether to trust a finding or not.” 

Beyond a Single Measure

The SCORE program analyzed research across social and behavioral sciences, testing findings in three key ways: reproducing original analyses, testing alternative analytical approaches, and conducting entirely new studies to replicate results.

Each method revealed different strengths and weaknesses. Some findings held up under one test but not another, highlighting what Nosek describes as the inherently complex nature of scientific discovery. And according to Nosek, the takeaway is that science advances not just by confirming what works but by identifying what doesn’t.

He likens the process to pruning the dead branches of a tree. “A lot of how science makes progress is eliminating options that aren’t viable,” Nosek said. “That can be really useful.” 

In other words, by better understanding which findings are more likely to hold up, researchers and funders can make more strategic decisions about where to invest time and resources.

Advancing a Core Mission

 For Christa Acampora, dean of UVA’s College and Graduate School of Arts & Sciences, the work reflects Arts & Sciences’ strategic goal of advancing research that makes an impact.

“It is not an overstatement to say that Professor Nosek’s work addresses one of the defining concerns of our time: how we can have trust in science,” Acampora said. “What makes this work so powerful is its commitment to truth-seeking through self-scrutiny — examining our own practices with openness, rigor and integrity. That kind of inquiry strengthens rather than undermines the foundations of knowledge and the trust that makes discovery possible.”

The project also aligns with UVA’s emphasis on collaborative, cross-disciplinary research. SCORE brought together experts from multiple institutions and fields, illustrating how large-scale cooperation can tackle some of science’s most fundamental challenges.

Bringing Complexity to the Classroom

For Nosek, the implications also extend directly to teaching. One of the biggest challenges, he said, is helping both students and the public understand uncertainty.

Rather than presenting science as a series of definitive answers, he hopes this work encourages a more realistic view of discovery as an evolving process. 

“Any individual study is just a single data point,” he said. “Real insight comes from accumulating evidence over time.”