Metascience

Metascience — also called meta-research or, informally, “the science of science” — is the empirical study of how research itself is designed, conducted, reported, evaluated, and rewarded. Rather than answering a first-order question within a specific discipline (does this drug lower blood pressure; does this material conduct electricity), metascience treats the research process as its own object of study: whether studies are adequately powered, how often results replicate, what share of published findings later prove false, how peer review and publication practices shape what gets published, and how funding and promotion incentives shape researcher behavior.

The field crystallized under this name over the last decade, building on earlier statistical critiques such as John Ioannidis’s 2005 PLOS Medicine paper “Why Most Published Research Findings Are False.” Dedicated infrastructure followed: the Center for Open Science (COS), founded in Charlottesville, Virginia in January 2013 by Brian Nosek and Jeffrey Spies, and Stanford’s Meta-Research Innovation Center (METRICS), launched in 2014 and co-directed by Ioannidis and Steven Goodman — both organizations fund and run empirical studies of research practice rather than substantive science in any one discipline. A widely cited 2015 PLOS Biology paper by Ioannidis, Fanelli, Dunne, and Goodman, “Meta-research: Evaluation and Improvement of Research Methods and Practices,” frames the field around five thematic areas: methods, reporting, reproducibility, evaluation, and incentives.

Research waste and reproducibility are two of metascience’s central, most-cited concerns. Iain Chalmers and Paul Glasziou’s 2009 Lancet analysis, “Avoidable Waste in the Production and Reporting of Research Evidence,” estimated that roughly 85% of biomedical research investment was avoidably wasted through poor question selection, weak design, incomplete reporting, and restricted access to results. Large coordinated replication efforts, such as the Center for Open Science’s Reproducibility Project: Psychology, apply the same empirical lens directly to a field’s actual reproducibility rate rather than any single study’s result.

Why this is metascience-adjacent for CASRAI. CASRAI does not conduct metascience research itself — it is a standards body, not a research group — but a meaningful share of its own work exists to supply the standardized, machine-readable infrastructure that metascience research depends on. CRediT (jointly stewarded by CASRAI and NISO as ANSI/NISO Z39.104-2022) turns “who did what on a paper” into structured, comparable data instead of free-text acknowledgments, which is exactly the kind of dataset a metascience study of authorship or credit allocation needs to exist before it can be studied at scale. The same is true of machine-actionable data management plans, persistent-identifier interoperability (ORCID, ROR, RAiD, DOI), and the Dictionary’s controlled vocabularies more broadly: each converts an aspect of the research process into a form that can be counted, compared, and audited across a field, not just described in prose in any one paper. Responsible-assessment initiatives such as DORA (the San Francisco Declaration on Research Assessment) and CoARA (the Coalition for Advancing Research Assessment), which push institutions away from journal-level metrics like the Journal Impact Factor toward broader, more direct evaluation of research and researchers, are themselves grounded in metascience findings about how those metrics distort incentives — and CASRAI’s standards work is part of the infrastructure that makes it practically possible to evaluate research assessment reform in the first place.

See also: Reproducibility, Reproducibility crisis, Research integrity, P-hacking, Registered report, Replication study.