Against Anonymising Meta-Scientific Data

There is an emerging obstacle to the replicability, credibility, and value of meta-science: the practice of anonymising study-level datasets, such as published articles, describing research artefacts. In this paper, we 1) highlight how this trend in meta-science is at odds with norms established in the broader field of research synthesis, 2) explain how this practice greatly reduces our capacity to verify, reproduce, or build on meta-scientific research, 3) explore common arguments and justifications for anonymisation, 4) provide counterpoints to these common justifications, and 5) demonstrate the adverse practical implications and epistemic costs for both meta-science specifically and science generally. If anonymisation continues as a practice, meta-science risks normalising a standard for itself that it would rightly criticise in any other area of empirical research, undermining its capacity to guide reforms as the underlying evidence is rendered unverifiable.