Daniele Fanelli

Who they are

Meta-science studies how research is produced, reported and corrected. The findings are statements about the state of a literature over time, and they are not claims about any individual researcher in it. That distinction carries most of the weight on this page.

The argument of the 2018 paper is that if science were deteriorating, the indicators that track deterioration should be rising, and across the series he assembles they are not. Retractions, misconduct findings, errata and detectable statistical reporting errors do not show the upward trend the strongest versions of the crisis narrative would predict. The paper does not claim that replication rates are high, and nothing in it contradicts the direct replication projects that found many published effects do not reproduce. Those are different measurements of different things, and conflating them is the main way this work gets misused in both directions.

This site holds him for that one paper and records nothing beyond what it supports.

Key papers

Fanelli, D. (2018), "Is science really facing a reproducibility crisis, and do we need it to?". Assembles trend data on misconduct findings, retractions, errata and statistical reporting errors, and finds the trends flat.

Key ideas

Flat trends are evidence about deterioration, not about reproducibility. The paper's indicators track misconduct and error correction rather than replication success.

The maximal crisis narrative predicts rising indicators. They are not rising, and that is the paper's central point.

Meta-science describes literatures. A finding about rates of error or retraction across a field is not an allegation about anyone whose work sits in that field.

Both misuses are available. The paper is quoted to deny a replication problem that it does not address, and dismissed for failing to address it. Neither is a fair reading.

---