Data manipulation

Introduction

TipDefinition

Data manipulation generally refers to cases where a study has been conducted but the reported data has been falsified in some way to make the result more impressive. We will also include here data fabrication, where an entire dataset may be invented. Both these ways of manipulating data correspond to research fraud (McIntosh et al. 2026).

GRIM consistency testing via ERROR (Estimating the Reliability & Robustness of Research) web application. Screenshot reproduced from the COSIG repository under CC BY-NC-SA license (Ozturk et al. 2026).

Key resources

Note“Detecting inconsistencies and fraud”

Review of techniques to detect possible data manipulation by Brown and Carlisle (2025).

NoteCOSIG

• Data forensics using Microsoft Excel: https://osf.io/2kdez/files/bz725
• Evaluating binary classifiers: https://osf.io/2kdez/files/pvr4a
• Evaluating multiclass classifiers: https://osf.io/2kdez/files/5x2rp
• Granularity-related inconsistency of means (GRIM): https://osf.io/2kdez/files/67bp4
• Standard deviation vs. standard error: https://osf.io/2kdez/files/hp4yd
• Multiple hypothesis correction: https://osf.io/2kdez/files/csxd5
• Data leakage: https://osf.io/2kdez/files/d8mzb
• Pseudoreplication: https://osf.io/2kdez/files/hyxvr

More information on COSIG: Ozturk et al. (2026)

Bibliography

Brown, Nicholas J. L., and John B. Carlisle. 2025. “Detecting Inconsistencies and Fraud in Research Data: Time for Authors to Share the Data Underlying Their Summary Statistics as a Matter of Course.” Anesthesia & Analgesia, ahead of print, December 16. https://doi.org/10.1213/ANE.0000000000007889.

McIntosh, Leslie D., Anna Abalkina, Achal Agrawal, et al. 2026. “Forensic Scientometrics (FoSci) Report 2026: Understanding, Detecting, and Documenting Manipulation in the Research Ecosystem.” Version 3. Preprint, Figshare, July 17. 1012969 Bytes. https://doi.org/10.6084/m9.figshare.32178456.v3.

Ozturk, Yagmur, Solal Pirelli, and Reese A. K. Richardson. 2026. “Introducing COSIG: The Collection of Open Science Integrity Guides.” PLOS Biology 24 (6): e3003817. https://doi.org/10.1371/journal.pbio.3003817.