Written and maintained by CASRAI Editorial Board
Last updated
Data Colada is a research-methods blog at datacolada.org, written by Uri Simonsohn, Joe Simmons and Leif Nelson, that publishes numbered posts on statistical evidence, data sharing and investigations of questionable research. This guide explains what the blog is, how its data-forensics posts work, and where it connects to research-integrity practice. It is aimed at research administrators, integrity officers and researchers who have seen it cited and want to read its posts critically.
What Data Colada is
The site’s tagline is “Thinking about evidence and vice versa.” Its posts are numbered (for example, post 98), and the homepage lists them as an index. The blog covers research methods, the sharing of data and materials, and analyses of published studies whose data look irregular. As read on 2026-11-07, the homepage also links to tools the authors maintain, including an R package called statuser and guidance on sharing materials through ResearchBox.
Who runs it and what it covers
The posts are written by Uri Simonsohn, Joe Simmons and Leif Nelson, either individually or as a team. Topics fall into three groups:
- Method and statistics posts on how analytic choices affect results. These sit close to topics CASRAI explains elsewhere, such as p-hacking, researcher degrees of freedom and p-curve analysis.
- Data-sharing and tooling posts, such as those about hosting data and materials on ResearchBox and OSF.
- Investigation posts that re-analyse the public data of a specific study and report anomalies.
How its data-forensics posts work
The clearest worked example is post 98, “Evidence of Fraud in an Influential Field Experiment About Dishonesty” (Simonsohn, Nelson and Simmons, posted 17 August 2021 and updated 23 April 2022). It concerns Study 3 of a 2012 PNAS paper by Shu, Mazar, Gino, Ariely and Bazerman, a field experiment with an insurance company (N = 13,488). According to the post, a 2020 PNAS follow-up by Kristal, Whillans and colleagues failed to replicate the paper’s lab studies and posted the field data on OSF; an anonymous team then analysed that file and brought it to the Data Colada authors.
The post names the checks it applied to the posted spreadsheet:
| Check | What the post reports looking at |
|---|---|
| Distribution | Reported odometer miles matched a uniform distribution from 0 to 50,000 (Kolmogorov-Smirnov p = .84) with no values above 50,000. |
| Rounding | Time 1 mileage values were rounded, Time 2 values were not. |
| Fonts | Rows in the spreadsheet were split roughly half and half between two fonts (Calibri and Cambria). |
| Sort order | The original sort order revealed pairs of near-identical records the post calls “driving twins”. |
| Simulation | A one-million-run simulation estimated how unlikely the observed pattern would be by chance. |
| File metadata | Properties of the spreadsheet file were examined as supporting evidence. |
The post states that it does not identify who altered the data, and it links the responses given by each of the four paper authors. Readers should treat the post as the authors’ analysis of a public file, with the original authors’ responses linked alongside it.
How to read and use the blog
- Open datacolada.org and use the homepage index to find a post by its number and title.
- Read the post’s own account of its data source (for instance, an OSF file) before its conclusions.
- Look for linked responses from the people whose work is discussed, as post 98 provides.
- Where a post links data or code on ResearchBox or OSF, open it to see exactly what was analysed.
- Check whether the post has an update date, since post 98 was revised after first posting.
What courts have and have not decided
One author of the 2012 paper, Francesca Gino, sued Harvard and the three Data Colada authors. According to the Harvard Crimson (12 September 2024), Judge Myong J. Joun dismissed her claims against Simonsohn, Nelson and Simmons in full, and dismissed her defamation, privacy and interference claims against Harvard. Her breach-of-contract claim against Harvard, concerning its disciplinary measures, and her Title IX claim remained. A procedural ruling on defamation claims is not a finding about the underlying data, and this page takes no position on any individual’s conduct.
Strengths and limits
- Strength: the methods are written out step by step and rely on files others can open, so readers can repeat the checks.
- Strength: the blog links responses from the people discussed rather than only the critique.
- Limit: it is an independent blog by three researchers, not a journal, regulator or research-integrity office; a post is not a formal finding of misconduct.
- Limit: forensic checks on a spreadsheet show anomalies in that file; institutions follow their own procedures, including their policies on questionable research practices and misconduct, to decide what they mean.
Related CASRAI pages
- The replication crisis and its history in psychology, the context for much of the blog’s work.
- Data dredging, a related analytic practice.
FAQ
What is Data Colada?
A blog at datacolada.org by Uri Simonsohn, Joe Simmons and Leif Nelson on research methods, data sharing and analyses of questionable research.
Is Data Colada a journal?
No. It is an independent blog with numbered posts, so it carries no peer review or formal adjudication of misconduct.
What was the outcome of the lawsuit against its authors?
Per the Harvard Crimson (12 September 2024), the judge dismissed the claims against the three Data Colada authors in full. Claims against Harvard partly remained.
Does the blog say who fabricated the data in post 98?
No. The post says it does not identify who made the alterations.
Sources
- Data Colada homepage, https://datacolada.org/ (read 2026-11-07).
- Simonsohn, Nelson and Simmons, “[98] Evidence of Fraud in an Influential Field Experiment About Dishonesty”, https://datacolada.org/98 (read 2026-11-07).
- Harvard Crimson, “Judge dismisses Gino lawsuit defamation charges”, 12 September 2024, https://www.thecrimson.com/article/2024/9/12/judge-dismisses-gino-lawsuit-defamation-charges/ (read 2026-11-07).
