Check every number in your paper before anyone else does.
We rebuild your analysis from your data and methods, compare it against every number you report, then review whether your conclusions hold up.
Per manuscript, with up to three passes for your revisions. Everything happens by email.
| Reported value | Published | Reproduced | Result |
|---|---|---|---|
| Analytic sampleFigure 1, flowchart | 9,412 | 9,412 | Matches |
| Adjusted ORTable 2, model 3 | 1.42 | 1.42 | Matches |
| PrevalenceWomen 50+, Table 3 | 12.8% | 11.3% | Differs |
| 95% CITable 2, model 3 | 1.18–1.71 | 1.18–1.71 | Matches |
Errors in published papers are common. Peer review often misses them.
A study of more than 250,000 p-values in eight major psychology journals found that half of the papers contained at least one p-value that didn't match its own test statistic. One in eight had an error large enough to change the statistical conclusion.
Those papers all passed peer review. Once yours is published, anyone can check it. Make sure it holds up first.
How it works
There's no account to create and no portal to learn. You pay once, send your files by email, and the report comes back the same way.
-
Pay for a review
Check out securely through Stripe. Use the email address you'll send your manuscript from.
-
Email your manuscript
Send the manuscript, supplementary files, and your code if you have it. Tell us which dataset you used.
-
Get your report
You'll receive a written report by email. Reply with your revised draft for a second or third pass within 90 days.
What the review covers
Every review answers two separate questions, because a paper can pass the first and still fail the second.
Part one
Do your numbers reproduce?
We write an independent implementation of your analysis from your data and methods section, then compare it with every value you report.
- Sample sizes and exclusions at each step
- Estimates, intervals, and p-values in every table
- Agreement between text, tables, and figures
- If you share code, whether it does what your methods say
Part two
Do your conclusions hold?
We review whether the analysis supports what the paper claims, and where possible we run the corrected analysis to show what changes.
- Model specification, adjustment sets, and survey design
- Variables that belong outside the model, such as mediators
- Sparse categories and unstable estimates
- Claims in the abstract that the results don't support
Findings are ranked as major, moderate, or minor, each with a specific recommended edit. The code and outputs behind every finding are available to you on request.
Pricing
One price per manuscript, covering the original draft and two rounds of revisions.
One manuscript
$99.99
Founding price
- Both parts of the review
- Up to 3 passes within 90 days of your first submission
- Written report with recommended edits
- Supporting code and outputs on request
Labs and groups
Prepaid manuscript credits at a discount, paid by invoice.
- Credits are valid for 12 months
- Any lab member you name can use them
- Purchase orders accepted
Free evaluation
See the depth of the review before you buy.
- One pass on a single section or key analysis
- One free evaluation per author
- Same email workflow as a full review
What we need from your data
The full review depends on whether we can get the data you used. You never need to send us data your agreement doesn't allow you to share.
Public or registration-based data
Datasets such as NHANES, DHS, CHARLS, or CDC WONDER. We obtain the data under our own access and run the full two-part review.
Restricted or private data
We review your code and statistics against the manuscript without re-running on the data. Your report will say clearly which checks this covers.
Questions
Who can submit a manuscript?
Only an author of the manuscript. When you submit, you confirm that you're an author. We can't review papers sent by editors or peer reviewers, because most journals don't allow manuscripts under review to be shared.
Is my manuscript kept confidential?
Yes. We don't share or publish anything about an unpublished manuscript. If your paper is published later, we'll only feature its review with your permission.
Your files are deleted 90 days after your last pass.
Do you use AI?
Yes. An AI-assisted pipeline builds the independent implementation and drafts the findings. A person reviews every report before it's sent to you.
How long does a review take?
Most reports arrive within 5 business days of receiving your files. Revision passes are usually faster.
Does a clean report mean my paper will be accepted?
No. The review checks whether your numbers reproduce and whether your analysis supports your conclusions. It doesn't predict what a journal or its reviewers will decide.
What if my submission is very large?
Very large submissions, such as extensive supplements with many models, may need a quote before we start. We'll tell you before any work begins.
What if you can't run the review?
If we can't get the data or can't run the analysis you describe, we'll tell you what we can check instead. If you'd rather not continue, you get a full refund.