Drop in your manuscript and the papers it cites. ResearchMaxer finds the exact passage behind each citation and highlights it in the PDF, so you and your co-authors can confirm every one before you submit.
Free start: your first 5 source papers. No card needed.
Introduction
Most journals now send proofs as PDF files, and many authors correct them without printing a page. Whether that choice changes how many errors are caught is rarely measured. In a randomized trial of 96 graduate students, those proofreading on paper found 71% of the planted errors, compared with 58% of those reading on screen. Yet in a survey of 412 corresponding authors, 62% said they now correct proofs only on screen. Errors missed at the proof stage are the most common reason journals publish a correction. Reading on screen for long periods also causes more eye strain than reading on paper. We therefore asked every author to check the final proofs in print, and we report here how many corrections they returned.
The source reports all three numbers in your sentence: 96 students, 71% on paper and 58% on screen.
The source gives both numbers in your sentence: 412 authors answered, and 62% of them correct proofs only on screen.
Partly true, but stretched. Missed proof errors are a common reason in the source (31%), but only the second most common: changes to the author list come first (38%). Your sentence calls them the most common.
What to do: Say “a common reason” or “the second most common reason”, or cite a source that puts proof errors first.
Nothing in this source is about eye strain or reading on screen. It is a study of how editors choose peer reviewers, so there is no passage to highlight.
What to do: Check that [4] is the reference you meant, and that this PDF is that paper. If both are right, cite a study that measured eye strain, or remove the citation.
Ninety-six graduate students were randomized, 48 to each group. Participants proofreading on paper found a mean of 14.2 of 20 planted errors (71%), compared with 11.6 of 20 (58%) on screen (difference, 2.6 errors; 95% CI, 1.4 to 3.8; P < .001). Median time to finish was 19 minutes on paper and 16 minutes on screen.
| Paper | Screen | |
|---|---|---|
| Errors found, mean of 20 | 14.2 | 11.6 |
| Share of errors found | 71% | 58% |
| Median time, min | 19 | 16 |
Of 1,030 corresponding authors invited, 412 responded (40%). Most received their proofs as a PDF file (94%), and 62% said they correct proofs only on screen; 27% print some pages and 11% print every page. Authors under 40 were more likely to work only on screen (71% vs 55%).
| n | % | |
|---|---|---|
| Only on screen | 255 | 62 |
| Print some pages | 111 | 27 |
| Print every page | 46 | 11 |
We reviewed 540 corrections published in 2024. Changes to the author list were the most common reason (204, 38%), followed by errors missed at the proof stage (167, 31%) and errors in tables or figures (113, 21%). The remaining 56 (10%) had other causes.
| n | % | |
|---|---|---|
| Author list changed | 204 | 38 |
| Error missed in proofs | 167 | 31 |
| Table or figure error | 113 | 21 |
| Other | 56 | 10 |
Nothing is highlighted: no passage in this source is about your sentence.
Editors at 38 journals described how they chose reviewers for 1,206 manuscripts. Most reviewers came from the editor’s own contacts (46%) or the manuscript’s reference list (29%), and 18% were suggested by the authors. Editors invited a median of 4 people to secure 2 reviews, and 1 in 3 invitations went unanswered.
| n | % | |
|---|---|---|
| Editor’s own contacts | 555 | 46 |
| Reference list | 350 | 29 |
| Suggested by authors | 217 | 18 |
| Database search | 84 | 7 |
Sample papers. The studies and their numbers are made up for this demo.
Drop in the Word or PDF file. ResearchMaxer reads the reference list and every sentence that cites a source.
Drop their PDFs, straight from your Zotero folder if you like. Citing a web page or a Word file? Save it as a PDF first. Each one is connected to its reference by DOI or title.
See the passage highlighted in the source, with a plain verdict and a support rating. A sentence that stretches its source, or cites one with nothing on it, comes with what to do. Mark it right or flag it.
Every highlighted passage is checked word for word against the PDF before you see it. If a quote is not in the paper, it is marked, not shown as fact.
The wrong paper cited for a method. A claim stated more strongly than the source. A finding about a different group. These pass a spell check and fail peer review.
Invite the people on your paper. Everyone sees the same highlighted passages and leaves their own mark and note on each citation.
ResearchMaxer finds and flags. It never edits your manuscript, and nothing counts as checked until a person says so.
A DOI, a PubMed ID, a link, an ISBN, or a whole reference in any style. ResearchMaxer looks it up and fills in the authors, journal, volume and pages. No reference list yet? Build one this way.
Bring your library in from Zotero, Mendeley, EndNote, RefWorks, Papers, Paperpile or PubMed, and send it back out as RIS, BibTeX, EndNote XML or CSL JSON.
Pick your journal and the list is formatted its way: AMA, Vancouver, APA, and the house styles of thousands of journals. Copy it into Word with the italics kept, or download a References page.
Paste the journal’s own reference rules and a few examples, and ResearchMaxer writes the style for you. Or upload a style file you already have.
Only you and the people you invite can open a paper. There is no public sharing and no directory.
Your manuscript and sources are read to find quotes for you. They are not used to train AI models.
Remove a file and it is removed from our storage. Delete a paper, or your whole account, and everything in it goes too.
ResearchMaxer is a citation checker for researchers. You upload your manuscript and the PDFs of the papers it cites. It finds the passage in each source behind each cited sentence, highlights it in the PDF, and gives a plain verdict (supported, partly supported, not supported, or nothing related) with a support rating.
It is the AI’s estimate, from 0 to 100%, of how likely a careful co-author would accept your sentence as a fair citation of the source. A sentence that reports the source’s data accurately scores 80% or more. One that is partly true but stretches the source, calling a result the most common when the source puts it second for example, scores 30 to 79%. One the source does not back scores lower, and a source with nothing on the subject scores under 5%. Use it to see where to look first. You and your co-authors decide.
The citation is marked “Nothing related” and nothing is highlighted, with a note on what the source is about instead. This usually means a slip in the reference number or the wrong PDF. Check that the reference is the one you meant and that the PDF is that paper; if both are right, cite a source that says it, or remove the citation.
It reads your reference list and every sentence that cites a source, connects each reference to its PDF by DOI or title, and finds the exact passage behind the sentence. Each quote is checked word for word against the PDF before you see it. Then you, or a co-author, mark the citation right or flag a problem.
Your manuscript as a Word (.docx) or PDF file. Each source you cite as a PDF: a journal article, book chapter, report or guideline. A web page, Word file or Google Doc works too once it is saved as a PDF. Scanned pages with no text you can select cannot be read yet. References can come in from Zotero, Mendeley, EndNote, RefWorks, Papers, Paperpile or PubMed, or from a DOI, PubMed ID, link, ISBN or a pasted reference in any style.
Yes. Open the page, press Ctrl+P (⌘P on a Mac), choose Save as PDF, and add that file. The PDF keeps what the page said on the day you checked it, which matters because pages change, and it works for pages behind a sign-in or a paywall that you can open yourself. ResearchMaxer does not fetch pages for you: many sites turn away automated readers, and the copy you saved is the one you actually read.
No. ResearchMaxer finds and flags. It never edits your manuscript, and nothing counts as checked until a person says so.
The free start covers your first 5 source papers, with no card needed. Scholar is $9 a month ($7 a month billed annually) for 50 source papers a month. Pro is $19 a month ($15 a month billed annually) for 200 source papers a month. Prices exclude tax.
Yes. Only you and the people you invite can open it, and your files are not used to train AI models.