Revision starts the moment a draft leaves your desk and returns with notes in the margin. For students and early-career researchers, the most difficult revision request is rarely "shorten this paragraph" or "fix the p-value." It is the quiet instruction to add new evidence: a reviewer asks for a more recent study, a supervisor points to a gap in the literature section, or you realize your own argument has outgrown the papers you originally cited.
This guide walks through a complete, evidence-aware workflow for adding new evidence during manuscript revision. You will learn how to diagnose what the manuscript actually needs, choose between new literature and new data, place the new evidence without breaking the existing argument, verify every reference, and know when the revision is ready to submit. The examples center on a thesis chapter and a supervisor-feedback cycle, but the same sequence applies to course papers and first manuscripts.
What does it mean to add evidence during revision?
Adding evidence is not the same as padding a reference list. Strong revision changes the manuscript's claims, support, or scope. ICMJE guidance for preparing a manuscript for submission emphasizes that authors should follow the selected journal's current instructions for article type, format, required statements, and submission materials. That framing matters here because new evidence must fit the manuscript type and the journal's format, not just the argument in your head.
In practice, evidence added during revision usually falls into three categories:
- New literature that fills a gap a reviewer or supervisor identified.
- New analysis or data from your own work, often prompted by a methods concern.
- Clarifying evidence from sources you already have, such as an unquoted passage from a paper already in your reference list.
Before searching for anything new, decide which category the request belongs to. A supervisor who says "the discussion ignores X" is asking for new literature. A reviewer who says "the effect may be driven by confound Y" is asking for new analysis or a tighter methodological justification.
How do I diagnose what the manuscript actually needs?
Start with the revision request itself. Put each comment into one of four bins:
- Claim gap: the manuscript says something it does not support.
- Coverage gap: relevant work is missing from the background or discussion.
- Method gap: the design, sample, or analysis needs strengthening.
- Reporting gap: required information is missing from the manuscript structure.
A claim gap usually requires you to either narrow the sentence or find a source that supports the specific claim. A coverage gap sends you to the literature. A method gap may require new data work before you can add any supporting prose. A reporting gap sends you back to the journal's author instructions, not to a database.
Write the diagnosis as one sentence per comment. For example, a supervisor's note reading "you need something on reproducibility here" might become: Coverage gap: the discussion omits recent methodological literature on rigor and reproducibility. That one-sentence diagnosis becomes your search brief.
When should I search for new literature?
Search only after you have a written diagnosis. Searching from a vague feeling of "I need more sources" produces a pile of loosely related papers that make the revision harder. Searching from a diagnosis produces a shortlist.
Use the diagnosis to build a search query around two or three core terms from the specific gap. If the gap concerns transparency and rigor in experimental design, search for those terms together rather than the broad topic of your paper. WisPaper's scholarly search is built for exactly this kind of targeted hunting: you can find academic papers through the WisPaper search interface and move candidates into your working set as you screen them.
A practical rule: stop searching when you have three to five strong candidates that speak directly to the diagnosed gap. More than that usually means the diagnosis was too broad.
Which new evidence belongs in my manuscript?
Not every relevant paper earns a citation. Apply three tests before you admit a source:
- Directness: does the paper make the specific point your sentence needs, or just a nearby point?
- Recency and fit: is it current enough for your field's expectations, and does it fit the manuscript type and journal format?
- Placement: do you know exactly which paragraph it supports?
If a source fails the placement test, set it aside. A paper that is interesting but has no clear home in the manuscript will either become an ornamental citation or force you to rewrite sections you had already stabilized.
The rigor piece also matters here. NIH's framing of scientific rigor emphasizes transparent design, methods, analysis, interpretation, and reporting. When you choose evidence, prefer sources that report enough of their method and analysis for you to describe them accurately in one or two sentences. A source you cannot accurately summarize is a liability during revision.
How do I organize new evidence without losing the thread?
Revision becomes chaotic when new PDFs, notes, and citation entries live in three different places. Bring them into one project space before you start writing.
A simple, repeatable setup is to keep every source you might use in one folder or library, with a short note attached to each entry saying which comment it answers. WisPaper's design supports this directly: save or upload candidate papers into My Library, and create a dedicated Research Project for the revision so the search results, saved papers, and notes stay together.
For a thesis chapter revision, this project might contain:
- The supervisor's feedback, copied into a note.
- The three to five candidate papers you found.
- A short mapping note linking each candidate to one reviewer comment.
- The revised section draft.
This keeps the evidence addition accountable: every new citation has a visible reason to exist.
How do I insert new evidence without breaking the argument?
Placement is the highest-risk step. A new source inserted in the wrong place can create a non-sequitur, duplicate a point already made, or shift the paragraph's topic sentence away from its actual function.
Work paragraph by paragraph. For each piece of new evidence, ask:
- What is the topic sentence claiming?
- Which sentence does the new source support?
- Does the source replace an old citation or reinforce it?
Then integrate in one of three ways:
- Replace a weaker or outdated source with the new one.
- Add the source beside an existing citation where two sources jointly carry the claim.
- Extend the paragraph by one sentence that introduces the new source and what it contributes.
Resist the urge to bolt a new paragraph onto the literature review just to house a source. If the source does not change a specific sentence, it is not yet evidence.
One analytical checkpoint is to reread the paragraph aloud without the new citation, then with it. If the paragraph's point changes, you have not integrated the evidence; you have started a new argument. Rewrite the topic sentence or move the source.
Can I ask questions against the papers I already have?
Sometimes the evidence you need is already sitting in a paper you read weeks ago, but you cannot recall which section made the point. Rather than rereading every PDF end to end, you can query the working set you have saved.
When candidate papers are uploaded to My Library or added from search results, WisPaper's Library QA lets you ask questions based on those papers. This is useful for a narrow revision task: checking whether a saved paper actually reports a subgroup effect, describes a protocol, or supports the sentence you want to write. It turns a memory problem into a retrieval problem, which is much faster.
Use this as a screening step, not as the final word. Ask a focused question, find the passage, and then open the paper to read the surrounding context before you cite it.
Worked example: responding to a supervisor's reproducibiity note
Maya is finishing the discussion of her thesis chapter on lab-based behavioral protocols. Her supervisor returns three comments. The one that stalls her is: "You mention rigor but never actually define what you mean operationally. Need at least one or two current sources here."
Step 1: Diagnose. Maya writes: Coverage gap: the discussion claims rigor as a strength but does not define rigor against current methodological literature.
Step 2: Search from the diagnosis. She searches for scholarly work connecting rigor, transparency, and experimental design rather than "behavioral protocols lab." She finds several candidates and screens them for direct discussion of methodological rigor and reporting.
Step 3: Organize. She saves three strong candidates to My Library and creates a Research Project called "Thesis Ch. 4 revision — rigor sources." Each saved paper gets a note with the sentence it may support.
Step 4: Verify before writing. She checks each candidate against the specific sentence she plans to write: does the source define rigor in a way that matches her operational description? She uses her Library QA working set to confirm which papers discuss reporting transparency directly, then reads those passages.
Step 5: Integrate. In the discussion, she replaces a sentence that merely asserts rigor with two sentences: one defining rigor in terms of transparent design and reporting, and one connecting that definition to her protocol. She cites two of the three candidates. The third is discarded because it made a broader reproducibility argument that did not fit the paragraph.
Step 6: Reference check. Before sending the revised chapter back, Maya reviews every BibTeX entry in the chapter. For the two new sources, she runs them through TrueCite, which screens BibTeX entries against academic databases to flag references that do not match real records. Both check out.
Step 7: Completion check. She rereads the paragraph aloud with and without the new citations. The paragraph's point stays consistent; the new sentences now carry the rigor claim with support. She marks the supervisor's comment as addressed and moves to the next one.
Revision checklist for adding new evidence
Use this checklist before declaring the revision complete.
| Stage | Checkpoint | Pass condition |
|---|---|---|
| Diagnose | Every supervisor or reviewer comment has a one-sentence gap label. | No comment is left as vague feedback. |
| Search | The search brief uses terms from the gap, not the whole paper topic. | Three to five candidates per gap maximum. |
| Screen | Each candidate passes directness, fit, and placement tests. | Every candidate has a specific target sentence. |
| Organize | All candidates live in one project or library with mapping notes. | Anyone can see why each source was added. |
| Integrate | Each new source replaces, reinforces, or extends a specific sentence. | No stand-alone citation exists without a supporting role. |
| Verify | Every new reference entry matches a real record. | No hallucinated or malformed citations remain. |
| Reread | Each changed paragraph reads cleanly with and without the new evidence. | The paragraph's point does not shift when the citation is removed. |
How do I know the revision is ready to submit?
Ready means the manuscript satisfies three conditions at once.
First, every revision comment has a visible response in the text or in a rebuttal note. If a request for new evidence produced no change in any sentence, the revision is not done.
Second, every new claim is narrower than its support. You should be able to point to the exact sentence or passage in each source that carries the claim. If a paragraph says more than its sources do, narrow the paragraph before submitting.
Third, the reference list is mechanically sound. New sources can enter with broken metadata, wrong years, or mismatched author lists. Screening new BibTeX entries through TrueCite before submission catches references that would otherwise reach a reviewer in a damaged state.
If all three conditions hold, the evidence addition served the manuscript rather than adding noise. That is the difference between a longer draft and a stronger one.




