Research Integrity

A research support service can easily slide into practices that damage trust: inflating claims, hiding limitations, producing text that misrepresents authorship or implying that results are guaranteed. We treat integrity as a design constraint for our service. This page explains the principles we follow, the requests we decline and the responsibilities that remain with clients. It complements the research integrity policy page, which sets out the commitments in a more formal structure.

We do not fabricate, falsify or selectively omit data. We do not generate invented results to fit a hypothesis, and we do not adjust analyses to reach a desired conclusion. If an analysis shows a result that conflicts with the client's expectation, we report it. If data are insufficient, we say so and describe what additional data would be needed.

Where we perform analysis, we document the steps in scripts or notes so that the client can reproduce and verify them. Good documentation also protects the client if questions arise later.

Authorship criteria used by many journals require substantial contribution to conception, design, data acquisition or analysis, drafting or critical revision, final approval and accountability for the work. Support services usually do not meet these criteria for the client's research and should be acknowledged where journal policy requires or permits it, rather than listed as authors, unless they genuinely meet the criteria.

We decline to produce complete manuscripts for clients to submit as their own work when the client did not carry out or understand the research. We support authors in writing and improving their own papers, and we ask clients to confirm that they have read and approve the final version.

Citations must be real, relevant and accurately represented. We do not invent references or cite sources we have not checked. Generative tools can produce plausible but false citations; any reference suggested by such a tool must be verified against the source. We ask clients to check all references in the final manuscript.

We also discourage excessive self-citation or citation manipulation intended to inflate metrics.

Figures must faithfully represent data. Permissible adjustments such as brightness changes must be applied to the whole image and disclosed where journal policy requires. Splicing, cloning or selective editing that alters the meaning is not acceptable. Plots should show the full range of the data where this matters and should not hide variability.

We re-draw figures from underlying data and keep the data files and scripts.

We do not copy text from other sources without attribution. Reusing one's own previously published text also needs care and, in many cases, citation or permission. Similarity checks are tools to detect problems, not proof of originality. Paraphrasing a source closely without citation is still plagiarism.

We help authors cite properly and restructure text, and we advise on journals' policies on text recycling.

Journals differ in their policies on generative tools, but common themes are that tools cannot be authors, that use must be disclosed and that authors are responsible for the content. We do not use generative tools to invent data or references. Where tools assist with language or drafting, the output must be reviewed by a person.

Clients should follow the policy of the target journal and their institution.

We ask clients to disclose conflicts of interest in the manuscript according to journal policy. We do not review work for which we would be a conflicted peer reviewer, and we do not accept requests to produce positive reviews for a client's paper or to manipulate peer review.

We avoid arrangements that would compromise independence, such as writing both a paper and an unaffiliated endorsement of it.

Before engaging any editing or research-support service, authors can ask: Will I remain the author who understands and can defend every claim? Will the service refuse to fabricate or alter data? Will references be real and checkable? Will the service respect confidentiality and tell me how my files are handled? Does it avoid guarantees of publication? Does it disclose limits of its competence? Will it put the agreement in writing?

We are happy to answer each of these in writing and invite clients to use the checklist with any provider.

Acceptable: an author asks for help restructuring a discussion section so that limitations are clear. Acceptable: an author asks for a review of the statistical analysis and wants to know whether the tests are appropriate. Acceptable: a team asks for help drafting a response to reviewers that addresses each comment with evidence. Not acceptable: a request to add references that were not read to make a paper look better supported. Not acceptable: a request to generate data to match a model. Not acceptable: a request to rewrite a published paper to evade similarity checks.

These examples are illustrative of how we draw the line. Borderline cases are discussed openly before work starts.

Recognised organisations publish guidance on authorship, peer review, data sharing, conflicts of interest and handling of misconduct allegations. Authors should consult the policies of their target journal, publisher, institution and funder, which may differ. Training in research integrity is offered by many universities and research councils.

We encourage students and early-career researchers to use these resources and to discuss concerns with their supervisors and research integrity officers.

Can you guarantee that a reviewer will not find problems? No; we can reduce avoidable problems but not remove all risk. Can you remove similarity-check flags? We can help you cite and restructure text legitimately, but we will not disguise copied text. Can you add citations to strengthen the paper? Only to sources that are relevant and that you have read or that we verify together. Can you make my results look better? No; we can help present them clearly.

Will you tell the journal that you helped me? You decide what to disclose according to journal rules; we will not misrepresent our involvement.