# What to Include in a Research Code of Conduct By the end of this guide, you will be able to draft a complete research code of conduct for a lab, department, or research group. You will know which sections belong in the document, what each section should commit people to, how to adapt a research code of conduct template to your setting, and how to roll it out so that people actually read and follow it. To follow along, you need a plain-text editor, your institution's existing policies on misconduct, [data management](/blog/guides/data-management-basics-principles-processes-and-best-practices), human subjects, and animal care, and optionally a Git repository or shared drive where you can version the document. Nothing else is required. If you are a student or early-career researcher without formal authority, this guide also shows you how to propose a code to your principal investigator in a way that is easy to accept. ## What a Research Code of Conduct Is, and What It Is Not ![Comparison table of institutional misconduct policy and a lab research code of conduct across who it binds, what it defines, and what it answers.](/article-images/illustrations/research-code-of-conduct-template-code-vs-policy.webp) A research code of conduct is a short, enforceable document that states how members of a research group agree to behave. It converts broad principles from [research ethics](/blog/guides/research-ethics-fundamentals-a-practical-guide-for-early-career-scientists), such as honesty, accountability, and fairness, into specific commitments with named owners and clear reporting paths. It is not the same thing as an institutional misconduct policy. Your university or institute almost certainly has formal rules that define fabrication, falsification, and plagiarism and that set out investigation procedures. Your code of conduct sits underneath those rules and makes them concrete for your group. It answers questions the institutional policy never will, such as who decides authorship on your next paper, where data live, and what a new graduate student does on day one. It is also not a substitute for training. Studies of responsible conduct of research (RCR) education show that training works best when it is active, case-based, and tied to clear learning objectives rather than passive compliance modules [1]. The code and the training reinforce each other. The code states the standard. The training builds the judgment to apply it. Finally, a code is not a one-time writing exercise. Research performing organizations that treat conduct as an ongoing cultural project, with accessible resources, clear reporting pathways, and repeated measurement, see measurable gains in researcher awareness of those resources [2]. Groups that treat the document as paperwork tend to end up with policies that are technically complete but poorly disseminated, a problem researchers themselves report when interviewed about their institutional climate [3]. ## The Core Sections to Include in a Research Code of Conduct A workable template has eight sections. Each section answers one question that will otherwise come up repeatedly and get answered inconsistently. | Section | Question it answers | Must-have elements | |--|--|--| | Scope and values | Who is bound by the code and what principles guide it | Applies to everyone with access to lab data, including visitors and collaborators | | Research integrity and misconduct | What counts as fabrication, falsification, plagiarism | Definitions, concrete examples, escalation path | | Authorship and credit | Who is an author and who decides | Criteria, timing of authorship discussion, contributor statements | | Data and records | Where data live, how they are named, how long they are kept | Storage location, naming convention, retention period, backup | | Human and animal subjects | What approval and consent are required before work starts | IRB or IACUC approval before data collection, documented consent | | Conflicts of interest | What must be disclosed and when | Financial, personal, and social media conflicts, disclosure process | | Reporting and publication | How results are communicated honestly | No spin, preregistration where applicable, AI-use disclosure | | Reporting concerns | What to do when something goes wrong | Named contact, alternative contact, no-retaliation statement | The subsections below explain what goes into each one and why. ### Scope and Values Open the code with two or three sentences that say who it covers. In a lab setting, cover everyone with access to data, notebooks, or code. That includes undergraduates, visiting students, collaborators, and core facility users. A code that only binds trainees while leaving senior members outside its scope will not be taken seriously, and junior researchers consistently identify the laboratory environment and pressures from principal investigators as barriers to integrity [3]. Follow the scope with a short statement of values. Keep it to honesty, accountability, transparency, and respect. Resist the temptation to write a philosophy essay. The rest of the document does the specific work. ### Research Integrity and Misconduct State the three standard categories of research misconduct in plain language and give one local example of each: 1. Fabrication, meaning making up data or results. 2. Falsification, meaning changing or selectively omitting data, images, or results so the record misleads. 3. Plagiarism, meaning using another person's text, ideas, code, or data without attribution. Then go one level deeper than most institutional policies do. Address day-to-day practices that fall short of formal misconduct but still damage the record. The clearest example is spin, the practice of presenting or interpreting results to make them look more favorable or significant than they are. Spin is a form of reporting bias that misleads readers, clinicians, and policymakers, and it accumulates across the literature into a distorted scientific record [4]. Your code should state plainly that selective reporting of outcomes, overstated claims in abstracts, and cherry-picked figures are violations even when the underlying data are real. ### Authorship and Credit Authorship disputes are among the most common and most preventable integrity problems in collaborative research. Published authorship guidelines vary widely between journals and institutions, which is exactly why a local code helps. A responsible authorship culture rests on three principles: transparency, credit, and accountability [5]. Write your authorship section to cover three commitments: - Authorship is discussed at the start of a project, not after the manuscript is drafted, and revisited when contributions change. - Every author meets agreed criteria for substantial contribution, and anyone who does not is listed in an acknowledgments or contributor statement instead. - Every author's specific contribution is recorded, for example in a structured contribution statement, so that credit and accountability are transparent. ### Data and Records This section prevents the most common reproducibility failure, which is not fraud but chaos. Specify: - The canonical storage location for raw data, for example an institutional server or approved cloud folder, and the rule that laptops are never the only copy. - A file naming convention, for example `2026-03-15_experimentA_plate2_raw.csv`, so files sort chronologically and experiments are identifiable without opening them. - A [lab notebook](/knowledge/molecular-biology/lab-notebook) policy, whether electronic or paper, including who can read it. - A retention period matched to your funder and journal requirements. - A rule that data underlying a published figure are preserved and shared on reasonable request. Formal training surveys show that many students have limited awareness of [data lifecycle management](/blog/guides/data-lifecycle-management-in-research-from-collection-to-preservation), so your code should not assume this knowledge exists. It should spell the lifecycle out. ### Human and Animal Subjects State the non-negotiable rule: no human or animal data are collected until the relevant approval, such as Institutional Review Board (IRB) or Institutional Animal Care and Use Committee (IACUC) approval, is documented, and no human participant contributes data without documented consent. This is worth being strict about because the failure mode is real and persistent. In a systematic review of craniosynostosis studies, IRB approval was reported in only 48 percent of included studies, and documentation of imaging consent appeared in only 3 percent, even though many studies used clinically obtained CT scans of children, a population where radiation exposure carries real risk [7]. A one-paragraph code section that requires approval numbers to be recorded in the project log before enrollment begins is cheap insurance against this entire class of problem. ### Conflicts of Interest A conflict of interest exists when personal, financial, or professional interests could bias, or appear to bias, professional judgment. In fields that intersect with industry, these conflicts are pervasive, and evidence shows inaccurate disclosure is common, with a large majority of authors in one surgical field failing to report conflicts correctly [8]. Cover three things: - What must be disclosed, including financial ties, royalties, personal relationships, and paid social media activity. - When disclosure happens, meaning at project start, at manuscript submission, and whenever circumstances change. - What happens after disclosure, meaning the conflicted person steps back from decisions where the conflict applies, such as selecting reagents from a company they consult for. ### Reporting and Publication, Including AI Use Commit the group to honest reporting. Add an explicit AI clause, because journals and publishers now increasingly expect transparency about artificial intelligence use and many follow frameworks such as CONSORT-AI for reporting AI-involved trials [9]. A workable clause has three parts: 1. Generative AI tools may be used for language polishing and coding assistance but not for generating data, analyses presented as the author's own, or peer review of others' manuscripts. 2. Any substantive AI use is disclosed in the methods or acknowledgments, with the tool and version named. 3. The human author remains fully responsible for all content, including checking every AI-suggested reference for existence and accuracy. ### Reporting Concerns and No Retaliation This is the section that determines whether the rest of the document functions. Research culture surveys show that significant improvements in researcher trust come specifically from improved awareness of reporting pathways for inappropriate behavior, not from policy text alone [2]. Include: - A named first contact, usually the principal investigator or lab manager. - A named alternative contact outside the direct reporting line, such as an ombudsperson, research integrity officer, or department chair, because the first concern is sometimes about the first contact. - A no-retaliation statement that covers good-faith reports made in error. - A statement of what the reporter can expect, meaning acknowledgment, confidentiality within limits, and escalation to institutional channels when needed. Perceived weak enforcement is a documented driver of misconduct, particularly among trainees [10]. If people believe nothing happens when rules are broken, the code is decorative. The alternative-contact mechanism is the single most practical feature for combating that belief. ## A Free Research Code of Conduct Template, Built Step by Step The following walkthrough builds a complete document in a version-controlled repository, which makes review and updates traceable. If you skip the Git parts and work in a shared document, every other step still applies. ### Step 1. Create the Skeleton Create a file named `CODE_OF_CONDUCT.md` in your lab handbook repository and add the eight section headings. In a terminal: ```bash mkdir lab-handbook && cd lab-handbook git init touch CODE_OF_CONDUCT.md ``` Then paste the skeleton into `CODE_OF_CONDUCT.md`: ```markdown # [Group Name] Research Code of Conduct Version 1.0 - Adopted [DATE] - Review due [DATE + 1 year] ## 1. Scope and Values ## 2. Research Integrity and Misconduct ## 3. Authorship and Credit ## 4. Data and Records ## 5. Human and Animal Subjects ## 6. Conflicts of Interest ## 7. Reporting and Publication, Including AI Use ## 8. Reporting Concerns ``` Expected output when you run `git status`: ``` Untracked files: CODE_OF_CONDUCT.md ``` Commit it: `git add CODE_OF_CONDUCT.md && git commit -m "Add code of conduct skeleton"`. The version header matters more than it looks. A dated version number lets everyone tell whether they signed the current code or an old one. ### Step 2. Fill Sections 1 and 2 Replace the headings with text. A worked example for a [molecular biology](/blog/careers/molecular-biology) lab: ```markdown ## 1. Scope and Values This code applies to everyone working in the Martinez Lab or with its data, including undergraduates, visiting researchers, and external collaborators. We commit to honesty in data and communication, accountability for our work, transparency in methods and decisions, and respect for colleagues and research participants. ## 2. Research Integrity and Misconduct We do not fabricate, falsify, or plagiarize. - Fabrication: inventing data, images, or results. Example: entering a band on a gel figure that no sample produced. - Falsification: altering or selectively dropping data to change a conclusion. Example: excluding outlier animals post hoc without a predeclared exclusion rule. - Plagiarism: using text, code, figures, or ideas without attribution, including reuse of our own previously published text without citation. We also treat selective reporting, overstated abstracts, and misleading figure presentation as violations, even when the underlying data are real. ``` ### Step 3. Fill Sections 3 through 6 Continue in the same style. Each commitment should be checkable. Compare these two versions of the same rule: Weak: "Authorship will be handled fairly." Strong: "Authorship is discussed at project start and documented in the project log. Any change requires agreement of all current authors before submission." For section 4, write the actual convention, not a principle about conventions: ```markdown ## 4. Data and Records - All raw data go to /lab-share/martinez/raw/ on the department server within 48 hours of collection. Laptops are never the only copy. - File naming: YYYY-MM-DD_experiment_condition_runA/B/C_operator.csv - Electronic notebooks are exported to PDF monthly and stored in /lab-share/martinez/notebooks/. - Data supporting published figures are retained for 7 years and shared on reasonable request, subject to participant privacy constraints. ``` For section 5, keep it to two enforceable sentences plus a bookkeeping rule: ```markdown ## 5. Human and Animal Subjects No human or animal data are collected before IRB or IACUC approval is documented, and no human participant contributes data without documented informed consent. The approval number and approval date are recorded in the project log before the first enrollment or procedure. ``` ### Step 4. Fill Sections 7 and 8 With Names, Not Roles The reporting section fails when it names job titles instead of people. Expected final version: ```markdown ## 8. Reporting Concerns If you have a concern about conduct, data, authorship, or treatment of people or animals: 1. Contact the principal investigator (Dr. Martinez, office 4.210) or, if the concern involves Dr. Martinez, contact the department ombudsperson directly. 2. Alternative contact outside the lab: [Name], Research Integrity Officer, [email], room [number]. Reports may be made confidentially. 3. Good-faith reports, including reports that turn out to be mistaken, will never result in retaliation. Anyone who retaliates against a reporter violates this code. 4. Anyone may escalate directly to institutional channels at any time. You are never required to report within the lab first. ``` ### Step 5. Review, Adopt, and Acknowledge Circulate the draft to every member of the group, including the most junior. Ask each person to mark any commitment they cannot actually meet, because an unmeetable rule trains people to ignore the document. Then adopt it. The minimum adoption ritual is a meeting where the code is read aloud or discussed, followed by a signed acknowledgment. The acknowledgment can be one line at the end of the document: ```markdown ## Acknowledgment I have read and understood this code of conduct. I agree to follow it and to raise concerns in good faith. Name: ______________________ Date: ____________ ``` Collect acknowledgments from current members within two weeks and from new members during onboarding, before they touch data. ### Step 6. Schedule the Review Put a recurring calendar entry one year from adoption to revisit the code. The review asks three questions: Did any provision turn out to be unworkable? Did any incident reveal a gap? Have institutional or funder policies changed? Update the version number and re-collect acknowledgments for any substantive change. ## How to Roll Out the Code So It Actually Works A signed document changes nothing by itself. Institutions that succeed at embedding responsible conduct of research combine several channels rather than relying on one. The Kenya Medical Research Institute, for example, reached more than 1,200 of its roughly 4,000 staff across 15 research centers by combining a train-the-trainer program, symposium sessions within existing conferences, center-specific training tailored to local needs, hybrid in-person and virtual delivery, and digital dissemination through bulletins and the institutional website [11]. A lab-scale version of the same logic: 1. Train one or two people deeply rather than everyone shallowly. Send one senior graduate student or postdoc to a research integrity workshop or case-discussion series, then have them lead the lab discussion of the code. 2. Discuss the code in a meeting, not by email. Real-world training that works uses active, dialogical, and case-based approaches rather than passive reading [1]. Bring one real anonymized dilemma from your field, such as an authorship disagreement or an ambiguous image, and work it through the code's provisions. 3. Attach the code to concrete moments. New project start means the authorship conversation. New member onboarding means the acknowledgment. Manuscript submission means a one-minute check of the AI and reporting clauses. 4. Repeat exposure. Digital dissemination through an intranet page, a pinned message, or a periodic reminder keeps the code visible without adding meeting time [11]. 5. Re-measure and revise. One research organization that surveyed staff before and after deploying targeted supports found that participation in training was associated with improved knowledge of integrity topics, authorship, and reporting mechanisms, and that open-ended responses shifted from emphasizing obstacles to emphasizing resources and constructive practices [2]. A three-question anonymous pulse survey once a year gives you the same signal at lab scale. One caution from the broader evidence: there is no single recipe that works everywhere. An analysis of university faculty found that research integrity emerged from several different combinations of personal and organizational factors, meaning institutions should tailor interventions to their specific people and context rather than copying a policy wholesale [12]. Your code should reflect your group's actual risks. A computational group needs strong code and data provenance rules. A clinical group needs consent and conflict-of-interest detail. A field ecology group needs permitting and community-consent language. ## Common Mistakes and Limitations - **Writing aspirational values without operational rules.** "We value integrity" appears in every document and prevents nothing. Every value statement should be followed by at least one behavior someone could be observed doing or not doing. - **Copying a template verbatim without localizing it.** A free research code of conduct template is a starting structure, not a finished document. Sections on human subjects that you never touch, or conflicts of interest clauses that ignore an actual industry relationship, signal that the document is not real. - **Making the code apply only to juniors.** If senior members are outside its scope, trainees learn that the rules are performative. Include everyone with access to the group's data and name, including the principal investigator. - **Leaving the reporting path ambiguous.** Naming only the person people may need to complain about is the classic error. Always include an alternative contact outside the direct line. - **Never revisiting the document.** An undated, unversioned code cannot be enforced or improved. Version numbers and dated reviews exist to prevent drift. - **Treating the code as a substitute for training or mentoring.** Formal training gaps in mentoring, project management, and data lifecycle skills persist even at well-resourced institutions, and a document alone does not close them [6]. Pair the code with structured onboarding and explicit mentoring expectations. - **Ignoring low-grade integrity problems.** Minor academic dishonesty that goes unaddressed predicts later problems, and students themselves cite weak enforcement as a reason misconduct persists [10]. Small violations are the cheapest ones to correct. - **Expecting the code to fix a structural problem.** If junior staff face insecure positions and extreme pressure to produce, a document will not remove those pressures, which researchers in one national study identified as core barriers to integrity [3]. Use the code to make the pressure discussable, and use institutional channels to address it. One limitation applies to this guide itself. A code of conduct cannot resolve every case, and specific incidents involving misconduct allegations, human or animal welfare, or legal exposure belong with your institution's research integrity officer, IRB, IACUC, or legal counsel, not with a lab-level document. ## Frequently Asked Questions ### What is the difference between a code of conduct and a code of ethics? A code of ethics states broad principles, such as honesty and respect for persons. A code of conduct translates those principles into specific, enforceable behaviors, such as where data are stored and when authorship is discussed. Most research groups need both, with the code of conduct doing the day-to-day work. ### How long should a research code of conduct be? Two to four pages. Longer documents go unread, and length does not improve enforceability. Aim for one checkable commitment per rule and cut any sentence that would not change anyone's behavior. ### Do we need a code of conduct if the university already has policies? Yes, because institutional policies define minimums and procedures but do not answer local questions such as notebook practices, authorship timing, or who to talk to first. The lab code sits underneath institutional policy and cannot contradict it. ### How often should we review the code? Once a year is a sensible default, plus immediately after any serious incident, major policy change, or change in funding requirements. Update the version number at every review so members know whether they signed the current document. ### What should a student do if a senior colleague appears to violate the code? Use the alternative contact named in the code, such as an ombudsperson, department chair, or research integrity officer, rather than staying silent or confronting the situation alone. Good-faith reporting is protected, and no one should be required to report within the direct chain of command first. ### Should generative AI use be covered in the code? Yes. State which uses are permitted, such as language polishing and coding help, which are prohibited, such as generating data or analyses presented as original work, and how substantive use must be disclosed. The human author remains responsible for all content either way. ### Is a signed acknowledgment really necessary? Yes. It converts the code from a reference document into an agreement, it gives new members an unambiguous onboarding moment, and it removes the defense of never having seen the rules. It takes one minute per person. ### Can I adapt a free template for my own lab or department? Yes, and you should. Use a template for structure and completeness, then replace every placeholder with your group's real names, storage locations, conventions, and risks, and have every current member review the draft before adoption. ``` ## Sources 1. [Fostering Research Integrity through Training: The training materials of four EU-funded projects through the lens of pedagogical underpinnings and three taxonomies of learning.](https://pubmed.ncbi.nlm.nih.gov/41287640/) 2. [Leadership to strengthen research culture and foster responsible conduct of research: A case study.](https://pubmed.ncbi.nlm.nih.gov/42708539/) 3. [Research Culture and Integrity in Japan: A Qualitative Study.](https://pubmed.ncbi.nlm.nih.gov/41831176/) 4. [Spin bias in scientific publishing: upholding integrity in research reporting.](https://pubmed.ncbi.nlm.nih.gov/42498761/) 5. [Creating a responsible authorship culture in science: Anchoring authorship practices in principles of transparency, credit, and accountability.](https://pubmed.ncbi.nlm.nih.gov/41811430/) 6. [Gaps and Opportunities in Responsible and Ethical Conduct of Research Training: A Survey of Faculty, Administrator, and Student Experiences in Higher Education Across the Mountain West of the U.S.](https://pubmed.ncbi.nlm.nih.gov/42130391/) 7. [Review of Routine Radiographic Imaging in Children With Craniosynostosis.](https://pubmed.ncbi.nlm.nih.gov/43002476/) 8. [Conflicts of interest in hand surgery: recognizing risks, navigating challenges and promoting integrity.](https://pubmed.ncbi.nlm.nih.gov/41355148/) 9. [Preserving scientific integrity in academic publishing: Navigating artificial intelligence, journal policies, and the impact factor as a quality indicator.](https://pubmed.ncbi.nlm.nih.gov/42222318/) 10. [Academic Integrity and Cheating in Dental Education: Prevalence, Drivers, and Career Implications.](https://pubmed.ncbi.nlm.nih.gov/41590189/) 11. [Institution-wide strategies for research integrity training: experience from the Kenya Medical Research Institute: a descriptive case study.](https://pubmed.ncbi.nlm.nih.gov/41952227/) 12. [Toward contextualized research integrity in higher education: a fuzzy-set analysis of individual and organizational configurations among university faculty.](https://pubmed.ncbi.nlm.nih.gov/42129265/)