Meticulous record-keeping is sound academic and scientific practice. It provides an accurate contemporaneous account of methods and observations that becomes a permanent reference for the researcher and the institution. It allows eventual sharing of information with others who will use the findings in their own research, try to replicate the outcomes, or apply the method to other situations. Accurate records also provide essential documentation for disproving any subsequent allegations of research misconduct, i.e., plagiarism, data fabrication, or data falsification.
Every researcher has an obligation to maintain a clear and complete record of data acquired to preserve accurate documentation of observed facts from which reports or conclusions will be made. As stated in the Research Data Management Policy RI 14, “University Members [must] maintain Research Record management practices that enable the reproduction and validation of Research Outputs upon the request of the University, funding agencies, government authorities, and other authorities who question their integrity.” In accordance with federal regulations, the University’s Research Misconduct Policy RI 07 indicates that the intentional destruction of research records or the failure to maintain and produce research records supporting a questioned publication or grant submission may be considered evidence of research misconduct.
In many fields of research, it is still standard practice to record data in ink in an indexed, permanently bound laboratory notebook with consecutively numbered pages. As electronic data generation becomes more integral to research, investigators must also manage the electronic files associated with experimentation. Principal Investigators (PIs) have a responsibility to establish clear practices for organizing, storing, securing, and reliably backing up electronic data and for ensuring that all lab personnel follow these procedures.
Electronic lab notebooks (ELNs) allow the direct entry of laboratory methods, observations, and results, and facilitate the seamless transfer and centralization of electronic data from a variety of instruments. They also allow for electronic validation signatures. The University offers to all members a cloud-based Electronic Research Notebook service that operates across many platforms and devices.
PIs should enforce for all lab personnel the experimental details that must be recorded as critical for documenting and understanding the studies. Dates of experiments, instrumental readings, and other observations should always be recorded, as they provide an essential link between distinct pieces of the research. They may also become important for establishing intellectual property and scientific priority. Research methods and statistical treatments should either be described or referenced by citation to other sources. Information on materials used and their sources should also be recorded.
Entries should never be erased or whited out from bound notebooks. If mistakes need correction, a thin line should be drawn through the erroneous entry so as not to obscure it. A correction with brief explanation should be written separately near the original entry and then dated and initialed. For ELNs, any alterations of previously logged data are automatically recorded and tracked for date, time, and personnel making the changes.
All essential information should be recorded contemporaneously with the production or observation of the data. Data obtained as printouts from instruments or computers should be appropriately labeled and pasted into print notebooks or uploaded to ELNs. Records too large to incorporate should be stored securely with their storage location referenced. If the research involves unique critical materials like cell lines, archeological artifacts, or synthetic chemical intermediates, they should be preserved, appropriately labeled, and recorded for their exact storage locations. Extensive data sets may be stored either as hard copy or electronically. Carefully documented definitions for codes should be included, together with rules for applying them to experimental, clinical, or field data and notes.
Researchers should acquaint themselves with the relevant quantitative methods available for processing data in their specific fields, including graphical and tabular methods of presentation, error analyses, and tests for reliability. Processed data should always be saved separately from the original data set.
Special care must be taken when using photo-images not to misrepresent the underlying data. For example, when using image-processing software like Adobe Photoshop to prepare a blot for viewing, it is improper and may constitute research misconduct to add or delete a band, to differentially adjust the intensity of one or more bands, to label an image from one experiment as representing a different experiment, to splice lanes without using a splice indicator, or to juxtapose pieces from different gels onto a single image. Such improprieties apply as well to histological images. Adjustments to the entire image, like altering white balance and contrast might be acceptable, whereas alterations to small portions of an image are rarely acceptable practice. Using images generated by artificial intelligence as representative data constitutes fabrication and therefore research misconduct.
When research involves human subjects, there is an added requirement that data be used, stored, and disclosed in a way that ensures individual privacy. As per the federal Health Insurance Portability and Accountability Act (HIPAA), access to health information identified with a particular subject is restricted. For purposes of analysis, human subject data are often coded and recorded with only code numbers identifying the individual subjects. Advancements in social and clinical sciences may require access to de-identified primary data collected by others. Moreover, original data must be made available to reviewers of research misconduct allegations. To satisfy these demands, the primary data – clinical or laboratory records, questionnaires, tapes of interviews, and field notes – should be available for review. Where possible, questionnaires should be stored without identifiers, using only code numbers to link them to computerized files. Records, including transcripts of taped interviews, can be redacted to remove names and other key identifiers. The rules and procedures for carrying out such redactions should be made available to those reviewing the data.
Best practice for managing data in the Arts and Humanities also begins with systematic documentation – using consistent file naming structures, systematic documentation practices, and a data dictionary to explain how data was collected and organized. Primary sources, transcripts, and field notes should always be preserved in original form in addition to any versions used for processing or annotation. As Arts and Humanities data often take non-traditional forms – sketches, recordings, performance videos, studio notes, material artefacts – these must be captured in ways that preserve both the object and its context. The iterative nature of creative practice means that version control and process documentation are especially important, capturing decisions and changes alongside final outputs. Where possible, physical materials should be digitized in the most appropriate formats (e.g., TIFF, FLAC) for long-term preservation. All data should be securely backed up and deposited in an approved institutional or disciplinary repository. Sensitive data involving human participants should comply with institutional ethical guidelines and relevant regulatory frameworks. When reporting their findings, researchers must be transparent about methodological limitations, interpretive choices, and subjective dimensions. Before disseminating research, appropriate permissions should always be secured to address ethical concerns such as copyright, intellectual property, and community-generated material.
Research funding is granted to the University rather than individuals, making the University accountable for the integrity of the data produced. Consequently, research data obtained in studies performed at the University of Pittsburgh using university resources and facilities, unless otherwise stipulated in an agreement, is owned by the University of Pittsburgh, not the University community members who generated, observed, collected, or recorded them. This includes the principal investigator of the research group and any visiting scholars. Members of the research team do have stewardship of the data and should not normally be denied reasonable access to it. If a copyright or patent application might emerge from a group project, a written agreement within the group should specify the rights, if any, of each group member to the intellectual property. A researcher who has made a finding that may be patentable should file an Invention Disclosure with the University’s Innovation Institute
Data must be shared with the public in accordance with funding agency requirements and data sharing plans in funded proposals. However, selective sharing of unpublished data with specific entities is generally prohibited. For example, it is unethical and a violation of conflict-of-interest principles for an investigator to selectively share unpublished data with a commercial entity unless a Data Use Agreement (DUA) or Sponsored Research Agreement (SRA) is in place.
A principal investigator who leaves the University must leave their original data but may be entitled to take a copy of the data to another institution to facilitate continued research. Investigators may only take original data when there is a written agreement with the University to make the data available on request within a stated time. In that case, a formal Agreement on Disposition of Research Data must be negotiated. The Office of Sponsored Programs (OSP) can assist with processing DUAs and Material Transfer Agreements (MTAs) for transfer of materials and data. Each student, postdoctoral fellow, or other investigator in a group project should come to an understanding with the research director or principal investigator, preferably in writing, about which parts of the project they might continue to explore after leaving the research group. Such an understanding should specify the extent to which they may take a copy of the research data with them. Co-investigators at other institutions may be entitled to access the data they helped to obtain, assuming their collaboration has been approved by the funding agency.
Unique materials prepared during research that can be proportioned, such as intermediate synthetic chemicals, autoradiograms, cell lines, and reagents, should be divided among members of a research group under the negotiated terms of MTAs. For non-divisible items, their allocation should be clearly stipulated in the agreement. The Office of Sponsored Programs facilitates the execution of such agreements.
Investigators have an obligation to the general scientific community to share data whenever possible. Science is a cooperative endeavor encompassing many people who pursue related research interests or wish to independently confirm reported outcomes. It is generally accepted that the data underlying a research publication should be made available to other responsible investigators who request it after the research results have been published or accepted for publication. Researchers who have access to unique sets of experimental or observational data, e.g., from satellites or from archeological or paleontological sites, have an obligation to either publish research results within a reasonable time or make the data available to others who may be able to do so.
The National Science Foundation (NSF) has a specific requirement that data, samples, physical collections, and other materials created or gathered with NSF-support be shared in a timely manner. The US Public Health Service (PHS) insists that not only data, but also unique materials (such as cell lines, cloned DNA, or reagents) developed with PHS funds must be made available to qualified individuals in the scientific community after the associated research results have been published or provided to the sponsoring agency.
The University of Pittsburgh requires that research data be stored securely for at least seven (7) years after completion of the project, submission of the final report to a sponsoring agency, or publication of the research, whichever comes last. Some agencies sponsoring research may specify a longer period for which data must be retained, and extended storage of some types of data may be required by laws or regulations. For example, the US Food and Drug Administration (FDA) requires that data associated with Phase I-III clinical trials be retained for a minimum of two (2) years following final approval of the respective drug or device, which is likely to be substantially longer than seven years after completion of the research project. Human subject research done on children must be retained until all subjects reach the age of 25.
In the absence of a specific agency regulation, a conservative rule is to retain data for as long as there is still scientific interest in the details of the research. Investigators should also bear in mind that research data may be requested at any time by the University as well as journal editors and publishers to address research misconduct concerns. Original data will always be the best defense against allegations of research misconduct.
Some types of data are expected to be deposited in a national or international databank, especially when they are so extensive as to preclude publication in a journal of record. Some examples are X-ray crystallographic data on protein structures, human genomic and proteomic data, and DNA microarray data. National Institutes of Health (NIH) has a Data Sharing Policy, effective January 25, 2023, that varies by the type of data collected.
For research in social sciences, the Inter-university Consortium for Political and Social Research has guidelines on preparing data for archiving.
Clinical trials done under FDA oversight, and those funded in part or fully by the NIH are required to register and report results on the ClinicalTrials.gov website. In addition, many medical journals require registration of clinical trials on Clinical Trials.gov prior to the enrollment of the first subject as well as results reporting.
Data that falls within the definition of ‘Controlled Unclassified Information’ (CUI) is subject to additional federal regulation. For more information, please refer to the University of Pittsburgh’s guidance on CUI.
Publication of research results is the primary means for communicating to the scholarly community so that researchers are informed of experimental outcomes and can build on those findings. Indeed, it is an ethical obligation for investigators at the University of Pittsburgh to make research findings accessible in a manner consistent with relevant publication standards. The reported data and methods should have sufficient detail to permit other researchers to replicate the results. Publications should be timely but not unduly hastened, as haste risks reporting results that have not been optimally confirmed and/or interpreted.
Some data may be sensitive, and a governmental funding agency may require approval to publish certain findings, such as those classified as Controlled Unclassified Information (CUI) under Part 2002 of 32 Code of Federal Regulations. Investigators must abide by any publication restrictions stipulated by regulations pertinent to their work.
A commercial sponsor of a research project may not prevent publication of experimental outcomes, although a delay of publication for an agreed upon period not to exceed 90 days, may be allowed to permit filing of a patent application (see Corporate Funded Agreements).
A group of journal editors, acknowledging the potential abuse of published research by perpetrators of bioterrorism, have suggested that the potential harm to society of publication can, on occasion, outweigh the potential benefits. Editors concerned about such risks might advise the authors of a submitted manuscript to modify or withhold publication and communicate their findings to the interested scientific community by other means.
Publications must give appropriate credit to all authors for their roles in the research when warranted. If more than one person contributes, the order of listed coauthors often reflects their relative contributions to the research, as described below. Many professional associations and journals have specified criteria for authorship. One standard in many of these statements is for each author to have contributed to or be prepared to contribute to:
- formulating the research question(s)
- interpreting the results
- writing the manuscript
- defending the publication against criticisms.
Other statements require meeting two or three of those criteria and, with respect to the last requirement, prescribe a more limited expectation—that each author should be prepared to defend those portions of the publication within their area of expertise. A person should not be listed as an author without their knowledge, permission, and review of the final version of the manuscript that includes all coauthors. Furthermore, the senior author has an additional responsibility to ensure that all raw data and analyses contributing to the findings in a publication have been examined and verified.
A procedure adopted by many journals and some universities or departments is to have each author sign a statement attesting to having read and approved the final manuscript and/or to having made a substantial contribution as detailed above. Departments or other academic units should consider drawing up statements of criteria and procedures for certifying authorship appropriate to their own units.
Persons whose contributions merit coauthorship in written reports should also be named in oral presentations, especially when abstracts or transactions of that conference’s proceedings will be published. Such persons retain their entitlement to authorship even after they have left the site where the research was conducted by the time a paper is submitted for publication.
Some federal agencies restrict co-authorships involving foreign collaborators and require that researchers obtain prior approval before conducting such studies and publishing the results. Investigators seeking further clarity on specific research projects or upcoming publications should consult the Office of Sponsored Programs.
Individuals who do not meet the criteria to be authors should not be given honorary authorship, also known as “gift” or “guest” authorship. Many standards for authorship suggest using alternative forms of acknowledgment within the paper for contributions that meet some but not all criteria for authorship or for technical assistance, providing research materials or facilities, etc. To avoid misunderstandings and recriminations, the inclusion and exclusion of research participants as coauthors should be made clear to all those involved in the research prior to submitting the manuscript. Ideally, these issues should be discussed and agreed upon prior to the start of research and revisited when new investigators join the project or other major changes occur to the scope or direction of the research.
Customs about author order vary with discipline. In many fields, the author list is ordered by the extent of contribution to the research and by seniority. Regardless of discipline, all coauthors must understand the basis for assigning the order of names and agree in advance to these assignments. Typically, the first author is the person who did the bulk of the data collection and analysis and who wrote the original draft. Co-first authors are increasingly common and reflect equal sharing of these activities.
Each paper should designate the senior or corresponding author, which might not be the same person. The senior author typically appears last in the coauthor list and is most often the principal investigator or lab head who directed the research and provided its primary intellectual anchor. The corresponding author is often the first or last person for a multi-authored manuscript and is responsible for communicating with the publisher or editor, for informing all coauthors of the status of review and publication, and for ensuring that all listed authors have approved the submitted version of the manuscript. If it makes sense and is agreed upon, some papers may have two corresponding authors.
Good faith disagreements over authorship — including questions of inclusion, order, or credit — may arise before, during, or after submission of a manuscript. Such disputes are separate from concerns about research misconduct and are best addressed promptly and directly by the individuals involved. Participants are encouraged to document the basis for their respective authorship claims, revisit the criteria for authorship in their field(s), and openly discuss these issues in an effort to reach consensus. When the parties cannot reach agreement on their own, they may request that the relevant Dean or Division Head serve as a mediator to facilitate a resolution. If mediation at this level does not resolve the matter, the parties may bring the dispute to the Authorship Committee, which is charged with reviewing unresolved authorship conflicts and offering guidance toward resolution. The Committee's recommendations are advisory in nature and are only binding if all parties agree in advance, prior to the Committee's review, to accept its determination as final.
In citing one’s own unpublished work, an author must be careful not to imply an unwarranted status of a manuscript. A paper should not be listed as submitted in anticipation of expected submission. A paper should not be listed as accepted for publication or in press unless the author has received a galley proof or page proof or a communication from an editor or publisher stating that publication has been approved, subject perhaps only to copyediting.
Researchers should not publish the same article in two different places without legitimate reason and must include appropriate citation in the later publication to the earlier one. In such cases, the editor must be explicitly informed of the prior publication when the manuscript is submitted. The same rules apply to abstracts. If there is unexplained duplication of publications without citation, sometimes referred to as self-plagiarism, a reader may be deceived as to the amount of original research data produced by that investigator.
It is improper in most fields to allow the same manuscript to be under review by more than one journal at the same time. Very often journals specify that a submitted work should not have been published or submitted for publication elsewhere, and some journals require that a submitted manuscript be accompanied by a statement to that effect.
An author should not divide a research paper that is a self-contained integral whole into several smaller papers merely for the sake of expanding the number of items in the author’s bibliography.
Publication of two papers representing different interpretations of the same data by different participants in the research is confusing to readers. The participants with differing interpretations of the same data should attempt to reconcile their differences in a single publication or present their alternative interpretations in the same paper.
"Self-plagiarism" also known as text recycling occurs when an author re-uses previously published text in later publications or manuscripts submitted for publication. Such re-use of one’s own words does not fall within the definition of research misconduct and, therefore, would not be investigated under research integrity policy. Generally, duplications of text describing experimental methods are not considered self-plagiarism. In the background, introduction, research description, results, and discussion sections, however, self-plagiarism is ethically dubious and may constitute a failure to meet the standards of professionalism in some disciplines. Self-plagiarism may also violate copyright law and therefore be subject to legal penalties.
The University expects scholars to follow the guidelines published by the Office of Research Integrity of the Department of Health and Human Services:
“Guideline 10: Authors who submit a manuscript for publication containing previously disseminated data, reviews, conclusions, etc., must clearly indicate to the editors and readers the nature of the previous dissemination. The provenance of data must never be in doubt.”
Further guidelines and examples relevant to the topic of self-plagiarism are available.
Most research funding agencies have proposed that all publications supported by federal funds be posted within a reasonable time in an electronically accessible form. For example, to advance science and improve human health, NIH makes the peer-reviewed articles it funds publicly available on PubMed Central. The NIH Public Access Policy requires scientists to submit final peer-reviewed journal manuscripts that arise from NIH funds to PubMed Central immediately upon acceptance for publication. The NSF requires that either the version of record or the final accepted manuscript in peer-reviewed scholarly journals and juried conference papers be deposited in a designated public access compliant repository and be available for download, reading and analysis free of charge no later than 12 months after initial publication. It is the investigator’s responsibility to be familiar with the accessibility of publication requirements from each of their funding agencies.
It is unethical to release to the media scientific information contained in an accepted manuscript prior to its publication or a presentation prior to its occurrence. This does not apply to sharing preliminary versions of research papers before peer-review by posting to online preprint servers. Exceptions to the prohibition against advance release may be made if a public health issue is involved and the editor or meeting organizers agree to an advance release.
