What are the best practices? (Has Best Practice)
From The Embassy of Good Science
U
- Maintaining RI standards
- Ethical and other frameworks
- Fostering a culture of research integrity
- Handling and preventing research misconduct
- Practices to strengthen integrity +
Understanding artificial intelligence ethics and safety: A guide for the responsible design and implementation of AI systems in the public sector +
This document provides end-to-end guidance on how to apply principles of AI ethics and safety to the design and implementation of algorithmic systems in the public sector. The guidance has set the foundation for The AI Ethics and Governance in Practice Programme curriculum which is composed of a series of eight workbooks, the first four published in 2023 and the second four to be published in 2024.
Each of the workbooks in the series covers how to implement a key component of the PBG Framework. These include sustainability, fairness, technical safety, accountability, explainability, and data stewardship. Each of the workbooks also focuses on a specific domain, so that case studies can be used to promote ethical reflection and animate the key concepts. +
The Belmont report '"`UNIQ--ref-0000000B-QINU`"' , the Declaration of Helsinki '"`UNIQ--ref-0000000C-QINU`"' and similar ethical and legal documents were enacted by governing bodies and professional associations.
'"`UNIQ--references-0000000D-QINU`"' +
V
The VLIR Mind the GAP podcast focuses in different episodes on:
* the concept of research integrity itself
* research data management
* corrections and retractions
* authorship
* the use of genAI in research
* how human vulnerability affects research integrity and vice versa +
The VLIR Mind the GAP podcast focuses in different episodes on:
* the concept of research integrity itself
* research data management
* corrections and retractions
* authorship
* the use of genAI in research
* how human vulnerability affects research integrity and vice versa +
Values and norms are core concepts in moral reflection about research integrity. For instance in Moral Case Deliberation (MCD), a method used to reflect on morally troublesome situations, investigation of values and norms is used to deepen the understanding of the situation at stake.'"`UNIQ--ref-000001E2-QINU`"' A core element in this investigation, guided by a facilitator, is analysing the case by looking at the values and norms of all relevant perspectives (persons involved in the case and/or participants in the MCD meeting). Which values motivate each person? How can these values be specified into norms? For example, a moral issue might concern supervision. Should one, as a supervisor of a PhD student, in preparing a response to a reviewer, give guidance and correct mistakes, or at some point take over the writing? A relevant value for the supervisor in the case might be: autonomy. The corresponding norm in the case could be: I should give the opportunity to the PhD student to try this herself. Another value might be: effectiveness. The norm related to this value in the concrete situation could be: the article should be accepted and published. In analysing the case from the perspective of the supervisor, the group becomes aware of these conflicting values. This may then give rise to a dialogue on what value is most important in this situation. This can lead to a conclusion on the most desirable norm and related course of action. Also, the dialogue might provide insights in how to deal with the conflicting value which turns out less important. How can one do justice to the value which will not be realized? In the example, efficiency might turn out to be the most important for the supervisor, meaning that, at a certain point, she will take over the writing. In order to do justice to the value of autonomy, the supervisor might, for instance, propose that the PhD student will get more responsibility for writing the response after submission of the next article. MCD can thus foster decision making, not by prescribing a rule, but by encouraging reflection and dialogue, enabling participants to achieve a new and richer view of the situation.
It is useful to differentiate between three different types of scientific values and norms: ''internal'' values and norms, ''external'' values and norms, and ''linkage'' values and norms.'"`UNIQ--ref-000001E3-QINU`"' Internal values and norms of science justify and guide the practice of science itself. Examples of scientific values of the internal kind are: truth, honesty, simplicity, consistency, coherence, economy, exactitude and completeness, openness, open-mindedness, confidence, originality and ‘interestingness’. External values and norms comprise general ideals and rules for action, which are relevant for science, but are not constitutive of the practice of science itself. Examples of external values are human (and animal) welfare notions, which are related to the wider social and cultural context in which scientists operate. Finally, we have a different set of values and norms guiding scientific research, which represent normative points of contact - linkage - between the research community and the community at large, between internal and external values and norms. Examples are requirements of fruitfulness and relevance.
'"`UNIQ--references-000001E4-QINU`"'
The tool prompts reflection on the five principles, and uses results-based planning to help you apply them to your research project.
Results-based planning (similar to the Theory of Change approach) is a technique that starts with the end goal of your project, asking: ''What is the change you want to see in the world?'' For each principle we have provided you with a broad, ultimate goal for change – an outcome. You are asked to apply this outcome to your own research context, and decide what it should look like for your project in practical terms.
The tool will then guide you through the next steps of this process: identifying the outputs that will help you to achieve your outcome, the actions needed to get there, and the necessary resources. +
The tool prompts reflection on the five principles, and uses results-based planning to help you apply them to your research project.
Results-based planning (similar to the Theory of Change approach) is a technique that starts with the end goal of your project, asking: ''What is the change you want to see in the world?'' For each principle we have provided you with a broad, ultimate goal for change – an outcome. You are asked to apply this outcome to your own research context, and decide what it should look like for your project in practical terms.
The tool will then guide you through the next steps of this process: identifying the outputs that will help you to achieve your outcome, the actions needed to get there, and the necessary resources. +
Values and norms are core concepts in moral reflection about research integrity. For instance in Moral Case Deliberation (MCD), a method used to reflect on morally troublesome situations investigation of values and norms is used to deepen the understanding of the situation at stake. '"`UNIQ--ref-00000199-QINU`"' A core element in this investigation, guided by a facilitator, is analysing the case by looking at the values and norms of all relevant perspectives (persons involved in the case and/or participants in the MCD meeting). Which values motivate each of the persons? How can these values be specified into norms for the person? For example, a moral issue might concern supervision. Should one, as a supervisor of a PhD student, in preparing a response to a reviewer, give guidance and correct mistakes, or at some point take over the writing? A relevant value for the supervisor in the case might be: autonomy. The corresponding norm in the case could be: I should give the opportunity to the PhD student to try this herself. Another value might be: effectiveness. The norm related to this value in the concrete situation could be: the article should be accepted and published. In analysing the case from the perspective of the supervisor, the group becomes aware of these conflicting values. This may then give rise to a dialogue on what value is most important in this situation. This can lead to a conclusion on the most desirable norm and related course of action. Also, the dialogue might provide insights in how to deal with the conflicting value which turns out less important. How can one do justice to the value which will not be realized? In the example, efficiency might turn out to be most important for the supervisor, meaning that at a certain point she will take over the writing. In order to do justice to the value of autonomy, the supervisor might, for instance, propose that the PhD student will get more responsibilty for writing the response after submission of the next article. MCD can thus foster decision making, not by prescribing a rule, but by fostering reflection and dialogue, enabling participants to achieve an new and richer view on the situation.
It is useful to differentiate between three different types of scientific values and norms: ''internal'' values and norms, ''external'' values and norms, and ''linkage'' values and norms . '"`UNIQ--ref-0000019A-QINU`"' ''Internal'' values and norms of science justify and guide the practice of science itself. Examples of scientific values of the internal kind are: ''truth, honesty, simplicity, consistency, coherence, economy, exactitude and completeness, openness, open-mindedness, confidence, originality and ‘interestingness’.'' External values and norms comprise general ideals and rules for action which are relevant for science, but are not constitutive of the practice of science itself. Examples of external values are human (and animal) welfare notions which are related to the wider social and cultural context in which scientists operate. Finally, we have a different set of values and norms guiding scientific research which represent normative points of contact - linkage - between the research community and the community at large, between internal and external values and norms. Examples are requirements of fruitfulness and relevance.
'"`UNIQ--references-0000019B-QINU`"'
[
- Planning research
-Collaboration
-Publication
-Avoiding misconduct +
Before the research commences, all researchers must be in agreement on a number of components. First of all, they should agree on what they are aiming for and the goals of their research as well as what is expected from them and their part of the research. Moreover, they should also plan for ways of meeting and communicating and how they will handle any cases of conflict. Also, all researchers should put forward any virtues that they wish to be followed, in order to ensure that everyone is respected and protected.
Researchers should be aware of differences in research practice and should try to reach consensus in the design of their research and how they will be implementing it. All researchers should take responsibility for their part of the work and hold themselves accountable for it. Also, one must communicate and seek advice in case of any difficulties in their work as after all, the purpose of a collaboration is to work together and help each other. It could also be beneficial for researchers to review one another’s work besides their own.
Finally, before any work is submitted, all researchers should give their consent and make sure that there was no breach of research integrity. Also, all collaborators should be given credit for all the parts that they contributed towards. +
The [https://wcrif.org/documents/354-montreal-statement-english/file Montreal Statement on Research Integrity] in Cross-Boundary Research Collaborations was developed at the 3rd World Conference on Research Integrity, 2013. '"`UNIQ--ref-000000A4-QINU`"' The statement details twenty separate responsibilities of individuals and institutions concerning general collaborations, management of collaborations, collaborative relationships, and outcomes of research.
'"`UNIQ--references-000000A5-QINU`"' +
'''Montreal Statement'''
[https://wcrif.org/documents/354-montreal-statement-english/file The Montreal Statement on Research Integrity] in Cross-Boundary Research Collaborations was developed at the 3rd World Conference on Research Integrity, 2013.'"`UNIQ--ref-00000024-QINU`"' The statement details twenty separate responsibilities of individuals and institutions concerning general collaborations, management of collaborations, collaborative relationships, and outcomes of research.
'''TRUST'''
TRUST is an EU funded project which aims to foster high ethical standards in research globally and to counteract the practice of “Ethics dumping” or double standards in research.
The TRUST project have developed a [http://www.globalcodeofconduct.org/ global code of conduct for research in research poor settings].
For the project website, click [http://trust-project.eu/ here].
'''Bridge Guidelines'''
The BRIDGE guidelines for good epidemiological practice in (global health) research foster high-quality epidemiological studies with impact where it is needed the most: in the local communities and local research systems where the research is conducted.
For more information see [[Resource:Edb02e27-f2e5-4b75-a78c-ec42e76011f6|here]].
'''Cape Town Preamble Statement on fostering Research Integrity through Fairness and Equity'''
This statement was drafted at the 7th World Conference in Research Integrity, held in Cape Town in 2022. The statement offers recommendations and principles for conducting research in a fairer and more equitable way, especially when the research in question involves collaborations between high and low- or middle-income countries.
The WCRI have developed [https://www.wcrif.org/downloads/main-website/cape-town-statement/580-cts-poster/file a poster summarising and explaining the preamble to the statement].
You can read the Cape Town Statement in full [https://www.wcrif.org/guidance/cape-town-statement here].
'''Lancet Global Health Initiative'''
In 2023, the Global Health team at the ''Lancet'' published a statement in support of the decolonising agenda in global health, and against "helicopter research" in particular. This statement outlines measures that can be taken to counter helicopter research, including a new Equitable Partnership Declaration that will be published alongside future articles where at least one author is from a high-income country, but in which the research itself was conducted in a low- or middle-income country.
You can find the statement [https://www.thelancet.com/journals/langlo/article/PIIS2214-109X(23)00278-4/fulltext here].
'''International Ethical Guidelines for Health-related Research Involving Humans'''
The 2016 guidelines published by the Council for International Organizations of Medical Sciences (CIOMS) offer guidelines for different aspects of research involving humans, to ensure it is conducted in an ethical manner.
Guidelines two and three concentrate explicitly on guideliens for the ethical conduct of research in low-resource settings and for the equitable distribution of benefits and burdens among partners in the conduct of research respectively.
You can find the full guidelines [https://cioms.ch/wp-content/uploads/2017/01/WEB-CIOMS-EthicalGuidelines.pdf here].
<br />
'"`UNIQ--references-00000025-QINU`"'
In its guidelines for editors, Elsevier states a number of possible situations which are considered to be a conflict of interest. Some of these are: co-authoring or working in the same department with some of the authors in the last three years, being a supervisor or supervisee of the author, having a personal relationship with the author, and having a direct financial interest or other professional benefit from the review. Another example is when you are asked to review a research submitted from a competing research team (Elsevier guidelines for conflict of interest in peer review provided in the tools section). Your own research experience and ambition may influence the way you see other teams’ work.
To handle this issue, not much can be done. If we would prevent everyone with potential conflict of interest to do a peer review, the quality of peer review would drop. Many researchers with knowledge and expertise can have a personal or professional connection with the authors, especially in a small and niche research area. Another option is blinding the reviewers, so that they do not know the names of the authors. Research has shown that reviewers often recognize the authors even when blinded, and blinding doesn’t mask the products or medicines used in research.'"`UNIQ--ref-00000136-QINU`"' However, any researcher asked to do a review should decline to do so if they have a COI. Clearly defined journal policies on this matter should also be put in place.
'"`UNIQ--references-00000137-QINU`"' +
The [[Resource:C386dbba-2f69-4257-89c2-903898cf1f12|ENERI Classroom]] as well as the ENERI [[Resource:F47b9bc7-c5a5-4b92-918b-438101bd9434|Research Ethics and Research Integrity Manual]] describe in more detail what COIs are and how they can be avoided and managed responsibly.
[[Theme:1fc5c5b6-6c30-4400-a79b-8838b5a041cc|Path2Integrity]] has developed a [[Resource:C13f2fea-2e63-4da9-8b9a-e4039c3dfde1|learning card]] (instruction for a learning unit) on COIs (card Y7).
The Research Ethics Program of the University of California, San Diego has developed an [http://research-ethics.org/topics/conflicts-of-interest/#summary educational resource on COIs].
The Office of Research Integrity (ORI) has collected [https://ori.hhs.gov/conflicts-interest-and-commitment resources on COIs and commitment].
The Committee on Publication Ethics (COPE) has developed [https://publicationethics.org/competinginterests guidelines] for addressing COIs in the realm of publication ethics.
The International Committee of Medical Journal Editors (ICMJE) has [http://www.icmje.org/disclosure-of-interest/ disclosure of interest form]. +
Different journals have different rules concerning editorial conflicts of interest. Some journals have no requirements, while others have strict rules (3). The International Committee of Medical Journal Editors (ICMJE) states that journal editors cannot participate in editorial decisions regarding submitted articles if they have a COI. The same rules apply to editorial staff and guest editors. Editorial staff must not use information gained through working with manuscripts for private gain. Editors should also regularly publish disclosure statements about potential conflicts of interests related to their own commitments and those of their journal staff. For more information, click [http://www.icmje.org/recommendations/browse/roles-and-responsibilities/author-responsibilities--conflicts-of-interest.html <u>here</u>].
The Committee on Publication Ethics (COPE) provided a case of an editor who submitted a manuscript to their own journal. This can often happen when the choice of journals is limited due to the highly specialised nature of the particular subject matter. In such cases, a well-documented and transparent process is necessary to minimise potential bias in the review process. This is usually done by requiring another associate editor to guide the peer review process. One could also have the manuscript anonymized (which is often impossible in a very narrow fields) and publish supplemental material to ensure transparent reporting of the peer review process. +
Good organization between the pharmaceutical companies, CRO and clinics where the research takes place. +
The use of DMPs in aeronautical and engineering projects in the 1960s expanded into other engineering and scientific fields in the late 1970 and 1980s. Their purpose was to help complex projects cope with their data management requirements during data collection or analysis stages.'"`UNIQ--ref-00000140-QINU`"' In the early 2000s the Organisation Economic Co-operation and Development (OECD) has recommended that “responsibility for the various aspects of data access and management should be established in relevant documents”, such as grant applications, and six major UK research funding bodies recommended that all funding bodies should mandate submission of DMPs with grant applications.'"`UNIQ--ref-00000141-QINU`"' Some of them have implemented the recommendation, among them Medical Research Council and the Wellcome Trust.'"`UNIQ--ref-00000142-QINU`"' Today, 86% of UK Research Councils and 63% of US funding bodies requires submission of DMPs with funding applications.'"`UNIQ--ref-00000143-QINU`"' DMPs have been created and promoted mostly by University libraries which have become “institutional centres of data management practice”.'"`UNIQ--ref-00000144-QINU`"' Two most known are the Digital Curation Centre and California Digital Library.
The Digital Curation Centre (DCC), a leading centre in digital information curation that gives advice and help on how to store, manage, protect and share digital research data, provides resources, such as online tools, guidance and training. One of them is DMP online, a free web-based tool that supports researchers in developing DMPs by providing them with templates and guidelines.'"`UNIQ--ref-00000145-QINU`"'
California Digital Library has also developed DMPTool, an open and international source project, that helps researchers, institutions, and funders in the USA in creating DMPs.'"`UNIQ--ref-00000146-QINU`"'
Science Europe, an association of Research Funding Organisations and Research Performing Organisations, has developed Practical Guide to the International Alignment or Research Data Management. This Guide, which provides a basis for development of data practices and management, has identified six main requirements for DMPs:
1. Data description and collection or re-use of existing data
2. Documentation and data quality
3. Storage and backup during the research process
4. Legal and ethical requirements, codes of conduct
5. Data sharing and long-term preservation
6. Data management responsibilities and resources.'"`UNIQ--ref-00000147-QINU`"'
'"`UNIQ--references-00000148-QINU`"'
The European Commission decided to run a pilot under Horizon 2020 the [http://ec.europa.eu/research/participants/data/ref/h2020/other/hi/oa-pilot/h2020-hi-erc-oa-guide_en.pdf Open Research Data Pilot] (ORD pilot). Which aims to improve and maximise access to and re-use of research data generated by Horizon 2020 projects. This initiative supports and requires the application of FAIR principles within the H2020 research projects, and therefore it strives to maximise the output and outreach of publicly funded research. +
Feeding back results from genomic studies is complicated because GWA is a research tool and is not designed for clinical diagnosis. In addition, most of the discoveries of whole genome methods identify genetic variants that explain very little in disease risk or quantitative trait variance. Feeding back raw data, would however, be completely misleading and not widely understandable. Providing participants with information about the general findings of research, such as publications based on the research, is an uncontroversial and welcome practice. +
