Perceptions and recommendations about research integrity and publishing ethics: A survey among Chinese researchers on training, challenges and responsibilities

Over the last few years there has been increased focus on research integrity and publishing ethics misconduct, especially the systematic manipulation of the publishing process driven by paper mills. This has led to an increase in retractions within many scholarly journals, and many of the retracted papers are from authors based in Chinese institutions, although this has affected authors and institutions globally (Candal-Pedreira et al., 2022; Li et al., 2024; Van Noorden, 2023; Zhu et al., 2024).

Given the current challenges with research integrity issues in publications from Chinese institutions, there has been a lot of focus on improving practices in managing research misconduct in China, and several research studies have shown that level of awareness about research misconduct issues can be low, highlighting the need for training (Han et al., 2023). In addition, the pressure to publish, gain promotions and funding introduces incentives to commit misconduct (Yu et al., 2021). A recent announcement from China’s Supreme People’s Court announced that they are focused on the importance of maintaining scientific research integrity “in accordance with the law”. Of note, they state there will be punishments for activities such as fabrication, falsification, buying, selling or faking scientific research results, data and submissions, and will “severely crack down on scientific research intermediaries buying and selling data and the paper industry chain”, which we take to refer to paper mills (China’s Supreme People’s Court, 2025).

Through our work, we are aware there can be a lack of full understanding about the roles and responsibilities of authorship, which therefore leads to recurring authorship issues (Alam & Wilson, 2023). Also, even though many researchers use third-party manuscript preparation/editing services, there is limited guidance available on how to distinguish between legitimate and unethical services. For example, while it is okay for a third-party service to suggest suitable journals to authors, if they guarantee publication in a specific journal for a fee, this is a strong indicator of an unethical service. While it is okay for the third-party service to help the authors curate their data and provide assistance on data deposition, it is unethical to actually provide “raw data” that has not been collected by the authors as part of their research. Another example is that third-party services can check the journals’ formatting requirements and reformat the manuscript accordingly, but it is not okay for the service to provide images, data (including figures and tables), or to add references of their own choosing. We are aware that some third-party service agents handle the submission process on behalf of authors. However, even though many journal submission systems may allow a submitting agent to assist authors in the administrative tasks, this is only generally deemed acceptable if their role and contribution as the agent is transparently declared within the submission systems as well as within the manuscript. Nevertheless, through our work, we are aware that unethical agents not only submit on behalf of authors (often without being transparent about their role), but also deal with all of the submission and editorial correspondence without consulting with or involving the authors. These types of services are not deemed to be ethical because the authors are responsible and accountable for the integrity of their submission, including the accuracy and thoroughness of editorial correspondence (which includes responding to peer reviewer comments). Unethical third-party agents (e.g., paper mills) are also known to manufacture papers by adding content, including data or selling authorship (COPE & STM, 2022). In our experience, sometimes changes to the content of genuine papers have been tampered with by the agent who has added content, including citations, without consulting the authors of that submission.

Guidance on acceptable uses of third-party services is outlined in the Blue Book on the pitfalls of using third-party editing agencies in scholarly publishing (Springer Nature & The Institute of Scientific and Technical Information of China, 2020). However, we are not aware of how widespread training and guidance on acceptable third-party manuscript editing services are made available to those conducting research and publishing in scholarly journals.

It is also well known that the pressure to publish can drive some researchers to seek out unethical paths to get published, without considering the long-term and widespread personal and societal consequences of doing so. This must be addressed, because bad actors such as paper mills operate in various ways to systematically manipulate the publishing process (COPE & STM, 2022; Nagarkar, 2024; Pérez-Neri et al., 2022) and so improving training and education in research integrity and publishing ethics is critical to prevent exploitation of researchers, pollution of the scholarly literature and research waste. To tackle research and publishing ethics misconduct issues widely, we need to work across stakeholders to ensure multiple perspectives and experiences inform the development of any potential solutions.

To address this issue across stakeholders, a Joint Research Integrity Lab was established in December 2023 between Taylor and Francis to provide the publisher research integrity and ethics perspective, and the National Science Library, Chinese Academy of Sciences (the CAS Library) to provide the scientometrics perspective, as well as the experiences of researchers based in China. Combining our experiences allows us to better understand from researchers what their attitudes and levels of awareness are regarding research integrity and publishing ethics issues, as well as awareness levels of the unethical practices of some third-party manuscript preparation/editing services.

To gain more direct insights on these issues from researchers in China, a survey was developed by the Joint Research Integrity Lab (authors listed on this paper) to gather anonymous responses from registrants of the CAS Library WeChat CAS Journal Ranking channel. This channel is one of largest online channels of Chinese researchers with more than 1.6 million followers, who represent a broad array of students, early career researchers, senior researchers and librarians across STM and social sciences.

The main questions we set out to answer were:

Within Chinese research institutions, what is the level of training and awareness about research integrity and publishing ethics across career stages/types?

Questions for the survey were originally developed in English, which were then translated into Chinese.

The Chinese version of the survey was distributed online between April 25th and 29th, 2024, via the CAS Library WeChat channel on CAS Journal Ranking. As discussed in section 8a, members of this channel already had a level of awareness and interest in research integrity, Responses were provided anonymously, and before answering the questions, respondents were informed that responses collected from the survey would be used for research purposes, to inform training and education programmes, and that the results would be published. The conduct of the survey was approved by CAS Library.

The survey consisted of 14 main questions, with a minimum estimated completion time of 60 seconds. As WeChat is a mobile app and responses to surveys can be clicked through without due care, any responses that were submitted in under 60 seconds were excluded from the analysis. Also, any responses that provided conflicting answers (indicating user error or random responses) were also excluded from the analysis.

The survey asked respondents to identify themselves as one of 8 career types/stages: Undergraduate student, Master’s student, PhD candidate, Researcher, Research manager, Researcher with research management responsibilities (this category will include deans), Librarian and Other.

All other questions were designed to gain insights into their levels of experience with publishing, access to training on research integrity and publishing ethics, levels of awareness of authorship responsibilities, levels of awareness of acceptable vs unacceptable third-party manuscript preparation/editing services, and their perceptions on who is primarily responsible for upholding research integrity and provide training on publishing ethics. A copy of the survey in English and Chinese is available as a supplementary file.

Between April 25th and 29th 2024, 10,087 anonymous survey responses were received. Of these, 7,438 responses were excluded as these were submitted within 60 seconds, and a further 872 responses were excluded due to conflicting answers. Therefore 1,777 responses were included in the analysis.

As shown in Figure 1a, the majority of respondents are Undergraduates (29.26%) followed by Master’s students (20.99%). Similar levels of respondents are PhD candidates or Researchers (18.68% and 18.74%, respectively), and less than 5% are Research managers, Researchers with research management responsibilities and Librarians. Only 1.8% of respondents were in the “Other” category, which contained a wide mix of different backgrounds and were therefore excluded from any further analysis involving separate respondent cohorts. Note that given the differences between the sample sizes of Undergraduates, Master’s students, PhD candidates and Researchers (29.26%-18.68%), and the sample sizes of Research managers, Researchers with research management responsibilities and Librarians (4.28%-2.53%), we have looked at the differences in responses between these two sets of cohorts separately where appropriate.

Figure 1a.

Survey respondent career stage/types and subject areas (1,777 respondents).

The majority of respondents are in the fields of social sciences (16.43%), medicine (12.72%) and engineering (11.59%). Respondents were also asked about their levels of concern regarding the impact of research integrity issues on the trustworthiness of research publications, and as shown in Figure 1b, most respondents stated they were very concerned (30.6%) or quite concerned (48.9%).

Figure 1b.

Respondents’ levels of concern about the impact of research integrity issues on the trustworthiness of research publications. Most respondents (48.9%) are “quite concerned”.

Publishing scholarly articles is necessary for researchers across most career stages, and the use of third-party manuscript preparation/editing services is also known to be common. Therefore, respondents were asked to provide information on their publishing experiences as well as whether they have access to training and education on research integrity and publishing ethics. Figures 1c and 1d show that respondents in the researcher cohort have the most access to training, experience with publishing, and the majority within this group have published more than ten papers and have used a third-party manuscript preparation/editing service. In contrast, the undergraduate cohort has the least access to training, even though they have experience with publishing papers and using third-party services.

Figure 1c.

Publishing activity and experience levels of respondents in publishing a paper. The lines show the difference between respondents who have previously published a paper (blue line) or who have not published a paper (purple line). For published authors (represented by the blue line), the volume of publications is represented by the bars.

Figure 1d.

Experience levels of respondents in using third-party manuscript preparation/editing services (dark green bars represent those who have previously used a third-party service, light green bars represent those who have not) and access to research integrity and publishing ethics training (dark orange bars represent those who have access to training and light orange bars represent those who do not). The lines show the difference between respondents who have previously published a paper (blue line) or who have not published a paper (purple line).

Out of all the respondents, just over half (55.4%) state they have access to training and education on research integrity and publishing ethics, meaning that a large amount (44.6%) do not. As shown in Figure 2a the majority of those who do, receive this via formal training (49%), although selflearning also forms a large part of the types of training respondents’ access (20%). Almost one-third of the 55.4% of respondents with access to training, receive this via a combination of formal training and self-learning (29%). As shown in Figure 2b, all respondent cohorts receive mainly formal training, but a higher proportion of undergraduates undertake self-learning. 2% of respondents state they only receive training by other means, which includes guidance from mentors, peer interactions, research practice, information on journal submission guidelines, and resources from China Hospital Research Integrity Alliance. Across all cohorts, a high proportion of respondents deem the training they do receive to be sufficient and timely.

Figure 2a.

Proportion of respondents receiving research integrity and publishing ethics training in the format of formal training, both formal and self-learning training, self-learning training, and other.

Figure 2b.

Levels of access to training (orange line) shown against the types of training received by each respondent cohort (out of 55.4% of total respondents). Their perceptions on whether the training they receive is sufficient and timely (green line) or not sufficient and timely (red line) is also shown.

To gain insights into not only access to training, but also whether this has an impact on understanding core publishing ethics standards, the survey included questions on two key areas: authorship behaviours and what types of third-party manuscript preparation/editing service are acceptable or unacceptable. The responses give insights into whether the levels of research experience, publishing experience, training and interaction with third-party service influence perceptions about ethical vs unethical practices. As shown in Figure 1a, the sample sizes of Undergraduates, Master’s students, PhD candidates and Researchers are similar (29.26%-18.68%), and the sample sizes between Research managers, Researchers with research management responsibilities and Librarians are similar (4.28%-2.53%), so we will look at the differences in responses between these two sets of cohorts separately.

4.1

Authorship responsibilities and behaviours

Understanding the responsibilities that come with authorship appears to be a common area of confusion. To find out more, the survey asked respondents about whether they are confused about this and whether they check journal authorship policies before submitting manuscripts. Of all respondents, 35.9% state they are confused (64.1% are not confused), and 58.3% do check authorship policies and criteria before submitting their manuscripts (41.7% do not).

As shown in Figure 3a in both cases, Master’s students are the most confused but are also the cohort to most frequently check journal policies before submission. In contrast, the Researcher cohort is less confused about authorship but also frequently checks journal policies. Undergraduates and PhD candidates report similar levels of confusion, but PhD candidates more frequently check journal policies. Between these four respondent cohorts, levels of publishing activity rise by seniority in career stages (i.e. Undergraduate-Master’s-PhD-Researcher). Access to training follows a similar trend - i.e. access to training is highest amongst the Researcher cohort, followed by Master’s students and PhD candidates, with the Undergraduates cohort having the lowest levels of access to training. This shows that higher levels of support are available to researchers compared to undergraduates. Also, given that the Master’s student cohort reports a higher level of access to training than PhD students, it could be the reason why they are more likely to check journal policies.

Figure 3a.

Respondents’ awareness of authorship criteria and policies compared against publishing activity and access to training.

Comparing the responses between Research managers, Researchers with research management responsibilities and Librarians, the highest levels of confusion about authorship as well as frequency in checking journal policies is reported by the Researchers with research management responsibilities cohort. This is also the cohort with a higher amount of publishing activity as well as access to training-once again indicating that training may raise awareness about the importance of checking journal policies.

4.2

Authorship misconduct practices

The survey also asked questions on whether respondents have ever engaged in two types of specific authorship practices that constitute gift authorship, adding an author even though they do not meet the authorship criteria and agreeing to be added as an author even when they do not meet the authorship criteria. Figure 3b shows that most respondents across all cohorts have not engaged in these two types of authorship practices (68%-83% and 77%-84%, respectively), however, a notable proportion of respondents in all cohorts have (17%-32% and 16%-23%, respectively). These practices are widely considered as misconduct (COPE Council, 2023; Misconduct, Taylor and Francis Editorial Policies, retrieved 2024), but to not unduly influence responses, these were not labelled as “misconduct” in the survey questions.

Figure 3b.

Authorship behaviours where respondents have added authors who do not meet authorship criteria, and where they have agreed to be added as authors even though they did not meet authorship criteria. This is compared against publishing activity and access to training.

Interestingly, even though levels of confusion about authorship vary across respondents (Figure 3a), the misconduct practices are similar across all cohorts, in particular the Master’s student, PhD candidate and Researchers cohorts. Comparing the responses between Research managers, Researchers with research management responsibilities and Librarians, the highest levels of the practice of adding authors who do not meet the criteria is reported by Research managers.

Another authorship practice the survey asked for responses about is whether any of the respondents have ever been added as an author to a paper without their knowledge. Figure 3c shows that the Master’s students and Researchers report a similar level of experiencing this, and respondents in the Research manager cohort have experienced this more frequently than the Researchers with research management responsibilities and Librarians. The question does not allow determination of whether this occurs due to the intentional practice of gift authorship or due to limited understanding about authorship criteria and responsibilities. However, as all respondent cohorts report having experienced this, it indicates it could be occurring due to multiple reasons.

Figure 3c.

Experience of respondents of whether they have ever been added as an author on a paper without their knowledge.

As authors in China frequently use third-party manuscript preparation/editing services to helppublish their research in international journals, and paper mills are known to market their services directly to researchers, we explored awareness levels of well-established services which are considered to be legitimate (e.g. language editing, translations, technical checks) vs unethical services which are often practised by paper mills (e.g. adding data, experimental images, citations and authors).

In the survey, 31% of all respondents confirmed they have used third-party manuscript preparation/editing services, while 69% have not. We sought to determine if perceptions of the acceptability of different types of services would vary across career stages, experience, and levels of access to training (Figures 4a, 4b, and 4c). Our analysis took into consideration that most respondents who have access to training do so via formal training programmes (Figure 2), and there are distinct differences in experience with publishing, access to training, and use of third-party services across career stages (Figures 1c and 1d).

Figure 4a.

Perceptions of acceptability of service for language editing and polishing, translation services, formatting and style changes and technical checks. Lines in orange show levels of access to training (Y= respondents who have used third-party services, N= respondents who have not used third-party services).

Figure 4b.

Perceptions of acceptability of service for submitting the manuscript on behalf of authors by declaring their role as the agent and handling all submission and editorial correspondence independently of authors. Lines in orange show levels of access to training. (Y= respondents who have used third-party services, N= respondents who have not used third-party services).

Figure 4c.

Perceptions of acceptability of service for the agent proactively adding authors of their own choosing, proactively adding citations of their own choosing, adding data and/or images of their own choosing, and writing part of or the full paper. Lines in orange show levels of access to training. (Y= respondents who have used third-party services, N= respondents who have not used third-party services).

The survey included 10 different types of specific service offerings and asked respondents to select which ones they thought are acceptable or unacceptable. Instances of non-response to specific questions were also recorded, as this indicates the respondent(s) was not sure about the answer, which therefore provides some insights into where confusion is occurring.

As research integrity and publishing ethics concerns can span different types of issues, the survey asked respondents to score their levels of concern over seven different types: plagiarism, unethical use of generative AI tools, paper mills, duplicate submissions, unethical authorship, data or image manipulation and peer review manipulation. As shown in Figure 5, respondents are “very concerned” about all these issues, but mainly over data/image manipulation (44.91%), peer review manipulation (43.84%) and plagiarism (42.71%). Interestingly, for duplicate submissions, there was similar distribution in responses, which may suggest there is confusion as to whether it is an unethical practice or if it may be seen as an opportunity to maximise the chance of acceptance. There is a similar distribution of responses for the use of generative AI tools, which may be due to confusion around its acceptability in manuscript preparation as well as emerging research applications.

Figure 5.

Levels of concern from all respondents over different types of research integrity and publishing ethics issues.

6.1

Does training have an impact on levels of concerns?

Figure 6 shows the levels of concerns of respondents on each of these seven issues. There is a notable difference between researchers and other respondent cohorts regarding levels of concerns regarding research integrity issues and access to training. The Researcher cohort has more experience, receives more training and therefore seems to be more aware (and therefore more concerned) about research integrity and publishing ethics issues. A similar pattern of responses is also seen in the Researcher with research management responsibilities cohort.

Figure 6.

Levels of concerns by each respondent group to plagiarism, unethical use of generative AI tools, paper mills, duplicate submissions, unethical authorship, data/image manipulation and peer review manipulation. Lines in orange show levels of access to training.

6.2

Does prior experience with using third-party manuscript preparation/editing services have an impact on levels of concerns regarding paper mills?

As paper mills are known to market their services as third-party manuscript preparation/editing services, we compared the levels of experience with using third-party services with the levels of concerns regarding paper mills. As shown in Figure 7, Researchers have the most experience in using these services and are also the most concerned about paper mills. This is also the group that has the most access to formal training, which indicates the positive impact training has upon raising levels of awareness.

Figure 7.

Levels of concerns by respondent groups about paper mills, compared with experience in using third-party manuscript services.

The final part of the survey asked respondents about which key stakeholders they believe are responsible for upholding research integrity standards and for providing training and education on publishing ethics. They were asked to answer “Yes” or “No” to each stakeholder group presented in the survey and were therefore able to select more than one stakeholder.

Overall, most respondents think researchers and institutions are primarily responsible for upholding research integrity standards (Figure 8a). When broken down by each respondent cohort (Figure 8b), it is notable how the Researcher cohort state that researchers are more responsible than institutions for upholding research integrity standards (78.08% vs 68.17%). Interestingly, this is scored lower by Research managers and Librarians (42.22% and 43.42%, respectively) who instead identify institutions as being primarily responsible (66.67% and 51.32%, respectively).

Figure 8.

Respondents’ perceptions on which stakeholders are most responsible for upholding research integrity standards (a) all respondents; (b) responses from each respondent group.

It is also noted that the more publishing experience respondents have (for example, PhD candidates and Researchers, Figure 1c), the more they feel that Journal Editors and Publishers should also be responsible for upholding research integrity standards.

As shown in Figure 9, when asked who is responsible for providing training on research integrity and publishing ethics, 72.7% of respondents think institutions are the most responsible (although fewer Undergraduates and Librarians share this view).

56.2% of respondents think funders should also provide training, and a similar proportion of respondents (55.2%) also think journal editors and publishers play a role in providing training, while 40.1% note that third-party neutral organisations (e.g. COPE) have a responsibility too.

Figure 9.

Respondents’ perceptions on which stakeholders are most responsible for providing training and education on publishing ethics (a) all respondents; (b) respondents groups.

Although publishing in scholarly journals is critical to the progress of research as well as the career of researchers, the survey reveals that across the 1,777 respondents, training in research integrity and publishing ethics is not widely offered at institutional levels. Only 55.4% of all respondents had any access to training, and an even smaller proportion to formal training. Where levels of access are higher to this type of training, the survey responses show it has a positive impact on levels of awareness of misconduct, understanding of basic publishing ethics and research integrity principles and increases likelihood of checking journal policies.

However, while those that do get training have stated this is timely and sufficient, the answers to authorship and third-party services questions reveal many areas of confusion across all respondents, including those who receive training. This indicates that current training content may not be sufficient. In addition, there are uneven levels of training across all cohorts, with higher levels of support being available to more experienced researchers compared to undergraduates.

Authorship concerns are a recurring issue for a variety of reasons, from unintentional misconduct (e.g. gift authorship due to not understanding authorship criteria) and intentional misconduct (e.g. buying and selling authorship), and responses from the survey confirm authorship responsibilities are not fully understood. Some respondents demonstrated either a limited understanding of ethical authorship practices, or they knowingly committed misconduct via gift authorship. Other studies have shown that early career researchers can also be coerced into gift authorship practices (Goddiksen et al., 2023), highlighting the need to reinforce publishing ethics training across all stakeholders. However, the survey also shows that even early career researchers experience being added as an author without their knowledge, so this could also be happening due to a lack of communication.

As publishers (SA, VB, JH, LW), we experience different types of misconduct every year, although this is increasingly of an intentional nature, primarily due to activities of paper mills (Alam, & Wilson, 2023; Alam, 2024). According to the responses from the survey, use of third-party manuscript preparation/editing services is common across all groups, but there is limited knowledge in recognising unethical services (e.g. adding authors, citations and data/images). This is highly concerning as we are aware that bad actors, such as paper mills, offer their services under a similar label. In practice, the Taylor and Francis Publishing Ethics and Research Integrity team have even experienced cases where authors admit a third-party agent handled all aspects of the submission and even inserted irrelevant citations without their knowledge.

Use of third-party services requires much better training across all researcher and student groups, indicated by the high level of respondents across all cohorts choosing not to answer the questions about acceptable types of services. Once again, undergraduates must be included in such types of training because the answers to the survey questions in Figure 4c show that a higher proportion of undergraduates deem unethical services to be acceptable.

The survey highlights the importance of establishing topical, accessible and well-designed training and education programs for all key stakeholders on publishing ethics and research integrity. It is important to ensure that training and education needs are being met for researchers at all levels, including undergraduates. Responses from the survey show that this cohort, who are also actively publishing and using third-party services, is underserved in terms of access to formal training programs. However, the fact that most of the respondents to the survey are undergraduates (29.26%) shows this is a topic they are keen to be engaged in. The undergraduate cohort also mainly undertake self-learning, again indicating a keenness to understand more, but the survey responses indicate they have a lower level of awareness of basic publishing ethics principles as well as recognizing unethical third-party services. This means undergraduates are more vulnerable to exploitation by bad actors, as well as more likely to commit unintentional misconduct.

Although all key stakeholders are perceived as being responsible for maintaining standards (including publishers), the survey respondents perceive research institutions as the main stakeholder most responsible for upholding research integrity training as well as providing access to training. As a publisher (SA, VB, JH, LW), we do offer training in publishing ethics and research integrity on a wide range of topics including authorship practices, plagiarism, dealing with competing interests, key ethics and integrity statements, citation practices, publishing ethics considerations for generative AI use, and general overviews of the publishing process. However in our experience, it can be unclear about who is responsible at the institutional level to ensure the right kind of training reaches the right people in the right format. For example, training sessions could be organised via librarians at institutions, but it is unclear what specific role regarding ensuring access to training they may or may not have. The training sessions we offer as a publisher are open to all. Since 2022 we have conducted over 65 online publishing ethics and research integrity training sessions, which reached over 16,000 participants globally. However, even though submissions to Taylor and Francis journals from researchers in China are amongst the highest, we note that registrations for the training sessions are markedly low. This lack of engagement could be due to several factors: lack of awareness, platform, delivery method (online instead of in-person), language of instruction, and timing. The survey findings highlight that further work is needed between publishers and institutions to establish not only the content of training but also to establish the most effective collaboration mechanisms and implementation pathways to ensure learning requirements are being met. Collaboration between publishers and librarians, or other relevant parties at institutions, in developing and delivering training would ensure greater uptake. One strategy could be to pursue a “train-the-trainer” approach, where publishers can guide librarians and other stakeholders through training materials so that they can implement training locally and more broadly at their institutions.

The survey responses provide insights into the low levels of access to training in research integrity and publishing ethics – as only 55.4% of all respondents had any access to training, and an even smaller proportion to formal training. Accessibility to formal training programmes is a pressing issue, and undergraduates should receive training early in their careers as they are already engaged in publishing activity as well as using third-party manuscript preparation/editing services and demonstrate a keenness to understand more. (Figures 1c and 1d).

Gaps in research integrity training, suitable resources and the need to improve publishing ethics awareness and training is a global issue (Crean et al., 2024; Evans et al., 2022; Pizzolato, 2020; Pizzolato & Dierickx, 2021), but to provide well developed, topical, adaptive and accessible training programmes, a clear understanding is required regarding remit, roles and responsibilities across different stakeholders. Specifically, investment is required for resources as well as expertise to develop and deliver training programmes and materials, provision of access to training, monitoring successful implementation of mandatory training, and establishing ongoing learning and development.

As most training is likely to be more successfully applied at the local level, train-the-trainer programmes are required to ensure appropriate expertise and core competencies exist at institutional/organisational levels. To do this a clear understanding of the roles and remits of the potential trainers within institutions, what their training needs and resource requirements are needs to be established. Apart from whomever has a formal remit in training-the-trainers, senior researchers (deans, PI, etc) -i.e. those who are responsible for research as well as supervising junior researchers-also need to ensure they are trained and play a crucial role in ensuring junior colleagues undertake training. As using third-party manuscript preparation and editing services is common, institutions should include clear guidance to their researchers about the acceptable and unacceptable types of service offerings they may come across. In response to the need for stakeholders across different industries to work together on tackling paper mills, United2Act was set up in 2023 (supported by COPE and STM) as a group of international representatives across 10 different types of stakeholders and 14 countries (United2Act, retrieved 2025). Several authors of this paper (SA, VB, JH, LW, ZS) are active members of United2Act, which includes five working groups focusing on practical solutions, including increasing education and awareness and facilitating dialogue between stakeholders through common language and shared responsibilities.

Publishers should continue to support cross-industry initiatives such as Think.Check.Submit (https://thinkchecksubmit.org/resources/) and United2Act to raise awareness globally. Publishers enforce publishing ethics policies and play a key role in ensuring the integrity of the scholarly record and so should also work closely with stakeholders to ensure publishing ethics training and education is timely, topical and accessible. As previously discussed in Alam 2023, publishers have developed clear pathways for handling ethical issues and allegations raised post-publication, but there is also a need to build clear communication channels between publishers and institutions.

Funders who provide resources for research should also provide funding for research integrity education and training (Tang, 2022), and to ensure compliance, we propose this should be a condition to be met by applicants to secure any type of research funding.

Based on our individual experience and perspectives, as well as insights from the survey, we propose the recommended roles for each key stakeholder in ensuring the provision and implementation of updated training on research integrity and publishing ethics (Table 1). How these recommendations can be implemented in practice is illustrated via a hypothetical case (Tables 2a and 2b).

The findings of this survey highlight the need for timely, accessible, fit-for-purpose training and education in research integrity and publishing ethics for researchers at all levels (including undergraduates), as well as those responsible for upholding research integrity standards. Core topics such as authorship responsibilities and working with ethical third-party manuscript preparation/editing services should be included in mandatory training. The gaps in training leave students and researchers vulnerable to exploitation by unethical organisations such as paper mills. Some of the responses also indicate misconduct may be occurring intentionally (e.g. giving or receiving gift authorship).

Research institutions and publishers need to collaborate closely to build training and outreach programmes in a more targeted and specific way based on these survey responses. Partnership and collaborations are required to enhance the overall quality of the research and publishing system, and specifically in China the General Offices of the Communist Party of China Central Committee and the State Council have also released policy recommendations on how to foster and promote research integrity practices amongst researchers (2018, 2019). Collectively, stakeholders need to devise a regulatory framework, which will also ensure enforcement of minimum standards of training (including provision of and delivery), consistently and across all stakeholders. To do this effectively, we need clear consensus on what each stakeholder is responsible and accountable for. These actions will help to collectively rebuild trust, improve the overall quality of the research and publishing system, which will ultimately ensure funding, research activity and publishing efforts are not wasted.