Aligning Scientific Values and Research Integrity: A Study of Researchers’ Perceptions and Practices in Four Countries

Abstract

Scientific values are considered to play a significant role in responsible conduct of research education, such as raising awareness, changing cognition, and altering behavior. However, there is still a lack of empirical evidence regarding the relationship between scientists’ subscription of scientific values and research integrity behaviors. This paper presents a cross-national study that examines researchers’ perceptions and practices regards research integrity. The results show correlations between value adherence, level of acceptance of research misbehaviors, and self-reported research misbehavior. The study also reveals significant variations in these variables among researchers from different countries, academic positions, age groups, and genders. This cross-national investigation offers valuable insights into researchers’ attitudes and behaviors regarding research misconduct, contributing to the promotion of ethical research practices worldwide and enhancing the credibility and integrity of scientific endeavors. Further research involving larger samples and more countries would provide deeper insights into the developments in researchers’ perceptions of research misbehavior.

Introduction

Promoting research integrity necessitates responsible conduct of research (also known as RCR) training within the realm of scientific inquiry (Kalichman, 2013; Steneck & Bulger, 2007). The objectives of such education are diverse, but they can be distilled into three stages: raising awareness, changing cognition, and altering behavior (Antes, 2014; Marusic et al., 2016; Steneck & Bulger, 2007). In essence, the aim is to effect a transformation from awareness to action (Resnik, 2014). Two approaches to governing research integrity emerge (Godecharle et al., 2014; Horbach & Halffman, 2017): first, there is a focus on addressing research misconduct, conveying the severe consequences of such actions, and imposing sanctions on wrongdoers as a deterrent; then a value-driven strategy encourages the positive exploration of responsible research conduct. The former, regulation-centric or norm-based approach can lead to a series of consequences (Devereaux, 2014), for instance inadvertently reducing the research integrity culture to mere compliance with the “don’t do” list of prohibited actions. Instead, the latter approach has gained increasing recognition in enhancing researchers’ moral reasoning capabilities and shaping their ethical attitudes.

Steneck defined research integrity as possessing and steadfastly adhering to professional standards, as outlined by professional organizations, research institutions, and, when relevant, the government and public (Steneck, 2006). Professional standards and principles such as honesty, respect, accountability, reliability, and fairness have garnered widespread recognition among scientists and have been acknowledged in important declarations and codes of conduct on research integrity, e.g., the European Code of Conduct for Research Integrity and the Singapore statement (All European Academies, 2017; World Conference on Research Integrity, 2010). Another widely recognized normative system among scientists is derived from Robert Merton’s framework, known as CUDOS (Communism, Universalism, Disinterestedness and Organized Skepticism). Universalism asserts that scientific claims should be judged solely on pre-established, objective criteria, without consideration of the personal or social attributes of the individuals proposing them, such as race, nationality, or class. Communism highlights the collective nature of scientific knowledge, emphasizing that findings are a product of social collaboration and should be openly shared with the community. Disinterestedness characterizes scientists as driven by a genuine passion for knowledge, curiosity, and an altruistic concern for the benefit of humanity, rather than personal gain. Lastly, organized skepticism advocates for a critical, detached evaluation of all claims, where judgment is suspended until beliefs are thoroughly examined against empirical and logical standards (Merton, 1973).

While Merton’s scientific ethos is regarded as a fundamental guiding framework for scientific research endeavors, in daily scientific practice, its guidance, like many other principles, can encounter various dilemmas (Ziman, 1996). This is because the principles may be too vague, and when faced with real-world challenges, researchers must interpret and apply the principles in specific contexts. Furthermore, scientific ethos faces additional issues, such as clashes with local cultures and conflicts arising from the incentive mechanisms in contemporary scientific innovation activities. Within the context of promoting a culture of research integrity, the instructive nature of scientific ethos is similarly questioned. For instance, although most researchers agree that the evaluation of scientific research outcomes should be based solely on the merits of the findings, sometimes the emphasis on maintaining interpersonal relationships in local cultures can erode this universalism and embrace its counter-norm, i.e., particularism. Moreover, while the scientific community presently advocates for the transparency and openness of research data, methods, and findings, the current research reward system paradoxically encourages researchers to maintain secrecy regarding their own scientific endeavors. This implies that our value judgments and behaviors indeed undergo changes due to situational variations (Toren, 1980).

Merton also recognized that these scientific norms are merely ideal and represent a consensus and moral imperative within the scientific community on how scientists should act. There exists a certain discrepancy between these norms and how scientists actually think and behave in reality. In terms of research integrity, there exists a gap between value recognition and decision making in particular situations. However, how substantial are these gaps? Is there a correlation between recognition of scientific value, attitude towards scientific misconduct, and actual behavior? Empirical research exploring the relationships among values, attitudes, and behaviors remains limited. Investigating scientists’ perceptions and behaviors can significantly aid in enhancing the effectiveness of RCR education.

In this study, we aim to explore views on scientific values and research integrity within a cross-national comparative framework. Currently, there is a lack of extensive research in this aspect. In addition, cross-national research is particularly valuable in presenting variations in the relationships between value adherence, cognition, and actual behavior in different cultural contexts, offering a more comprehensive perspective. Furthermore, it highlights commonalities and differences between countries and reveals certain countries’ issues. Therefore, the objectives of this study are as follows:

• To examine the associations between the level of adherence of Merton’s scientific ethos, attitudes toward research misbehavior, and level of self-reported misbehavior;

• To compare these variables among researchers from different countries.

Method

This study builds upon the authors’ previous research, conducted in 2017 and published in 2018 (Li & Cornelis, 2018), which focused on a cross-cultural comparison of perceptions of research misbehavior in China and Belgium. It examined how these perceptions might be influenced by cultural differences, research environments, and the individual characteristics of researchers. Consequently, our current investigation includes four countries: Belgium, China, the Netherlands, and Vietnam. These countries were chosen based on several considerations. Firstly, we aimed to continue investigating China and Belgium to enable longitudinal comparisons over time. The inclusion of the Netherlands and Vietnam was motivated by cultural and geographical similarities—specifically, the cultural similarity between Flanders in Belgium and the Netherlands, and between China and Vietnam. By adding these two countries, we sought to further verify our research hypotheses. Secondly, the selection of these countries is also related to the progress of national research integrity policies and research cultures. Globally, the emphasis on research integrity is primarily concentrated in a few developed countries, notably the United States, the United Kingdom, and some European Union nations (Aubert Bonn & Pinxten, 2019). Recently, China has intensified its efforts to regulate research integrity, whereas some developing countries like Vietnam have shown relatively lower levels of engagement in international research integrity forums, such as World Conference on Research Integrity (Barr et al., 2020). This diversity allows us to observe the extent of differences in values and behaviors among researchers from different countries. Thirdly, the selected countries represent different levels of economic development. Belgium and the Netherlands are developed economies, while China and Vietnam are at varying stages of development. This economic diversity enhances our understanding of research behavior across different socio-economic contexts. Lastly, the degree of international collaboration differs among these countries. The extent of such collaboration can impact researchers’ behaviors and values, including their interpretation of scientific ethos and research integrity rules. Additionally, the accessibility to social networks and feasibility of conducting research in these countries further facilitated our choice.

Survey Design

The data collection method employed in this study involved a web-based questionnaire. The questionnaire’s content was derived from our previous study and pertinent literature. Comprising four primary sections, the questionnaire’s design is as follows.

(1) The first section collects demographic and professional information, including research fields, academic positions, gender, age groups, and years of research engagement. This section facilitates both inter-country comparisons and the identification of potential discrepancies among the participants from different groups.

(2) Then the second section comprises four questions aimed at exploring participants’ subscription of Merton’s four principles of scientific ethos. To address this, we drew inspiration from some well-established studies (Anderson, Martinson et al., 2007; Anderson et al., 2010; Toren, 1980; Cheng & Tang, 2015; Brooker & Allum, 2024), and calculated a composite measure, termed value adherence, representing participants’ commitment to scientific norms. We described 4 types of behaviors corresponding to Merton’s norms and asked respondents to indicate how they personally feel scientists should act, focusing on ideal practices rather than actual behaviors. Although this approach may not fully capture the complexity of Merton’s normative framework, it provides a useful reference for analysis. For example, in assessing universalism, we asked, “scientists should evaluate research solely based on its intrinsic merits, such as adherence to widely accepted standards in the field, rather than on the identity, status, race, gender, or age of the researcher.” Respondents rated their agreement on a five-point scale, ranging from “No, never should” to “Yes, always should,” with an additional option for further explanation (“None of the above. Please explain your view”). The responses were numerically coded, with a value of 1 assigned to “No, never should” and a value of 5 to “Yes, always should”. The average score across all questions in this section was then computed for each participant, indicating the extent of their subscription of scientific ethos. Higher scores indicate a greater commitment to Merton’s scientific norms.

(3) The third section focuses on the perception of 15 research misbehaviors, which were extracted from the 32 questions used in the previous study, with the aim of evaluating participants’ individual views on various common research misbehaviors. These 15 malpractices are FFP (fabrication, falsification, and plagiarism), fabrication of qualifications in grant applications, data selection, inappropriate authorship, improper peer review, selective citation, duplicate publication, concealing conflicts of interest, failing to disclose research methods, data, or results, breaches of research ethics, and redundant publication. The selection of these items was based on two considerations. Firstly, they represent some of the most concerning forms of research misbehaviors identified in existing literature. For instance, Martinson et al. (2005), through six focus-group discussions, highlighted these behaviors as being of greatest concern to scientists, with many of them being highly likely to be sanctionable. Similarly, Bouter et al. (2016) engaged experts in the field of research integrity to rank 60 types of research misbehaviors across four dimensions, i.e., frequency, impact on truth, impact on trust, and preventability. The 15 misbehaviors investigated in this study were consistently identified as major problems in modern science. Additionally, Haven et al. (2019) focused on the dimensions of impact and frequency, revealing variations in the importance of certain activities across different scientific disciplines. Given that this study extends beyond the biomedical and life sciences, it also incorporates issues pertinent to the humanities and social sciences. Secondly, the breadth of the selected items was considered. The 15 behaviors encompass the most severe forms of misbehaviors, as well as more common but problematic and detrimental practices. The aim was to provide a comprehensive coverage of research misbehaviors that spans the entire research process and includes a range of practices varying in type and severity.

Participants were presented with a five-point scale ranging from “totally unacceptable” to “totally acceptable” to indicate their level of acceptance. Additionally, an “allow text entry” option was provided for participants unable to make a selection, enabling them to explain their viewpoints. As this section explores participants’ attitudes toward research misbehavior, the variable is named “level of acceptance,” where a score of 1 represents “completely unacceptable” and a score of 5 reflects “completely acceptable.” The average score across all questions in this section was then calculated for each participant, indicating their level of acceptance of research misbehavior. Higher scores indicate a greater acceptance to research misbehaviors.

(4) The fourth section concerns self-reported research misbehaviors. Participants were asked whether they had engaged in the given behaviors, and response options ranged from “never” and “one or two times” to “sometimes” and “frequently.” To account for disciplinary variations and the relative infrequency of certain research misbehaviors in some fields, a “don’t apply” option was provided. Given the explicit nature of the instructions for this section, no “text entry” option was offered. This section assesses participants’ actual research conduct, and thus, the variable is named “misbehavior level.” A response of “one or two times” is coded as 1, “sometimes” as 2, “frequently” as 3, and other responses as 0. Subsequently, the scores for all eight questions were summed for each participant, resulting in a variable ranging from 0 to 24. A higher value denotes a higher self-reported frequency of engaging in research misbehavior.

Language

The questionnaire was designed in English as the primary language, and then it was translated into local languages by team members. All authors are proficient in multiple languages, and the translated versions underwent review and piloted testing by native speakers to ensure accuracy and comprehension. Notably, many researchers from universities of the Netherlands and Belgium are often from abroad and possess English proficiency, while the majority of researchers in China and Vietnam can read English. Thus, both the local language (Dutch, Chinese, and Vietnamese) and English versions of the questionnaire were provided for their convenience in making their selection.

Survey Dissemination

The questionnaire was created on two platforms, Qualtrics and Wenjuanxing, the latter being chosen specifically to facilitate the participation of researchers in mainland China. Multiple distribution methods were employed to enhance the diversity of participants: First, we selected top-ranked research-intensive universities from four countries as the target institutions for survey distribution. The three authors collected email addresses of faculty members from the official university websites or through direct contact with the institutions. Then, personalized survey invitations were sent via the Qualtrics platform to increase the response rate, totaling 9682 email invitations across four countries. In the invitation emails, participants were encouraged to share the questionnaire with interested researchers, colleagues, or their PhD students. Additionally, we reached out to university and departmental heads to solicit their assistance in distributing the survey. Finally, personal social networks were leveraged to promote the survey further. The questionnaire distribution began in mid-May 2023 and concluded by the end of June 2023.

Research Ethics

Ethical approval was exempted by the local ethics committee, given the nature of the procedures and routine care being administered. Prior to inviting participants to engage in questionnaire completion, we presented them with a cover letter containing comprehensive information about the research, including an informed consent section. This cover letter provided detailed explanations of the research’s objectives, participation procedures, associated risks, data use and storage, privacy protection measures, and information about the researchers involved. Participants indicated their awareness of and consent to participate by clicking the “participate” button. Participants retained the option to withdraw from the study at any point, with the assurance that their data would neither be retained nor used for any further purposes.

Data Analysis

For data analysis, SPSS version 29 was used. Descriptive statistics were used to present the participants’ demographic information and the distribution of responses for the value subscription, level of acceptance, and misbehavior level items among various groups. Differences across groups in these three aspects were analyzed using the Kruskal-Wallis Test and Mann-Whitney U Test, and the correlation analysis among the three variables was conducted using Spearman’s rho.

Results

Descriptive Analysis

A total of 765 valid questionnaires with a completion rate of 100% were collected. Among them, 225 are from Chinese universities, 196 from Vietnamese universities, 179 from Belgian universities, and 165 from universities of the Netherlands. The following table (Table 1) illustrates the distribution of participants from each country in terms of research fields, academic positions, gender, age groups and research experience.

Table 1.

Participants’ demographic and professional information

Research field	Belgium (N = 179)	China (N = 225)	the Netherlands (N = 165)	Vietnam (N = 196)	Total (N = 765)
Natural sciences & engineering	79	137	28	34	278
Life & Medical sciences	15	14	20	34	83
Social & Behavioral sciences	67	53	64	77	261
Arts & Humanities	18	21	53	51	143
Position
PhD student	78	3	33	28	142
Postdoc	40	1	14	5	60
Assistant Professor	14	53	41	105	213
Associate Professor	16	74	14	14	118
Full Professor	22	90	51	3	166
Policymaker	2		1	5	8
Administrator		4		15	19
Others	7		11	21	39
Gender
Male	101	169	96	77	443
Female	78	55	66	119	318
Other		1	3		4
Age
Under 30	58	9	15	12	94
30–39	58	93	45	61	257
40–49	32	68	31	101	232
50–59	14	53	31	17	115
Above 60	17	2	43	5	67
Year in research
Minimum	0	5	1	0
Maximum	60	44	58	48
Mean	11.38	16.09	20.98	14.75
Standard deviation	11.23	8.141	14.912	8.017
Total	176	222	163	196	765

It can be observed that there are certain variations among participants from different countries in terms of research fields, academic positions, gender, and age groups. Regarding research fields, the majority of participants from China and Belgium come from the disciplines of Natural Sciences & Engineering and Social & Behavioral Sciences, while the distribution is more even for the groups of the Netherlands and Vietnam, with slightly more participants in Social & Behavioral Sciences and Arts & Humanities. Most participants from China are faculty members, while Belgium has a higher proportion of younger researchers, i.e., doctoral students and postdoctoral researchers. The distribution of participants from the Netherlands is relatively balanced across academic positions, while Vietnam has a larger number of assistant professors. The distribution of age groups aligns to some extent with the academic positions. For example, the majority of Chinese participants are above 30, whereas participants from Vietnam are mainly concentrated in the age of 30–49. Finally, there is a significant difference in the years of research experience among the four countries, with Belgium having the lowest mean value, followed by Vietnam, China, and the Netherlands. This discrepancy is related to the aforementioned academic positions and age groups. Due to the above substantial variations in characteristics among individuals from different countries, it is essential to exercise caution when making excessive inferences based on country comparisons.

The result showed that universalism received the highest level of subscription while organized skepticism received the lowest. The results also revealed a considerable consensus among participants from all four countries concerning the most unacceptable research misbehaviors. Both falsification, fabrication, and plagiarism (also referred to FFP) and non-adherence to research ethics of human participants were considered among the top five most unacceptable behaviors by participants from all four countries. Similarly, in line with our previous survey results, Chinese participants perceived “fabrication of personal qualification and achievement in funding application” as the most unacceptable behavior, while participants from the other three countries rated it relatively less unacceptable. Participants from the Netherlands considered “non-disclosure of conflict of interest " as highly unacceptable, whereas participants from China and Vietnam rated it as less unacceptable.

Regarding the self-reported research misbehaviors, participants reported the lowest frequencies of committing non-adherence to research ethics and FFP, with 5.5% of participants admitting to ethical violation in research and 6.9% admitting to fabrication at least one time. Self-reported rates of committing falsification and plagiarism were slightly higher, at 14.2% and 18.6%, respectively. This trend aligns with the level of acceptance observed in the third section. The highest rates of self-admitted research misbehavior were found in “co-authoring out of mutual benefit” (48.8%) and “deleting data point based on pure gut” (43.1%). In terms of participants’ country of origin, Vietnamese participants reported the highest frequency of self-admitted research misbehavior.

Furthermore, the text-entry options provided in the second and third sections offered participants the opportunity to express diverse perspectives. Regarding the value of disinterestedness in the second section, respondents mentioned that alongside personal ambitions and pursuit of reputation, their research is also driven by a desire to contribute to society and advance scientific knowledge. However, the concept of personal contribution to scientific knowledge appears somewhat ambiguous, indicating that self-interest and non-self-interest are not entirely mutually exclusive and can coexist. In the third section, participants expressed various alternative viewpoints concerning certain research misbehaviors. For instance, a contentious discussion emerged around the statement “Dropping some data (e.g., outliers) because one felt those data were inaccurate.” Some comments pointed out that statistical rules exist for handling outliers and other exceptional data, suggesting that judgments should be made based on specific circumstances while maintaining transparency in research reporting. Regarding the statement “Mention of the supervisor as co-author solely out of respect,” divergent opinions were also evident. Some argued that the supervisor should contribute to the research in any case and, therefore, deserves co-authorship. Others emphasized that the term “supervisor” implies the individual initiated and supervised the research, justifying their inclusion as a co-author. Nonetheless, comments highlighted the importance of considering the actual situation, such as the funding source of the project, the core ideas of the research, and the supervisor’s level of involvement throughout the process.

Comparison of Participants’ Value Adherence, Level of Acceptance of Research Misbehavior and Misbehavior Level

Table 2 presents results of comparative analyses within groups, i.e., country, research field, academic position, age, and gender across three variables. Notably, Chinese participants showed the lowest value adherence, significantly lower than the other three countries, while Vietnamese participants also displayed a lower value adherence compared to their Dutch and Belgian counterparts. Remarkably, Vietnamese participants exhibited the highest level of acceptance of research misbehavior, significantly surpassing the other three countries, whereas Chinese participants demonstrated higher level of acceptance than their colleagues from the Netherlands. Additionally, Vietnamese participants reported the highest level of self-reported committed research misbehavior.

Table 2.

Mean scores across groups and results of within-group comparative analysis (country, research field, position, age, and gender)

	Value adherence (Mean)	p*	Level of acceptance (Mean)	p*	Misbehavior level (Mean)	p*
Country
Belgium	4.27	< 0.001	1.88	< 0.001	2.16	< 0.001
China	3.70		1.97		2.01
the Netherlands	4.24		1.76		1.60
Vietnam	3.96		2.25		4.97
Research field
Natural sciences & engineering	3.96	0.05	1.91	< 0.001	2.46	0.194
Life & Medical sciences	4.07		1.86		2.54
Social & Behavioral sciences	4.07		2.02		2.76
Arts & Humanities	4.02		2.10		3.24
Position
PHD	4.19	< 0.001	2.01	< 0.001	3.14	< 0.001
Post-doc	4.22		1.88		2.12
Assistant Professor	3.98		2.04		2.81
Associate Professor	3.80		2.06		2.77
Full Professor	4.00		1.76		1.78
Others	4.06		2.21		4.30
Age group
Under 30	4.11	< 0.001	2.00	< 0.001	2.63	< 0.001
30–39	3.97		1.98		2.62
40–49	3.97		2.08		3.44
50–59	3.98		1.92		2.18
Above 60	4.32		1.65		1.49
Gender
Male	4.00	0.048	1.93	0.022	2.46	0.002
Female	4.06		2.03		3.07

* The comparative analyses within countries, research fields, positions, and age groups were conducted using the Kruskal-Wallis test, while the comparative analysis for gender groups was performed using the Mann-Whitney U test

The differences among disciplines in these three variables were relatively minor compared to the countries. However, distinctions were evident in value adherence and level of acceptance, while no significant differences emerged in self-reported research misbehavior. Specifically, participants from the Natural Sciences and Engineering discipline exhibited lower value adherence compared to those in the Social and Behavioral Sciences, and the latter group displayed higher level of acceptance compared to both the Natural Sciences and Engineering and Life Sciences participants.

Regarding age and academic position, the results revealed similar patterns across these three variables. Participants aged above 60 exhibited significantly higher value adherence, along with significantly lower level of acceptance and misbehavior level compared to other age groups. On the other hand, participants in the 40–49 age group reported significantly higher self-reported misbehavior level compared to those aged above 50. Concerning academic positions, PhD and post-doc exhibited the highest value adherence, while associate professors showed the lowest value adherence. Full professors’ level of acceptance of research misbehavior and misbehavior level were significantly lower than other groups, excluding post-doc.

Lastly, the analysis found that male participants’ mean scores across all three variables were consistently lower than those of their female counterparts, meaning that though the male demonstrated a lower level of adherence to scientific values than females, they found research misbehaviors less unacceptable and reported a lower level of violation of research integrity.

Correlations Between Value, Attitude, and Behavior

We examined the relationships among these three variables, and the results revealed significant correlations among them. Firstly, there was a negative correlation between value adherence and the level of acceptance of research misbehavior (correlation coefficient of -0.337, p < 0.001). Secondly, there was a positive correlation between the level of acceptance and the misbehavior level (correlation coefficient of 0.510, p < 0.001). This indicates that as the level of acceptance of research misbehaviors increases, the frequency of self-reported research misbehavior also tends to increase. Lastly, there was a negative correlation between value adherence and the misbehavior level, albeit with a relatively low correlation coefficient (-0.181, p < 0.001). This suggests that a higher adherence to scientific values is associated with a lower frequency of self-reported research misconduct. However, it is crucial to note that the lower correlation coefficients indicate a weak relationship, and other factors are likely influencing the associations among these variables.

Furthermore, to examine whether these patterns held true within each country, correlation analyses among the three variables for each country were conducted. The results aligned with the overall trend, indicating significant correlations among these variables within all countries, although the correlation coefficients slightly differed. Overall, the strongest correlation was observed between the level of acceptance and the misbehavior level, whereas the correlation between value adherence and the misbehavior level was relatively weaker.

Longitudinal Comparison of Researchers’ Level of Acceptance

This study also examined whether there have been changes in researchers’ attitudes towards research misbehavior between the survey we conducted in 2017 and the current study. The comparison was made on the data from Belgium and China, as these were the countries surveyed in both periods. The comparison results revealed that there was no significant difference in the level of acceptance of research misbehavior among researchers in Belgium between the two surveys. This suggests that the attitudes of the Belgian participants towards research misbehavior remained relatively stable over time, and the level of acceptance remained low. On the other hand, significant differences were observed in the level of acceptance among Chinese participants between the two surveys. More specifically, the level of acceptance among Chinese participants in the current study was significantly lower than in 2017. It is essential to consider that there might be variations in the sample composition between the two surveys. Therefore, a further comparison was made between groups, such as academic positions, age group, and gender, between the two surveys for both countries. The comparison revealed that Chinese participants experienced changes in all three groups (gender, age, and academic positions) regarding the level of acceptance of research misbehavior. The mean level of acceptance in the current study was significantly lower than in 2017, indicating that Chinese researchers find now less acceptable of research misbehavior compared to the previous survey.

Discussion

This study explores the relationship between scientific value adherence, attitudes towards research misbehavior, and self-reported committed misbehaviors within a cross-national research framework. The results revealed significant correlations among these three variables, with the strongest correlation observed between attitudes and behaviors. In the cross-national comparison, differences in these three variables were identified among researchers from different countries, gender, age, and academic positions.

The findings regarding the relationship between value adherence, attitudes, and behaviors are consistent with previous research (Anderson, Martinson et al., 2007; Stone et al., 2009). According to the theory of planned behavior, there are three factors influencing behavioral intentions: attitudes towards the behavior, subjective norms, and perceived behavioral control (Ajzen, 1991; Beck & Ajzen, 1991). In this study, we examined two of these factors: researchers’ attitudes towards research misconduct and their personal commitment to scientific values, both of which demonstrated significant correlations with self-reported research misbehaviors. These results hold valuable implications for promoting RCR education and training. On one hand, the strong correlation between attitudes and behaviors suggests that organizations can implement strict regulations, conduct thorough investigations, and enforce strict punishments to communicate the negative consequences of research misconduct to researchers. On the other hand, the association between researchers’ adherence to scientific values and their attitudes opens up opportunities to foster positive changes. Emphasizing and rewarding ethical behavior, as well as encouraging responsible research conduct, can effectively influence researchers’ attitudes. This aligns with some European research projects, such as Virtue based ethics and Integrity of Research: Train-the-Trainer program for Upholding the principles and practices of the European Code of Conduct for Research Integrity (also known as VIR2UE) (Evans et al., 2021), which emphasizes ethics education to provide researchers with critical thinking and opens communication opportunities to address research integrity issues within specific contexts. By increasing researchers’ awareness of research integrity and transforming their previous “indifference” attitudes, such initiatives aim to create a culture of research integrity.

Both FFP and non-adherence to research ethics of human participants were among the top five most unacceptable behaviors, as perceived by participants from all four countries. This suggests that the narrow definition of research misconduct proposed by Office of Research Integrity of the United States has gained recognition across different countries (Office of Science and Technology Policy, 2000), as certain research norms are considered universal and fundamental. Interestingly, the results from both this study and the one in 2017 indicate that Chinese participants rated “fabrication of personal qualification and achievement in funding application” as the most unacceptable behavior. This finding may be attributed to the high pressure and competition that Chinese researchers face when applying for research funding. As the competition intensifies, strategies that may lead to unfair competition are met with intense resistance. This is also closely related to China’s research integrity policies, such as the National Natural Science Foundation of China (NSFC) publicly disclosing and handling cases of research misconduct in funding applications. A study examining 204 cases processed by NSFC from 1999 to 2009 found that nearly half (48.18%) of the cases involved fabricating qualifications and the foundation of research work (Chen & Fang, 2011). In recent years, the NSFC has significantly intensified its research integrity review process for projects it funds, imposing more severe penalties for violations. Applicants found to be in breach of the regulations may face severe consequences, including the return of grant, and a temporary ban on applying for similar grants for several years (Tang, 2022; Wang & Liu, 2019). As a result, these prohibitive norms have a profound impact on researchers’ attitudes towards research integrity in China.

The self-reported rates of engaging in research misconduct in this study are relatively higher compared to other reported studies (Hofmann & Holm, 2019; Kaiser et al., 2022; Necker, 2014; Zeljic, 2021). Fanelli’s meta-analysis reported that less than 2% of participants admitted to fabrication and falsification (Fanelli, 2009), and 1.7% admitted to plagiarism (Pupovac & Fanelli, 2015). Xie’s meta-analysis found a self-reported research misconduct rate of 2.9% and a questionable research practices rate of 12.5% (Xie et al., 2021). These differences may be attributed to the fact that our study did not explicitly ask participants to report instances of research misconduct within the last few years, as some other studies have done. Instead, our study relied on participants’ recollections of their research behaviors throughout their careers, which could lead to higher self-reported rates of research misconduct. Additionally, the inclusion of Vietnam and China, as developing countries, in our study might contribute to the differences, as the previous meta-analyses were predominantly conducted in developed countries (Pupovac et al., 2017). Therefore, differences between countries, including researchers’ understanding of research misconduct, institutional and national research integrity policies, and even cultural variations could contribute to this phenomenon.

The cross-national comparisons identified significant differences among countries in terms of their value adherence, level of acceptance of research misbehavior, and misbehavior level. It is obvious that researchers from the Netherlands and Belgium displayed greater adherence to research integrity standards, while researchers from China and Vietnam, especially Vietnam, deviated from these standards. These findings align with our previous study in 2017, which also identified significant differences in the perception of research misconduct between researchers from two countries. The commonalities between the Netherlands and Belgium extends beyond linguistic and cultural aspects, as they share significant cooperation in research integrity initiatives. For instance, the high-profile case of Diederik Stapel at Tilburg University in the Netherlands in 2011 significantly influenced research integrity practices (Stroebe et al., 2012), especially in the Flanders region of Belgium. Consequently, the Flanders region established its own committee on research integrity, modeled after the Netherlands and other countries, and issued corresponding regulations. Additionally, some collaborative projects between universities from both countries have indirectly contributed to enhancing research integrity awareness among their researchers. However, while research integrity has predominantly been studied in developed countries and internationally prolific developing countries (e.g., India, China), some less focused countries, therefore, received less attention to research integrity. Our study found that participants from Vietnam exhibited the highest level of acceptance and self-admitted research misbehavior, significantly surpassing the other three countries. This situation in Vietnam calls for immediate action on two fronts. Firstly, the Vietnamese government should elevate its commitment to research integrity, recognizing its significance for scientific progress and societal development. Secondly, Vietnamese researchers should actively engage in research related to research integrity, laying the groundwork for evidence-based policy development and implementation.

This study revealed consistent trends among different age groups, academic positions, and years of experience in research, wherein senior researchers (especially full professors and those above 50) displayed responses more aligned with research integrity requirements compared to their younger counterparts. These findings are in line with previous research (Hofmann & Holm, 2019; Jordan & Gray, 2013; Tijdink et al., 2014). This difference in responses may be attributed to several factors. On one hand, senior researchers often enjoy career stability and may not face the pressures related to job hunting, promotion, and job security, which are common challenges for mid-stage and early-stage researchers. On the other hand, senior researchers possess a wealth of research experience, making them better equipped to handle complex scientific practices and employing effective strategies to address challenges in research activities. However, the trend does not follow a linear progression with increasing age or academic position. Interestingly, mid-stage researchers (associate professors and those aged 40–49) exhibited higher misbehavior levels and lower value subscription. This could be attributed to the increased responsibilities mid-stage researchers bear, such as seeking career advancement, securing research funding, maintaining lab operations, and mentoring junior researchers. The heavy workload and pressure they face might contribute to their misbehavior tendencies (Wright et al., 2008). Existing studies hold the view that mentorship plays a crucial role in research integrity education and socialization, especially for junior researchers (Anderson et al., 2007; Wright et al., 2008). Nevertheless, our findings suggest that even researchers in mentorship roles may experience research misconduct issues. Therefore, as suggested by other studies (Evans et al., 2022; Pizzolato & Dierickx, 2021), we believe that RCR education should not solely focus on junior researchers but should extend to researchers at all stages of their career journey.

The longitudinal analysis in our study has brought encouraging insights, particularly concerning the changing attitudes towards research misbehavior among Chinese participants in recent years. Previously characterized by a neutral and tolerant stance, Chinese researchers now exhibit a more resolute position, reflecting a greater emphasis on research integrity. This observed shift is not coincidental. Over the past few years, the Chinese government has issued several high-level policy documents urging higher education institutions and research organizations to strengthen research integrity. Key initiatives include the issuance of Opinions on Further Strengthening the Construction of Scientific Research Integrity by the Central Committee and the State Council in 2018, as well as the rules on Investigation and Handling of Untrustworthy Research Behavior jointly issued by twenty-two ministries in 2022, all signaling a zero-tolerance approach to research misconduct. Consequently, we have witnessed significant changes in research integrity efforts at Chinese universities, such as the formulation of protocols to address research misconduct, the implementation of research integrity education for students, regular ethics training for faculty, retrospective self-audit of research misconduct, and even the revocation of degrees for theses found to have plagiarism issues dating back over a decade (Tang, 2019; Wang & Liu, 2019). These initiatives serve as a wake-up call for researchers in Chinese academia, elevating the importance of research integrity. Nonetheless, it is essential to interpret these results with caution, taking into account potential response bias, sample differences and other external factors that may have influenced the shifts in attitudes over time.

This study has several limitations. Firstly, due to the use of multiple survey distribution methods, we were unable to estimate the exact size of the population reached by our survey. However, based on the number of email invitations sent and the final number of participants, we can assume that the response rate was relatively low, with an actual response rate of less than 8%. This may be attributed to the sensitive nature of the questions concerning self-reported research misconduct, a challenge commonly faced in surveys on similar topics. While these approaches introduced difficulties in accurately estimating the population size, they also contributed to a more diverse dataset. We have acknowledged these limitations in our analysis and encourage readers to interpret the findings, particularly those related to cross-national comparisons, with appropriate caution. Future research could adopt more controlled sampling techniques and longitudinal designs to further validate and extend our results.

Secondly, this study acknowledges an inherent limitation in the “level of misbehavior” section, stemming from the reliance on self-reported data, which may result in underreporting of research misconduct. Participants might underreport their own misconduct due to factors such as social desirability bias, fear of repercussions, or psychological discomfort, thereby affecting the reliability of the findings. Additionally, the questionnaire design may have influenced participants’ responses, as placing the “self-reported committed research misbehavior” questions immediately after the section on perceptions of research misbehavior could prompt participants to downplay the frequency of their own misconduct to maintain cognitive consistency. Furthermore, as noted by our reviewers, the discrepancy between reported and actual behavior could also be influenced by various factors, such as country, age, position, and gender. The issue of underreporting is an inherent challenge in empirical research, and we can approximate the true extent of misconduct through better study design and conducting cross-study comparisons. Therefore, we compared our findings with previous studies and meta-analyses, noting that our results indicate a higher prevalence of misconduct than those reported in other studies. Nonetheless, readers should interpret these findings with caution, as the true prevalence of misconduct may still be underrepresented. Future research could mitigate this limitation by employing alternative methods to enhance data accuracy, such as offering incentives to encourage participants to provide truthful responses. Studies have shown that reward mechanisms can reduce the likelihood of underreporting questionable research practices and improve the reliability of self-admission data (John et al., 2012).

Thirdly, the phrasing of the survey questions might have influenced participants’ understanding and responses. As noted in the feedback received from participants, researchers from different disciplines and research paradigms may interpret the questions differently. To minimize such differences and reduce the potential influence of biased wording, we took great care to design the questions with neutral language. However, some inherent variability might still exist. Lastly, small sample sizes in certain subgroups could affect the accuracy of certain analyses. To address this issue, we considered reorganizing the groups to avoid small sample sizes. Despite these limitations, this study provides valuable insights into researchers’ attitudes and behaviors regarding research integrity across different countries. Future research can build upon these findings to overcome the identified limitations and expand our understanding of research integrity in diverse academic settings.

Conclusion

This cross-national study investigated the relationships between value adherence, attitudes, and behaviors regarding research integrity among researchers from different countries. The results confirmed the correlations between three variables, denoting necessities and importance of aligning scientific values and research integrity. Our findings also revealed significant variations in their perception and behavior across the participating countries, career stages, age groups and academic positions. Overall, our study contributes to the growing body of research on scientific integrity by providing empirical evidence of cultural and contextual variations in researchers’ attitudes and behaviors. The insights gained from this study can inform the development of tailored interventions and educational programs to promote responsible research conduct across diverse academic settings. As the global scientific community strives for greater transparency, rigor, and ethical conduct in research, our findings can help shape policies and practices aimed at enhancing research integrity worldwide.

Author Contributions

Dan Li contributed to the conceptualization, research design, data collection and analysis, and manuscript writing. Le Thu Mach contributed to the research design, data collection, and manuscript revision.Gustaaf Cornelis contributed to the research design, data collection, and manuscript revision.

Funding

This research was supported by the 2023 Hunan Provincial Social Science Fund and the Degree & Postgraduate Education Reform Project of Central South University.

Data Availability

The relevant data are available upon request from the corresponding author.

Declarations

Competing Interests

The authors declare no competing interests.