Based on an overview of RM definitions issued by different research institutions across various research disciplines (Gilbert & Denison, 2003), early definitions tend to be inclusive and generic, not specifying what is RM and what is not. For instance, in the 1990s, the US National Science Foundation simply defined RM as serious deviations from generally accepted practices in a particular scientific discipline (Bornmann, 2013; Buzzelli, 1993). This kind of definition leaves more room and freedom to work against all potential forms of aberrant conduct (Bornmann, 2013).

The most widely used definition, coined by the Office of Science and Technology Policy (OSTP) of the US White House (The Office of Science and Technology Policy, 2000), refers to ‘fabrication, falsification, or plagiarism in proposing, performing, or reviewing research, or in reporting research results’. This definition specifically points out what is not acceptable, and what is an offence against the rules of good scientific practice. In this definition, fabrication is making up data, processes, and results and claiming them as real. Falsification means the manipulation of research data, materials, equipment, or processes to yield a better research outcome. Plagiarism refers to untruthful reports of research processes and results or using other people’s work or ideas without acknowledging the source.

In recent years, the international academic community has observed a surge in instances of RM (Martinson et al., 2005; Wang & Zhu, 2023). The persistent occurrence of RM has significantly undermined the credibility and value of scientific research, eroded public trust in the scientific enterprise, and compromised the motivation and morale of scientists to engage in innovative endeavours (Martinson et al., 2005; Nelson & Lubchenco, 2022; Seadle, 2017; Xie et al., 2021). To address the issues of RM, several studies have paid attention to preventing, detecting, correcting, and reporting RM (Bonito et al., 2012; Buljan et al., 2018; Gillespie, 1991; Gunsalus, 2019; Kumar, 2010; Wiedermann, 2018), as well as cultivating and repairing RI (Mejlgaard et al., 2020; Tang, 2019; Titus et al., 2008). Specifically, numerous measures have been implemented, such as integrity training and education, targeting fake peer reviews, strengthening management of science foundations, and providing guidance and policy (Cyranoski, 2017; Y. Gong, 2005; Mejlgaard et al., 2020; Michel, 2010; Yi et al., 2019).

The measures of preventing RM are often developed without a deep understanding of the factors of RM, which are characterized by ‘secrecy and complexity’ in research (DuBois et al., 2013, p. 321). A comprehensive understanding of the underlying factors of RM can facilitate the prediction of potential RM, the design of targeted interventions, and the formulation of policy decisions and educational initiatives, among other benefits (Fanelli et al., 2015; Farahat, 2022; Horbach et al., 2020; Horbach & Halffman, 2019; Mardani et al., 2020).

As reported in the existing studies, RM factors can range from individual personality traits, and organizational characteristics to professional causes (Committee on Responsible Science et al., 2017; Fanelli et al., 2022; Feng et al., 2024; Franzen et al., 2007; Horbach et al., 2019; Sovacool, 2008). Some researchers map these three typologies of RM factors to micro, meso, and macro levels respectively, indicating that the factor of RM is a multilevel problem (Faria, 2015; Fierz et al., 2014; Mardani et al., 2020; Schocker et al., 2021). Individual narrative of RM factors suggests the existence of ‘scoundrels among scientists’ (Sovacool, 2008, p. 276), informing that individual personality traits such as narcissism, overconfidence, and a lack of ethical values can be major factor of RM (Cairns et al., 2021; Krstić, 2015; Tijdink et al., 2016). According to Cairns et al. (2021), Mardani et al. (2020), and Tourish and Craig (2020), organizational factors such as institutional pressures, leadership style, and organizational culture can lead to RM. Additionally, professional factors stem from ‘professional competitions and pressures among scientists’ (Feng et al., 2024, p. 7). They exist across organizations and throw threats to the whole academic system, including publication bias, the current research funding system, the pressure to publish (Buljan et al., 2018; Golden et al., 2023; Gopalakrishna et al., 2022; Horbach & Halffman, 2019; Rodrigues et al., 2023), etc. The specific RM factors of these three main typologies, according to Feng et al. (2024), are shown in Table 1.

Besides the above three typologies of RM factors, Davis (2003) developed a conceptual model of RM factors supplementing cultural factors and situational factors. The five typologies of RM factors in Davis’s model have an overlapping and parallel relationship with each other as shown in Figure 1. Davis’s (2003) model suggested that diverse typologies of RM factors can take effect together on RM. Situational factors refer to stressful and extrinsic circumstances that can encompass situational challenges to scientists and easily induce them to commit RM (Committee on Responsible Science et al., 2017). Since situational factors can be essentially traced back to professional competitions and pressures among scientists (Davis et al., 2007; Faria, 2015), it can be recognized as a combination of organizational factors (i.e. organizational situation) and professional factors (i.e. professional situation) to some extent.

Davis (2003) brought culture into the research landscape of RM factors, suggesting that scientists may engage in RM in some ways that they may find appropriate but that are not appropriate in other cultures. Factors such as attitudes towards competition, success, and hierarchy (Aubert Bonn & Pinxten, 2021; Buljan et al., 2018), as well as the emphasis on individual achievement versus collective well-being (Mirshekary & Lawrence, 2009; Parnther, 2022), can contribute to the occurrence of RM and have roots in culture. The influence of culture on RM has been studied with diverse research methods, such as the comparative approach to understand its universality and specificity (Stephens et al., 2010; Williams et al., 2014), literature analysis to construct a comprehensive landscape of cultural contexts (Rodrigues et al., 2023), and theoretical discussions to expand RI from the cultural-practice perspective (Valkenburg et al., 2021). Culture may influence people’s behaviours in RM by predisposing them to certain actions, shaping their values, influencing their knowledge acquisition, and structuring distributions of responsibility and accountability (Valkenburg et al., 2021). Despite significant studies exploring the cultural influence on RM, the whole research landscape of cultural factors affecting RM is still fragmentary and obscure. Furthermore, the influencing mechanism between cultural factors and RM behaviours remains underexplored.

Some established cultural theories have been used to explain the cultural factors of RM. For instance, Sellin’s (1938) theory of culture conflict has been employed to explicate the factors for the divergent ethical standards among researchers from different nations when facing similar RM issues (Davis et al., 2007; Jesilow et al., 1992; Parnther, 2022). As well, Merton’s (1938) theory of social structure and anomie posits that scientists are inclined to engage in RM when societies excessively prioritize socially desirable outcomes (e.g., economic growth) over reasonable and lawful means. For instance, the ‘winner-take-all system’ in academia prioritizes breakthroughs in scientific discovery, contributing to an atmosphere in which some scientists attempt to commit RM (Casadevall & Fang, 2012, p. 892; Stern et al., 2014). Nevertheless, these theories only offer one-sided interpretations for the cultural factors affecting RM and does not clearly clarify the influencing mechanism of cultural factors on RM.

To conceptualize these cultural factors of RM and comprehend the encountered cultural challenges, G. Hofstede’s cultural dimension theory is adopted as a theoretical lens. This choice is motivated by the cultural dimension theory’s (a) stability over time (G. J. Hofstede, 2015), (b) prevalence across various national cultures globally albeit to varying degrees (G. J. Hofstede, 2015), and (c) application and effectiveness in numerous RM studies to explore potential factors (e.g., Mirshekary & Lawrence, 2009; Olesen et al., 2017; Parnther, 2022; Williams et al., 2014). The initial framework of cultural dimension theory only contained four dimensions (G. Hofstede, 1984). In response to critiques of Western bias, G. Hofstede and Bond (1984) introduced a fifth dimension to specifically address Eastern cultural traits, making the cultural framework highly relevant to the RM issues globally addressed in this paper (Zhou & Nunes, 2013). Until 2010, G. Hofstede had added a sixth dimension to the cultural dimension theory with the cooperation of Michael Minkov (G. Hofstede, 2011; G. Hofstede et al., 2010).

The latest cultural dimension theory consists of the following six dimensions:

Power distance reflects how countries ‘handle the fact that people are unequal’ (G. Hofstede et al., 2010, p. 55). It refers to the degree to which less powerful individuals accept and anticipate unequal distribution of power (G. Hofstede, 1994). Societies with high levels of power distance tend to have rigid hierarchies, a strong respect for authority, and a reluctance to challenge those in positions of power.

It is worth noting that certain existing studies (Coseglia et al., 2023; Faria, 2015; Haven & van Woudenberg, 2021) have included working or organizational culture in the category of cultural factors affecting RM. However, working, or organizational culture primarily and directly contributes to external factors (e.g., organizational factors, professional factors) rather than internal motivations of people’s behaviours. On the other hand, working or organizational culture represents a situational ‘group identity’ that changes when the working environment changes rather than a typology of ‘culture’ (G. J. Hofstede, 2015, p. 224). Hence, this study does not consider working or organizational culture as the primary cultural factors. Nevertheless, if certain kinds of working or organizational culture are rooted in the cultural dimensions outlined in G. Hofstede’s theory, it is still worthy of discussion in the present study.

The risk of RM is higher in countries that are more hierarchical and less regulated (Fanelli et al., 2015, 2017). In terms of geographic distribution, RM cases are more often reported in some Asian countries like China, Indonesia, and India (Aubert Bonn & Pinxten, 2019; Putra et al., 2023; Rodrigues et al., 2023), which demonstrate high authority hierarchy cultural traits (G. Hofstede et al., 2010, p. 60). The literature analysis revealed some specific manifestations of strong authority hierarchy leading to RM, including pressure to deference, fear of challenging authority, limited transparency, lack of accountability, etc.

Strong authority hierarchy is closely related to a culture of deference to those in positions of authority (G. Hofstede et al., 2010, p. 67), such as principal investigators and senior researchers in academia (Yu et al., 2021). Subordinates working in strongly hierarchical structures are more likely to trust and obey their leaders, as well as recognize their leaders’ views and behaviours (Z. Gong et al., 2019; Graham et al., 2018; Peltokorpi, 2019). This has led some students to blindly believe that the information provided by their supervisors is entirely accurate, then directly utilizing this information without critical verification and consequently engaging in RM. The notorious case of Diederik Stapel serves as a classic example of this phenomenon (Retraction Watch, 2016). Furthermore, reports of RM by junior scholars are less likely to be spoken up, let alone be effective in the cultural environment of strong authority hierarchy, as they may fear negative consequences for challenging those with stronger authority (Buljan et al., 2018; Horbach et al., 2020). Limited transparency and openness on RM will therefore be intensified, making it easier for RM to occur without intervention.

More seriously, power-holders in a situation with strong authority hierarchy even have the privilege to, by deliberately misrepresenting information, dictate what is plagiarism, whether plagiarism belongs to the category of RM or not, as well as who should be punished due to RM (Horbach et al., 2019; Putra et al., 2023). In this regard, the lack of accountability and oversight in the research process will happen since those in positions of authority are less likely to be held accountable for unethical practices (Velayutham & Perera, 2004).

The collectivist culture has a significant impact on academic behaviours (Farahat, 2022; Mirshekary & Lawrence, 2009; Parnther, 2022). In collectivist cultures, researchers are connected with some critical social elements like trust, confidence, and cooperation, which are conducive to the emergence of RM behaviours (D. W. Chapman & Lindner, 2016; Tourish & Craig, 2020). Some collectivist cultures prioritize loyalty to peers and affiliated institutions as more important than compliance with RI principles (Bertram Gallant, 2015). This cultural trait has led to an environment where RM is facilitated by institutional protection of wrongdoers and informants’ reluctance to divulge any information about RM for the safeguarding of the collective image (Schulz, 2015; Tourish & Craig, 2020). Empirical evidence from Williams et al. (2014) supported this viewpoint by demonstrating that students from collectivist cultures (e.g., the United Arab Emirates and some other nations from the Middle East) were less inclined to take RM as serious violations compared to those from individualist cultures (e.g., the United States).

The diverse cultural landscape within and between countries contributes to varying perceptions of ethically troubling behaviours (Fonseca et al., 2016). Specifically, foreign researchers often face distinct pressures to deviate from established norms in their home society, which do not take some behaviours as RM (Davis, 2003; Davis et al., 2007). This can be attributed to the belief of foreign researchers in the values of their home society out of a collectivist mindset (Parnther, 2022).

However, collectivist culture presents a dual impact on RM. Scientists in collectivist environments exhibit higher levels of self-esteem (Rehman & Waheed, 2014), which is advantageous for RI due to their confidence in research and the desire to protect their reputation and that of their groups (Tijdink et al., 2016). Additionally, the incidence of RM may decrease with an increasing number of co-authors in a collectivist culture due to enhanced self-control over the content and origin of research (Horbach & Halffman, 2019).

The self-interested culture has emerged from literature analysis as an important cultural factor to perpetuate RM behaviours. Scientists in the individualist pole are inclined to do some challenging work including RM to satisfy ‘a personal sense of accomplishment’ (G. Hofstede et al., 2010, p. 92), especially when the benefits from RM outweigh the loss (Tourish & Craig, 2020).

Completely different from collectivist cultures, in self-interested cultures, the occurrence of RM can rise when the average number of co-authors on a publication increases since responsibility among authors becomes diluted (Bennett & Taylor, 2003; Horbach & Halffman, 2019; Sun, 2013). Self-interested researchers, who often prevail in placing individual interest above collective interest (G. Hofstede et al., 2010, p. 130), may engage in RM for personal gain disregarding the reputation of their collaborators and their collective (Committee on Responsible Science et al., 2017; Gillespie, 1991; G. Hofstede, 1984; Tourish & Craig, 2020). Contrary to the popular opinion and the stereotype that collectivists like Asians are more likely to plagiarize due to their cultural norms, studies conducted by Kasler et al. (2021) and Martin (2012) support the notion that individualists are more prone to plagiarism due to their pursuit of personal interest.

The thematic analysis demonstrated that gender roles could play an impact on RM behaviours (Aubert Bonn & Pinxten, 2019; Fanelli et al., 2022; Rodrigues et al., 2023). Academia is characterized as a classically masculine society (Bagilhole, 2002; Fotaki, 2013), in which the aggressive, competitive, and assertive psychological characteristics of male scientists predispose them to seek status and take risks through RM (Fanelli et al., 2015, 2017). Data from the US Office of Research Integrity indicated a higher incidence of RM cases involving male scientists compared to their female counterparts (Fang et al., 2013). Other studies (e.g., Ghias et al., 2014; Gopalakrishna et al., 2022; Mirshekary & Lawrence, 2009; Yu et al., 2021) also illustrated a greater tendency for male scientists or students to engage in RM behaviours. Kaatz et al. (2013) attributed this phenomenon to the implicit gender schema, which potentially contributed to a lower threshold for male scientists to engage in RM, influenced RM case investigations, and affected the distribution of funding between male and female scientists.

While female scientists statistically exhibit fewer RM cases than males, this does not negate the potential influence of femininity on RM. Conversely, feminine cultural traits may influence a collaborative and nurturing approach (G. Hofstede et al., 2010, p. 139), which could contribute to authorship issues such as ghost, guest, or gift authorship during the process of information dissemination (Rodrigues et al., 2023; Smith et al., 2020; Yu et al., 2021).

Although gender can play a role in RM, the statistical relationship between gender roles and RM is not established universally across scientific disciplines (Dannhoferová et al., 2022; Fanelli et al., 2015). Males continue to dominate a significant portion of the academic society and power structure today, suggesting that the link between gender roles and RM may be ‘modulated by career or status’ (Fanelli et al., 2015, p. 11). The extent to which gender role contributes to RM also varies across countries, indicating the need for further research in this area (Farahat, 2022).

The literature analysis revealed that nations with low uncertainty tolerance have a high acceptance of RM behaviours. Several comparative studies (e.g., Chapman & Lupton, 2004; Salter et al., 2001) between different countries supported the impact of low uncertainty tolerance on RM. Iranian students, for instance, displayed a greater acceptance of serious RM compared to Australian students. This is reflected in Iran’s highest dimension ranking of ‘uncertainty avoidance’ at 59, indicating a lower tolerance for uncertainty, while Australia’s score was 51, indicating a higher tolerance for uncertainty (Mirshekary & Lawrence, 2009). Individuals in societies with low uncertainty tolerance tend to take actions to ensure certainty and are easily caught into “high stress and high anxiety” (G. Hofstede et al., 2010, p. 203), which contribute to RM mainly through certain ways emerged from literature analysis, including reluctance to innovation, avoidance to work insecurity, education practices emphasizing memorization, etc.

The low uncertainty tolerance culture tends to resist the innovative process in organizations (G. J. Hofstede, 2000), as it creates an environment where individuals are uncomfortable with ambiguity and prefer established routines and procedures. This resistance to change and uncertainty can lead to a reluctance to challenge established norms and practices, as people may fear the potential negative consequences or failure that could arise from trying something new (G. Hofstede et al., 2010, p. 169). Consequently, researchers working in such cultures may find it challenging to engage in innovative research (Jam et al., 2018; Kaasa & Vadi, 2010), as they are often discouraged from taking risks or deviating from the status quo (Kaasa, 2016). This fear of negative outcomes and the pressure to conform can sometimes push researchers towards engaging in RM practices, such as data manipulation and fabrication (Kaasa & Vadi, 2010). However, the direct link between the fear of reporting negative results and the engagement in RM is not well-established and requires further investigation.

Additionally, job insecurity is a typical performance of low uncertainty tolerance (Jam et al., 2018; Roll et al., 2015). It, in scientific work, is viewed as a trigger for RM (Davis et al., 2007) because individuals with a low tolerance for uncertainty prefer structured and secure lives, and may feel threatened by job insecurity. Such individuals thus are more likely to resort to RM to enhance their academic performance and reduce insecurity (Afsar & Masood, 2018; Lee et al., 2018; Staufenbiel & König, 2010), especially when the academic society is becoming hypercompetitive.

Education practices prioritizing memorization, replication, and rote learning are accepted by some national cultures like China (G. Hofstede, 1986), which reflect a pursuit of stability and norms, consistent with the cultural trait of low uncertainty tolerance. Although the ability to memorize lectures or source material is often awarded (Madray, 2013), these education practices are admitted to be one of the major factors of plagiarism (Hu & Lei, 2012; James et al., 2019; Rodrigues et al., 2023; Shi, 2006). Scientists who were raised in these education practices are often less tolerant of uncertainty and, therefore, more likely to engage in plagiarism (Dannhoferová et al., 2022; Rodrigues et al., 2023; Vučković et al., 2020). On the one hand, individuals from cultures that rely on memorization are more likely to condone RM behaviours like unattributed information sources in papers due to the lack of critical thinking (James et al., 2019; O’Sullivan & Guo, 2011). On the other hand, replication of original information, instead of being regarded as plagiarism, is sometimes accepted by certain cultures with low uncertainty tolerance as respect for original work, while rational synthesis may be classified into the queue of disrespect (Gregson & Kell, 2009).