Introduction

Research integrity is the performance of research to the highest standards of professionalism and rigour, and in an ethically robust manner (Desmond & Dierickx, 2021). Failure to adhere to research integrity may lead to research misconduct. According to The Office of Research Integrity (ORI), research misconduct is defined as “fabrication, falsification, or plagiarism in proposing, performing, or reviewing research, or in reporting research results” (The Office of Research Integrity, 2022). Fabrication is the reporting of findings that were completely constructed and had never occurred during the study process (Elsayed, 2020). Meanwhile, falsification involves the alteration of data, research materials, equipment, or processes for fraudulent purposes to obtain the desired outcome (Reisig, Holtfreter, and Berzofsky, 2020). Plagiarism is the copying or appropriation of other people’s written, artistic, or other creative work, either partly or completely, without indicating the authorship or source of the original (Masic, 2014).

Research misconduct is of great concern and the attention to research integrity has been growing over the past decades. Fanelli (2009) shows that there is an average of around 2% of scientists admitted having falsified, fabricated, or modified results or data at least once, and over 14% of scientists observed these acts of misconduct being done by their colleagues. Among these statistics, research misconduct was reported more frequently by clinical, medical, and pharmacological researchers compared to researchers from other fields (Fanelli, 2009). The higher prevalence of research misconduct in health-related research as compared to research in other fields negatively influences the healthcare sector. The factors which might influence the research practice of researchers include their level of knowledge and awareness regarding research misconduct, as well as their acceptance of unethical practices in research misconduct (Qoronfleh, 2020). A study among the NIH-funded researchers in the United States suggested that mentoring on research ethics decreased engagement in research misconduct (Anderson et al., 2007). Satalkar and Shaw (2018) also found that low awareness of research misconduct compromised research integrity among researchers in Switzerland. However, the influence of these factors on the responsible conduct of research is still unclear in Malaysia, especially among pharmacy academics and students. Therefore, we aimed to investigate the relationship between research misconduct, its knowledge, awareness, and acceptance which focuses on the Malaysian pharmacy education community, i.e., the educators/ researchers, and students.

We hypothesized that, H1: there is a negative relationship between the level of knowledge regarding research misconduct and perceived research misconduct in workplace/institution among pharmacy academics and students in Malaysia; H2: there is a negative relationship between the level of awareness on the terminologies used in describing research misconduct and perceived research misconduct in workplace/institution among pharmacy academics and students in Malaysia; and H3: there is a positive relationship between the level of acceptance to unethical practices in research misconduct and perceived research misconduct in workplace/institution among pharmacy academics and students in Malaysia. Higher level of acceptance means higher threshold for considering research misconduct as acceptable.

Methods

Ethical Consideration

An institutional ethics committee approval was obtained from the Universiti Malaya Ethics Committee (UMREC) with the reference number, UM.TNC2/UMREC_1170.

Setting

A cross-sectional study which uses an online questionnaire was carried out in 20 pharmacy schools in Malaysia. Developed in English, the online questionnaire has six parts (Part A – F) and was adapted from different sources. The questionnaire was made available from the 20th of February 2021 to the 21st of May 2021 in Google Form and a convenient sampling method (i.e., the researchers promoted and announced the study in the available platforms and participants take part in it if they wish to participate) was used in the recruitment of participants. To be legitimate to participate, questions were included in the demographic session (Part F) to solicit participants’ background and experience, as individual needs to be either pharmacy academics, pharmacy undergraduate students or pharmacy postgraduate students in Malaysia. We did not ask for identities to make sure anonymity can be guaranteed. All participants must have some minimal exposure to research as an inclusion criterion. For example, undergraduate students must have experience with at least a semester of research project. Those who do not meet the criteria were removed from the data analysis. However, not all participants are required to have undergone an ethics training. All participants must be able to understand English and provide consent online. A pilot test was conducted among the first 30 responses collected, and they were excluded from the final data analysis. The participants were not given any form of payment or incentive to participate.

Survey

Part A of the survey measured the level of knowledge and awareness of the terminologies regarding research misconduct, and it was based on a study by Ababneh et al. (2020). In their study, the level of awareness of terminologies was measured using a 3-Point Scale “aware (I know)”, “not aware (I do not know)” and “not sure”. In our survey, the measurement was modified into a 5-Likert Scale, ranging from “1-Not Familiar at All” to “5-Very Familiar”. Examples of terminologies are “Informed consent”, “Declaration of Helsinki”, “Plagiarism”, “Falsification”, and “Fabrication”. This part of our survey reported a Cronbach Alpha of 0.859.

Part B measured the level of knowledge regarding research misconduct, and it was in a 5-Likert Scale, ranging from “1-Strongly Disagree” to “5-Strongly Agree”. An example of items is “Publication ethics in research is an essential element of paper writing.” Part B of our study reported a Cronbach Alpha of 0.771.

Part C measured the level of acceptance of unethical practices in research misconduct. In the study by Ababneh et al. (2020), it was measured with a 5-Likert Scale, ranging from “1-Very Acceptable” to “5-Definitely Unacceptable”. We modified one of the items, which was “Any manuscript considered for publication only if has not been published previously” to “Manuscript is considered for publication although it has been published previously”. Other examples of unethical practices given include “Publishing results that belong to someone else” and “Using someone else’s words or ideas without giving proper credit”. Part C of our survey reported a Cronbach Alpha of 0.755.

Part D consisted of four items to detect and assess common method variance (CMV). Examples of the items are “I’m always willing to admit it when I make a mistake” and “I sometimes try to get even rather than forgive and forget”. Common method variance is assessed as this study employs a self-report questionnaire to collect data. Hence, the concern arises when both “the dependent and focal explanatory variables are perceptual measures” resulting from the same respondent (Chang et al., 2010; Fischer & Fick, 1993).

Part E measured the level of perceived research misconduct in the workplace/institution. It was adopted from the section “Prevalence of Scientific Misconduct” by Broome et al. (2005) which reported a Cronbach Alpha of 0.84. The items measured solely the respondents’ perception and experience of any research misconduct at their workplace, and no direct performed or observed act was involved. As any form of research misconduct record is private and confidential, hence, the researchers are unable to obtain this record from any of the respondents or their organization without the respondent’ consent. The level of perceived research misconduct was measured using the 4-Likert Scale, ranging from “1–Never” to “4–Frequently”. The examples of item to the question “How frequently do you think the following research misconduct practices happen in your workplace/institution?”

Finally, the respondent’s demographic and general information, such as gender, age, ethnicity, nationality, employment status, years of teaching experience, experience in research ethics training/project exposure, and the number of research ethics classes attended/conducted were collected in Part F.

Recruitment and Sample size

A priori G* power analysis was referred to generate the minimum sample size required to detect the medium effect size with 0.95 desired statistical power level, 4 latent variables, and 36 observed variables (Kang, 2021; Soper, 2022). The minimum sample size required to detect an effect was 207. Hence, a total of 260 participants were targeted for this study, with the assumption that less than 30% of incomplete or missing data may be excluded from the study.

Data Analysis

Data collected were analysed using the Partial Least Square-Structural Equation Modelling (PLS-SEM) with the Smart-PLS 3.0 software. Two stages of PLS, measurement model and structural model assessment, and a bootstrapping method of 5000 samples were performed to determine the significance of the path coefficients and the factor loadings of the constructs in the study (Hair et al., 2018; Hair et al., 2014).

Results

Demographic

Of the 393 participants, most of the respondents were female (72.8%) and aged between 18 and 25 years (77.6%). Considering the ethnicity, approximately half of the respondents were Chinese (46.3%) and most of the respondents were Malaysian (92.4%). There was a total of 30 respondents (7.6%) who are non-Malaysian who either study or work in Malaysia as a pharmacy academic or student. Table 1 shows that out of 393 respondents, 62 respondents were academics (15.8%), while the remaining ones (331) were Pharmacy students (i.e., 77.3% was reported as pharmacy undergraduate and 6.9% was reported as postgraduate student). Among the 393 respondents, 66 of them reported that they have teaching experience. Among the 66 respondents, mostly (72.7%) had one to ten years of teaching experience (Table 1).

Table 1 Participants’ Demographic and Research Exposure Among the 393 respondents
Full size table

Perceived Research Misconduct

Table 2 shows the frequency of perceived research misconduct in the respondents’ workplaces or institutions. For this study, we only collected respondents’ perception if any research misconduct was performed/observed at their institutions. We did not measure the actual behaviour of research misconduct recorded or reported officially as disciplinary misconduct. Around one-third of the respondents (34.9%) reported that plagiarism happened occasionally in their workplace while 30% of them reported it happened frequently. A total of 31%, 46.3%, and 45.3% of the respondents reported that they had never observed the practice of falsification of data, intentional protocol violations related to subject enrolment, and intentional protocol violations related to procedures in their workplace or institution, respectively. Around half of the respondents had never observed any coercion of potential subjects (50.6%) and double billing for study procedures (51.4%) in their workplace. Moreover, most of the respondents reported that they had never observed selective dropping of data from ‘outlier’ cases; falsification of biosketch, resume and reference list; disagreements about authorship; and pressures from a study sponsor to engage in unethical practices (38.2%, 38.7%, 40.7% and 48.1%, respectively).

Table 2 Respondents’ Perceived Research Misconduct in Their Workplace/Institution
Full size table

Construct Validity and Reliability

Internal consistency reliability, convergent validity, and discriminant validity of the model were assessed in the measurement model. One item in awareness on terminologies regarding research misconduct construct (A7) and one item in acceptance to unethical practices in research misconduct construct (AC7) were deleted as they reported low factor loading (< 0.40). Six items were deleted to maintain the Average Variance Extracted (AVE) larger than 0.5 – two items in knowledge and awareness regarding research misconduct construct (K1, K4), and four items in acceptance to unethical practices in research misconduct construct (AC2, AC3, AC4 and AC5).

The factor loading of the remaining items ranged from 0.443 to 0.879. According to Hair et al. (2014), factor loading equal to or greater than 0.4 is acceptable if the summation of leading results contributing to the AVE score is larger than 0.50. Our AVE was reported to range from 0.501 to 0.612, which was greater than 0.50. Thus, we had a total of 28 remaining reflective items in the measurement model (Table 3).

Table 3 Reliability and Convergent Validity of the Measurement Model
Full size table

The reliability of the constructs was assessed using composite reliability (CR) and Cronbach’s alpha. The CR ranged from 0.829 to 0.940, and Cronbach’s alpha ranged from 0.755 to 0.929, all of which surpassed the boundary of 0.70, signifying strong reliability among the constructs. It was also concluded that convergent validity was achieved in this study with the reported factor loading, AVE, and CR.

Discriminant validity was examined by using the Heterotrait-Monotrait (HTMT) ratio with 0.85 as the threshold (Henseler et al., 2015). The readings of the correlations between all factors were not above the threshold of 0.85. Hence, discriminant validity was justified in our study (Table 4). In addition, the common method variance (CMV) was assessed by comparing the original R-square with R-square with added CMV items. The difference between them was 1.6%, which was less than a 10% difference from the original model, showing that our dataset does not have a problem with CMV. The bootstrapping method was used to perform the structural model assessment. The squared multiple correlations (R2) were stated to determine the explanatory power of the structural model (Chin et al., 2003; Sarstedt et al., 2017).

Table 4 Direct Relationships Between Each Tested Variable
Full size table

Relationship Between Knowledge, Awareness, Acceptance, and Perceived Research Misconduct

Figure 1 shows that both awareness of terminologies regarding research misconduct, and acceptance of unethical practices in research misconduct explained the 10.8% variance in perceived research misconduct in the workplace or institution of pharmacy academicians and students in Malaysia. Awareness of terminologies regarding research misconduct is the strongest predictor of perceived research misconduct among pharmacy academics and students in Malaysia, with a positive relationship (ß=0.281, p < .05), followed by the acceptance of unethical practices in research misconduct, which is having a negative relationship with perceived research misconduct (ß=-0.225, p < .05). There is no statistically significant relationship between knowledge and awareness regarding research misconduct and perceived research misconduct in our study (ß=0.044, p > .05), as shown in Table 4.

Fig. 1
figure 1

Structural model of proposed study

Full size image

Multigroup Analysis

We have divided the sample (n = 393) into 2 groups to conduct the comparison testing. The first group comprised of the pharmacy academics and postgraduate students (n = 64), while the second group is made up of only the undergraduate students (n = 236). We found that multigroup analysis is robust for the data since partial measurement invariance is established between the groups in the permutation test. Table 5 shows that there is a significant difference (p < .001) between the academics and undergraduate students pertaining to the acceptance to practices in research misconduct (β = 0.519, p = .008). When we referred to the individual group’s analysis, the acceptance to practices in research misconduct is not significantly associated with perceived research misconduct among research academics (β = 0.202, p = .200) while the relationship is found significant among the undergraduate students (β = -0.317, p < .001). This shows that the undergraduate Pharmacy students have a higher threshold for considering research misconduct as acceptable.

Table 5 Multiple Group Analysis (Academics and Postgraduates vs. Undergraduates)
Full size table

Discussion

According to the Ministry of Education of Malaysia, the female students are performing well in Science, Technology, Engineering, and Mathematics (STEM) subjects, from primary schools up to university (Tienxhi, 2017). Hence, it is no surprise that in Malaysia, we have no shortage of female talent in Science, Technology, and Innovation (STI). Hasan et al. (2010) also reported in their report, that more than two-thirds of the final-year Bachelor of Pharmacy (BPharm) students from both public and private universities in Malaysia, were female. In Malaysia and most other countries, female students constitute a greater proportion than male students who enrolled in undergraduate pharmacy courses. This is similar to a study in Jordan by Jarab et al. (2021) in which the female respondents outnumbered male respondents by 5 to 1 margin.

The most frequent perceived research misconduct in the workplace of our respondents is plagiarism, followed by falsification of data, selective dropping out of “outlier” cases, and disagreement about authorship. According to Abu Farha et al. (2021), a total of 72 pharmacy postgraduate students (69.9%) in Jordan reported that they have committed plagiarism during their studies. According to Drolet et al., (2023), the highly competitive and performance-based research atmosphere has caused researchers to engage in “unethical behaviour that reflects a lack of research integrity”. Some participants, during the semi-structured individual interview, indicated that competition for grants and scientific publications serves as a factor that encourages researchers to falsify their research output (Drolet at al., 2023). This is different from the finding of a previously published review article (Armond et al., 2021). According to the study, among 388 publications which were related to research ethics and research misconduct, the most frequently reported act of research misconduct was falsification (46.8%), followed by fabrication (45.4%); and non-adherence to laws and regulations which involves data protection, informed consent, and research ethics committees’ approval (14%). Our finding is also different from a previous study which reported on the involvement in research misconduct among 100 investigators who engaged in HIV research in Kenya (Were et al., 2020). The most frequently reported research misconduct in the study was the selective dropping of “outlier” data (32.9%), followed by disagreements about authorship (31.7%), fabricating data (23.5%), and plagiarism (22%). The much higher prevalence of perceived conduct of plagiarism among the pharmacy academics and students in Malaysia requires serious attention. The reasons which led to this unethical practice should be identified to develop strategies to mitigate this practice. Perhaps, incorporating ethics content into the courses across the pharmacy curriculum can be performed to ensure the undergraduate and postgraduate pharmacy students are prepared to address ethical issues (Stratton, 2020).

Our finding shows that there is a statistically significant positive relationship between awareness of terminologies regarding research misconduct and the perceived research misconduct in the workplace or institution of Malaysian pharmacy academics and students. This finding is contrary to our hypothesis, where we hypothesized that there would be a negative relationship between the level of awareness regarding research misconduct and the perceived research misconduct. Olesen et al. (2020) showed that the lack of understanding, knowledge, and awareness of research integrity were some of the factors that contribute to research misconduct. Lehobye (2010) also reported that novice researchers were more likely to perform plagiarism due to a lack of awareness about plagiarism. However, this is contrary to our findings. This might be due to our reported results of perceived research misconduct not only including the misconduct being committed by the respondents themselves but also including the research misconduct being observed by the respondents in the workplace. Lind and Lepper (2007) revealed that the higher the awareness and sensitivity to research misconduct, the more readily a person was able to address the occurrence of research misconduct. Hence, the higher the awareness of research misconduct, the higher the perceived research misconduct. Since the increase in the awareness regarding research misconduct increases the possibility of educators/researchers identifying the occurrence of misconduct, education on research ethics, as well as strategies that involve protection of whistle-blowers, should be considered as tactics to prevent research misconduct (Horton et al., 2014; Olesen et al., 2019).

Our finding demonstrates that there is a statistically significant negative relationship between the acceptance of unethical practices in research misconduct and the perceived research misconduct in the workplace among pharmacy academics and students in Malaysia. This is contrary to our hypothesis, where our hypothesis suggested a positive correlation. This might be due to the respondents’ perception that some unethical practices were acceptable, hence they were unaware of the incidence in their research environment, which eventually led to a low perceived research misconduct. Our results show that a total of 29.6% of respondents felt that it was acceptable to change a few words in a phrase from the source material into their own words when doing research writing, and a total of 31.6% of respondents had a neutral stand on this unethical misconduct. Paraphrasing is defined as “taking the idea of a sentence or passage and putting it into your own words” (Gasparyan et al., 2017). The act of paraphrasing does not apply to “merely copy the sentence and replace or change a few words to be different from the original” (Gasparyan et al., 2017). Besides, around half of the respondents rated that it was acceptable or remained neutral to consider publishing a manuscript which had been published previously but clearly it is not at all accepted.

From this study, knowledge and awareness regarding research misconduct are found to have no statistically significant relationship with perceived research misconduct. This shows that the level of knowledge and awareness is not a predictor of perceived research misconduct in our study. This contravenes other previous studies such as the study by Adeleye and Adebamowo (2012). Olesen et al. (2017) also supported emphasizing research ethics-related subjects among academics, students, and researchers to enhance the current knowledge and awareness regarding research misconduct among academics and students to prevent misconduct in research. Moreover, Fisher and Partin (2014) showed that the lack of knowledge of the rules of an appropriate citation could potentially result in plagiarism.

On top of the findings above, we found that the percentage of contribution of the two predictors in our study, which are the awareness of terminologies regarding research misconduct and the acceptance of unethical practices in research misconduct, to the perceived research misconduct is only 10.8%, which is considered low. This indicates that there are other predictors which are not being studied in our study that contribute to research misconduct. Other factors that contribute to research misconduct include a research environment which does not support ethical research practice, weak institutional regulatory frameworks, lack of oversight by institutions, and lack of facilities in an institution (Li & Cornelis, 2018; Xie et al., 2021). Besides, pressure to publish, personal or behavioural factors, lack of training in ethical research, and lack of mentorship might also contribute to research misconduct (Xie et al., 2021). Future research that studies other predictors of research misconduct is recommended to examine their correlations with research misconduct.

The Malaysian Pharmacists Society (MPS) is “the national association for pharmacists formed in 1967 to promote and maintain the honour and interests of the profession of pharmacy in the country” (Malaysian Pharmacists Society, 2023). MPS is thus highly recommended to enforce the education of integrity amongst the pharmacy students to uphold and enhance the standard and ethics of the profession. On the global scale, the International Pharmaceutical Federation (FIP) has approved the “FIP Statement of Professional Standards – Codes of Ethics for Pharmacists” in 2014. It was mentioned in the document that “pharmaceutical practice is not restricted to the provision of direct clinical care, but also includes working in a non-clinical relationship with consumers, patients, or carers; in the discovery, development, manufacturing of medicines, in maintaining the supply chain or in procurement, in general management or administration, in education, research, advisory, regulatory, or policy development roles”. The code of ethics will therefore apply to pharmacists in all practice settings, including the pharmaceutical scientists (The International Pharmaceutical Federation, 2014).

Limitations of the Study

Our study explores self-reported perceptions and perceived research misconduct thus introducing the possibility of social desirability bias. The same concern has been also raised by Kadayam Guruswami et al. (2022). In this study, it is suspected that the undergraduate students would report a high prevalence of plagiarism. Given the predominance of undergraduate students in this cohort, this perhaps contributes to the apparently anomalously high prevalence of plagiarism in the current study. This certainly serves as a limitation to our study. Hence, more academics should be included in the future study. In addition. our reported results of perceived research misconduct which not only includes the misconduct being committed by the respondents themselves but also includes those being observed by the respondents in the workplace affected the conclusion of our study. Besides, interpretation of the finding shall be cautious too as this is a cross-sectional study focusing on just a group of accessible participants and hence, the result could not be generalized. Moreover, another limitation of our study is respondents’ gender imbalance where majority of our respondents are females. Further, the anonymity nature of the survey has hindered us from following up with the respondents to ensure greater participation and improve the response rate.

Conclusion

This study was conducted to determine the relationship between perceived research misconduct, its awareness, knowledge, and acceptance among the pharmacy community (i.e., the academics and students) in Malaysia. Our study found that the awareness on terminologies regarding research misconduct had a positive relationship with perceived research misconduct, whereas acceptance to unethical practices in research misconduct had a negative relationship with perceived research misconduct. However, the percentage of their contribution to perceived research misconduct is low. Knowledge and awareness regarding research misconduct were found to have no correlation with perceived research misconduct. It indicates that research misconduct’s awareness, knowledge and acceptance might not be the main predictors to research misconduct, at least among the pharmacy academics and students in Malaysia. It is suggested that future research on other predictors of research misconduct is highly recommended to investigate the significant contributing factors of research misconduct other than awareness and knowledge. By knowing the contributing factors, effective strategies to combat research misconduct can be developed.