**Hypothesis 5 (H5).** *There is a positive relationship between policies for lecturers and the scientific research output of lecturers.*

With reference to resources for scientific research, Ref. [38] presented resourcesrelated factors, including technology and equipment, libraries, and research funding. The work by [13] revealed that accessibility of international scientific documents, experimental devices or tools, software used for research purposes, and accessibility of research funding sources are the determinants in the performance of international publishing in Vietnam. The study by [18] demonstrated that research funding is considered to be the most important

factor for research results, followed by human resources, journal and library resources, and facilities. Similarly, several other studies concluded that one of the institutional factors for improving scientific research output is financial support [5,7,20]. For this reason, the final hypothesis was proposed:

**Hypothesis 6 (H6).** *There is a positive relationship between resources for scientific research and the scientific research output of lecturers.*

According to the literature review and hypotheses proposed, a research model was developed with the following factors: scientific research objectives of HEIs, leadership, decentralization, policies for lecturers, support for scientific research activities, and resources for scientific research. Additionally, to determine the difference between population groups in research output, the research model included the following factors (i.e., control variables): gender, age, degree and academic title, experience, position (managers or non-managerial lecturers), place of graduation (abroad or domestically), and scientific area. Figure 1 depicts the proposed research model.

**Figure 1.** Proposed research model.

#### **3. Methods**

#### *3.1. Research Design*

A questionnaire was constructed based on the literature review and was adjusted based on the interview results with five experts. The final questionnaire consisted of 49 items, not including the participants' demographics section, which included 41 statements of governance factors and eight items of lecturers' research output. Participants were requested to rate a 5-point Likert scale, whereby 1 = totally disagree, 2 = somewhat disagree, 3 = neither agree nor disagree (neutral), 4 = somewhat agree, and 5 = totally agree.

The data were collected from April to September 2020. Firstly, a pilot study including 82 observations was implemented before distributing the online survey (via Microsoft Forms) and the offline survey. Next, the questionnaire was revised based on the Cronbach's Alpha value from the initial pilot test. In the formal survey stage, a non-probability sampling method was implemented, and a total of 413 responses were collected from 12 HEIs in the North of Vietnam. The 15 bias observations were eliminated. Finally, there were 398 observations valid for further analysis, of which 313 observations (78.6%) were from the online survey via Microsoft Forms (in the period of social distancing due to the COVID-19 epidemic in Hanoi, in April 2020), while 85 observations (21,4%) were from the offline survey (after the period of social distancing; from May to September 2020). Table 3 shows the descriptive statistics of the participants' demographics.


**Table 3.** Descriptive statistics of participants' demographics.

According to Table 3, the ratio of males to females is 42.5% and 57.5% respectively. In terms of age, the majority of lecturers (48.7%) fall into the 31–40 age group, 30.4% fall into the 41–50 group, 12.6% fall into the 22–30 group, and only 8.3% of the lecturers are over 50 years old. Qualification in the survey sample is high; up to 52.8%, 31.4%, 14.8%, and 1% of lecturers have the academic title of Doctor, Master's degree-holder, Associate Professor, and Professor, respectively. The number of lecturers who have more than six years of working experience accounts for quite a high proportion (70.4%), while 11.6% and 10.6% of lecturers have 3–6 and 1–3 years of working experience respectively. Only 7.5% of the lecturers have less than one year of experience. Regarding scientific research areas, social science takes the highest proportion at 55%, while natural science accounts for 45%. The ratio of abroad to domestic graduation is similar, with 47.7% and 52.3% respectively. The number of lecturers without managerial positions accounts for a higher percentage than those with managerial positions, with 63.3% and 36.7% respectively.

#### *3.2. Data Analysis Techniques*

Firstly, the *t*-test and ANOVA approaches were adopted to determine the differences in scientific research output between lecturer groups, according to their demographic characteristics. Next, Partial Least Squares Structural Equation Modeling (PLS-SEM) was employed to achieve the research objectives. This approach has several advantages. PLS-SEM was recommended in the initial period of theory development to access and verify the exploratory research models [46]. It also has several benefits for cause-and-effect analysis in behavior studies [47]. PLS-SEM has been the best alternative to CB-SEM in cases wherein there is little background theory available [30,48] even though PLS-SEM is not as effective as CB-SEM in model fit evaluation. Furthermore, there have not been any common

theoretical frameworks in previous studies on the factors influencing the research output of lecturers [25]. In addition, a new scale (i.e., decentralization) was added to the research model proposed in this study. For these reasons, it is appropriate to utilize PLS-SEM.

PLS-SEM is based on two main steps, namely, measurement model assessment and structural model assessment [46]. Within this study, the SmartPLS 3.3.3 application of PLS-SEM was used to assess the measurement model, the convergent, discriminant validity, and composite reliability. Finally, the bootstrapping technique analyzed the t-statistics for the path coefficients to assess the importance of the hypothesized connections.

The factors were coded as follows: OBJ = scientific research objectives of HEIs, DEC = decentralization, LEA = leadership, POL = policies for lecturers, SUP = support for scientific research activities, RES = resources for scientific research, and OUT = research outputs (Appendix A).

#### **4. Results**

#### *4.1.* t*-test and ANOVA*

Table 4 presents a summary of *t*-test and ANOVA results to demonstrate the difference between participants' research output among demographic variables.


**Table 4.** Summary of and *t*-test and ANOVA results.

The difference in the scientific research results of lecturer groups is summarized hereafter. Regarding gender, there is a difference in scientific research output between the sexes, shown by the sig. of the *t*-test being <0.05. More specifically, the mean value shows that male lecturers have higher scientific research output than female lecturers.

Concerning age, there is a difference in scientific research output between age groups, shown by the sig. of the Welch test being <0.05. More specifically, the mean value of age

groups shows that scientific research output decreases in the order of the following age groups: 41–50, 31–40, >50, and 22–30.

In terms of academic title and degree, there is a difference in scientific research output among lecturer groups with different qualifications, shown by the sig. of the F test being <0.05. More specifically, the mean value of the groups shows that scientific research output increases gradually by the level of academic title and type of degree.

In terms of lecturer experience, there is a difference in scientific research output among lecturer groups with different seniorities, shown by the sig. of the Welch test being <0.05. More specifically, the mean value shows that scientific research output increases gradually with seniority.

Regarding the scientific research area, there is a difference in scientific research output among lecturer groups, shown by the sig. of the T-test being <0.05. More specifically, the mean value shows that scientific research output is higher for the field of natural sciences than it is for the social sciences.

Concerning managerial position, there is a difference in scientific research output between lecturer groups, shown by the sig. of the T-test being <0.05. More specifically, the mean value shows that scientific research output is higher for lecturers who do not hold managerial positions compared to those who do. This is likely because lecturers who hold managerial positions have busier schedules and less time allocated for conducting research.

In terms of abroad graduation, there is a difference in scientific research output among lecturer groups, shown by the sig. of the T-test being <0.05. More specifically, the mean value shows that scientific research output is higher for lecturers who graduated abroad.

#### *4.2. Measurement Proposed Research Model Assessment*

The psychometric properties of the scales measuring the considered dimensions of scientific research output were first examined through the Exploratory Factor Analysis (EFA) procedure, before being included in the PLS-SEM model. The study procedure was carried out using SmartPLS software version 3.3.3. The SEM model included six constructs, namely, scientific research objectives, decentralization, leadership, support for scientific research activities, policies for lecturers at HEIs, and resources for scientific research. To assess the measurement model, the convergent and discriminant validity and composite reliability were considered.

In Table 5, for this measurement model, all of the quality criteria were met, since all factor loadings, Cronbach's alpha values, composite reliability (CR) values, and average variance extracted (AVE) values were above the recommended thresholds (0.7, 0.7, 0.7, and 0.5, respectively) [24,49–51].

Discriminant validity is the degree to which items distinguish between constructs. Using the Fornell–Larcker criterion, the results indicate that the square root of the average variance extraction is greater than the inter-construct correlations. Regarding the crossloadings criterion, the factor loadings of each item or indicator must be greater than the rest of its cross-loadings to ensure the discriminant validity of the construct [51]. Seven criteria were used to test discriminant validity, including Fornell–Larcker and cross-loadings. Table 6 illustrates the results of the discriminant validity measurements. In the first column, the square root of the extracted variance that appears in the upper part in parentheses must be greater than the correlations that appear in the following lines of the same column. This criterion was applied for each column. Table 7 shows the fulfillment of this criterion for all the subscales, demonstrating the discriminant validity of the tested instrument.


**Table 5.** Convergent validity and reliability.

**Table 6.** Discriminant validity.



**Table 7.** Hypotheses testing.

Note: OBJ: scientific research objectives of HEIs; DEC: decentralization; LEA: leadership; SUP: support for scientific research activities; POL: policies for lecturers; RES: resources for scientific research; OUT: scientific research outputs.

As the goal of SEM-PLS is to explain the endogenous latent variance, the key target is to have a higher R square. The greater the value, the better the explanatory power of the model [52]. The authors of [53] argued that a value of R square greater than 0.26 is considered substantial, as a rule of thumb. The results obtained in this present restudy show that the R square value for scientific research output was 0.684, which was acceptable. The corresponding results are presented in Figure 2.

**Figure 2.** Confirmatory Factor Analysis Result.

### *4.3. Testing Research Hypotheses*

In the structural model, the relevance and significance of all the direct and indirect effects were assessed, examining the path coefficients, associated t-statistics, and their biascorrected confidence intervals, which were computed through a bootstrapping procedure. The study conducted the test with a sample size of bootstrapping N = 5000 [46,54]. The proposed hypotheses were considered statistically significant at the 99%, 95%, and 90% reliability levels.

The bootstrapping technique was used to analyze the t-statistics for the path coefficients to assess the importance of the hypothesized connections [55,56]. The *p*-value is a constant measure of evidence, but it is usually dichotomized into highly important, marginally important, and not statistically important at conventional levels, with cut-offs at *p* ≤ 0.01, *p* ≤ 0.05, and *p* > 0.10, respectively [57].

Table 7 shows that scientific research output (*p* < 0.01) shares a significant relationship with scientific research objectives, leadership, support for scientific research activities, policies for lecturers, and resources for scientific research. This means that five hypotheses in the conceptual model were fully supported (Hypothesis 1, Hypothesis 3, Hypothesis 4, Hypothesis 5, and Hypothesis 6). Decentralization had no direct effect on scientific research output (*p* = 0.217 > 0.01), thus Hypothesis 2 was not supported. Among these variables, resources for scientific research (RES) was the most effective factor (β = 0.471, t = 16.025, *p* = 0.000). Policies for lecturers (POL) and support for scientific research activities (SUP) had the second strongest influence on scientific research output (β = 0.359, t = 10.872, *p* = 0.000; β = 0.263, t = 9.587, *p* = 0.000, respectively). Scientific research objectives of HEIs (OBJ) and leadership (LEA) were also significant factors that affected scientific research output (β = 0.151, t = 4.768, *p* = 0.000; β = 0.121, t = 3.306, *p* = 0.000, respectively).

#### **5. Discussion**

#### *5.1. Findings and Implications*

The results reveal that "resources for scientific research" have the most significant impact on scientific research output by lecturers among the six given governance factors. This reflects the reality of such resources, especially the limited facilities of universities in Vietnam, with narrow university campuses that lack synchronization. Financial resources are also limited in terms of both volume and procedures because most Vietnamese HEIs remain partially dependent on the state's budget; there exist only 23 autonomous universities [58] with better resources.

This finding supports the results of prior studies, from a university governance perspective. By empirical evidence, Ref. [5,7,17,18,37,38] proved that resources for scientific research, including space, equipment, information systems, databases, expenses, funds, and colleagues with good research capacity had a positive influence on lecturers' scientific research output. In Vietnam, Ref. [13] revealed that factors such as the accessibility of international scientific documents, research data, experimental devices or tools, software, and funding sources played an important role in international publishing. The work by [16] demonstrated that the main barriers to publication in Vietnam are funding and time for research.

Therefore, the implication is that HEI managers should pay careful attention to ensuring adequate resources for scientific research. First of all, it is vital to improve research space, equipment, free information systems, databases, and digital libraries for scientific research activities at HEIs. In terms of finance, HEIs should attract more non-state budget revenues and establish funds for internationally-indexed publications, intellectual property applicants, and the commercialization of scientific research outcome. In addition, HEIs should reform the mechanism of budget allocation for scientific research in particular, at both the university and faculty levels.

Secondly, the obtained results demonstrate that "policies for lecturers" had a significant influence on lecturers' scientific research output. This finding provides a comprehensive assessment of how policies for lecturers affect their research outcomes. In fact, the income of lecturers at Vietnamese HEIs that are not yet autonomous is generally low, according to the general regulations of the State [59]. The regulations on workload and rewards for lecturers are all "one-size-fits-all" policies [14]. The finding of this research is consistent with the previous studies. However, provided with a governance perspective, the study has become more relevant for university administrators and management researchers. For instance, the work by [11,20,37,38,60] affirmed that policies related to income, recruitment, remuneration, evaluation, reward, retention, training, and development of teaching staff had a positive impact on the scientific research results of facilities. In Vietnam, Ref. [12,13] pointed out that "time for research purpose" affected the research productivity of Vietnamese social scientists.

Thus, it is recommended that university administrators pay close attention to developing and improving policies for lecturers in order to enhance the overall effectiveness of

scientific research. In Vietnam, the time available for research purposes must be taken into account by HEI managers and policy-makers. Sabbatical leave, which is popular in developed countries as a time to focus on research, should also be considered by Vietnamese HEIs. Furthermore, "tailor-made" incentives for different types of lecturers should be implemented in Vietnamese HEIs. Lecturers with high research output should be rewarded and paid differently compared to others with relatively lower research output.

Thirdly, the results of testing hypotheses show that both "support for scientific research activities" and "scientific research objectives of HEIs" had a positive effect on lecturers' scientific research output. This is also in line with previous research. The authors of [19] and [29] argued that scientific research support had the most significant effect on institutional factors. The work by [20] demonstrated that setting clear common goals in scientific research was one of the factors that had the strongest impact on the results of scientific research by lecturers. The influence of an organization's scientific research objectives on the faculty's scientific research results has also been tested and recommended by [6,19,22,32,61].

Therefore, the implication offered here is that HEIs should actively plan long- and medium-term strategies for scientific research activities, with the positioning of clear scientific research results, along with mechanisms and resources to ensure the validity of the strategies. They should not simply offer annual plans according to the force of administrative procedures. This solution can potentially increase scientific research results in a sustainable and focused manner.

Additionally, HEIs should pay close attention to creating an environment and culture that values and supports scientific research, building and developing research groups, and encouraging collaboration between scientists inside and outside the organization. The managers of Vietnam's HEIs also need to pay attention to administrative reforms, scientific research management mechanisms, and digital transformation to actively support scientific research activities.

Fourthly, the factor "leadership" had a positive influence on lecturers' scientific research output, but it had less of an impact compared to the above-mentioned factors. This finding is consistent with previous studies, but it has more significance when applied to university governance particularly. The authors of [9,20,30] demonstrated that institutions with academic excellence had leaders who could link scientific fields, select and train young faculty members, encourage and develop new scientific ideas, attract funds, build an environment to promote research and creativity, and set and disseminate goals.

Therefore, the implication for university governance is to further improve the awareness of HEIs' leaders of the importance of scientific research in the current context. They need to have a sense of regular and clear communication about the organization's scientific research goals, always recognize and appreciate the results of scientific research, and be fair in resource allocation. The selection and appointment of managers in departments, laboratories, faculties, or universities should have specific criteria that are suitable for the particular job (i.e., high creativity and different from ordinary administrative management). In addition, in the current integration context, HEIs also need to pay attention to fostering and improving management skills towards international standards for university managers.

Fifthly, "decentralization" is a new factor that was added in the research model, and was the only factor that did not have an impact on lecturers' scientific research output. Due to the relevant findings in previous studies, this factor was tested and included in the research model. For example, Ref. [20] demonstrated that participative governance was a characteristic of organizations with high scientific research achievements. The authors of [30] clarified that autonomy in building a research team and staff evaluation policy each had a good influence on the results of scientific research. This result can be explained and supported by several prior studies. The theory of management Y [62] noted that it is impossible to have complete autonomy to achieve both organizational goals and individual needs. Therefore, there is a need to decentralize at an appropriate level. The authors of [63]

argued that decentralization does not mean sharing power to the extent that senior leaders in the organization do not know the important rights of subordinates.

Interestingly, to the best of our knowledge, this may be the first study to quantify "decentralization" in relation to university lecturers' research results. It has experimentally proven whether decentralization affects the research performance of lecturers. Although the results of testing the model and hypothesis H2 show that this factor does not affect the scientific research output of lecturers with a significance level <=0.05, this finding still forms a theoretical and practical contribution as a basis for formulating appropriate university governance policies.

Overall, this study has built a model and verified the influence of governance factors on the scientific research output of lecturers. The results obtained can act as a guideline to aid university managers in improving governance by prioritization of resources for scientific research, policies for lecturers, support for scientific research activities, setting scientific research objectives, and leadership.

Sixthly, the *t*-test and ANOVA results indicate that there is a difference in the scientific research output between groups of lecturers. More specifically, the scientific research output by males is more than that by females. The output of natural sciences is more than that of social sciences. The output by lecturers who do not hold managerial positions is more than that by those who hold such positions. The output by graduates from universities abroad is more than that by graduates from domestic universities. The output by those with advanced degrees is more than that by those with relatively low ones. The output by higher seniority is more than that by lower seniority. Finally, the output by the middle-aged group is more than that by the elderly and younger lecturer groups. The difference between these groups is also mentioned in prior studies [7,11].

Thus, it is recommended that HEI managers formulate suitable policies for different groups of lecturers. For example, the *t*-test and ANOVA results show that lecturers in the natural sciences have higher scientific research output compared to those in the social sciences. Hence, HEIs should have appropriate resource investment policies for each field, and during each period. To rapidly increase scientific research results in the short term, HEIs should focus on attracting and prioritizing more investment for groups of lecturers with higher scientific research output, such as groups with doctoral degrees or higher (especially associate professors and professors); those who graduated abroad; those in the 31–40 and 41–50 age groups; those with at least three years of seniority; and those in the natural sciences. However, to ensure sustainable and balanced development in the long term, HEIs need to pay attention to the scientific research activities of the remaining groups. In addition, the *t*-test results show that the group that does not hold a management position had higher scientific research output than the other. Therefore, the implication is that HEIs should change their mindset about appointing managers: excellent scientists should manage research groups, laboratories, and centers of excellence, without the necessity of holding other administrative positions that limit time and scientific research performance.

Finally, this study has a novel contribution in terms of research methodology on this topic. As the literature showed, it is difficult to find a study that adopted PLS-SEM, even though this approach was considered to be the best alternative to CB-SEM in cases where there is little theoretical background available [30,48]. Moreover, in this study, PLS-SEM was employed in conjunction with the bootstrap technique, *t*-test, and ANOVA to provide detailed insights on this increasingly important topic. In Vietnam, to the best of our knowledge, this is the first study that utilized the SEM approach with such a large population and primary data for this particular scientific issue.

Furthermore, compared with other works in the literature, this work can be considered a rare empirical study that has focused on testing the influence of governance factors (i.e., not institutional factors in general) on the scientific research output of university lecturers. Although the impact of management activities on employee performance is no longer a new research topic, the number of studies focusing exclusively on governance factors

affecting the scientific research productivity of lecturers is lacking. To date, most of the studies only consider the institutional, environmental, and personal factors that affect the results of scientific research.

In summary, this study not only offers theoretical contributions, but practical contributions as well, helping policymakers, managers, and lecturers to take appropriate solutions for groups of lecturers in order to enhance scientific research performance.

#### *5.2. Limitations and Suggestions for Further Studies*

All studies have their limitations, and this study is no exception. Firstly, this study uses a non-probability sampling method, so PLS-SEM has not been analyzed by different groups of lecturers (according to demographic characteristics) to understand the influence of governance factors on scientific research output by the groups. Meanwhile, non-parametric measures (Kruskal–Wallis Test, Mann–Whitney U Test, etc.) that are considered more suitable for non-probability sampling have not been utilized.

Second, this study explores the influence of governance factors based solely on lecturers' perceptions of their scientific research output. In other words, this is a cross-sectional study, so it is impossible to compare the change of scientific research output in reality, according to specific time and space milestones.

Third, this study assumes that governance factors have a direct influence on scientific research output. However, in reality there are other mediating factors, such as the behavior, motivation, and attitude of lecturers. Therefore, it is recommended that intermediary factors (both as independent and dependent variables) be added to the research model. Additionally, the review of the literature and practice also showed that there are many other governance and personal factors that affect scientific research output which have not been tested in this study.

Finally, the scope of this research has not focused on research-oriented universities in Vietnam in particular. Therefore, caution should be taken when generalizing the results of this study to all types of HEIs.

In the future, it is more reliable and worthwhile if researchers deploy the probability sampling method, thereby applying PLS-SEM for different sample groups and employing ANOVA more deeply for each sample group. It is possible to use time string data or array data (not cross-sectional data) to conduct empirical research to clarify the relationship between governance factors and scientific research results, which can be expressed after a long time. In addition, it is necessary to explore the relationship between governance factors and the attitude, motivation, and behavior of lecturers with respect to scientific research and scientific research output. It is also possible to combine different approaches (i.e., other individual factors and governance factors) to explain why, in the same environment and with the same governance, lecturers have differing scientific research output. Finally, it is necessary to expand the scope of the research according to groups of HEIs, which can be divided by their research or practice orientation, autonomy, whether they are multidisciplinary or focus on a single field, and whether they are public or private universities.

#### **6. Conclusions**

This study has proposed and empirically tested a research model on governance factors affecting the scientific research output of university lecturers. It has adjusted the measurement criteria (observed variables) of the governance factor scales and scientific research output according to the context of Vietnam. The results show that the level of influence of governance factors on lecturers' scientific research output, in order from strongest to weakest, is as follows: resources for scientific research, policies for lecturers, support for scientific research activities, scientific research objectives of HEIs, and leadership. A new scale of "decentralization" has been added to the analytical framework, but it was not statistically significant regarding its impact on lecturers' research output.

In addition, the study examined the difference in the scientific research output of lecturer groups according to their demographics, including gender, age, degree and academic title, experience, position (manager or not), place of graduation (abroad or domestic), and scientific area. The obtained results reveal that there are differences in the research output between lecturer groups. Hence, they provide empirical evidence for implications in management decision-making and policy-making.

Overall, the obtained results can guide HEIs in evaluating the current status of scientific research activities and the governance factors that influence lecturers' scientific research output, thereby helping these HEIs take appropriate measures to enhance their scientific research achievements in today's knowledge-based economy.

**Author Contributions:** Truong, H.T. and Le, H.M. have mainly contributed for giving the paper idea, writing, reviewing and editing. Le, D.A.; Nguyen, H.T.; and Nguyen, T.K. contributed for interviewing and collecting the data, questionnaire designing and result calculating. Do, D.A contriuted for reading and giving the comment and conclusion parts of the paper. All authors have read and agreed to the published version of the manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare that they do not have any conflict of interest.

#### **Appendix A Description of Scales**



