Use of Knowledge Management to Enhance International Research Collaboration

Abstract

With globalization and rapid changes in the international research environment from technological advancement, political instability, and economic crisis, knowledge management (KM) is crucial to help research institutions operate international research collaboration (IRC) effectively and sustainably. This study uses systematic literature review to extract key KM factors for IRC enhancement. Exploratory factor analysis and structural equation modeling methods are performed to confirm KM factors and explore how key KM and IRC factors relate to each other. Several KM strategies are established based on study results to achieve sustainable IRC development. The results show that five key KM factors, including knowledge sharing (KS), knowledge creation (KC), knowledge retention (KR), knowledge storage (KT), and knowledge utilization (KU), influence each other. They have both direct and indirect impacts on IRC. The KU factor is crucial for immediate IRC improvement. Research institutions can use existing knowledge and resources to address current IRC opportunities. For example, personnel with IRC experience can act as coaches and mentors to facilitate activities, and integrating IRC into career paths can be beneficial. Activities related to KC, KR, and KT should support KU implementation. Setting up a task force, improving organizational structure, and engaging management in KM can help achieve better IRC performance. The KS factor should be emphasized for sustaining IRC. The plan should involve activities to raise the processes of knowledge sharing effectively. The study results provide guidelines for research institutions aiming for sustainable IRC in the long term.

1. Introduction

1.1. International Research Collaboration

The world is rapidly evolving into a global community, where common challenges and opportunities await beyond borders. International research collaboration (IRC) has proven to be crucial for tackling global problems, such as the emergence of COVID-19, by involving researchers from different countries to seek solutions for humanity and explore new boundaries of technological knowledge (Bump et al., 2021). Hückstädt and Leisten (2024) added that IRC is leading to scientific progress, innovative scientific knowledge, technological skill, and supportive resources. Successful IRC requires an adequate R&D budget to promote scientific publications, research mobility, and joint research projects. Public investment in R&D stimulates behavioral changes within research institutions and firms, enhancing research productivity, industry outcomes, and IRC (Ul Haq et al., 2020). Direct government support and R&D tax incentives encourage organizations to engage in long-term IRC by reducing financial risks and attracting dynamic actors to the R&D environment. Sustained and well-structured R&D investment catalyzes scientific advancement across sectors and nations.

Aksnes and Sivertsen (2023) stated that IRC tends to be more successful in developed countries. Over the last two years, R&D expenditures in Great Britain, Japan, and the USA were up to three times higher than those allocated in Thailand (see Figure 1). In Thailand, the National Science and Technology Development Agency (NSTDA), the country’s largest R&D institute, faces several challenges in boosting IRC. The common challenges are language and cultural barriers between collaborative researchers. Intermittent communication among cross-country team members frequently leads to confusion, misunderstandings, and inaccuracies. In addition, the working personnel in organizations often change. Vasantham and Aithal (2022) stated that a high turnover rate undermines overall organizational performance by reducing productivity, creating excessive workload and stress, and diminishing the information and knowledge of organizations’ intellectual properties.

1.2. Knowledge Management

As IRC is a complex and dynamic issue requiring workforce and precise knowledge, knowledge management (KM) has been adopted to achieve sustainable IRC. It is a multidisciplinary approach, focusing on creating, sharing, and managing knowledge (Girard & Girard, 2015). It is increasingly being recognized as a critical organizational resource for facilitating innovation. Sharma and Kaur (2016) suggested that KM can support effective decision-making and organizational competitiveness. The emergence of KM allows a shift from an industrial-based economy to a knowledge-driven economy, where the ability to manage knowledge effectively becomes a key determinant of long-term success (Martelo-Landroguez & Cepeda-Carrión, 2016). It is a structured approach to managing intellectual assets, leveraging explicit and tacit knowledge to create value for organizations. While explicit knowledge is codified and easily transferable, tacit knowledge is deeply embedded in individuals and requires socialization for an effective transfer (Smith, 2001). A theory-like knowledge-based view (KBV) emphasizes knowledge as the most valuable asset in firms (Grant, 1996; De Moortel & Crispeels, 2024). It explains that knowledge creation and integration mainly come from tacit knowledge. It is used to support comprehensive investigations in international business and university–industry research collaboration (Stoian et al., 2024).

1.3. Knowledge Management in International Research Collaboration

A number of studies have utilized KM in enhancing IRC. For example, Costa and Monteiro (2016) mentioned that KM databases benefit future IRC activities through effective collaborations. Ling (2011) and Zamiri et al. (2019) added that information and knowledge sharing among team members can mitigate confusion and misinterpretation of collaborative activities and raise IRC performance. Information and knowledge sharing between different researchers from various backgrounds can raise IRC performance through the co-authorship of publications and research findings (Numprasertchai & Barbara, 2005).

Good KM is crucial for successful international collaboration in various areas, such as government policies, education, and businesses. Haaland and Kind (2008) found that the government can use R&D subsidies as a KM tool to help local firms compete globally. Coordinating these subsidies between countries may lead to better IRC results. Li-Hua (2007) stated that the sharing of personal experience and cultural differences may improve education systems between Asian and European countries. Rodríguez et al. (2018) commented that knowledge sharing between partners from neighboring countries helps to improve professional and technology services. Papadaki and Polemi (2008) developed a KM platform, combining knowledge codification with personalization strategies to support information security risk management. Nguyen-Viet et al. (2025) stated that the Consultative Group for International Agricultural Research (CGIAR) network was initiated utilizing KM strategies, such as centralized data repositories, harmonized research protocols, and participatory platforms, to enable the systematic sharing of datasets, models, and tools between international teams. The network emphasizes codified (e.g., standardized datasets, policy briefs, and technical guidelines) and tacit (e.g., expert network, capacity-building workshops, and stakeholder engagement sessions) knowledge to bridge knowledge gaps and foster inclusive and impactful IRC. Goffin et al. (2011) applied KM concepts to support international project management by emphasizing the creation and utilization of a knowledge-based framework for managing complex and cross-border projects. It was suggested that international project managers could capture knowledge during project processes using standardized templates, store it in a structured database, and reuse it for future projects.

1.4. Problem Statement, Research Aim, and Research Objectives

Despite several studies on KM and IRC, the interrelationships between key KM factors and IRC have been rarely studied. To plan IRC using KM, it is important to understand the relationships between KM factors and IRC. This study uses exploratory factor analysis (EFA) to identify key KM factors, IRC factors, and their associated items in Thailand’s R&D institutions. Structural equation modeling (SEM) then examines how these factors directly and indirectly relate to IRC. This study’s results are expected to guide how KM can enhance IRC, helping policymakers to create better strategies for sustainable IRC.

In summary, this research study aims to examine interrelationships between KM factors and their associations with IRC to propose strategic planning for IRC enhancement. The research objectives are set.

• Performing a systematic literature review (SLR) to extract key KM factors, IRC factors, and their associated items;
• Developing a questionnaire survey using key KM factors, IRC factors, and their associated items from SLR results;
• Performing a pilot test through interviews to screen and adjust the questionnaire survey prior to collecting data for analyses;
• Finalizing the questionnaire survey based on pilot test results to be used for data collection;
• Collecting data for the analyses;
• Performing preliminary analysis to screen data for the analyses;
• Performing EFA to confirm key KM factors, IRC factors, and their associated items;
• Performing SEM to examine relationships between key KM factors and their effects on the IRC factor;
• Proposing strategic plans related to key KM factors to enhance IRC.

Figure 2 shows the research framework. This study reviews literature on KM and IRC issues worldwide. Key KM factors and their associated items are identified from literature. A questionnaire survey is developed and tested with experts in pilot testing. This helps to finalize the survey used for data collection. The collected data are analyzed with preliminary tests, including normality and outliers, to ensure suitability for further analysis. Screened data then undergo EFA to confirm groupings of KM and IRC factors. Measurement and structural models are utilized to confirm correlations and directions between KM and IRC factors. These steps lead to the final model of KM and IRC factors. The analysis results are used to develop strategic plans to achieve sustainable IRC operations.

2. Theoretical Framework

2.1. KM Factors and Their Associated Items

To systematically list key KM and IRC factors, we performed a systematic literature review. The Scopus and Thai Citation Index (TCI) databases were used to extract key factors. Keywords, including “KM and international collaboration” or “KM and international cooperation” or “KM and inter-organizational collaboration” or “KM and international research collaboration” and “factor” were input into the databases to extract related publications. The search was limited to the past 10 years (2015–2025) and English- and Thai-language journal publications. Five hundred ninety-six articles were extracted and screened to categorize key KM factors using key themes. Grouping was finalized with frequency analysis to obtain five KM factors, representing over 90% of the total frequencies, confirming their importance in KM-related studies (Hosking, 1990) (see

Table 1. Summary of five KM factors extracted from the literature.

Key KM Factor	Number of Articles	Frequency (%)	Cumulative (%)
Knowledge sharing (KS)	121	20.30	20.30
Knowledge creation (KC)	116	19.46	39.76
Knowledge storage (KT)	108	18.12	57.88
Knowledge retention (KR)	97	16.28	74.16
Knowledge utilization (KU)	95	15.94	90.10
Knowledge acquisition	24	4.03	94.13
Knowledge capture	16	2.68	96.81
Knowledge retrieval	11	1.85	98.66
Knowledge distribution	8	1.34	100.00

).

The five extracted KM factors and the IRC construct were further examined by adding keywords like variable, item, attribute, and factor name to the Scopus database, one factor at a time, to list the associated items. For example, the keywords “knowledge sharing,” “items,” “variables,” and “attributes” were input, and three items mostly cited as associated items of the KS factor were listed: competency, networking, and organizational culture. The details of KM and IRC factors and their associated items are summarized below.

2.1.1. Knowledge Sharing (KS) Factor

Knowledge sharing refers to exchanging and distributing tacit and explicit knowledge to others, creating a distribution of knowledge flow and diverse expertise and perspectives (Kamasak & Bulutlar, 2010; Yeşil et al., 2013; Castaneda & Cuellar, 2020). It includes three associated items.

• Competency (KS1): The interpersonal skill that identifies issues and shares relevant knowledge to initiate and promote collaborations (Kamasak & Bulutlar, 2010). Having staff with suitable backgrounds and competencies ensures a sustainable IRC.
• Networking (KS2): The interaction among people within and beyond the same organization to exchange expertise and knowledge (Castaneda & Cuellar, 2020). Activities like consortia and mobility programs may be initiated to expand networking between organizations and their international partners (Puljak & Vari, 2014).
• Organizational culture (KS3): A culture where personnel share knowledge to achieve organizational goals (Castaneda & Cuellar, 2020).

2.1.2. Knowledge Creation (KC) Factor

Knowledge creation is achieved by combining existing insights with new information from the knowledge flow. New knowledge gained creates innovation and work improvement (Merx-Chermin & Nijhof, 2005; Esterhuizen et al., 2012). It includes six associated items.

• Common knowledge (KC1): The shared understanding and experiences that foster collaboration and decision-making (Castaneda & Cuellar, 2020).
• Feedback (KC2): The delivery process that includes insights to refine actions and improve strategic decisions based on past outcomes (Ibrahim & Al-Shara, 2007). Strategies like after-action review may be implemented to retrieve positive and negative feedback for further improvement (Shuffler et al., 2018).
• Organization structure (KC3): The structure that accommodates knowledge flows and collaboration dynamics (Castaneda & Cuellar, 2020).
• Teamwork (KC4): Teamwork is required to achieve goals and create new knowledge (Castaneda & Cuellar, 2020).
• Training and education (KC5): Skill development to support ongoing organizational learning and innovation (Ibrahim & Al-Shara, 2007).
• Trust (KC6): The willingness to embrace vulnerability and differences among collaborators to create new knowledge effectively (Mayer et al., 1995; Ford, 2004).

2.1.3. Knowledge Storage (KT) Factor

Knowledge storage refers to systematic processes of capturing, organizing, and preserving organizational knowledge in structured repositories to enable efficient retrieval and reuse. It involves converting explicit (e.g., documents, databases, and manuals) and tacit (e.g., expertise and experience) knowledge into accessible formats in KM systems. It emphasizes technical and procedural mechanisms that support the codification and structuring of knowledge assets. It comprises four associated items.

• Documentation (KT1): Structured information and documentation are required in KM practices for future learning and collaboration (Egeland, 2017).
• Mutual benefits (KT2): The gained understanding that benefits individuals and organizations by preserving IRC knowledge (Wikström et al., 2018).
• Shared resources (KT3): Data and workforce should be shared between IRC parties to prevent knowledge loss (Tung-Ching et al., 2016).
• Supporting policies (KT4): KM-related policies are required to store knowledge effectively (Ford, 2004).

2.1.4. Knowledge Retention (KR) Factor

Knowledge retention refers to an organization’s capability to maintain the availability and continuity of valuable knowledge when facing employee turnover, technological changes, and organizational restructuring. It focuses on preventing knowledge loss and ensuring that critical knowledge remains usable to support long-term learning and decision-making. It is achieved through a combination of human-centered practices (e.g., mentoring and succession planning) and system-based approaches (e.g., KM systems) (Girard & Girard, 2015; Marsh & Stock, 2006; Papa et al., 2020). This factor comprises seven items.

• Funding (KR1): Adequate budget for KR initiatives is required to ensure knowledge continuity (Egeland, 2017).
• Human resource management (KR2): The planned workforce policy to support IRC and prevent knowledge loss (Papa et al., 2020).
• Incentives (KR3): Incentives like financial and recognition programs boost employee engagement in KM implementation (Egeland, 2017).
• IT support (KR4): Technological infrastructure and digital tools for knowledge preservation, such as management databases, activity logs, and data pools, may be used to effectively store, retain, and utilize knowledge (Ibrahim & Al-Shara, 2007).
• Management commitment (KR5): Management plays a crucial role in retaining KM-related activities (Egeland, 2017).
• Measurement (KR6): Quantitative and qualitative indicators used to assess the effectiveness of KR strategies and guide continuous improvement (Tung-Ching et al., 2016). Tools, such as balanced scorecard and key performance indicators, may be used to measure IRC activities (Kaplan et al., 2010).
• Shared risks (KR7): Shared responsibility among personnel to minimize knowledge loss and ensure continuity during personnel changes (Egeland, 2017).

2.1.5. Knowledge Utilization (KU) Factor

Knowledge utilization refers to using knowledge to drive efficiency, innovation, and competitive advantages. Utilizing existing knowledge may generate new knowledge that enhances efficiency, innovation, and collaboration. It consists of four associated items.

• Career path (KU1): Successful IRC should be used as an indicator for a career path (Lazarova & Taylor, 2009).
• Competitiveness environment (KU2): Utilization of knowledge enables organizations to sustain and strengthen their competitiveness (Zaim et al., 2018).
• Leadership (KU3): Effective knowledge utilization requires leaders who promote, share, guide, and foster IRC activities (Egeland, 2017).
• Strategy and purpose (KU4): KM initiatives should be clearly listed in the organization’s strategic plans (Egeland, 2017). Junior researchers may utilize strategies like a community of practices (CoP) to exchange viewpoints, and the successes and challenges of IRC activities (Pyrko et al., 2017).

2.2. IRC Factor and Its Associated Items

Successful KM implementation may lead to successful IRC, which can be measured through items associated with the IRC factor. To systematically list IRC’s associated items, we performed a systematic literature review with Scopus and TCI databases. Keywords, including “international collaboration” or “international cooperation” or “inter-organizational collaboration” or “international research collaboration” and “items” or “variables” or “attributes” were input into the databases to extract related publications. The search was limited to the past 10 years (2015–2025) and English- and Thai-language journal publications. Four hundred ninety articles were extracted and screened to categorize associated items of IRC. Grouping was finalized with the frequency analysis to achieve five associated items, representing over 90% of the total frequencies (see

Table 2. Summary of five associated items under the IRC factors extracted from the literature.

IRC Factor	Number of Articles	Frequency (%)	Cumulative (%)
Publication (PB)	134	27.35	27.35
Financial indicator (FI)	112	22.86	50.21
Intellectual property (IP)	88	17.96	68.17
Mobility (MT)	75	15.31	83.48
Collaborative document (CD)	50	10.20	93.68
International visibility	31	6.33	100.00

). The details are as follows.

• Publication (PB): Effective IRC produces publications collaborating with multiple authors from different institutions and countries (World Intellectual Property Organization, 2024).
• Financial indicator (FI): In a good IRC environment, internal and external funding fosters innovation and technological advancement and transforms academic findings into marketable products (World Intellectual Property Organization, 2024).
• Intellectual property (IP): This IRC indicator refers to the amount of jointly owned patents by organizations and companies, universities, and research institutions (World Intellectual Property Organization, 2024).
• Mobility (MB): The number of international visitors visiting researchers/interns at the host organizations and counting participation in international collaborative activities (World Intellectual Property Organization, 2024).
• Collaborated documents (CD): The number of proposals for international grants, signed R&D contracts, memoranda of understanding (MOUs), and memoranda of agreement (MOAs) with international partners (World Intellectual Property Organization, 2024; National Science and Technology Development Agency, 2023).

In summary, the five KM factors comprised 24 associated items, whereas the IRC factors consisted of five associated items. They were used to develop a questionnaire survey to collect data for EFA and SEM analyses.

3. Materials and Methods

The research steps of this study are shown in Figure 3 to explain the steps following research objectives and methods used in each step.

3.1. Systematic Literature Review

The SLR method was used to extract key KM factors, IRC factors, and their associated items in this study. This method is used to examine themes that can reveal what lies underneath the viewpoints of datasets (Chinda & Chinda, 2025). It has been used in various studies. For example, Chinda and Chinda (2025) conducted an SLR to identify carbon neutrality implementation methods in major carbon-intensive industries in Thailand. They concluded four implementation methods, including renewable energy, carbon sequestration, carbon capture and storage, and clean energy vehicles, mostly used in carbon-intensive industries (i.e., electricity, chemical, construction, and transportation sectors) in Thailand. Amaechi et al. (2025) applied an SLR method to KM for project risk management in construction and concluded three key factors, namely project management, KM, and risk management. According to Chinda and Chinda (2025), the steps of conducting an SLR include selecting appropriate databases, inputting keywords relevant to the research topic, selecting publication types, publication years, and languages used in the publication, manually screening articles, and extracting key factors using relevant themes. Key KM factors, IRC factors, and their associated items extracted from SLR were used to develop a conceptual model for data collection and analyses.

3.2. Data Collection Method

The items associated with KM and IRC factors extracted from SLR were used to develop a questionnaire survey for data collection. A questionnaire survey was chosen because of its ability to reach a large and geographically dispersed population. They are mostly used for analyzing attitudes, behaviors, and perceptions of respondents. They are cost-effective and can be adjusted to reduce respondents’ bias (Rowley, 2014; Taherdoost, 2016; Brace, 2018). In this study, the questionnaire survey comprised three sections: general information, KM factors, and IRC factors (See Supplementary Materials). The general information section comprised five questions about the respondents’ information, including affiliations, educational background, job titles, work experiences, and involvement in IRC. The respondents were from research institutes, universities, and partners in Thailand and held various positions, from operational officers to top managers. The KM section contained statements on 24 items associated with five KM factors. Respondents were required to rate their agreements using a five-point Likert scale, where one was “strongly disagree,” three was “neither agree nor disagree,” and five was “strongly agree” (Garcia & Mollaoglu, 2020). Similarly, the IRC section comprised statements following five IRC associated items to be rated using the five-point Likert scale. An example of a questionnaire survey is in

Table 3. Examples of statements in the questionnaire survey.

No.	Statement	Scale
1.	Being able to communicate fluently in English or the languages of international partners supports IRC.	1	2	3	4	5
2.	Quickly adapting to different cultures of international partners supports IRC.	1	2	3	4	5
3.	Offering a clear career path to an upper managerial level position supports IRC.	1	2	3	4	5
4.	Knowing the regulations of international collaborative programs supports IRC.	1	2	3	4	5
5.	Knowing competitors’ strategies supports IRC.	1	2	3	4	5

.

3.3. Pilot Test

Pilot testing was performed using the interview method, as it could be arranged at a suitable time and in preferred languages. Pilot testing aimed to gain insights and in-depth opinions of respondents. It was conducted with selected experts to ensure the suitability of statements in the questionnaire survey before performing data collection. Statements in the questionnaire survey (following items associated with KM and IRC factors, see Table 2) were used in the interviews. Experts were requested to screen details of the survey and provide suggestions, if any, to improve the survey quality. Comments were summarized and used for questionnaire survey adjustment to achieve final survey for data collection. A total of 30 experts participated in the pilot test. According to Bujang et al. (2024), 15–30 experts are considered sufficient for assessing the reliability of a questionnaire. Most were researchers from research institutes with master’s and Ph.D. degrees. They had cross-border research experience as international project managers and facilitators, and had been involved in international research communities for up to six years. They were requested to screen the details of the survey and provide suggestions, if any, to improve the survey quality. The pilot test results suggested adjustments as follows.

• Provide the definitions of KM and IRC factors at the beginning of the survey to enhance the understanding of the statements.
• Incorporate questions about formal and informal meeting statements in the KS2 variable. This aligns with Nishimoto and Matsuda (2007), who stated that crucial information can be gathered through face-to-face interactions and formal meetings.
• Add an open-ended question at the end of the survey to gather insights from the respondents.

After the adjustments, the final survey was distributed to the respondents via email, LINE application, and social media platforms in response to the COVID-19 pandemic (i.e., March–June 2023).

3.4. Preliminary Analyses

The data collected were required to be screened prior to performing statistical analyses. This study utilized normality and outlier tests. The normality test ensured that the identified factor structures reflected the valid latent constructs. It was assessed using skewness and kurtosis values, in which the skewness value of ±2 and kurtosis value of ±7 confirmed the normal distribution (Martynova et al., 2018). The outlier test used boxplots to visualize data distribution, where potential outliers scattered outside the interquartile ranges (Hair et al., 2017).

3.5. Exploratory Factor Analysis

In this study, the screened data were analyzed with EFA to confirm key KM and IRC constructs. This is a multivariate statistical method that reduces a large set of items into a smaller set of factors (Watkins, 2018; Rogers, 2022). It identifies latent constructs and reviews underlying relationships between observed items that may not be immediately apparent. It has been widely applied across numerous fields, most notably in social and behavioral sciences, such as psychology, education, health sciences, economics, business, and construction (Finch, 2024). Various KM-related studies adopted the EFA method to group items into key factors for planning. For example, Abdul Rauf et al. (2020) used EFA to validate a multidimensional instrument measuring the KM component in Malaysian public higher education. Ehido et al. (2020) adopted EFA to test the dimensionality of organizational commitment and job performance measures among academics. Hartono et al. (2017) examined KM strategies in Indonesian construction firms and concluded that the codification strategy of documenting and storing knowledge in databases, manuals, and information systems positively influences a firm’s performance.

Data suitability, factor extraction, and factor rotation must be identified before performing the EFA. The Kaiser–Meyer–Olkin (KMO) and Bartlett’s Test of Sphericity are commonly used to ensure data suitability, in which the values of ≥0.6 at p < 0.05 are considered acceptable (Hair et al., 2017). Factor extraction involves identifying the underlying latent factors that explain the shared variance among observed items. Principal component analysis was used as the extraction method in this study as it can transform correlated items to uncorrelated components, preserving directions of maximum variance and minimizing overlapping variance among items (Hair et al., 2017). Factor rotation is a technique for enhancing the interpretability of extracted factors by adjusting factor loadings, while maintaining the overall explanatory power of the model. This study considered the varimax rotation method, as it assumes that factors were uncorrelated and had a simple structure for interpretation (Vitharana & Chinda, 2019). Key KM factors extracted from the EFA were further tested for reliability to confirm their constructs. Hair et al. (2017) stated that the Cronbach’s alpha values should be ≥0.70 to accept the constructs’ reliability.

3.6. Structural Equation Modeling

The KM and IRC factors achieved from the EFA were tested with SEM to examine direct and indirect effects between them. It is a comprehensive statistical technique used to assess complex relationships between observed items and latent factors. It combines elements of factor analysis and multiple regression, enabling researchers to model intricate causal paths and test theoretical constructions simultaneously (Rezaei, 2024). It has the capacity to analyze direct and indirect effects in a model, making it ideal for validating hypotheses in social sciences, education, and behavioral research (Rashno & Syahmansouri, 2025). It allows for the simultaneous analysis of multiple dependent relationships, making it robust and flexible (Khoddami et al., 2025).

The method comprises measurement and structural models (Martynova et al., 2018). The measurement model ensures that the observed items represent their latent constructs (i.e., KM and IRC factors) (Hair et al., 2017). In contrast, the structural model examines the directions of relationships between KM and IRC constructs (Byrne, 2016). The measurement and structural models are confirmed using fit indices, namely chi-square/df (χ²/df), comparative fit index (CFI), and root-mean-square error of approximation (RMSEA) (Byrne, 2016). The acceptable levels of the indices are ≤3.0 for χ²/df, ≥0.90 for CFI, and ≤0.08 for RMSEA (Hair et al., 2017). To improve the model fit, if required, modification indices (MIs) shown in the analysis output are considered as they suggest factors and items to be correlated to improve the model fit. However, the literature must confirm the added correlations to ensure their interpretation.

The relationships between KM and IRC factors in the measurement and structural models could be classified as weak, existing, moderate, or strong relationships through correlation and path coefficients. According to Hair et al. (2017), correlation and path coefficients below 0.3 are considered weak relationships, and those less than 0.1 should be removed. Values between 0.3 and 0.5 indicate existing relationships, 0.5–0.7 indicate moderate relationships, and those above 0.7 show strong relationships between the two factors (Martynova et al., 2018; Hair et al., 2017; Byrne, 2016; Mahdieh et al., 2024; Soltani et al., 2025).

4. Results

4.1. Data Collection and Preliminary Analysis Results

According to Jackson (2003), a sample size of at least 10 times the number of observed items is considered adequate for EFA and SEM analyses. Contrarily, Weston and Gore (2006) recommended a minimum sample size of 200 for SEM analysis. In this study, a minimum sample size of 300 datasets was required for reliable statistical analysis in this study (Martynova et al., 2018). The non-probability sampling method (purposive sampling) was used in this study as it is a common method for qualitative research (Palinkas et al., 2015; Etikan et al., 2016). Targeted respondents were management and personnel who worked in research institutions, research universities, and business sectors with R&D departments. They had knowledge and experience in KM and IRC.

Three hundred forty sets of questionnaire surveys were launched from March to June 2023, and three hundred twelve sets were returned, representing 91.76% of the response rate. Respondents were from research institutes, universities, and private partners (see Figure 4). Half of them worked at the operational levels, a quarter were junior managers, and over 20% were in top management positions. Most held Master’s and Ph.D. degrees, with more than half having over three years of work experience. These factors represented the suitability of the respondents in providing data for analyses, covering perspectives from all working levels in research organizations. The 312 datasets underwent normality and outlier testing. The skewness (ranging from 0.34 to 1.73) and kurtosis (ranging from 0.02 to 3.31) values confirmed the normal distribution of the 312 datasets; however, 2 datasets were identified as outliers as they had boxplots above the acceptable levels. As a result, they were removed, resulting in the final 310 datasets for EFA.

4.2. Exploratory Factor Analysis Results

The 310 screened datasets were tested for data suitability using KMO and Bartlett’s tests. The value of 0.94 at p ≤ 0.001 indicated high shared variance among items and confirmed sufficient inter-variable relationships, proving the suitability of the 310 datasets for EFA. The EFA was performed with the principal component analysis method, varimax rotation, and minimum factor loadings of 0.4 to group key KM and IRC factors (Cruthaka, 2019). The results (see

Table 4. Extracted KM and IRC factors from the EFA.

Item	Factor Extracted
	(1) KS	(2) KC	(3) KT	(4) KR	(5) KU	(6) IRC
KS3	0.85
KS2	0.84
KS1	0.68
KC5		0.83
KC4		0.82
KC3		0.71
KC2		0.68
KC6		0.68
KC1		0.63
KT4			0.84
KT3			0.74
KT2			0.69
KT1			0.64
KR2				0.80
KR5				0.78
KR6				0.77
KR3				0.75
KR1				0.75
KR4				0.72
KR7				0.68
KU1					0.81
KU2					0.78
KU3					0.75
KU4					0.63
CD						0.90
MT						0.90
FI						0.86
IP						0.84
PB						0.79

) confirmed 24 KM attributes and associate KM factors with a total variance of 64%. Factor 1 included KS3, KS2, and KS1, with factor loadings of 0.85, 0.84, and 0.63, confirming the KS factor. Factor 2 grouped six items under the KC factor previously hypothesized, with factor loadings ranging from 0.63 to 0.83. Factor 3, the KT factor, was confirmed to have four associated items with factor loadings ranging from 0.64 to 0.84. Factor 4 included seven items under the KR factor with factor loadings from 0.68 to 0.80. Factor 5 comprised four items, confirming the KU factor. Lastly, the IRC factor was confirmed with five items with factor loadings ranging from 0.79 to 0.90.

The common-method bias was performed using Harman’s Single-Factor test by forcing all KM items into one factor. The results confirmed that the bias did not affect the results significantly, with a total variance of 44.85% (i.e., less than 50%) (Jakobsen & Jensen, 2015). The extracted KM and IRC factors were further confirmed with the internal consistency, and the results (see

Table 5. Cronbach’s alpha, CR, and AVE values of KM and IRC factors.

Factor	Cronbach’s Alpha	CR	AVE
KS	0.70	0.84	0.63
KC	0.82	0.87	0.53
KT	0.70	0.82	0.53
KR	0.87	0.90	0.56
KU	0.72	0.83	0.56
IRC	0.91	0.93	0.74

) showed Cronbach’s alpha values ranging from 0.7 to 0.91, confirming the reliability of the KM and IRC factors extracted from the EFA. In addition, the composite reliability (CR) and average variance extracted (AVE) were calculated to confirm the groupings of KM and IRC factors. The results in Table 4 confirm that all factors have CR values over 0.7 and AVE values over 0.5, thus confirming the validity of the extracted factors (Raykov, 1997). The Heterotrait–Monotrait (HTMT) ratio was also used to verify that the extracted factors are distinct. The results in

Table 6. HTMT ratios.

Factor	KS	KC	KT	KR	KU	IRC
KS	-
KC	0.44	-
KT	0.40	0.40	-
KR	0.46	0.46	0.43	-
KU	0.42	0.42	0.34	0.44	-
IRC	0.54	0.54	0.50	0.57	0.52	-

confirm the groupings with HTMT values less than 0.9 (Rönkkö & Cho, 2022).

4.3. Structural Equation Modeling Results

4.3.1. Measurement Model Results

The confirmed KM and IRC factors were explored with the measurement model test to examine their correlations and interrelationships using two-headed arrows (see Figure 5). Fifteen hypotheses were set as follows.

• H1 (KS ↔ KC): Knowledge sharing has a relationship with knowledge creation.
• H2 (KS ↔ KT): Knowledge sharing has a relationship with knowledge storage.
• H3 (KS ↔ KR): Knowledge sharing has a relationship with knowledge retention.
• H4 (KS ↔ KU): Knowledge sharing has a relationship with knowledge utilization.
• H5 (KC ↔ KT): Knowledge creation has a relationship with knowledge storage.
• H6 (KC ↔ KR): Knowledge creation has a relationship with knowledge retention.
• H7 (KC ↔ KU): Knowledge creation has a relationship with knowledge utilization.
• H8 (KT ↔ KR): Knowledge storage has a relationship with knowledge retention.
• H9 (KT ↔ KU): Knowledge storage has a relationship with knowledge utilization.
• H10 (KR ↔ KU): Knowledge retention has a relationship with knowledge utilization.
• H11 (KS ↔ IRC): Knowledge sharing has a relationship with international research collaboration.
• H12 (KC ↔ IRC): Knowledge creation has a relationship with international research collaboration.
• H13 (KT ↔ IRC): Knowledge storage has a relationship with international research collaboration.
• H14 (KR ↔ IRC): Knowledge retention has a relationship with international research collaboration.
• H15 (KU ↔ IRC): Knowledge utilization has a relationship with international research collaboration.

The model was run, and the model fit was assessed using fit indices (see

Table 7. Fit index results.

Fit Index	Acceptable Value	Baseline Model	Best-Fit Measurement	Best-Fit Structural
CMIN/DF	≤3.0	3.16	2.50	2.23
CFI	≥0.90	0.85	0.90	0.92
RMSEA	≤0.08	0.08	0.07	0.06

). The results revealed the requirement to adjust the model to improve the model fit. Model adjustment was performed using the MI values suggested in the model output. They suggested several correlations to be added to improve the model fit. Over 50 MI values were suggested, and correlations were added, one at a time, until all fit indices were in acceptable ranges to avoid overfitting. A total of six highest correlations (KR5 ↔ KU3, KS1 ↔ KU2, KC1 ↔ KU2, KT4 ↔ KU4, KC1 ↔ KS1, and KR3 ↔ KR1) were added to enhance the fitness of the model (see

Table 8. MI values to improve the model fit of the measurement model.

Suggested Correlation	MI Value
KR5 ↔ KU3	48.80
KS1 ↔ KU2	36.56
KC1 ↔ KU2	32.90
KT4 ↔ KU4	31.08
KC1 ↔ KS1	29.32
KR3 ↔ KR1	23.30

). These correlations were confirmed by various studies. For example, Amoako-Gyampah and Meredith (2018) stated that senior management’s commitment supports leadership by providing essential resources, establishing clear goals, and facilitating open communication for IRC (i.e., KR5 ↔ KU3). Danko and Crhová (2025) mentioned that knowledge sharing is essential, specifically at the managerial level, to enhance information flow through horizontal and vertical perspectives, foster innovativeness, and boost organizations’ strengths and competitiveness (i.e., KS1 ↔ KU2). Cristache et al. (2025) added that organizations can use common knowledge exchange tools to nurture new ideas, insight information, and understandings to form collective expertise to enhance competitiveness (i.e., KC1 ↔ KU2). Szalma et al. (2010) mentioned that KM’s supportive policies should be integrated into organizational plans to improve transnational research and practical applications (i.e., KT4 ↔ KU4). Sanderson et al. (2022) stated that sharing interpersonal skills strengthens teamwork and team performance (i.e., KC1 ↔ KS1). Chatzifoti et al. (2025) stated that adequate funding and incentive support is essential for KM implementation to motivate and engage employees in KM activities (i.e., KR3 ↔ KR1).

After the adjustment, the measurement model was reanalyzed, and the best-fit measurement model was achieved (see Figure 5) with the best-fit indices in Table 7. The results in Figure 5 support 15 hypotheses, confirming interrelationships between KM and IRC factors (see

Table 9. Results of measurement model test on hypotheses.

Hypothesis	Detail	Result	Correlation Coefficient
H1	KS ↔ KC	Supported	0.85
H2	KS ↔ KT	Supported	0.94
H3	KS ↔ KR	Supported	0.87
H4	KS ↔ KU	Supported	0.84
H5	KC ↔ KT	Supported	0.93
H6	KC ↔ KR	Supported	0.92
H7	KC ↔ KU	Supported	0.97
H8	KT ↔ KR	Supported	0.95
H9	KT ↔ KU	Supported	0.93
H10	KR ↔ KU	Supported	0.98
H11	KS ↔ IRC	Supported	0.65
H12	KC ↔ IRC	Supported	0.77
H13	KT ↔ IRC	Supported	0.75
H14	KR ↔ IRC	Supported	0.78
H15	KU ↔ IRC	Supported	0.77

). The best-fit measurement model revealed strong relationships between five KM and IRC factors, specifically KR ↔ KU, KC ↔ KU, KR ↔ KT, and KS ↔ KT, with correlation coefficients of 0.98, 0.97, 0.95, and 0.94, respectively. Farand et al. (2025) stated that using concept inventories in engineering helps students to retain key ideas and utilize them in their careers (i.e., KR ↔ KU). In the IT sector, AI-driven tools help employees to retain and use skills more effectively (Singhal & Rastogi, 2025). Finseth et al. (2025) mentioned that virtual reality helps people to create and use new knowledge from their experiences to solve real-life situations (i.e., KC ↔ KU). Dumitra et al. (2025) stated that collaboration creates knowledge and becomes valuable when applied in problem-solving. Igbinovia and Ikenwe (2018) emphasized that repositories, databases, and content management systems are crucial for effective knowledge sharing (i.e., KS ↔ KT). Omigie et al. (2019) agreed that well-organized knowledge supports continuous dissemination between educators and students.

The best-fit measurement model was further tested with the structural model to identify directions of relationships between KM and IRC factors.

4.3.2. Structural Model Results

In the structural model test, one-headed arrows replaced two-headed arrows previously used in the measurement model. The literature review was performed to hypothesize directions of relationships between them (i.e., 15 directions of relationship). The search was performed in the Scopus database by inputting keywords related to each pair of relationships, such as KU, KR, influence, relationship, and affect, to extract journals related to the KU and KR factors and their relationships. The extracted journals were screened to conclude the explicit direction of the relationship between these two factors (i.e., KU → KR or KR → KU). The literature review results confirmed 11 definite and 4 indefinite directions of relationships between KM and IRC factors, as they were highly depicted in the literature. In contrast, four directions of relationships (i.e., KS ↔ KT, KC ↔ KT, KC ↔ KU, and KT ↔ KU) were found to be possible both ways in the literature (e.g., KS → KT and KT → KS). For example, Moradi et al. (2020) highlighted that shared knowledge should be stored appropriately for future use (i.e., KS → KT). On the contrary, Olivera (2000) stated that stored knowledge may be restrained for future sharing to create feedback systems that enhance collective intelligence (i.e., KT → KS). Pei et al. (2025) mentioned the significance of a collaborative environment where new knowledge leads to intelligent classification and semantic tagging that are upgraded in the storage systems (KC → KT). On the other hand, Zaim et al. (2019) stated that KT supports the readiness of new knowledge creation by providing accessible channels, such as databases and intranets (i.e., KT → KC).

Based on the above, 19 hypotheses were set. The details are as follows.

• H16 (KS → KC): Knowledge sharing influences knowledge creation. KS plays a pivotal role in facilitating KC across diverse organizational contexts. Chan et al. (2024) concluded that KS stimulates innovation and creates new knowledge among Southern Anhui entrepreneurs. Simukonda (2024) added that KS and resource pooling within clusters significantly enhance SMEs’ capacity for KC and innovation.
• H17 (KS → KR): Knowledge sharing influences knowledge retention through structured communication and documentation practices. Shared knowledge becomes embedded within organizational memory systems, ensuring its persistence beyond individual contributors (Sumbal et al., 2017).
• H18 (KS → KU): Knowledge sharing influences knowledge utilization when researchers openly exchange knowledge. It facilitates the adaptation and integration of knowledge into future studies, enhancing the practical utilities of scientific findings (Rahimli, 2012). KS mechanisms, such as workshops, shared databases, and collaborative publications, help to translate raw data and theoretical insights into actionable knowledge (Donate & Sánchez de Pablo, 2015).
• H19 (KS → IRC): Knowledge sharing influences international research collaboration because it enables ideas and expertise across borders and enhances trust, communication, and mutual understanding among international partners, creating a sustainable IRC (S. Wang & Raymond, 2010).
• H20 (KC → KR): Knowledge creation enhances knowledge retention by systematically generating new insights for the institutional memory that spread through collaborative processes, such as co-authorships, research workshops, and shared experimentation, reducing knowledge loss risks (Awan & Khalid, 2015).
• H21 (KC → IRC): Knowledge creation supports international research collaboration because it encourages researchers from diverse backgrounds to engage in collaborative ventures (Gui et al., 2019).
• H22 (KR → KU): Knowledge retention serves as a foundation upon which new research is built, streamlining processes and supporting knowledge utilization across a collaborative environment (Sumbal et al., 2017).
• H23 (KR → KT): Effective knowledge retention strategies influence robust knowledge storage infrastructure to prevent knowledge loss. It drives institutions to develop formal storage mechanisms, such as digital libraries, databases, and institutional repositories, for research continuity (Durst & Zieba, 2019).
• H24 (KR → IRC): Knowledge retention supports international research collaboration by preserving strategic, operational, and cultural insights for future collaboration. The prior collaboration protocols, partner preferences, and legal frameworks can be reused to streamline new collaborations and avoid pitfalls (Lucas, 2006).
• H25 (KT → IRC): Knowledge storage significantly enhances international research collaboration by providing centralized knowledge access, retrieval, and codification platforms. It allows research teams to navigate time zones and language barriers more effectively and support intellectual properties (S. Wang & Raymond, 2010; Iatridis & Schroeder, 2016).
• H26 (KU → IRC): Knowledge utilization supports international research collaboration activities when knowledge is embedded into career progression systems, such as promotion criteria, international mobility programs, and performance metrics. With this practice, researchers are incentivized to utilize collaborative knowledge (Jonkers & Tijssen, 2008).
• H27 (KS → KT): Repeated knowledge sharing helps to validate and refine knowledge before it is stored, thus improving quality. Knowledge sharing across teams helps organizations to build collective memory, so knowledge is not lost when staff leave (Alavi & Leidner, 2001; Antunes & Pinheiro, 2020).
• H28 (KT → KS): In R&D operations, knowledge storage helps to store new knowledge, making it available for further knowledge sharing through publications and staff mentoring (Xiao et al., 2021).
• H29 (KC → KT): Knowledge creation can support knowledge storage by providing insight and criteria for new knowledge to be properly stored and reducing loss when projects end (Versiani et al., 2024).
• H30 (KT → KC): Cheng et al. (2023) stated that knowledge storage supports accessible, well-structured formats of knowledge and assists research teams to effectively create new findings.
• H31 (KC → KU): Jacobi et al. (2022) commented that transdisciplinary knowledge co-creation increases knowledge utilization, leading to sustainable transformation.
• H32 (KU → KC): Knowledge utilization in university–industry collaboration in lower-income countries accumulates knowledge creation and new findings, such as using practice-based knowledge and lessons learned from industries to generate new research topics (García-Hurtado et al., 2022).
• H33 (KT → KU): Dei et al. (2024) mentioned that effective knowledge storage is needed for utilizing knowledge in learning organizations. If storage is not usability, KM practices cannot be fully utilized.
• H34 (KU → KT): Collaboration between universities and industries leads to the utilization of existing knowledge to create new knowledge through joint projects, which must be systematically stored for future use (Abu Sa’a & Yström, 2025).

Structural models are tested with different patterns of directions of relationships. Each structural model is analyzed and adjusted using MI values to achieve best-fit indices. Path coefficients less than 0.1 are removed as they represent very weak relationships (Ringle et al., 2020). The best-fit indices of the structural models are compared to select the best-fit structural model that represents the directions of relationships between KM and IRC factors (see Table 7 and Figure 6). It is noted that the best-fit structural model displays 13 directions of relationships (see

Table 10. Hypothesis results of structural model test.

Hypothesis	Detail	Result	Path Coefficient (Standardized)
H16	KS → KC	Supported	0.88
H17	KS → KR	Supported	0.28
H18	KS → KU	Not supported	-
H19	KS → IRC	Supported	−0.59
H20	KC → KR	Supported	0.67
H21	KC → IRC	Supported	0.11
H22	KR → KU	Supported	0.13
H23	KR → KT	Supported	0.53
H24	KR → IRC	Not supported	-
H25	KT → IRC	Supported	−0.39
H26	KU → IRC	Supported	0.61
H27	KS → KT	Supported	0.47
H28	KT → KS	Not supported	-
H29	KC → KT	Not supported	-
H30	KT → KC	Not supported	-
H31	KC → KU	Supported	0.53
H32	KU → KC	Not supported	-
H33	KT → KU	Supported	0.93
H34	KU → KT	Not supported	-

), as 6 directions are removed.

The best-fit structural model (or the final model of KM and IRC factors) reveals two strong path coefficients (i.e., ≥0.7) between the KT and KU (KT → KU) and KS and KC (KS → KC) factors. These strong relationships are confirmed by the literature. For example, the SpExoDisks project developed an integrated database and web portal to consolidate the internal workflow and input from external researchers, centralizing real-time data. The database improved collaboration and analytical accuracy, increasing transparency in knowledge utilization for current and future stakeholders (i.e., KT → KU). Wheeler et al. (2024) and Kilar et al. (2025) agreed that archival sources support interdisciplinary learning, research continuity, and knowledge utilization among researchers. Corbí et al. (2025) mentioned that sharing knowledge between researchers from different backgrounds and disciplines creates new ideas (i.e., KS → KC). Byberg and Crimi (2025) demonstrated that team-based learning is essential for integrating emerging technologies like AI and machine learning and creating organizational learning. Kamasak and Bulutlar (2010) added that knowledge sharing is a strategy to convert tacit knowledge into innovations.

Moderate relationships are found in KC → KR, KU → IRC, KS → IRC, KC → KU, and KR → KT, with path coefficients ranging from 0.53 to 0.67. Effective KC processes, e.g., strong teamwork, constructive feedback systems, and shared risks, enable KR mechanisms, such as incentives, infrastructure development, and shared responsibilities (i.e., KC → KR). Ali et al. (2021) mentioned that collaborative team dynamics and trust-based communication contribute significantly to employees’ willingness to document and preserve team knowledge. Kohnová et al. (2019) stated that high-performing teams foster a continuous learning culture that accelerates the utilization of knowledge (i.e., KC → KU). Trust is crucial for effective knowledge utilization between team members as it raises collaboration and drives performance (Reagans et al., 2016; Mas-Machuca & Costa, 2012). Kim (2006) stated that researchers are more likely to participate in IRC when knowledge is effectively utilized, resulting in increased co-authored publications and project mobility (i.e., KU → IRC).

Interestingly, the KS factor has a negative moderate relationship with the IRC factor (path coefficient = −0.59). The experts involved in the interviews agreed that, in Thailand, oversharing research information and insights between researchers from different divisions and organizations may sabotage collaborated activities like research proposal submissions for international grants. For instance, junior researchers who seek to advance their careers may unintentionally release crucial information about international project funding to senior researchers, who may have adjusted proposals to suit the required targets. This may result in junior researchers losing their opportunities to receive funding. This is referred to the high-power distance dimension in the Hofstede’s Cultural Dimensions Theory, where junior researchers are reluctant to express opinions or argue with senior researchers due to a strong seniority culture in Thailand (Żemojtel-Piotrowska & Piotrowski, 2023). This is consistent with Castaneda and Ramírez (2021), who commented that a strong power distance discourages open knowledge exchange, as data flow becomes centralized. This is specifically in countries with vertical collectivism, e.g., China and Russia, resulting in slow and limited research collaboration. Zhang et al. (2026) added a negative effect of the power distance dimension on knowledge sharing, pinpointing that a few inventors who possess the concentrated innovative information in the host country can hinder the cross-border knowledge sharing to the local creation. Another case study is releasing research prototypes prior to attaining IPs. This results in losing rights of ownership and commercialization to large companies. This could be explained by the short-term orientation dimension in Hofstede’s Cultural Dimensions Theory, where researchers fear being humiliated from information leaking and decide not to fight for ownership (Żemojtel-Piotrowska & Piotrowski, 2023; Memon & Shahid, 2024). Hawamdeh and Qatamin (2021) agreed that short-term orientation has a negative effect on the intention to share knowledge in Jordanian higher-education institutions. Jin (2012) commented that long-term orientation is one of the five dimensions of Hofstede’s Cultural Dimensions Theory that has a positive effect on staff’s willingness to share knowledge.

The negative relationship between KS and IRC is also supported by several studies. Tang and Marinova (2020) commented that excessive knowledge sharing in new product development leads to diminishing creativity and performance due to information overload, resulting from cognitive stimulation and interference effects. Villena et al. (2011) added that an extensive information exchange with partners may result in dependency and inefficiency in IRC. Thomas (2025) commented that while knowledge sharing can foster collaboration and improve effectiveness, it may trigger counterproductive behaviors like knowledge hiding, hoarding, and sabotage. This is specifically in power asymmetries, competition, and unclear incentives systems like NSTDA. It is explained by the individualism dimension in Hofstede’s Cultural Dimensions Theory, where researchers focus mainly on themselves to survive in a highly competitive environment (Żemojtel-Piotrowska & Piotrowski, 2023). Ritala et al. (2018) added that sharing knowledge among employees is associated with risks like knowledge leaking and innovation hinders, specifically when it is not handled properly. The normalization of knowledge sharing can provoke knowledge hoarding behaviors from perceived unfairness (Evans et al., 2015).

Table 11. Total effects of five KM factors and their effects on the IRC factor.

Direct Relationship	Direct Effect	Indirect Relationship	Indirect Effect	Total Effect
KS → KC	0.88	-	-	0.88
KS → KR	0.28	KS → KC → KR	0.59	0.87
		(0.88 × 0.67)
KS → KT	0.47	KS → KC → KR → KT (0.88 × 0.67 × 0.53)	0.31	0.93
		KS → KR → KT (0.28 × 0.53)	0.15
KS → KU	-	KS → KC → KU (0.88 × 0.53)	0.47	1.05
		KS → KR → KT → KU (0.28 × 0.53 × 0.93)	0.14
		KS → KT → KU (0.47 × 0.93)	0.44
KS → IRC	−0.59	KS → KC → KU → IRC (0.88 × 0.53 × 0.61)	0.28	0.34
		KS → KC → IRC (0.88 × 0.11)	0.10
		KS → KC → KR → IRC (0.88 × 0.67 × 0.45)	0.27
		KS → KC → KR → KT → KU → IRC (0.88 × 0.67 × 0.53 × 0.93 × 0.61)	0.18
		KS → KC → KR → KT → IRC (0.88 × 0.67 × 0.53 × −0.39)	−0.12
		KS → KR → IRC (0.28 × 0.45)	0.13
		KS → KT → KU → IRC (0.47 × 0.93 × 0.61)	0.27
		KS → KT → IRC (0.47 × −0.39)	−0.18
KC → KR	0.67	-	-	0.67
KC → KT	-	KC → KR → KT (0.67 × 0.53)	0.36	0.36
KC → KU	0.53	KC → KR → KT → KU (0.67 × 0.53 × 0.93)	0.33	0.86
KC → IRC	0.11	KC → KR → KT → KU → IRC (0.67 × 0.53 × 0.93 × 0.61)	0.20	0.47
		KC → KR → KT → IRC (0.67 × 0.53 × −0.39)	−0.14
		KC → KR → IRC (0.67 × 0.45)	0.30
KR → KU	0.13	KR → KT → KU (0.53 × 0.93)	0.49	0.62
KR → KT	0.53	-	-	0.53
KR → IRC	0.45	KR → KT → IRC (0.53 × −0.39)	−0.21	0.54
		KR → KT → KU →IRC (0.53 × 0.93 × 0.61)	0.30
KT → KU	0.93	-	-	0.93
KT → IRC	−0.39	KT → KU →IRC (0.93 × 0.61)	0.57	0.18
KU → IRC	0.61	-	-	0.61

shows direct and indirect effects among the five KM factors and IRC. Interestingly, KS → KU and KC → KT factors were found to have no direct, but strong indirect, relationships through other KM factors. Specifically, the KS factor highly influenced the KU factor through the KC and KT factors. Sharing experiences among team members enhances the learning capacity of the team, creating knowledge flow, retaining its loss, and supporting the utilization of the knowledge cycle to solve problems (i.e., KS → KC → KU) (Y. Wang et al., 2015; Ahn et al., 2019). When members from different countries share their specific expertise and mentor junior members, meaningful information is created and stored as mutual knowledge to achieve the missions and goals of IRC (i.e., KS → KT → KU) (Larkan et al., 2016). Nonaka and von Krogh (2016) stated that knowledge is created through social interaction and becomes impactful when retained. Durst and Wilhelm (2012) commented that failure to store knowledge systematically leads to severe knowledge attrition, because critical knowledge remains solely in individuals’ minds, rather than being codified as organizational knowledge (i.e., KC → KR → KT).

5. Discussion

The best-fit structural model of KM and IRC factors indicates that the KU factor strongly influences IRC and could be encouraged to enhance IRC. Though the KC, KR, and KT factors have direct relationships with IRC, their effects are weak and moderate. In contrast, their relationships with the KU factor are strong; therefore, they should be planned to support KU’s implementation. The KS factor, having strong direct and indirect relationships with other KM and IRC factors, could be planned to achieve a sustainable IRC.

5.1. Strategic Plans Related to the Knowledge Utilization Factor

IRC may be enhanced through action plans related to the KU factor by utilizing existing resources. Examples of activities are as follows.

• Assigning personnel with IRC experience as mentors to facilitate IRC activities (i.e., KU3 item with the highest loading value of 0.71, see Figure 5). Researchers who have been awarded with IRC grants or who have participated in IRC projects may be selected as mentors for new recruits and junior researchers to provide comments on grant searching, proposal development, and how to get international partners. This will encourage team members to strategically navigate through the intercultural environment and mingle ideas with regional and global issues. The sessions should be scheduled regularly to capture new ideas and brainstorm for better IRC activities (Knight, 2025; Villarente & Durante, 2025). This mentoring program may lead to better IRC in terms of publications, IPs, and research proposals (i.e., PB, IP, and CD items under the IRC factor).
• Setting up a CoP (i.e., KU4 item) to generate insights and deep expertise of participants, resulting in crucial material to create action plans. The sessions should be scheduled regularly to capture new ideas and brainstorm for better IRC activities (Pyrko et al., 2017). For example, NSTDA deployed a CoP as an objective-oriented space to exchange specific information, e.g., international grants, international proposal submission, and international project management. Grants like Horizon Europe, the EU’s key funding program for research and innovation, which has complex regulations and high fail rates, are explained in the CoP to share tips, tricks, and hints to raise possible success rates (i.e., CD and MB items under the IRC factor). This CoP helps to elaborate the complexities of the international collaborative grant environment that may be driven by various networks from different cultures. Lessons learnt from successful IRC projects (e.g., matching the right partners and adjusting to different work cultures) may be discussed with those achieved from unsuccessful IRC projects (which may come from ineffective communication between different languages and complexities of grant protocols) to minimize failure points and increase success rates.
• Integrating IRC activities as a complement to the personnel’s career path (i.e., KU1 item) (Nguyen & Tran, 2025). For example, one IRC project proposal submission may be counted as twenty points compared with ten points of a domestic project proposal submission, and one co-authorship publication with at least two international partners may earn double points compared with a single-authorship publication. The points received may be included in the annual performance report and considered for promotion rankings. This will encourage researchers and support teams to search for more IRC projects.

KU-related activities should be planned carefully when facing political instability and economic crises when budget is highly cut-off. Research institutes may prioritize their expertise and initiate research topics that attract potential international partners to attain the government’s involvement and fully utilize a limited budget (Cristache et al., 2025; Knight, 2025; Villarente & Durante, 2025). The CoP members should come from various areas to express meaningful ideas in project meetings. Thailand, for instance, has know-how in biotechnology, agriculture, cultural tourism, medical, and healthcare. These topics may be proposed in IRC project initiation, as they have always been listed in the national development plans regardless of political changes and economic swing.

5.2. Strategic Plans Related to the Knowledge Creation, Knowledge Retention, and Knowledge Storage Factors

Strategies related to the KC, KR, and KT factors may be set as follows.

• Launching IRC policy as part of organizational policies (i.e., KT4 item with the highest factor loading of 0.78). The policy must be co-developed and regularly updated with international partners to reflect evolving ethical, technological, and geopolitical realities and access new research opportunities and grants (i.e., FI item under the IRC factor) (Ford, 2004). The NSTDA has initiated international research committees as a part of the IRC policy to supervise and review IRC activities, thus lowering potential failures.
• Engaging senior management (i.e., research department director and above) in IRC activities to ensure effective decision-making that aligns with organizational policy and reduces possible conflicts (i.e., KR5 item with factor loading of 0.72) (Van Rensburg et al., 2014). Senior management typically has extensive experience and a network that can guide directions for researchers in IRC projects.
• Setting up the IRC task force, including researchers and supporting teams, to enhance and explore IRC projects and activities (i.e., KC4 and KC3 items, with KC4 being the item with the highest loading of 0.79). The task force members must have specific job functions related to IRC activities. They must work constantly with their international counterparts to share useful information and create common knowledge about regulations, cross-cultural customs, and research trends that align with collaborative goals and expectations (i.e., MB and PB items under the IRC factor) (Al-Husseini & Elbeltagi, 2018). Knowledge learnt must be stored in tangible forms (e.g., written documents, digital files, audio, and video clips) for future reference (Coakes, 2006).
• Updating statuses of IRC activities to teams, such as the amount of grants, progress of projects, allowances for visiting and working abroad, and risks of legal and financial clauses in collaboration agreements (i.e., KR6, KR4, and KR7 items) (Setyanto et al., 2025). A performance management tool, i.e., Microsoft Project and a balanced scorecard, may be applied to measure and follow-up IRC activities (Kaplan et al., 2010). AI-driven collaboration platforms, such as Microsoft Teams, Google Meet, Zoom Workplace, and Slack, may be considered as communication channels for a quick discussion and problem solving (Reclaim, 2025).
• Improving organizational infrastructures, such as laboratories, offices, supplies, and databases (i.e., KR4 and KT3 items). Supplies should be freely accessible to drive engagement. Databases should be regularly updated, and confidential information, such as personal information, should be securely stored. The databases should have specific functions to track the progress of IRC activities, measure outputs, and record personnel commitment for incentives and annual promotions (Singh & Pradhan, 2024). Some AI tools, like cloud-based systems (e.g., Google Cloud, Microsoft Azure, and IBM Cloud), Notion AI, Dropbox, and Google Workspace, may be applied to enhance accessibility to previous data (onsite and offsites), minimize information technology costs, and assist in real-time updating (Reclaim, 2025). Having an effective organizational infrastructure may lead to a systematic documenting system and future collaboration (i.e., CD item under the IRC factor).
• Implementing the after-action review to collect and analyze feedback from successful and unsuccessful IRC activities to decrease failures and increase the success rate of future IRC (i.e., KC2 item) (Shuffler et al., 2018). AAR is a human-resource tool used for performance improvement by capturing lessons learnt and tacit knowledge from staff. It involves four processes: plan, outcome, root cause, and improvement action identification (UNICEF, 2015). This feedback analysis helps to reduce inter-institutional misunderstandings, enhance transparency, and minimize workflow bottlenecks (Kezar, 2006; Gibson et al., 2019).

5.3. Strategic Plans Related to the Knowledge Sharing Factor

To achieve sustainable IRC, knowledge must be part of organizational culture, where team members can share, create, retain, and utilize relatable knowledge to solve problems. Tacit and explicit knowledge should be systematically shared through networks, e.g., the IRC department, MOUs, MOAs, and project management directories. These networks should be enlarged to cover local and international partners. Strategies related to the KS factor may be set as follows.

• Transforming a non-learning organization culture to a learning organization culture (i.e., KS3 item with the highest factor loading of 0.84). This process could be achieved using regular communication about the importance of the learning organization values (Lindqvist et al., 2023). Though cultural transformation takes time and effort, it will strengthen the organizational knowledge-sharing system (Mohiuddin et al., 2022; Lee & Han, 2024). For example, the NSTDA has established a co-working space, encouraging researchers from different departments to share knowledge and expertise in a friendly environment, enhancing a learning culture in the organization.
• Organizing consortia, mobility programs, and digital communities of practices through existing regional and international bodies, such as the Association of Southeast Asian Nations (ASEAN), Asia-Pacific Economic Cooperation (APEC), Asia-Pacific Tele-community (APT) (i.e., KS2 item). These may assist in locating new potential collaborators and updating global trends (i.e., MB, PB, and FI items under the IRC factor) (Puljak & Vari, 2014).
• Recruiting new talent with backgrounds in international relations, international business, and international research projects to ensure effective communication and effectiveness of IRC (i.e., KS1 item) (Reis da Silva et al., 2024).

6. Conclusions

IRC is crucial for increasing technological advancement, improving quality of life, strengthening national competitiveness, and tackling global problems. Research organizations must manage IRC effectively. This study examines relationships between KM implementation and IRC in Thai R&D institutions. Five key KM factors, namely KS, KC, KR, KT, and KU, were extracted from the literature and used in the questionnaire survey for data collection. The survey was conducted with research institutions, universities, and research partners in Thailand. The data collected were screened and analyzed with the EFA to confirm key KM constructs and their associated items. The results confirmed 5 KM factors with 24 associated items, whereas the IRC factor was associated with five items. The confirmed factors were further examined with the SEM to identify the interrelationships between KM and IRC factors. The results showed strong relationships, directly and indirectly, among the five KM factors, specifically, between the KT and KU, KS and KC, and KC and KR factors. Strong indirect relationships were found between the KS and KU, KC and KU, KT and KU, and KS and KT factors. In addition, the five KM factors were confirmed to influence IRC enhancement, specifically, the KU, KC, and KR factors. This proves the importance of KM implementation in raising IRC performance and achieving sustainable development.

The study results were used to develop strategic plans for R&D institutions. Activities related to the KU factor should be promoted to tackle ad hoc or urgent IRC activities using current resources. It is recommended that administrators set the highest priorities in setting up mentoring and CoP teams to brainstorm, mentor, and guide directions to award IRC projects using the existing know-how of senior researchers. Activities like organizing site visits for international researchers may be initiated to exchange ideas and mentor new researchers. Administrators should also develop indicators to measure IRC achievement and link them to career growth. Once these strategic plans are implemented, activities related to the KC, KR, and KT factors may be implemented to support the KU factor and enhance IRC. Administrators should set up an IRC team, compiling members and managers from various departments. The IRC team is responsible for boosting IRC activities, exploring new opportunities, analyzing positive and negative feedback from previous projects, and making IRC-related decisions. To achieve sustainable IRC, administrators should consider activities related to the KS factor, e.g., expanding potential collaborators, recruiting new talent, and allocating adequate budget for IRC-related investment.

The study results support research objectives as follows.

• Five key KM factors (i.e., KS, KC, KT, KR, and KU) are crucial for IRC enhancement;
• There are direct and indirect relationships between KM and IRC factors;
• KM strategic plans are proposed to achieve sustainable IRC.

The study results contribute to the body of knowledge. They confirm using KM to enhance IRC and provide new insights.

• This study utilizes EFA and SEM techniques to systematically capture interrelationships between KM and IRC factors, enhancing understanding and assisting in developing implementation plans;
• The study results guide KM implementation in assisting organizations in planning their resources to achieve IRC targets.

The study results may be applied and adjusted to be used in other research institutions in Thailand and neighboring countries to foster an effective IRC environment. There are some limitations in this study. Data were collected from research institutions, universities, and partners in Thailand, and international partners were omitted. Including international partners may yield better results. Private research organizations may be included to cover different working cultures. The distribution of respondents is bottom-heavy due to a common organizational imbalance between executives and operations, and conflicts of schedule of top managers during the data-collection process. Different channels for data collection may be used to retrieve insights from the management level. Critical external factors like political instability and economic crisis were not included in this study. In many developing countries like Thailand, the government is frequently reformed, resulting in changes in academic-related policies and strategies. Economic crisis from disputes between world’s leading countries may affect IRC plans and implementation. Geopolitical policy and regional disputes may sabotage new partner formation and suspend ongoing IRC projects. Future studies are required to examine how these external environmental turbulence acts as a moderator between KM processes and IRC success. It is also recommended that comparative analysis be performed between local and international public and private organizations to enhance sustainable IRC strategies. Complex relationships between KM and IRC factors may be further explored using simulation tools, such as system dynamics modeling and optimization models, to effectively plan for long-term IRC improvement.