Introduction

In recent decades, and particularly over the last few years, the environment in which organizations operate is increasingly changing, and the way each entity deals with this reality depends on how knowledge is generated, transferred, and managed internally and externally (García-Sanchez et al., 2019). Universities are not alien to this situation, as their role as an actor in society has changed over the last decade, involving them in undertaking new activities that have reshaped their fundamental purpose, whereby they help solve problems in their community and environment. Thus, universities have gone from being isolated knowledge-production institutions to becoming know-how and know-why institutions (Pan et al., 2021; Zhang et al., 2020).

In this sense, this change in university systems, with an emphasis on research-oriented policies and a quality profile expressed through academic production and indexed journals, has led to the growing importance of scientific research and high-impact research (Mitchell et al., 2022), particularly in developing countries. However, high-quality journals are concentrated in a few countries. In this context, research groups can play a fundamental role by converting the findings of their scientific research into results that can be transferred to society, thus contributing to a country’s economic progress (Martín-Alcázar et al., 2019). For these reasons, it is of great interest to investigate the determinants of results in research groups, especially in developing countries.

Social capital theory states that collaboration between agents has a positive impact on performance (Augusto Felício et al., 2014; Petrou & Daskalopoulou, 2015; Zhang & Shih, 2022). According to this theory, collaboration between researchers in research groups is an essential factor for the results of such groups, because it allows academic researchers to learn from each other and to reach new scientific findings. Therefore, the analysis of social capital as a determinant of knowledge sharing and scientific results is crucial (García-Sanchez et al., 2019; Pan et al., 2021; Lin & Huang, 2023). In this sense, the management and organizational literature has demonstrated the effects of social capital on diverse organizational performance measures (Dimitriadis, 2021; Martínez-Pérez & Beauchesne, 2018; Ruiz-Ortega et al., 2016). However, although theoretically accepted, few studies have provided empirical evidence of the effects of social capital on research outcomes (Gonzalez-Brambila, 2014).

For this reason, scientific collaboration has now become an important research topic (Salimi et al., 2022). Sonnenwald (2007) mentions that scientific collaboration is related to the interaction between two or more scientists that promotes the exchange of knowledge and the fulfillment of previously agreed objectives. In this sense, collaboration contributes to the generation of social capital reflected in the increase in production of better quality, visibility, and impact. At the same time, these relationships allow for access to equipment, infrastructure, and the generation of important skills that strengthen research processes (Andrade et al., 2009; Beaver & Rosen, 1978; Martín-Alcázar et al., 2020a). Academics can develop their research with closed groups within their university or may open up groups to various collaborations, either inside or outside the country (Kontinen & Nguyahambi, 2020). Therefore, collaboration is particularly important because the processes of generating high-quality knowledge and products greatly depend on synergistic interactions between actors in the system (Acosta et al., 2020).

The advantages of collaboration justify that the members of the research groups work together to approach and develop activities and share resources, in order to promote the flow of knowledge among their members (García-Sánchez et al., 2019; Martín-Alcázar et al., 2020b). These relationships can be generated internally through cohesive groups (bonding capital) or through cooperation with external networks (bridging capital).

Physical proximity is an essential aspect for cohesive social relationships since strong ties are an important factor in the transmission of tacit knowledge, which requires frequent contacts and face-to-face relationships (Kale et al., 2000). When research team members possess bonding capital, in terms of strong ties, trust, cognitive proximity, and centrality, research productivity and impact are enhanced (García-Sanchez et al., 2019). However, in the field of management, perverse effects of excessive bonding capital have been detected, such as internal blocking, groupthink, network inertia, or information redundancy (Ruiz-Ortega et al., 2022).

With regard to external and cooperative networks, the access to novel and diverse knowledge attributed to bridging capital allows new ideas to be identified, helping detect opportunities for research and integrating different research approaches to complement the advantages of bonding capital and mitigate its disadvantages in order to achieve high-impact research results (Gonzalez-Brambila, 2014; Pan et al., 2021).

Despite the importance of both types of collaboration (Lin & Huang, 2023; Salimi et al., 2022; Zhang & Shih, 2022)—bonding capital and bridging capital—we find a gap in the literature in relation to how the link between these phenomena influences the results of research groups. Thus, we pose the following research question: How does the relationship between bonding and bridging social capital influence the scientific results of research groups?

Furthermore, drawing on the literature on social capital, we consider that the relationship between bonding capital and the scientific results of research groups will not be linear, since, from a certain level of bonding capital onwards, negative effects of this type of social capital will emerge (Ruiz-Ortega et al., 2022). Therefore, the main aim of this study is to analyze the curvilinear effect (inverted U-shaped) of bonding capital on the scientific results of research groups and the moderating effect of bridging capital on this relationship. The article has five main contributions: (1) it contributes to the literature on social capital, responding to the demand for a more detailed examination of the balance between the advantages and disadvantages of bonding capital and its interaction with bridging capital. (2) The paper contributes to enriching the study of the most controversial aspects of social capital in the context of research groups in universities. (3) While most studies have addressed these issues in developed countries, this work approaches the dynamics of collaboration and the scientific results of the research groups of higher education institutions in a developing country, namely, Colombia. (4) We focus on engineering research groups, which, in Latin American university systems, tend to be identified with technological transfer rather than scientific output and the impact of bonding and bridging capital is different to other areas of knowledge (Gonzalez-Brambila, 2014). 5) In contrast to studies focused on scientific productivity, this paper centres on the relational factors that determine scientific results represented in the publication of research articles.

Literature Review

Strong interest and a broad debate have recently arisen in the field of organizational studies on the effects on results of homogeneous versus heterogeneous work groups (Lin & Huang, 2023; Salimi et al., 2022), and how these impact the level of performance (Zhang & Shih, 2022). This debate has shifted to the field of scientific research, with several authors stressing how certain characteristics of the configuration of groups—size, cohesion, multidisciplinarity, diversity, common objectives, among others—can affect not only scientific productivity (Bozeman et al., 2013; Martín-Alcázar et al., 2020b) but also research impact (Li et al., 2013; Mitchell et al., 2022).

In the case of universities, the objective of which throughout history has been to create and transfer knowledge through researchers, these relationships can be created in internal networks or outside them, as part of their institutional strategy, taking into account that they have to simultaneously respond to many types of pressures. Among such pressures are the new demands of social and economic actors (Salaran, 2010; Mitchell et al., 2022), who expect universities to propose solutions that respond to changes in the environment. These challenges require universities to expand their knowledge management activities through research (Martín-Alcázar et al., 2019). In this sense, research groups have been called upon to convert their scientific research results into fundamental inputs for the social welfare and technological and economic progress of a country, generating key, valuable knowledge. In this way, they bolster their reputation and academic excellence (Fullwood et al., 2013). In this context, the development of new scientific knowledge as a contribution to the generation of solutions in the real environment has evolved, no longer being based exclusively on individual work but on collaboration and cooperation between researchers from different areas (Gonzalez- Brambila, 2014; Lin & Huang, 2023). To this end, university research groups work, on the one hand, to strengthen the cohesion of the relationships between their members—bonding capital—and, on the other, to establish cooperative relationships with diverse and distant research groups—bridging capital. In this way, they create and develop strong capacities that enable them to obtain high-impact results.

Social Capital Theory

The term social capital was initially used to explain how the satisfaction of individuals’ social needs is influenced by community commitment. The social capital theory defines social capital as the social relations between a group of people that form a social unit, possessing a stable network of familiarity and recognition (Bourdieu, 1986). Nahapiet and Ghoshal (1998:243) define social capital as “the sum of the actual and potential resources embedded within, available through, and derived from the network of relationships possessed by an individual or social unit.” Thus, the capital of many people who benefit from each other accumulates, allowing individuals to achieve different advantages (Coleman, 1988; Lin, 2001).

Burt (2017) analyzes the concept of social capital through his theory of structural holes, arguing that these allow a bridge link to be established between an agent and other agents that participate in different information flows, thanks to the non-redundancy of the links. Thus, the benefits of information can primarily be obtained in large and diverse networks (Pan et al., 2021).

Putnam (2000) also contributes to developing the theory of social capital, differentiating between bonding capital, which refers to the reciprocity and solidarity that occurs between people with similar characteristics, and bridging capital, which refers to the connection produced between diverse networks. Bonding capital has generated greater debate, as it is fundamental in creating a strong sense of belonging to the group, but can also generate redundancy of information and antagonism with members who do not belong to the group. Bridging capital is oriented towards establishing contact with agents in other networks to enhance the transfer of novel information. In this way, bonding capital focuses on internal relationships that facilitate the pooling of resources for capacity building and knowledge transfer. Meanwhile, bridging capital refers to the interaction with external agents for the acquisition of novel information (Zhang & Shih, 2022).

There is a large body of literature in the field of management on the positive and negative implications of each of these two types of social capital, although fewer studies address their joint effect (e.g., Martínez-Pérez & Beauchesne, 2018). The literature on social capital also mentions that the relationships of teams affect the amount and availability of knowledge resources of the members of a group and, therefore, have a direct influence on their performance and can affect the development of new products (Chung & Jackson, 2012; Nahapiet & Ghoshal, 1998). In this sense, social capital plays an important role in the performance of any organization, and in its absence, human resources cannot evolve to efficiently generate results (Fallah Tafti et al., 2020).

Hypotheses

Bonding Capital and Scientific Results of Research Groups

Bonding capital is generated in cohesive, dense, and strongly linked structures derived from nearby networks (Ahuja, 2000; Smith et al., 2005), focusing on the internal relationships of individuals (Adler & Kwon, 2002; Ruiz-Ortega et al., 2022). Bonding capital offers organizations two main advantages: the exchange of high-quality information and tacit knowledge. In this way, it stimulates knowledge transfer and protection in inter-organizational settings (Dyer & Nobeoka, 2000; Kale et al., 2000). Thus, bonding capital builds on the advantages of dense networks (Coleman, 1988), since these cohesive networks have a positive effect on the production of social norms that facilitate trust, generating advantages linked to the knowledge of the different agents to share information (Tsai & Ghoshal, 1998).

Dense, cohesive, or closed networks are those in which the different members are connected (Coleman, 1990), share information from complex relationships that drive innovation, and allow academic researchers to learn from each other, leading to new scientific discoveries. Consequently, the creation of a research group is a major challenge, as new knowledge is only the result of the individual work of researchers and the relationships established between them (Martín-Alcazar et al., 2019; Lin & Huang, 2023).

Several studies have reported that the bonding capital of the authors of research studies, in terms of tie strength, efficiency, centrality, and cognitive proximity, favors the productivity of their papers and their impact (Abbasi et al., 2011; Li et al., 2013). Bonding capital recognizes the links and relationships between individuals or research groups with common characteristics belonging to the same institution. This type of social capital focuses on these links between individuals and, specifically, on the characteristics that provide cohesion and facilitate the pursuit of collective goals (Martínez-Perez & Beauchesne, 2018). Therefore, researchers who build bonding capital from cohesion in the group outperform researchers in a more isolated position, which, ultimately, significantly affects scientific productivity, since it influences the ability of scientists to benefit from the construction of bonding capital by establishing collaborations between members of research groups. In this way, researchers tend to act as guardians of knowledge and resources, generating an advantage by having immediate access to new or improved knowledge (Burt, 2017). The density of networks and the strong ties between members of a research group, associated with bonding capital, generate the trust necessary to access high-quality knowledge and the exchange of tacit knowledge (Dyer & Nobeoka, 2000; García-Sánchez et al., 2019). In this context, valuable knowledge is disseminated among group members in a more accurate and timely manner, cooperative motivation is generated among group members, and opportunistic behavior is avoided. For all these reasons, several authors highlight the positive effects of bonding capital, leading research groups to obtain significant scientific results (Martínez-Alcázar et al., 2020b).

In contrast to the extensive literature on the positive effects of bonding capital, other studies have detected perverse effects of excessive bonding capital, such as internal blocking, groupthink, network inertia, or information redundancy (Ruiz-Ortega et al., 2022; Zach & Hill, 2017). This negative effect of bonding capital on scientific performance has also been identified in the context of research groups (Gonzalez-Brambila, 2014). Because members of the same team, who are connected through a high frequency of interactions, share increasingly homogeneous information, the difficulty of looking for new information increases, enhancing the risk of developing redundant and obsolete processes (Exposito-Langa & Molina-Morales, 2010; Martín-Alcázar et al., 2020a). These perverse effects make it difficult to obtain novel information and generate combinations of knowledge, creativity, and innovation, which are especially necessary to produce research results. Members of research groups that maintain frequent, closed, and recurrent relationships with each other share experiences and approaches. This can make it difficult to search for novel information beyond the research group, to identify research opportunities and to develop novel publications. Therefore, the knowledge flowing within the research group network may become redundant and obsolete, reducing the group’s ability to successfully produce high-impact publications.

In summary, bonding capital has both advantages and disadvantages, depending on its level. We expect that when bonding capital is moderate, there will be a net positive effect on scientific results, because the trust generated will allow research group members to exchange tacit knowledge and access valuable information, without generating opportunistic behavior. However, for high levels of bonding capital, above a certain threshold, the dark side of bonding capital emerges, which reduces the ability of research groups to obtain high-impact scientific results, as they are exposed to redundant information and knowledge that is irrelevant to their output (Tiwana, 2008). In this way, the positive effects that are initially profitable due to close relationships are reduced when the connections are too dense (Ruiz-Ortega et al., 2022). Therefore, we predict an inverted U-shaped curvilinear relationship between the bonding capital of research groups and their scientific results. Thus, research groups with a certain level of bonding capital have the necessary trust among their members to generate increasingly scientific results. There are risks, however, for groups that develop excessive bonding capital because they have limited access to new and varied information, suffering from internal blockage, inertia, and redundant information in attempting to develop research for publication in a very competitive context. Drawing on these arguments, we propose the following hypothesis:

Hypothesis 1

Bonding social capital has an inverted U-shaped relationship with the scientific results of research groups.

The Moderating Role of Bridging Social Capital

The literature on social capital has highlighted the restrictions of bonding capital in generating new knowledge, leading to disruptive innovations, which are more commonly associated with bridging capital (Martínez-Pérez et al., 2019). This type of social capital refers to cooperation through the development of external networks, which help researchers make contacts in the outside world. Bridging capital has its origin in weak ties (Granovetter, 1973) and structural holes (Burt, 2017). Bridging capital is more effective in generating novel and valuable outcomes, as bridging social distances involves more diverse and meaningful members. Granovetter (1973) proposed that new information is obtained through casual relationships with weak ties, not through personal friendships where strong ties are necessary because although these ties generate a more significant number of interactions, some are redundant. On the contrary, weak ties allow diverse researchers to unite in discovering opportunities and to bring diverse approaches to developing quality research.

However, different positions exist in the study of bridging capital. Some authors mention that this type of capital must not be excessive, either. In this regard, Dahlin et al. (2005) suggest that the most diverse teams generate more knowledge only to a point at which the benefits begin to decrease. On the one hand, Cummings et al. (2013) point out that although there is a tendency for large research groups to be more productive than smaller ones, productivity decreases since members come from different disciplines. Meanwhile, Lee et al. (2015) mention that teams with diverse knowledge can access broader understanding. Moreover, they are also exposed to decreasing effects that will impact team results as heterogeneity increases. Furthermore, White et al. (2020) reports that relationships with external members must go beyond the capacity for building to explore the interactive generation of international scientific collaborations between researchers.

The literature widely reports that weak ties support the dissemination of novel information (Martínez-Pérez & Beauchesne, 2018). Burt (2017) considers that the advantages of bridging capital derive from the weakness of links and the existence of structural holes. Making an analogy, structural links and holes allow researchers to connect with distant environments, considering that members of different groups may have different results and complementary resources. This is in line with the fact that when research groups cooperate with different researchers, they expose themselves to interaction with new actors and access unique and diverse experiences and ideas, which provide competitive advantages in terms of radical new developments (Zheng, 2010). Following these arguments, Rodriguez and Gonzalez-Brambila (2016) find that structural holes favor the citation impact of research papers.

When, from their central position in the network, scientists maintain an adequate level of bridging capital, the knowledge and resources that flow from other scientists in the network increase as they build their position, establishing ties with external communities (Reagans & McEvily, 2003). This may mean that an increase in the possibility of absorbing, recombining, and transforming knowledge, in turn, increases the opportunities for high-impact scientific discoveries (Lin & Huang, 2023). The access of members of research groups to individuals with different knowledge, experience, and skills helps scientists to expand the scope of their results. In parallel, bridging capital serves scientists to rapidly spread their output to multiple communities, thus increasing its impact.

Rotolo and Messeni (2013) point out that the effect of bonding capital can be moderated by the number of ties established between communities. More specifically, the access to novel and diverse knowledge attributed to bridging capital allows new ideas to be identified, helping detect opportunities for research and integrating different research approaches to complement the advantages of bonding capital and mitigate its disadvantages in order to achieve greater research results (Gonzalez-Brambila, 2014).

Thus, structural holes allow a research group to access heterogeneous networks, promoting the diversity of ideas, creating new perspectives, and developing research outcomes (Reagans & Zuckerman, 2001). Thus, bridging capital reinforces the positive effect of bonding capital on the development of research articles. To the extent that groups can take advantage not only of the specific tacit knowledge of bonding capital but also of the complementarity of resources of the different agents and the generation of novel and disruptive ideas inherent to bridging capital, research results will be greater, and of better quality, impact, and visibility. However, above a certain threshold of bonding capital, the excessive cohesion and closure of research groups leads them to distrust external contacts and resist valuing their contributions, interpreting them as opportunistic behavior. In this context, high bridging capital can generate costs involved in coordinating and maintaining external contacts that are not capitalized on, given the differences in objectives and the cultural distance between external and internal agents, accentuating the problems derived from excessive bonding capital (Martínez-Pérez et al., 2021). This can lead to high bridging capital accentuating the negative effect of bonding capital on the development of research articles.

In summary, we propose that the balance between the advantages and disadvantages of strong ties and dense networks on research results varies according to the number and diversity of relationships established through the structural holes developed by research group members (Martínez-Pérez & Beauchesne, 2018). Building on the previous hypothesis, which establishes a curvilinear (inverted U-shaped) effect between bonding capital and research outcomes, we propose that the combination with bridging capital accentuates both positive and negative effects of bonding capital on research results. Based on these arguments, we propose the following hypothesis:

Hypothesis 2

The curvilinear (inverted U-shaped) relationship of bonding capital with the scientific results of the research groups will be more pronounced for a high level of bridging capital.

Figure 1 shows the proposed model.

Fig. 1
figure 1

Proposed model

Full size image

Methodology and Data

Data

First, we revised the existing literature on the Web of Science and Scopus databases related to the main topics of our research. Scientific or technological research groups in Colombia constitute the population under study. A research group is defined as a set of people who meet to investigate, formulate one or more problems of their interest, framed in a strategic plan, and produce new knowledge, development, technological innovation, training of human resources, and social appropriation of knowledge (Castrillon-Muñoz et al., 2020; Garcia-Sanchez et al., 2019; Martin-Alazar et al., 2020a).

For the National Science and Technology System of Colombia to validate the existence of a research group, it must meet the following requirements within a window of observation time: It must be registered in the GrupLAC system of the ScienTI Platform of the Ministry of Science and Technology, have a minimum of two members, one or more years of existence (declared age), be endorsed by at least one institution registered in the InstituLAC system of the Platform ScienTI, and have at least one ongoing research project, technological development, or innovation project. The group’s leading researcher must have a bachelor’s degree, a master’s degree, or a PhD. The group must have an output of new knowledge or the results of technological development and innovation activities, equivalent to a minimum of one product per declared year of existence; this must involve the social appropriation and circulation of knowledge or products resulting from activities related to the training of human resources in CTeI (Ministry of Science and Technology, 2020).

According to the information provided by the Ministry of Science, Technology, and Innovation of Colombia, the last time the indicators of the research groups were measured, in 2019, there were 5772 research groups, classified into the large areas of knowledge defined by the Organization for Economic Cooperation and Development (OECD): Natural Sciences, Engineering and Technology, Medical and Health Sciences, Agricultural Sciences, Social Sciences, and Humanities. The science database in figures administered by the Ministry of Science and Technology (2020) was used.

In this study, we focus on analyzing the research groups identified within the Engineering and Technology area of knowledge, with a total of 1124 groups. According to Rodriguez and Gonzalez-Brambila (2016), the production of knowledge and the dynamics of publications varies considerably across areas of knowledge, and we thus decided to focus on this specific field, considering the scarcity of studies on this group and the nature of the knowledge and performance that engineers create. Likewise, a specific area was taken as a selection criterion, based on Kulczycki (2017) and Patton (2018), who argue that the comparison of the productivity of active groups in different areas may be erroneous due to the differences in the number of publications. Furthermore, Engineering and Technology is the field with the greatest diversity of geographical location, as these groups operate in many different Colombian cities.

In this sense, information was collected on the number of existing groups, area, size, category and geographical location, and collaboration profile (Ministry of Science and Technology, 2020). Additionally, a data filtering process was carried out, consulting the SCIENTI platform, specifically the Gruplac, which provides the resume of each research group. It collects the information on the scientific production of the research groups of Colombia, as well as the profile of collaboration. These were analyzed and expressed with the value of the indicator of cohesion (bonding capital) and cooperation (bridging capital) and in percentages, which represents the profile. Additionally, the control variables of age, group size, and area were also analyzed to understand their relationship with the development of scientific production (Li et al., 2013; Martín-Alcázar et al., 2020b). Finally, the data were studied through a descriptive and correlational analysis and a hierarchical regression analysis. Figure 2 shows the steps followed.

Fig. 2
figure 2

Methodology design

Full size image

Variables

Scientific Production: Research Articles (Dependent Variable)

A product is generated by a group when one or more of its members are authors or co-authors of the product and authorize their link to the production of the research group to which they are attached. As the dependent variable, this study analyzes the scientific articles as results of activities to generate new knowledge, understood as original and previously unpublished production, published in a journal that has been evaluated and endorsed by peers as a significant contribution to knowledge in the area. The articles are categorized according to the impact factor of the Journal Citation Report (JCR) and SCImago Journal Rank (SJR) citation systems. Thus, the model takes the impact factor calculated for the journals in the same area of knowledge in the JCR, when the journal is categorized in the Web of Science Index (Clarivate Analytics), or in the SJR, an index whose source of information is Scopus (Elsevier). When a journal is indexed in both bibliographic citation indexes (JCR and SJR), the Ministry of Science and Technology, during the measurement information analysis process, selects the index and the knowledge area in which the journal obtains the highest position according to the quartiles (Ministry of Science and Technology, 2020).

Bonding and Bridging Capital (Independent Variables)

Bonding Capital (Group Cohesion Indicator). For each research group, the cohesion indicator is calculated. In so doing, we seek to assess the existence of joint work among the members of the group. In this sense, bonding capital is considered as a resource located in the internal networks of the groups from the social capital perspective (Martín-Alcázar et al., 2019). To calculate this indicator, the Ministry of Science and Technology (2020) considers the co-authorship of a new knowledge product as clear evidence of a collaborative connection between authors. This indicator is calculated as:

$$IC=Authors/Products-1$$

where “authors” corresponds to the total number of authors participating in the same production and “products” is the number of products associated with the group.

Bridging capital (Group Cooperation Indicator). For each research group, the collaboration factor is calculated, the aim of which is to demonstrate work between groups. In this sense, bridging capital depends on the personal trajectory and is built exclusively according to the specific needs of the groups (Kallio et al., 2010; Singh, 2007). To calculate this factor, the Ministry of Science and Technology (2020) considers the co-authorship of a product as clear evidence of a collaborative connection between authors linked to different groups.

$$ICoop=(Total\; number\; of\; related\; groups)/Products-1$$

where the “total number of related groups” is the research groups where the co-authors of new knowledge products are linked, and “products” is the number of products in the group.

Control Variables

The control variables included in the study were the age and size of the group and the area of knowledge. Age was measured as the difference between the year of data collection (2019) and the year of group creation. This variable was included to control for the impact of age and experience on the research results (Lee et al., 2015). Size was measured by the number of members of the research group to control for its impact on the expected results. The area of knowledge was included to control for the impact of specialization on scientific output. In this sense, within Engineering and Technology, there are 9 areas of knowledge: (1) civil engineering, (2) electrical, electronic, and computer engineering, (3) other engineering and technology, (4) chemical engineering, (5) environmental engineering, (6) mechanical engineering, (7) medical engineering, (8) materials engineering, and (9) biotechnology.

Results

The descriptive statistics and bivariate correlations are shown in Table 1. The hypotheses were tested by means of hierarchical regression analysis, following the recommendations in Haans et al. (2016). The regressions are presented in Table 2.

Table 1 Descriptive analysis and correlations
Full size table
Table 2 Hierarchical regression analysis
Full size table

The control variables of age, group size, and area of knowledge were initially entered into a base model. The data in the base model reveal that the variables of age (β = 0.056; p < 0.05), group size (β = 0.440; p < 0.01), and knowledge area (β = 0.082; p < 0.01) have a positive and significant influence on scientific results, and in particular, on the publication of research articles by research groups. Regarding Hypothesis 1, the variables of bonding capital and bonding capital2 were introduced in the next step (curvilinear effect model). To corroborate a quadratic relationship, we followed the procedure proposed by Lind and Mehlum (2010). Figure 3 shows that, as proposed, in the curvilinear effect model, bonding capital2 has a negative and significant effect (β = −0.244, p < 0.01) and produces an increase of 0.034 in the R2 with regard to the base model.

Fig. 3
figure 3

Curvilinear effect model: inverted U-shaped curve

Full size image

The sign of the regression coefficient β of the quadratic term represents the direction of the curve of the effect of bonding capital on the publication of research articles. In this case, the negative effect of the quadratic term suggests that bonding capital has an initial positive effect, but that this effect becomes negative beyond a certain level—inverted U-shaped. Thus, the curve can be seen to initially present a positive slope and then change and become negative. Finally, the turning point is in the data range. Therefore, the data confirm the existence of an inverted U-shaped relationship, as shown in Fig. 3, and we can thus accept Hypothesis 1.

In order to test Hypothesis 2, we added the interactive term between bonding capital2 and bridging capital to the regression. In this case, by including the interactive term (interactive effect model) over the curvilinear effect model, the R2 increases by 0.007. The results suggest that bridging capital moderates the curvilinear relationship between bonding capital and high-impact results. The coefficient of the interactive term describes how bridging capital moderates the non-linear relationship between bonding capital and high-impact results.

The results obtained in the interactive effect model corroborate Hypothesis 2. The interactive effect between the quadratic term of bonding capital and bridging capital has a negative effect on the publication of research articles (β = −0.155; p < 0.01), meaning that bridging capital strengthens the curvilinear relationship between bonding capital and the publication of research articles. Thus, the curvilinear relationship is more pronounced when bridging capital is high and less pronounced when it is low. Figure 2 shows the results in different situations.

Graphically, the negative effect reinforces the inverted U-shaped curve. Figure 4 shows that, with high levels of bridging capital, the inverted U-shaped curve is more pronounced, such that the positive and negative effects of bonding capital on the publication of research articles are accentuated.

Fig. 4
figure 4

Interactive effect model

Full size image

Discussion

The results obtained have allowed us to achieve our aims. Thus, firstly, we have verified the existence of a curvilinear inverted U-shaped relationship between bonding capital—cohesion—and the results of the research groups. These results confirm that collaboration between members of a group is an antecedent of research output, and, in particular, of research articles (Gonzalez-Brambila et al., 2013). This inverted U-shaped relationship implies that the positive effects of cohesion will initially predominate. However, beyond a certain level, the negative effects outweigh the positive ones, generating a negative effect on the publication of scientific articles. In this sense, contrary to studies that suggest collaboration with members of the group has a positive and linear influence on results (Birnholtz, 2007), our findings indicate that, for publications, low or medium levels of collaboration will have a positive impact, but above a certain level, for high levels of collaboration, there will be a negative impact on publications. These results confirm the appearance of the dark side of bonding capital, highlighted by several authors (Tiwana, 2008; Ruiz-Ortega et al., 2022), which generates information redundancy and poorer knowledge when relationships are too dense and frequent. Furthermore, our results confirm that bridging capital, that is, cooperation with members of other groups, has a negative moderating effect, which implies a larger focal distance of the inverted U-shaped curve. In this sense, the more pronounced slope of the curve reflects that cooperation with other research groups allows members of these groups to make better use of their internal collaboration networks for the development of scientific research. Therefore, research groups with higher levels of cooperation with other groups will be able to take better advantage of collaboration among their members to generate scientific publications. However, according to Martínez-Pérez et al. (2021), in contrast to the works that focus on the advantages of bridging capital to compensate for the problems of excessive bonding capital, our results suggest that excessive cohesion of research groups can lead to distrust about the value of the contributions of external contacts and about their possible opportunistic behaviors, accentuating the negative effect of bonding capital on the development of research articles.

In relation to the control variables, the results show that age and size have a positive and significant impact, indicating that, as stated in previous studies (Lee et al., 2015), larger groups with greater experience tend to carry out more impactful research. Additionally, the group’s area of knowledge also has a positive and significant influence. To better understand this result, we performed an ANOVA analysis that revealed significant differences between some groups in relation to the publication of articles. Specifically, the fields of materials engineering and chemical engineering show higher results than the rest of the groups in these publications.

Conclusions

Main Results

Scientific research is essential for the economic progress of a country, especially in developing nations (Bhargava, 2016). This study responds to the demand for studies that analyze the relational determinants of scientific production (Martín-Alcázar et al., 2020b) and specifically of the article publication (Rodriguez & Gonzalez-Brambila, 2016). This work has fulfilled our proposed objectives, namely, to analyze the curvilinear relationship between bonding capital and the publication of research articles, and the moderating effect of bridging capital on this relationship. Previous studies in this line have established that social capital positively influences knowledge creation (Siadat et al., 2012), the effectiveness of collaborations with companies (Steinmo, 2015), and scientific productivity (Gonzalez-Brambila, 2014). In relation to the analysis of scientific publications, previous studies have established the influence of the different dimensions of social capital—structural, relational, and cognitive—on the quality and quantity of publications (Gonzalez-Brambila et al., 2013) and the influence of strength, efficiency, and centrality on these results (Abbasi et al., 2011). Complementarily to these contributions, our study focuses on the determinants of scientific production, focusing on the key role of social capital, specifically its dimensions of bonding capital—cohesion—and bridging capital—cooperation.

The main contribution of this paper is that it delves into the controversial relationship between bonding capital and outcomes of research groups—publication of articles, identifying an inverted U-shaped curvilinear relationship between these two variables. Thus, for research groups, we find that for low levels of cohesion between members—bonding capital—the level of publications tends to increase, taking advantage of the frequency, trust, and shared values of group members. However, as this level of internal collaboration increases, problems arise from information redundancy, inertia, and myopia, which prevent access to novel information and constrain the development of groundbreaking research, limiting the publication of articles. Another important contribution of the study is the understanding of the important role of cooperation with other research groups—bridging capital—in promoting this type of publication, which is reflected in its significant moderating effect. Thus, our results have verified that external cooperation allows access to diverse information that complements the capabilities of the members of the group and generates new ideas based on a wider range of skills (Gonzalez-Brambila, 2014; Lin & Huang, 2023; Pan et al., 2021). However, high levels of bridging capital can also generate costs involved in coordinating contacts due to the cultural distance between external and internal agents, accentuating the problems derived from excessive bonding capital (Martínez-Pérez et al., 2021).

The originality of the study is shown in the linking of different types of social capital—bonding and bridging capital—to test their impact on the results of the research groups and in the proposal and testing of a moderate curvilinear relationship, which allows us to advance in this line of research.

Theoretical and Practical Implications

With regard to the theoretical contributions of the study, we have been able to delve deeper into the complex relationship between collaboration and performance (Salimi et al., 2022; Steinmo, 2015), supported by the Social Capital Theory (Petrou & Daskalopoulou, 2015; Zhang & Shih, 2022). Specifically, we have examined the controversial consequences of the two basic dimensions of social capital—bonding and bridging—addressing their joint effects, which have been scarcely analyzed in the literature (Lin & Huang, 2023; Zhang & Shih, 2022). In addition, in contrast to the authors who have highlighted the advantages of bonding capital and bridging capital, we identify and justify perverse effects for excessive levels of both dimensions (Martínez-Pérez et al., 2021; Ruiz-Ortega et al., 2022). This study complements the proposal of Martín-Alcazar et al. (2020b) related to the impact of social capital on the results of research groups. It highlights, in line with Oh et al. (2006), that group effectiveness is maximized through optimal configurations of internal and external relationships, which can complement existing capabilities and provide access to new networks of contacts that increase the level and quality of publications.

In terms of practical implications, the results allow us to make a number of recommendations for members of research groups. Firstly, researchers should promote the collaborations with members of their group that allow them to take advantage of the trust and knowledge shared to develop research. This should complement cooperation with researchers from other groups, allowing them to detect new research trends, and new capacities to increase their publications. In this sense, we recommend that research group leaders encourage their members to attend congresses, conferences, and seminars and to undertake research stays, as these can be tools to favour the access of researchers to new cooperation networks that help complement the positive effects of bridging capital and to limit the negative effects of excessive bonding capital. Finally, the leaders of research groups must also avoid excessive costs of coordinating and maintaining relationships with external groups with very different objectives and values, in order that members of the research group do not distrust external contacts and hinder scientific production.

Limitations and Future Lines of Research

Despite the precautions taken, this study has certain limitations that could affect the generalization of its results. Firstly, it is a cross-sectional study that provides us with information at one point in time. In addition, we have only analyzed scientific and technological research groups in a developing country, namely, Colombia. Therefore, caution is needed in extending the conclusions to other areas of knowledge or countries.

Based on these results, we propose a series of future avenues of research. In line with the proposal of Martin-Alcazar et al. (2020a), we consider it interesting to propose social network analysis as a complementary methodology to represent and analyze in depth the social structures of research groups. Furthermore, we propose extending the analysis to other contexts and complementing it with the analysis of new variables.

Note: Both the bivariate correlations and the tolerance and variance inflation factor (VIF) statistics indicate that there are no multicollinearity problems.