1 Introduction

Principal investigators (PIs), as leaders in the R&D context, need to meet different requirements compared to those working in other areas (Keller, 2017). Their role is crucial in the achievement of the research goals and realising other a range of impacts (Cunningham, Mangematin, et al., 2016; Menter, 2016). Therefore, PIs require a specific human capital (HC) in order to manage a research team (RT) successfully and to ensure the continuous improvement of research outcomes (Bozeman et al., 2001; Cunningham et al., 2018). HC comprises all the knowledge, abilities and skills of the individual (Becker, 1964), and it is considered an intangible resource (Martin-Sardesai & Guthrie, 2018). PIs must have certain abilities and skills beyond technical expertise or knowledge, such as the ability to adapt to changes, critical thinking skills, motivation, creativity, initiative or perseverance (Ansett, 2005; Carl, 2020; Cunningham et al., 2014). Unfortunately, PIs do not always find institutional support in the process to acquire the HC needed, so they usually need to learn on the fly, which is far from optimal (Cunningham, 2019; Cunningham et al., 2014).

At present, there is a growing stream of research studying academic PIs in the R&D context from multiple points of view (Kastrin et al., 2018; O’Kane et al., 2017). For instance, some studies have focused on investigating the responsibilities that a PI has in the R&D context (Cunningham et al., 2019; Mangematin et al., 2014), or examining the key role they play in the process of obtaining public funding (O'Kane et al., 2017). Further research has been undertaken on the roles that PIs play in the three missions of the university and also as a value creator taking a quadruple helix approach (Carl, 2020; Cunningham et al., 2018; O'Kane et al., 2020). Despite this, there is yet no analysis based on the PI’s HC (Käpylä et al., 2010).

Consequently, the literature calls for a deeper understanding of the PI role by studying the specifics of their HC as a way to better understand their behaviour as RT leaders (Boehm & Hogan, 2014; Cunningham & O’Reilly, 2018). According to Wadhwa et al. (2017), the leaders’ HC may influence strategic decisions, such as decisions related to innovations in the organisation, and PIs are certainly not an exception (Herrmann & Datta, 2002). Thus, studying PIs’ HC could help us understand more about how and why certain PIs achieve certain outcomes.

Munshaw et al. (2019) have already linked the HC of researchers to the achievement of patents. However, the role of PIs is much more complex than that of other researchers and their HC, therefore, is expected to be different, which is precisely the question motivating this research: what HC do PIs need? This study addresses that question not only by exploring which skills, knowledge and abilities are required for the PI position, but also by trying to stablish whether they can be found in different combinations, i.e., whether different profiles can be identified, which may facilitate the classification and comparison of PIs. This classification could also provide meaningful insights into PIs performance. Contrary to the ‘more is better’ assumption (Munshaw et al., 2019; Ployhart et al., 2014), our findings revealed that higher levels of HC are not necessarily connected to better results, suggesting that a ‘too-much-of-a-good-thing’ effect (Pierce & Aguinis, 2013) could be taking place. This is also in line with Kor and Sundaramurthy's (2009) study, which suggests that an excess of HC could derive in negative synergies.

Deepening into the study of the PI’s HC is therefore essential, not only for improving RT management and leadership, but also for designing public research programmes, since it could provide valuable insights into the selection of those PIs most suitable for the position, depending on the relevant circumstances in each case (Cunningham et al., 2018). The remainder of this paper is structured as follows. The following section reviews the role of the PI and contains an approach to the HC theoretical background. Subsequently, the sample and the methodology utilised are described. The next section presents and discusses the analysis of the results. Finally, conclusions, limitations and a research agenda are drawn.

2 Theoretical background

2.1 Principal investigators

Since the R&D environment is so challenging and unique, the appearance of the PI is increasingly noticeable (Guerrero & Urbano, 2012). In fact, PIs are widely acknowledged as a key player in this context, as it was recognised in the triple helix model (Menter, 2016; O’Kane, 2018), and further highlighted in the quadruple helix model, where they influence the rest of the actors developing the innovation process (Carl, 2020). Cunningham, O’Reilly, et al. (2016) asserted that ‘PI’ is a commonly used term in the academic literature as well as in the publicly funded institutions. Del Giudice et al. (2017) stated that the PI is responsible for the design and implementation of the RT research programme. In this study, we define the concept of PI as:

A heart of value creation through development of knowledge that can be appropriated and utilized by other triple helix actors. This PI driven value creation can result in a number of scientific, economic and societal impacts and gains that contribute to a joint production motivation of the triple helix (Cunningham, Mangematin, et al., 2016, p. 780).

At present, academia has turned into a highly competitive environment due to both the pressures on scholars to obtain results and the increasing specialisation of academic research (Degn et al., 2018; Leahey, 2016). As a result, academic researchers are driven to collaborate more and more with each other in order to survive in this highly competitive environment and overcome the publish-or-perish mandate (Kastrin et al., 2018). In this sense, the formation of multi-diverse research teams is increasingly common and the PI role is crucial to lead and manage this collaboration between researchers in order to achieve a common objective (Antes et al., 2019b; O’Kane et al., 2017).

Becoming a PI is a noticeable evolution for a scientist, moving up to a leadership role (Kolb et al., 2012). This transition process is definitely a difficult one (Tregoning & McDermott, 2020). Not only is the PI role crucial for the R&D context, but it is also an essential step in the academic career progression of a scientist (Cunningham et al., 2019). So much so, that this forward step in their progression completely transforms their individual career and, even more, their trajectories (Cunningham et al., 2018; Kastrin et al., 2018). For that reason, becoming an RT PI brings prestige for a scientist among other scientists (Cunningham et al., 2014). It also provides them with the benefit of being able to plan their research agenda, and their scientific productivity is bound to grow (Feeney & Welch, 2014). Whereas some authors highlighted obtaining funds for RT research activities as the most relevant responsibility of PIs (Cunningham et al., 2021; O’Kane et al., 2017), others asserted that PIs also manage the available resources, enable everyone to accomplish their aims, brake boundaries and span them (Cunningham et al., 2019; Cunningham, Mangematin, et al., 2016; Mangematin et al., 2014). Moreover, PIs also have to accomplish some academic responsibilities, such as supervising, mentoring and teaching (Boehm & Hogan, 2014). In every single scientific field there is a consensus on how important the PI is, not only in the RT internal dynamics, but also in enhancing knowledge and technology transfer beyond RTs (Catalán et al., 2019; Lin et al., 2016; O’Kane et al., 2015). PIs constitute an important key actor in making university-industry engagement more viable (Cunningham, O’Reilly, et al., 2016; Menter, 2016; O'Kane, 2018). Indeed, PIs are better positioned to act as an effective link to overcome any potential barrier between industry and academia than anyone else (Del Giudice et al., 2017). Their role may even be pivotal to match the different timeframes that exist for industry and academia when both are involved in a partnership, based on the application of the principles of an iterative approach (Albats et al., 2018). Nonetheless, this role is not without its complexities, so the PI needs to be more than just a good researcher (Cunningham et al., 2015; Foncubierta-Rodríguez et al., 2020; Mangematin et al., 2014). The requirements that PIs have to face are specific and challenging (Casati & Genet, 2014; Kidwell, 2013). Actually, after becoming PIs they carry out additional functions related to the leadership and management of the RT as well as the research lines (Cunningham et al., 2019). Therefore, acquiring managerial capabilities is essential not only in order to address internal RT issues and to improve RT outcomes, but also to enhance RT external relationships, such as strengthening the bridge between industry and academia (Boardman & Ponomariov, 2014; Boehm & Hogan, 2014; Cunningham et al., 2015).

Being the PI in an RT is a much more complex role because it involves assuming many more responsibilities, and therefore, many more competences are required in order to be effective (Cunningham, O’Reilly, et al., 2020). Not only do they have to have the competences of being a scientist, but also other competences such as managing their project, leading their RT, managing all the relations with industry or other stakeholders, and acquiring the needed resources (Casati & Genet, 2014; Cunningham, 2019; Cunningham et al., 2017; Kidwell, 2014; Mangematin et al., 2014). In light of the above, it seems clear that PIs need some additional competences on top of the ones that are normally expected for scientists.

Consequently, considering how influential the PI role is, it is crucial to address the competencies needed by those who are leading research projects and teams (Cunningham & O’Reilly, 2018; Cunningham et al., 2018). Some studies have analysed different dimensions of academic researchers’ HC, although none have focused on the ones performing the role of PIs. Therefore, there is still a lack of consensus about the PI role at a micro-level, particularly in determining factors or characteristics that could allow establishing suitable profiles of PIs. In this regard, analysing and identifying their HC could provide meaningful insights.

2.2 Principal investigators and human capital

So far, it has been clearly established that being a PI requires having a specific HC to manage an RT successfully in order to ensure the continuous improvement of research outcomes (Bozeman et al., 2001; Cunningham et al., 2018). However, there is more to a certain PI’s HC than simply a set of competences that are needed to fulfil their role. According to Wadhwa et al. (2017), the leaders’ HC may influence strategic decisions, such as decisions related to innovations in the organisation, and PIs are certainly not an exception (Herrmann & Datta, 2002). Thus, studying PIs’ HC could help us learn more about how and why certain PIs achieve certain outcomes.

Naturally, PIs do not work in isolation and each member of the RT contributes to the success of the research project. However, it is the PI who, ultimately, decides who joins the team, and their ability to combine the experience and HC provided by the new members with the HC that already exists in the group is key to get the maximum potential out of these new incorporations and to guarantee that the research project can be accomplished with that particular combination of HC (Pan et al., 2020). In other words, as Kidwell (2014) puts it, PIs mobilize resources based on their vision of what their RT should be, its internal operation and the achievement of the objectives.

HC will be defined here according to the differentiation of the SKA dimensions−Skills, Knowledge and Abilities (Ployhart, 2015). Considering how particular the R&D context is, PIs usually require a range of skills, knowledge and abilities quite different to that of other managerial positions (Keller, 2017).

Skills are those qualities which contribute towards achieving better results in research (McNie et al., 2016). For instance, in public-funded research, skills such as time management and time allocation have a significant impact on the results of the study (Cunningham, O’Reilly, et al., 2016). Researchers’ creativity provides flexibility and adaptability to any variation in the initial research planning, which can also impact the results of the study (Bazeley, 2010; Marie, 2008). Developing appropriate protocols to support the study, publishing studies in high-impact journals, communicating effectively, leading and managing are some of the key skills required by individual scientists doing research (McNie et al., 2016). In the case of PIs, managerial skills–such as those needed to control the RT resources availability or to implement an RT plan– are obviously important (Cunningham, O’Reilly, et al., 2020), but these need to be complemented with a range of entrepreneurial skills (Miller et al., 2018). Lastly, diversity entails a positive effect on RTs as it provides them with the possibility of having a broader range of HC, experiences and ideas (Huang & Lin, 2006; López-Fernández & Sánchez-Gardey, 2010). Therefore, the scientist in the PI role is required to have the skill to manage cultural diversity and to collaborate through a wide range of disciplines (Cunningham, O’Reilly, et al., 2020).

Knowledge is defined as the training completed by researchers. Taking into account the wide variety of existing research fields, this training may vary considerably throughout the predoctoral and postdoctoral periods, depending on each particular case (Bozeman et al., 2001). Ployhart and Moliterno (2011) defined it as the “understanding of principles, facts and processes” (p. 134). Moreover, researchers’ past experiences are also considered a form of knowledge for scientists (Hitt el al., 2001). As stated in the literature, potential future publications might depend both on the academic knowledge of the scientists and on the influence that the postdoctoral training has been able to exert (Lovitts, 2005; Su, 2014). In this sense, all the knowledge acquired throughout the scientist's education and training will be to their benefit and will be reflected in their scientific outcomes (Jacob & Lefgren, 2011). In fact, both an in-depth study of the topic which is being explored and developed, and the knowledge of the methodological aspects contribute to this purpose (Bazeley, 2010). The differentiation between know-how –i.e. research methods and technical aspects– and know-that –i.e. theoretical training in a particular scientific field– will be used in this research to define the knowledge dimension of the HC (Bozeman et al., 2001), since it is helpful to highlight areas where the PI role is more challenging than that of other researchers. For instance, in the case of know-that, apart from a sound knowledge of their area of expertise, PIs also need to have knowledge related to technology transfer, or research commercialisation (Cunningham, Dolan, et al., 2020). In the case of know-how, PIs past experiences might prove crucial, since scientists do not normally receive any specific training in order to occupy the PI role, which means they usually have to learn on the fly (Cunningham, 2019; Cunningham et al., 2014).

Abilities address those distinctive features of scientists relevant to their specific field of study (Lindberg & Rantatalo, 2015). Among the academic hard skills, which any researcher should have, are those of being rigorous in the research process, being able to present and publish the findings of their research, and being able to propose hypotheses (McNie et al., 2016). Additionally, fostering collaboration within the RT, as well as analysing the findings resulting from the study, are also abilities considered essential for any scientist (Bozeman et al., 2013; Marie, 2008). However, in the case of PIs, these abilities are not enough. Their role demands that PIs also develop a range of soft-skills, such as the ability to create a safe environment for every member of the RT to communicate effectively and to share their information, experiences or ideas (Antes et al., 2019b). On the same note, PIs also need to develop the ability to influence and motivate people inside and outside their RTs. This applies not only to their RT members, but also to external partners, either in an academic or industrial environment, such as the technology transfer office and also managing governance arrangement (Cunningham, O’Reilly, et al., 2020).

In sum, PIs are researchers (Mangematin et al., 2014). When carrying out their responsibilities, PIs must have certain abilities and skills beyond technical expertise or knowledge (Carl, 2020; O’Reilly & Cunningham, 2017; Othman et al., 2019). Every PI has distinctive SKA that make them behave differently when making decisions, and thus they influence the makeup and the outcomes of their RTs (O’Kane et al., 2015). These individual competencies (SKA) are considered valuable intangible resources (Martin-Sardesai & Guthrie, 2018). There is an accepted body of research on HC that asserts that ‘more is better’, so that the best option for individuals is to have more and better SKA (Ployhart, 2015). Nonetheless, as a consequence of the drawbacks associated with possessing an excess of HC, rather than having an appropriate pool of SKA which are necessary to fulfil the objectives (García-Carbonell et al., 2018), there are also limitations to this perspective. For instance, Kor and Sundaramurthy (2009) suggest that an excess of HC could derive in negative synergies. Accordingly, it would be valuable to analyse PIs’ HC to pinpoint different patterns based on several combinations of these distinctive SKA.

3 Methodology

3.1 Data collection and sample

An empirical work was undertaken in order to identify features that shape the PI's HC, which could be fundamental in the achievement of macro-, meso- and micro-level objectives in the R&D context (Cunningham et al., 2018; Käpylä et al., 2010). Due to the absence of empirical studies on academic HC, a two-phase research approach was developed. First, an exploratory research based on an expert panel was developed in order to identify items and design the survey. Then, an exploratory research approach was applied. In doing this, an inductive methodology was carried out through a cluster analysis developed on a sample comprised of PIs.

In the first stage, based on the Delphi technique, a panel of experts was created in order to design the survey. This technique relies on the anonymous exchange of opinions among experts, informed by their experiences and their knowledge of the existing literature regarding the topic at hand, until a common consensus is reached (Hsu & Sandford, 2007; Landeta, 2006). The PIs comprising this expert panel were selected among the members of the Andalusian Research Plan, by fulfilling two conditions: 1) they should belong to different research fields to ensure no biases − 6 from Social and Legal Sciences, 8 from Health Sciences, 10 from Engineering and Architecture, 18 from Sciences and 20 from Arts and Humanities − , and 2) they should have been the RT PI of a submitted project which had achieved competitive public funding (Okoli & Pawlowski, 2004). A total of 134 experts fulfilling both conditions were contacted. The panel was finally comprised of 62 experts (46% rate response), all of them having an extensive background in managing and leading RTs and research projects. The original purpose of implementing the Delphi technique was to conduct a larger research project, covering not only the human capital of researchers, but also the social and organisational capitals, which, in turn, constitute the intellectual capital, and how this intellectual capital is managed by the university. For this study, we have focused on human capital only, since this is the foundation of all the other competences that PIs need to develop in order to face the many responsibilities to which they will be committed while they are in the PI role (Ployhart, 2015; Wadhwa et al., 2017). Following the expert discussion process by the Delphi technique, this panel of experts identified 22 HC items after three rounds of exchanging opinions. Then, a draft questionnaire was distributed among the experts for their feedback. Each item included in the draft questionnaire was coded with a five-point Likert scale (1 = total disagreement, 5 = total agreement), adding a space where the experts could make their suggestions as to how to improve any of the items. Once all the pre-test surveys were received, the final questionnaire was developed.

The survey was emailed to PIs of RTs at Spanish public universities, with experience in developing research projects with national or international competitive public funding in any scientific field (Engineering and Architecture, Social and Legal Sciences, Health Sciences, Arts and Humanities or Sciences). In order to promote a high response rate, apart from contacting each PI directly, the heads of departments of the PIs and each university’s vice-rector for research and transfer affairs also received an email explaining our research objectives and calling for support by prompting their RT PIs to participate. Our final sample was comprised of 224 PIs of RTs, which represents a response rate of 42% (Table 1). The examination of the descriptive statistics and the characteristics of the sample allowed considering it representative of the population, and the possibility of non-response bias was excluded (χ2scientific fields = 5.173, sig. = 0.270; χ2age = 23.509, sig. = 0.946; χ2seniority in the university career = 5.991, sig. = 0.112).

Table 1 Descriptive Statistics of the Sample
Full size table

Additional relations between certain variables of the sample were also examined, such as that between the gender of the PI and their performance or the average performance in each scientific field at both PI and team level (Table 2). In this study, performance was measured by calculating the h-index (Hirsch, 2005). The h-index is a measure combining the number of citations and the number of publications, concentrated in a single indicator (Hirsch, 2005) which can be measured not only at an individual but also at a team level, and both figures are already available in SciVal (Colledge & Verlinde, 2014). Significant differences were found in this regard. First, there is a discrepancy between the gender of the PI and their individual h-index which, on average, is higher for male PIs. Similarly, there are also clear differences based on the scientific field at both levels of the h-index, i.e., there are certain scientific fields whose productivity is much higher. In particular, Science, Health Sciences, and Engineering and Architecture are the areas where the h-index mean is vastly greater, at both PI and team level. By including these measurements into our analysis, we would broaden our vision of the context and the evaluation of the resultant profiles could be more complete.

Table 2 ANOVA research productivity of PIs and RTs, gender and scientific field
Full size table

Considering the large number of HC items to be assessed−22 items−and also in order to make the cluster analysis more feasible, a preliminary dimension reduction was undertaken (Hair et al., 2010). An exploratory factor analysis was conducted, obtaining five HC factors (Table 3). In this process, non-significant items were excluded (I consider myself an observer, I can autonomously develop research, I know how to conduct research (thesis, research projects, etc.), I consider myself an altruistic person, and I am able to identify research topics in my research context). Findings revealed and confirmed the SKA model approach based on the Kaiser–Meyer–Olkin (KMO) index value, which was 0.86. Therefore, the resultant five HC factors are applicable to the variables studied (Ployhart, 2015). The eigenvalues of the five factors are higher than one, so they fulfil the latent root criterion. Moreover, these factors constitute 67.359% of the total variance and are consistent with the requirements stipulated in the literature (Hair et al., 2010). Cronbach’s alpha value was 0.864, which is considered high and means that the scale is reliable (George & Mallery, 2003). The resulting factorial model fits correctly to explain the data, since Bartlett's test of sphericity is significant (p < 0.001).

Table 3 Exploratory factor analysis
Full size table

The first factor (F1) is composed of four items related to the essential knowledge needed to develop research activities and, thus, it was labelled scientific educational training (Table 4). The second factor (F2), composed of five items comprising the necessary abilities to carry out an academic research, was labelled investigation abilities (Table 4). The third factor (F3) was labelled self-mastery skills (Table 4) because it is composed of three items, which referred to self-management and self-control. The fourth factor (F4) is composed of three items, which correspond to skills associated with more flexible responses to changes, such as being creative and having initiative and motivation. Accordingly, it was labelled openness-to-change skills (Table 4). The fifth factor (F5) is composed of two items related to being critical and being able to accept criticism from others. For that reason, it was labelled self-analytical skills (Table 4).

Table 4 HC factors of the PIs
Full size table

3.2 Cluster analysis

Based on the five HC factors identified in the previous section, our study then carried out a cluster analysis of the data in order to determine whether or not distinct PI profiles emerged, with the aim of stablishing a typology of PIs which could help to better understand their characteristics, in the light of empirical evidence. In order to make this typology useful and actionable, the different PI profiles identified had to be as internally homogeneous yet externally different from each other as possible; while also being conceptually interpretable (Schmitt et al., 2007). For this purpose, before carrying out the hierarchical cluster analysis, it was necessary to establish the number of profiles according to the sample (Ketchen & Shook, 1996). After the implementation of the dendrogram and the assessment of its results, it was concluded that the optimal number of clusters was three (Ketchen & Shook, 1996). Then, after verifying that all five factors of the HC of PIs were considered significant according to ANOVA tests, a K-means cluster analysis was conducted. From the K-means cluster analysis, three different PI profiles were identified with 59 cases (CL1), 128 cases (CL2) and 37 cases (CL3) respectively (Fig. 1).

Fig. 1
figure 1

PIs profiles of the RT and cluster analysis

Full size image

To extend our knowledge of these profiles, the connections between clusters and h-index at both PI and team level were also analysed. There were variations between clusters at both levels of the h-index (see Table 5). All the information that has emerged from the data obtained by applying the ANOVA test to the variables may be highly valuable in explaining the different PI profiles.

Table 5 ANOVA h-index of PIs and clusters and h-index of the RT and clusters
Full size table

4 Results

The results of the empirical analysis provided three different clusters, which led to the identification of three distinct PI profiles that were named Research-oriented PIs, Accomplished PIs and Management-focused PIs respectively. Among these three profiles, two of them–Accomplished PIs and Management-focused PIs–are in sharp contrast with each other. While the former consistently shows high or extremely high values in all five HC factors, the latter shows low or extremely low values in most of them. The third profile –Research-oriented PIs– could be considered somewhat in between the other two, showing a combination of rather high values in certain HC factors together with low or moderate values in the remaining ones. The composition of each profile in terms of gender and scientific fields is shown in Table 6.

Table 6 Gender and Scientific Field Distribution
Full size table

Research-oriented PIs profile is comprised of 59 scientists in the PI role, mainly belonging to Science (35.6%) or Engineering and Architecture (27.1%) scientific fields, and displays the lowest representation of women (only 23.7%). This profile has been labelled Research-oriented PIs (CL1–dashed line in Fig. 1) because they consider that they have the highest scientific educational training − theoretical and methodological knowledge. They regard themselves as highly trained to carry out research in their own scientific field, having the necessary methodological knowledge to undertake it. These PIs also rated themselves as having the highest self-analytical skills and, as a consequence, they present a high capacity for criticism, not only from themselves but also from others. They consider themselves to be highly skilled in investigation abilities, although they did not display the highest scores. Nonetheless, among these abilities, they believe themselves skilled enough to be able to interact easily with other researchers. Contrastingly, they also have the lowest scores in the remaining factors that are not strictly connected to research activities, namely the openness-to-change skills and the self-mastery skills. These preliminary results could suggest that they do not consider themselves very creative or as having too much initiative and, more importantly, that they might not be considered disciplined scientists and could show a lack of organisation in their research. However, this PI profile has the highest h-index mean, at both individual and RT level (Table 5).

Accomplished PIs (CL2–dash-dotted line in Fig. 1) is the most abundant profile, since it is comprised of 128 scientists in the PI role. As in the former cluster, most of them belong to Science (31.3%) or Engineering and Architecture (25.8%) scientific fields. Almost 40% of them are women, making this profile the group with the highest representation of female scientists in the PI role. This profile is comprised of all of those PIs who consider that they have high values in all of the different HC factors. In this sense, this profile shows the highest scores in three of the five HC factors: openness-to-change skills, investigation abilities and self-mastery skills. These preliminary scores might suggest that they believe that they are creative and they are able to network with other researchers, besides being motivated for research. These openness-to-change skills might enable them to be flexible with the changes that come up in their scientific fields, in order to adapt their research agenda (O’Kane et al., 2015). It is noteworthy that this is the only group of PIs who highly rated themselves in self-mastery skills, since they positively believe that they are accurate and disciplined. Moreover, not only do they regard themselves as having the necessary scientific educational training − both theoretical and methodological knowledge − to carry out an investigation in their area of expertise, but also as able to obtain and manage the information required for this purpose. Strikingly, even though this cluster has shown the highest scores in several HC factors − self-mastery skills, investigation abilities and openness-to-change skills − it has not been enough for them to obtain the highest h-index mean results, either on a personal or on a team level (Table 5).

The Management-focused PIs profile (CL3–solid line in Fig. 1) is the smallest cluster, which is comprised of 37 scientists in the PI role, of whom 27% are women. Within this cluster, even though the Engineering and Architecture scientific field is well represented (29.7%), more than 40% of the PIs belong to the two least productive scientific fields−Arts and Humanities and Social and Legal Sciences–, which may be connected to the fact that this profile shows the lowest h-index values at both individual and team level. Apart from that, they also showed the lowest score in three of the five HC factors–scientific educational training, investigation abilities and self-analytical skills–, suggesting that they might lack essential skills required to produce high-impact research outcomes. This deficit contrasts with a rather high value in openness-to-change skills and a moderate value in self-mastery skills, which implies that, even though they do not consider themselves as particularly well trained in theoretical and methodological grounds, they do regard themselves as highly creative and motivated, as well as reasonably disciplined and organised. This combination of characteristics might indicate that the PIs who comprise this profile are not mainly focused on research and that they might devote most of their time to managerial and administrative tasks, prioritising certain aspects of the PI role –such as the allocation of time and resources– over the rest, and that is why they were labelled Management-Focused PIs.

All the profiles are predominantly composed of PIs–over 70% of them–who have a long experience (which means more than 16 years of experience in a PI role), although in the case of the Accomplished PIs profile the proportion almost reaches 80%.

5 Discussion

It is becoming increasingly common for RTs to have a multidisciplinary composition (Cummings & Kiesler, 2005; Tyran & Gibson, 2008). So much so, that this is often an explicit requirement in order to obtain public funding (O’Kane et al., 2017). However, the management of this knowledge-intensive multidisciplinarity is challenging (Harney et al., 2014). This diversity within RTs demands that whoever is to direct and manage them should have a wide range of competences –SKA– that define their HC (Cunningham et al., 2018; O’Kane et al., 2020). Naturally, each PI will have a particular mix of competences in their HC, with varying degrees of proficiency in each aspect, that account for differences in behaviour, which can have significant consequences when they affect their decision-making (Herrmann & Datta, 2002; O’Kane et al., 2015; Wadhwa et al., 2017). Despite this, there is no previous attempt in the literature to study the PI role at a micro-level from the perspective of their HC, and that is precisely the question that this research tries to answer: what HC do PIs need?

5.1 PI profiles and performance

When differences in HC have to do with the way PIs make decisions, this could influence the final outcomes of their RTs and, by extension, their performance (Antes et al., 2019a; Cunningham et al., 2017; Ebrahimi & Azmi, 2015). In this sense, our analysis also revealed significant differences in h-index values among the three profiles, at both individual and RT level (Table 5). At an individual and RT level, Research-oriented PIs achieve the best results, followed closely by Accomplished PIs, while Management-Focused PIs obtain considerably lower values. So much so, that the maximum h-index of a PI in the Management-Focused cluster is less than half the maximum h-index of the other two clusters.

Among the three different PI profiles, Research-oriented PIs not only have the highest personal h-index mean, but they are also capable of leading their RTs to achieve the highest h-index mean (Table 5). These findings strikingly contrast with the ‘more is better’ assumption, which expects that researchers with higher HC will obtain higher outcomes, because HC is considered a valuable intangible resource (Martin-Sardesai & Guthrie, 2018; Munshaw et al., 2019; Ployhart et al., 2014). However, our findings are more in line with the conclusions of the study by Garcia-Carbonell et al. (2018) where they suggested the benefits of an adequate combination of SKA to achieve the objectives, in contrast to having an excess of HC.

Furthermore, the contrast between Research-oriented PIs and Accomplished PIs is sharpest in their scores of openness-to-change skills –the highest ones in Accomplished PIs and the lowest ones in Research-oriented PIs–, which might suggest that an element explaining their performance could be based on differences in the way each profile interprets all the information from the ongoing research and in their understanding of the R&D context (Jarratt & Stiles, 2010). Additionally, the fact that both profiles score rather high in scientific educational training seems to agree with other studies that based their PI selection only on the technical expert role (Clarke, 2002; Huang & Lin, 2006).

From the relationship between the PI profiles and their individual and RT performance, it may be inferred that those PIs who displays high theoretical and methodological knowledge could have a positive influence on their personal and RT outcomes. For an effective leader, being technically well prepared is a required quality, because it ensures that they are both trusted and respected (Paulsen et al., 2009; Sapienza, 2005). On the same note, their capacity both to accept criticism and to interact easily with other researchers are key skills for diminishing conflict and motivating RT members (Croucher et al., 2020; Sapienza, 2005).

At the opposite end of the spectrum is the Management-focused PI profile. They presented the poorest outcomes at both individual and team level (Table 5). The fact that they showed the lowest scores in scientific educational training, investigation abilities and self-mastery skills could seriously hinder their ability to produce high-quality results or even to identify which hot topics in their scientific fields need to be explored. If this is the case, they should surround themselves with well-trained scientists in their RTs to compensate for these weaknesses if they are to succeed with the increasing level of competition to publish (Kastrin et al., 2018). Another reason might be that this profile is highly composed of PIs belonging to the fields of Social and Legal Sciences, as well as Arts and Humanities. As is known, these two scientific fields generally tend to show lower values of h-index than other research areas, for instance, due to the preference of publishing books with the results of their research rather than papers, among other factors (Hirsch & Buela-Casal, 2014). On top of that, there is also the possibility that their low scores in those HC factors may be motivated by a decision to focus on the development of other competences, such as managerial skills, boundary spanning or brokering science (Cunningham et al., 2015; Mangematin et al., 2014).

Strikingly, the differences in performance among clusters do not seem to be affected by seniority in the university career as has been stated in the literature (Lee et al., 2005; Quinones et al., 1995; Smith et al., 2018). In our study, this factor was measured from the year of the first publication in order to ascertain whether it could have an influence on the h-index results (Carter et al., 2017), but it turned out to be non-significant. Age of the PIs is another characteristic that does not seem to be different among clusters. Actually, the mean of the age is quite similar among them, 33 for Research-oriented PIs, 36 for Accomplished PIs and 35 for Management-Focused PIs.

5.2 PI profiles and gender

Regarding gender and trying to fulfil the public policy call for attention on gender, studying research profiles (Cunningham et al., 2021; Larsen, 2011; Thursby & Thursby, 2011), our results show significant differences in performance between male and female PIs (Table 2). On average, male PIs have a 23% higher h-index mean than female PIs, which is consistent with other studies (Carter et al., 2017). This might be due to the higher teaching and administrative loads than female PIs have in comparison with male PIs (Acton et al., 2019). Nevertheless, it is noteworthy that the presence of women in the different scientific fields and clusters is markedly uneven. On the one hand, women are grossly overrepresented in the Arts and Humanities scientific field (50%), which is the least prolific one. On the other hand, their presence in the Accomplished PIs cluster is also substantially higher than among Research-oriented PIs or Management-focused PIs (38.2, 23.7 and 27%, respectively) but that does not seem to affect the overall performance of this cluster, which is only slightly lower than that of Research-oriented PIs. This is in line with the findings of Menter (2020) which suggest that research activities benefit from a surplus of female scientists and that the effects are clearer when a certain threshold of gender diversity is exceeded.

All in all, this emphasises the suggestion that there are scientific fields that are predominantly male-dominated, which should be taken into account together with the fact that, even though there has been an increase of female representation in higher education (Mayer & Rathmann, 2018), their presence in senior structural positions in the university, such as PI, is blurred (Cunningham et al., 2021; West et al., 2013). Furthermore, there are gender differences in success obtaining funding which hinders the economic resources they can have for their RTs and their research programmes (Acton et al., 2019).

6 Conclusions

There have been some previous studies already trying to define typologies or categories of PIs (Cunningham, O’Reilly, et al., 2016; McAlpine, 2016; O’Kane et al., 2015). For instance, O’Kane et al. (2015) looked at sorting different PI profiles from a funding perspective and also in terms of their strategic behaviour, i.e., whether they were reactive or proactive. Their study resulted in four different profiles: research adapters, research pursuers, research designers and research supporters. Based on their responsibilities, Cunningham, O’Reilly, et al. (2016) identified ten different roles of the PIs, among which are the following: research strategists, team leaders, knowledge brokers, resources managers and mentors. In this study, we contributed to the existing HC literature, as well as to the PI literature, by studying different PI profiles based on their HC, which had not been previously attempted.

HC has long been praised due to its many advantages and beneficial effects (Martin-Sardesai & Guthrie, 2018; Ployhart, 2015) and, since it is normally regarded as a good thing, it is often understood that the more, the better. However, there is another research approach, more in line with the results of the present study, which takes into account the ‘too-much-of-a-good-thing’ effect, whereby having huge amounts of a valuable resource may not always be synonymous of better outcomes (Cavazotte & Paula, 2021; Foncubierta-Rodríguez et al., 2021; Pierce & Aguinis, 2013).

The findings of our study concur with this latter proposal and support the idea that the optimal HC is dependent on the particular situation and profile and that, therefore, there is no ‘one size fits all’ (Antes et al., 2019b; Garcia-Carbonell et al., 2018; Smith et al., 2018). Additionally, our findings also highlight that it may be erroneous to assume that any given PI, simply because they have reached the PI position, already possesses the necessary HC to perform their duties or upon which they can build further competences that may be crucial to achieve their goals. This latter fact is something that should be considered by public institutions, who may want to include these recommendations as complementary conditions and take them into account when assessing the possibility of funding a certain submitted project.

This is not to say that the decision of funding any given project should be exclusively based on the HC analysis of the PI who is to lead and manage it. Quite the contrary, what is proposed here is that it be regarded as one among many other indicators used in the decision making for funding, much as a certain size or composition of the RT currently are (Cummings & Kiesler, 2005). The PI’s HC is a foundation upon which they can develop further competences that are much needed nowadays, such as industry interaction, commercialisation or technology transfer. Therefore, having an in-depth knowledge of the PI’s HC will be helpful for both universities and research centres in order to identify those PIs who best adapt to them, to their research approach or to their policies on the future lines of research that they will undertake. Even though the literature has not explicitly examined the process whereby PIs learn to carry out their functions, it is suggested that they learn by doing (Cunningham, Mangematin, et al., 2016; O'Kane et al., 2017). In this sense, it will also be beneficial for any training proposals offered to scientists in the PI role in order to be better prepared for the challenges they address, opening the way for them to become more tailor-made proposals rather than generalist suggestions, since it has been demonstrated throughout this analysis that PIs are not homogeneous based on their HC. A final application of the findings of this study is that they may facilitate the self-assessment of PIs, thus enabling them to identify those HC factors –scientific educational training, investigation abilities, self-mastery skills, openness-to-change skills and self-analytical skills– where they can further improve, according to the objective to be achieved. Therefore, it could be considered critical to establish development policies and training practices to improve the PI’s HC factors which remain suboptimal (Youndt & Snell, 2004). However, the aim should not be to have very high levels in each of the HC’s factors, but to enhance those that are considered essential (Pierce & Aguinis, 2013). Indeed, PIs could take advantage of the multidisciplinarity of the RTs that they lead by using the HC that other researchers in their teams can provide in order to complement their own and, thus, they could focus only on the areas that would benefit the most to the whole RT (Hollenbeck et al., 2004; Mathieu et al., 2014). This latter possibility opens up some interesting future lines of research that could provide further insights into RTs, such as the impact that PIs have on RT dynamics, the pattern or profile of HC that the PI should have in order to obtain certain RT outcomes, the mechanisms whereby PIs could transition from one profile to another depending on the kind project or the nature of their RT, or the influence exerted by the PI’s leadership style or their managerial competences. Accordingly, this study may prove helpful in a number of different applications. At the individual level, it could be used by scientists in the PI role–as well as by those who envision becoming one– to assess their strengths and weaknesses in terms of their HC and, thus, pinpoint those areas suitable for enhancement or improvement. In the case of universities, it may assist in appraising the HC that is available within the institution and its potential. For policy makers, this classification may be helpful insofar as it could be considered when assessing the possibility of funding a certain submitted project or even included in the framework conditions for obtaining competitive public funding.

Finally, our study is not without its limitations, which are detailed as follows and should be taken into consideration when drawing conclusions. There is a limitation in relation to self-reporting bias, since all respondent data was collected individually and PIs were self-assessed. In future studies, including at least one survey of the supervisors of the RT PIs should be considered, to avoid the only data available being the responses of PIs themselves. As an additional limitation, the contextual characteristics of the Spanish public research system constrain the possibility of generalising the results to other countries and other nationalities. Thus, for future studies, exploring this analysis in other countries or other cultures is suggested. Lastly, another limitation is related to the measure of the performance of PIs and RTs, since there are certain concerns about the use of the h-index to compare several researchers and different scientific fields. For this reason, it is proposed that other measures be considered in future investigations, or that the h-index be complemented with measures of commercialisation issues and research productivity (Hirsch, 2010).