Mentorship is defined as a dynamic and reciprocal relationship in a work environment between a mentor and a mentee, aimed at promoting the career growth of both (1). Models of career success emphasize mentoring as a key element that impacts satisfaction for both mentor and mentee (2, 3). Notably, these models depict mentoring as an organizational effort, one with structures, processes, and norms as defined and practiced by the employees of the organization.

Research organizations increasingly recognize the importance of mentoring, not only for employees' career development but for organizational success. Mentoring influences employees' career trajectories (4) as well as work satisfaction, which can directly affect their productivity (5). As a result, successful mentoring can increase an organization's achievements. Research mentorship has been found to be pivotal in enhancing the competencies of healthcare professionals (and in turn, their organizations) in evidence-based practice and research methodologies, providing not only professional development but also emotional support that boosts self-confidence and skill acquisition among mentees (6, 7). For mentors, mentoring relationships are shown to increase their career success and satisfaction (8). This high value of mentorship is well recognized by many research funding agencies, exemplified by their requirement that applying organizations clearly demonstrate the ability to provide high-quality mentorship (9).

Organizations trying to improve their mentorship structures, processes, and norms must first accurately assess their current mentoring practices. Specifically, it is important to understand (i) existing mentoring relationships and how they are structured, (ii) factors that impact employees' perceptions of and participation in mentoring, and (iii) employee preferences for mentoring practices, both existent and aspirational. It is also important to recognize that mentorship may be formal or informal, where formal mentorship is a structured relationship in which an organization pairs a mentor and mentee as part of an official mentoring program, while informal mentorship arises naturally through shared interests and personal rapport between a mentor and mentee (10). However, there are no established methods to evaluate mentoring practices within a research organization. Research comes with its own specific nuances of necessary mentoring for academic scholarship that differ from other industries, such as navigating ethics review processes and grant writing, and thus it needs a tailored approach to evaluating its mentoring practices.

Furthermore, although some studies have examined mentorship structures beyond the traditional dyadic pairing between one mentor and one mentee (11, 12), they have conceptualized such non-dyadic structures primarily as representations of how specific mentees and/or mentors are related, rather than as representations of mentoring relationships within a research organization. This latter organization-level representation of employee relationships has been adopted by organizations targeting process improvements, including in the form of assessing employee social networks [i.e., sets of individuals and groups connected by similarities and their interactions with one another (13)] to leverage existing relationships for the improvement efforts (14). Even though it is recognized that mentorship can be viewed through this lens of who is connected to whom in a mentorship social network (15), we are unaware of social network analysis being applied to examine mentorship within research organizations.

Therefore, we developed and deployed a cross-sectional survey to assess mentorship among employees at a two-site federally funded research organization. We then used social network analysis to understand the organization's current mentoring relationships, practices, and attitudes. We outline below the steps we took and our findings, followed by a discussion of the potential utility of our methods for other research organizations seeking to assess and enhance their own mentoring practices.

Of the 193 investigators and staff invited to participate, 120 (62.2%) responded to the survey. Table 1 summarizes respondent characteristics. Fifty-two (43.3%) and 68 (56.7%) respondents were investigators and staff, respectively, and 83 (69.2%) respondents had been at the organization for less than 10 years.

In this project, we examined mentorship experiences of investigators and staff at a federally funded health research center. We used a cross-sectional survey to inquire about their formal and informal mentoring connections, satisfaction with mentorship, and reasons for mentoring-related challenges. Findings indicated that a larger proportion of investigators had at least one mentor, at least one mentor within the Center, and at least one mentee, while staff had a larger average number of mentors outside the Center. Satisfaction with mentorship was higher for investigators than for staff, and an analysis of formal and informal mentorship networks showed that the mean number of formal mentees was significantly different for respondents who (i) had been at the organization for 0–3 years vs. 10–19 years, (ii) were early- or mid-career investigators vs. late-career investigators, and (iii) were mid- or late-career investigators vs. staff. The mean number of informal mentors within the Center was significantly higher for investigators than for staff. Across all of the examined networks (investigators, formal mentorship; staff, formal mentorship; investigators and staff, formal mentorship; and investigators and staff, formal and informal mentorship), we found that, in terms of mentoring relationships, (i) network members are sparsely interconnected relative to the maximum level of interconnectedness possible (low network density), (ii) networks are highly decentralized (low clustering coefficient), (iii) network members are generally connected to a small proportion of other members (low degree centrality), and (iv) network members have large degrees of separation from most other members (low betweenness centrality).

For investigators, a commonly perceived barrier to both receiving and providing mentorship was limited time availability of the individuals who could fulfill the mentor role. Another common barrier to providing mentorship was not being asked to serve as a mentor, suggesting that there may be an opportunity to formally implement a process to assess mentorship needs and availability to identify matches that might not be apparent otherwise. Mentor-mentee matching processes have been gaining increased attention in many non-research oriented entities (18, 19), and research organizations may benefit from adapting such processes for their use.

Although mentoring relationships in research are formalized for research career training initiatives, including fellowships (20, 21), less is known about the extent to which such relationships are common and expected of individuals in other career tracks (e.g., staff) within the research work environment. Common staff-reported barriers to mentoring relationships were not knowing how to start the process of looking for a mentor and not feeling confident to serve as a mentor. These hurdles may be addressable through strategies identified by a 2024 global evidence synthesis aimed at improving research mentorship in resource-limited settings (22). Two of the strategies may be particularly relevant—(i) encouraging peer mentorship, which provides exposure to mentorship without the pressure of guiding a junior colleague's career and (ii) nurturing a culture where mentees are expected to serve as mentors to others, so that such expectations could help build a cascade of mentorship over time. Another promising strategy is a structured mentoring program published in 2023 that specifically focuses on research staff in a university setting (23). The program encompasses mentor training activities, explicit appointment of experienced staff to facilitate creation of mentor-mentee matches, and non-binding permission for the mentor and mentee to each reassess their matches and dissolve the mentoring relationship if the match is not meeting needs. Each of these aspects of the program could contribute to increasing the confidence of staff in getting involved with mentorship.

Our network analysis found that satisfaction with mentoring is correlated with more and closer connections to other closely connected network members, and in the case of the combined formal and informal mentorship network, also with serving as a bridge between other members. Thus, potential approaches to foster additional connections among investigators and staff within an organization can draw on strategies with strong evidence from the fields of organizational management and operations research. These strategies that could contribute to strengthened relationships among co-workers include creating forums for open communication between multiple levels of the organization (24, 25) and opportunities for collaborative problem-solving that call for innovative solutions to be devised as a multi-perspective team (e.g., efforts to improve work processes that bring together colleagues in various organizational roles and get them to together undergo experiential learning) (26). Feeling connected to co-workers is crucial for employees' motivation and well-being (27, 28), which could in turn fuel their willingness to take part in new initiatives for improved mentoring practices stemming from this work. Warranting further investigation is our finding that, for the formal mentorship network, satisfaction with mentoring was negatively correlated with serving as a bridge between other members. This finding may stem from network members’ experiences of high effort associated with formal mentorship (e.g., required mentorship meetings even when there are many other demands on their time) (29, 30).

Although our project's findings suggest that the abovementioned strategies may be beneficial to improve mentorship experiences of investigators and staff, an important limitation of our work is its cross-sectional focus on one research organization. However, our goal in sharing this work is to communicate not only our findings but also the survey-based approach that we took to examine mentorship practices, which can be adapted and applied by other research organizations seeking to assess their own employees' mentorship experiences. Six additional limitations are also worth noting. First, we included data from respondents who did not answer every survey question. Even though there were no noticeable patterns that suggest that the missing data were not at random, there may be characteristics of respondents that we did not collect that impacted which questions were skipped by whom. Second, findings may be subject to researcher bias, as we may have interpreted improvement opportunities to be addressable by strategies related to ones that we know to have been considered for implementation by the participating research organization. To demonstrate the strategies' actual impact, further work is needed to assess change over time resulting from implementing one or more of those strategies. Third, we conducted multiple statistical comparisons across subsets of our data. Although this may increase the likelihood that the statistically significant relationships that we found are due to chance, we did not adjust the significance level accordingly when reporting our findings, given the primary focus of this work to explore potential relationships between variables to generate hypotheses that can be more formally tested in subsequent work. Fourth, although we report on how satisfaction with mentoring is correlated with the network metrics, this project did not include examining how mentoring practices differ between more and less connected network members. Further work to examine such differences may help identify additional recommended mentoring practices for consideration by research organizations. Fifth, our survey methodology did not include formally (i) conducting pilot testing of the developed survey prior to full deployment, (ii) analyzing open-ended responses that were collected as part of the survey, or (iii) being guided by social learning theories or organizational mentorship models in designing the survey [although our work did build on how mentoring relationships have been viewed as social networks (15)]. Formal incorporation of such methodological steps may have enabled a higher-quality survey instrument, as well as a deeper understanding of both the reasons behind our findings and how they explicitly compare to relevant conceptual theories and models. Sixth, because the investigators and staff who were invited to participate in the survey were based on Center-wide email lists at the time of survey administration (rather than on the organization's personnel records), we did not collect data on non-respondents' role at the Center, site, years at the organization, or career stage. This lack of data hinders our ability to accurately gauge the extent to which the different member groups are represented by the survey participants.

This work adds to extant literature by examining mentoring experiences of both investigator and staff in a research organization, highlighting differences between these employee groups in their engagement with mentoring and reported barriers to mentoring involvement. Both our survey-based approach and our findings are especially timely given the increased attention that mentorship is receiving as a critical requirement for successful careers, motivated employees, and workplace productivity. Future work is needed to delve deeper into similar and unique needs that research organizations have for mentoring compared to other types of organizations, as well as to develop a firmer understanding of how best to measure and address equitable improvement of mentoring experiences for employees from diverse backgrounds.