Formal, Informal, and Research Mentorship in STEMM

Only a few of the studies in STEMM have examined informal mentorship, and there appears to be no systematic studies comparing the processes of mentorship and outcomes of formal versus informal mentorship in STEMM.⁴ The research that has been done indicates that formal and informal relationships may offer complementary and overlapping forms of support. One study, for example, found that graduate students receive mentorship support from formal relationships with advisors or dissertation committee members and also from informal relationships such as friends, family, and peers (Griffin et al., 2018). Research has shown, too, that members of UR groups in STEMM often find it more difficult to gain access to the benefits of informal mentoring relationships.

Mentoring relationships with dissertation committee members would be considered “formal,” while mentoring relationships with other faculty and staff who do not have any supervisory or evaluative responsibilities would be considered “informal,” as long as they emerge spontaneously based on personal connections or interactions. A primary research advisor for both undergraduate and graduate research experiences in STEMM will have supervisory and evaluative roles similar to formal mentors, but a research advisor is not always assigned and most mentees have some say in the research advisors they choose. In fact, students and faculty often enter into these relationships based upon mutual interests and respect. Furthermore, not all formal relationships that students have with research advisors, thesis advisors, dissertation committee members, and other higher education professionals are “mentoring” relationships. For such relationships to be mentorship, the mentee must perceive them as providing career and psychosocial support functions and ideally they would be characterized by trust and responsiveness in the form of a working alliance (Montgomery, 2017; Schlosser and Gelso, 2001, 2005). Moreover, the various levels of formality in STEMM mentoring relationships may have different outcomes, a finding that echoes studies outside of STEMM showing that formal and informal mentorship can provide complementary forms of support (Desimone, 2009; Desimone et al., 2014; Erickson et al., 2009).

Given the nature of mentorship in STEMM in higher education, it may be more useful to delineate mentoring relationships by their goals and contexts rather than by their level of formality (NASEM, 2017c; Pfund, 2016). For example, much of the research on mentorship in STEMM examines mentorship in the context of research. While trust and responsiveness may develop over time in a way that shifts the relationship with a research advisor from advising to mentorship, mentees occasionally choose research advisors based on scant information. In addition, not all mentees and research advisors consider their relationship to be mentorship (Hayward et al., 2017; Schlosser and Gelso, 2005). Future research on mentorship in STEMM could clarify the nature and evolution of the mentoring relationships being investigated.

Formal and Informal Mentorship Outside of STEMM

Despite the lack of research on formal and informal mentoring relationships in STEMM contexts, meta-analytic work on mentorship in workplace settings and academic settings in general—including but not limited to STEMM—suggests that mentees view informal relationships as more effective than formal mentoring relationships, but the differences are small in magnitude (Eby et al., 2013). Furthermore, formal mentorship is important when mentees might not otherwise have the same access to informal mentorships, an important consideration for UR students.

Studies from outside STEMM suggest that informal mentorship may be more effective than formal mentorship programs in affecting job satisfaction and compensation outcomes (Ragins and Cotton, 1999). However, the effect sizes, while robust across studies, are small (Eby et al., 2013). Overall, formal, assigned mentorship of employees is not necessarily less effective than informal mentorship, with mentee satisfaction with their mentoring relationship appearing to account for positive outcomes, not the formality of a relationship (Ragins et al., 2000). Further, job satisfaction is a fundamentally different outcome than the development of research and career skills that occurs via mentorship in STEMM training. Thus, findings about formal mentorship in other workplace settings may not translate to research mentorship even though research mentorship shares many characteristics of formal mentorship.

Mentees in informal mentoring relationships within organizational settings report receiving higher levels of career and psychosocial support and having higher-quality relationships than do individuals in formal mentorship programs (Chao et al., 1992; Inzer and Crawford, 2005). Because informal mentoring relationships form through personal and professional respect and admiration between mentor and mentee, and sometimes result in mentors and mentees sharing more identity characteristics with one another, mentees in informal mentoring relationships report being more satisfied with their mentors than mentees in formal relationships. As a result, mentees may develop greater trust with their mentor in an informal relationship and identify with them to a greater extent than mentees in formal relationships, thereby reporting a higher-quality relationship (Hadjioannou et al., 2007; Inzer and Crawford, 2005; Kram, 1985a; Nemanick, 2000; Ragins, 1997). The extent to which these findings from organizational behavior research translate to mentees in STEMM and academic contexts has yet to be determined. Studies of students in STEMM do, nonetheless, indicate that trust and identification may be particularly important for mentees from UR groups (Carroll and Barnes, 2015; Denson et al., 2015; Muller et al., 2012). As such, positive psychosocial support activities found in informal mentoring relationships (Ragins et al., 2000), such as social interaction, role modeling, and friendship, may facilitate trust and identification for UR mentees.

Mentorship in Triads

Sociologists have long recognized that triads—a group of three people—are the smallest non-dyadic social group that has the potential to experience the full range of social relations, both positive and negative (Caplow, 1956; Krackhardt and Handcock, 2007; Simmel, 1964). Qualitative accounts of mentorship triads have shown that mentees gain distinct forms of career and psychosocial support from different mentors (Dolan and Johnson, 2010; Griffin et al., 2018). When three people are involved in a relationship, scenarios that are not observable in dyads can arise, such as competition and coalitions (Burt, 2009; Simmel, 1964).¹² Possible types of mentoring triads are discussed in Box 4-3.

BOX 4-3

Mentoring Triads.

A handful of studies have examined one type of mentorship triad that is commonplace in STEMM at research universities: an undergraduate researcher, the graduate student or postdoctoral associate who provides day-to-day guidance on research, and the faculty member who is head of the research group. One survey of approximately 800 undergraduate life science researchers found that undergraduates reported experiencing a range of triadic mentorship structures (Aikens et al., 2016). A second study examined the two most common of these mentoring triads: open triads with undergraduate-postgraduate and postgraduate-faculty interactions but no direct interactions between the undergraduate student and faculty member, and closed triads, in which there are interactions among all three members (Aikens et al., 2017). This study found that students with different identity characteristics, such as gender, race, or ethnicity, experienced different mentorship configurations, which partially explained differences in their outcomes. For example, men were significantly more likely than women and UR students were significantly more likely than White or Asian students to report being in closed triads.¹³ For women, being part of an open triad mentorship structure appeared to have a negative effect on the development of their scientific identity, intentions to pursue a science, technology, engineering, and mathematics (STEM) Ph.D., and scholarly productivity. For UR students, a closed triad mentorship structure appeared to have a positive effect on the development of their scientific identity, intentions to pursue a STEM Ph.D., and scholarly productivity. Asian students, meanwhile, reported lower scientific identity and were less likely to intend to pursue a STEM Ph.D., both of which were unrelated to their open triad mentorship structure. Furthermore, undergraduates in dyads with faculty mentors reported similar outcomes as undergraduates in closed triads but superior outcomes to undergraduates in open triads (Joshi et al., 2019). In all of these studies, the effects of mentorship structure on undergraduate outcomes were significant but small.

Another study of a largely triadic system looked at the experiences and growth of biomedical Ph.D. programs engaged in a unique partnership between the intramural program at the National Institutes of Health (NIH) and Ph.D.-granting universities. In that structure, Ph.D. students were co-mentored by a faculty member at NIH and at a participating university. One finding from this study was that the co-mentored students were able to develop more quickly, acquired more complex research management skills, and became more independent (McGee and DeLong, 2007). Another finding was that mentors behaved differently with these students, allowing them more autonomy and working closely with co-mentors, largely for the benefit of the student. There was no indication mentors changed their mentorship styles for other Ph.D. students in their research groups.

Collective or Group Mentorship

Group-based mentorship is distinctive because it involves the collective development and cultivation of communities of mentors, including multiple mentees who themselves serve as mentors for each other (Bradley et al., 2017; Comer et al., 2017; de Janasz and Sullivan, 2004; Dodson et al., 2009; Eby, 1997; Huizing, 2012; Ireland et al., 2018; Ko et al., 2014; Kroll, 2016; Martinez et al., 2015; Thomas and Hollenshead, 2001; Varkey et al., 2012). Such efforts can be accomplished in person (Allen and Joseph, 2018; Comer et al., 2017; Dodson et al., 2009; Eby, 1997; Ireland et al., 2018; Ko et al., 2014; Martinez et al., 2015; Thomas and Hollenshead, 2001; Varkey et al., 2012), online through social media and other digital platforms (Columbaro, 2009; Gareis and Nussbaum-Beach, 2007; Gregg et al., 2016; Wolfe and Gregg, 2015), and using mixed online and in-person approaches (Martinez et al., 2015). Mentorship groups can span levels of expertise and cross disciplines (Dodson et al., 2009; Horner-Devine et al., 2018; Reeves et al., 2019; Thomas et al., 2014). Group mentorship can be thought of as an application of the community-of-practice concept, which is defined as a group with a mutual focus that improves through regular interactions (Wenger et al., 2002). Although there has been little direct investigation of group mentorship in STEMM as it relates to career or educational stage, group mentorship involving individuals at different educational stages has the potential to provide developmentally adapted mentorship (Dodson et al., 2009; Montgomery et al., 2014).

Mentorship groups can be affinity based, meaning that the group comes together around a common identity, such as African American women in STEMM.¹⁴ Affinity-based mentorship groups have the potential to create a microclimate that provides critical support for individuals experiencing isolation and invisibility due to their identities (Comer et al., 2017; Martinez et al., 2015; Smith et al., 2014; Tuitt, 2010). A handful of studies of particular mentorship groups have shown that affinity-based mentorship groups have been used successfully to support group members, including individuals from UR groups in STEMM who are interested in advanced degrees (Allen and Joseph, 2018; Dodson et al., 2009). For example, group mentorship among women scholars has been shown to help participants build skills, self-efficacy, and career satisfaction (Martinez et al., 2015; Varkey et al., 2012). Group mentorship among African American male undergraduate and graduate students has also been shown to help build skills, improve academic success and persistence in research, and value the communal goals of the group (Dodson et al., 2009). Another study highlights the value of cohorts for providing peer support and socialization opportunities among deaf mentees that would otherwise be absent in a strictly dyadic mentoring relationship (Majocha et al., 2018).

In particular, collective mentorship in affinity groups can produce guidance that disrupts negative influences existing in historically White spaces (Allen and Joseph, 2018) and traditionally male spaces (Thomas et al., 2014), as well as leading to advocacy (Thomas et al., 2014). Additionally, group mentorship may confer benefits for women and UR students as a means of addressing the scarcity of traditional “senior” mentors.¹⁵ Several features have been proposed for ensuring the effectiveness of group mentorship, specifically that groups intentionally focus on equitably supporting strengths (Allen and Joseph, 2018; Kelly and McCann, 2014), providing social support (Mondisa and McComb, 2015), and prioritizing self-definition and self-valuation (Dodson et al., 2009).¹⁶

Peer and Near-Peer Mentorship

Peer mentorship groups, in particular, may promote collaboration, provide mentees with psychosocial and career—specifically academic—support, increase dedication to a STEM major, and increase retention (Holland et al., 2012; Tenenbaum et al., 2014; Zaniewski and Reinholz, 2016). Peer or near-peer groups may also serve to enhance self-efficacy and diminish feelings of isolation (Driscoll et al., 2009; Thomas et al., 2014). This outcome is supported in part through shifting the focus from mentor-centered power hierarchies to mentee-centered peer sharing and support (Bynum, 2015; McDaugall and Beattie, 1997; Wilson et al., 2012).

In the absence of available mentors, or to supplement effective mentors, UR STEMM students are likely to mentor each other or form mentoring groups with peers who are at approximately the same stage of career development, an approach referred to as near-peer or step-ahead mentoring.¹⁷ One comparative study that examined traditional, peer, and step-ahead mentoring relationships in the organizational setting found that employees in traditional mentoring relationships had the highest job satisfaction. From a theoretical perspective, this may result from the fact that compared with peers and step-ahead colleagues, traditional mentors have greater access to power and influence, which translates into better career outcomes (Ensher et al., 2001). Nonetheless, peer and step-ahead mentoring groups can be an important approach for addressing the lack of STEMM UR faculty.

Because peer mentors in these types of mentoring relationships share an important identity (e.g., being a UR doctoral STEM student), they may benefit from having greater levels of interpersonal comfort. A 2005 study of health care and technology employees examined the role of interpersonal comfort in mentoring situations (Allen et al., 2005). Institutions can integrate near-peer mentoring into their programs (see Box 4-4). Furthermore, whenever a mentor is in a more advanced position than the mentee, the mentee can learn vicariously from the mentor (Williams et al., 2016a).

BOX 4-4

Near-Peer Mentoring in the Fisk-Vanderbilt Master's-to-PhD Bridge Program.

Network Mentorship

Mentorship networks—the constellations of mentoring relationships and resources that a mentee taps for support—have gained increasing recognition both within and outside of STEMM (Higgins and Kram, 2001; Long et al., 2014; Sorcinelli and Yun, 2007; van Emmerik, 2004). While much of the research on this mentee-centered structure has occurred outside of undergraduate and graduate student career stages or outside of STEMM, they provide some insights that are likely to be applicable for STEMM mentorship. For example, among UR faculty members, a mentorship network can offer less hierarchical, more relational, and more reciprocal mentorship (Yun et al., 2016). Networks can also serve as critical for the provision of support, affirmative spaces, and accountability (Hernandez et al., 2017). One means of ensuring equitable access to the elements of mentorship networks involves the intentionality and accountability of institutional leaders (Beach et al., 2016; Ko et al., 2014; Lloyd-Jones, 2014; Montgomery, 2018a; Turner et al., 2011; Whittaker et al., 2015).

Longitudinal studies of mentorship outside of STEMM have found improved long-term outcomes for mentees based on engagement with mentorship networks, as opposed to the effective support of short-term goals observed in traditional hierarchical dyads (Higgins and Kram, 2001; Higgins and Thomas, 2001). Specific tools to promote building and cultivating mentorship networks intentionally are emerging (Montgomery, 2017). The formal inclusion of a network mentorship into STEMM programs may have challenges in coordinating accountability or other aspects of mentorship.

Pre-Medical Mentoring and Advising

At the undergraduate level, most, but by no means all, pre-medical students are also STEM students and would therefore benefit from the same mentorship as other nonpre-medical students. For those pre-medical students who participate in undergraduate research opportunities, mentorship considerations in the context of research would be similar to or the same as for other STEM undergraduates. Similar to other STEM fields, the transition in medical education from undergraduate to further study such as medical school or an M.D.-Ph.D. program is a particularly vulnerable period and can highly depend on the competencies of the mentors that are involved. Crucially, seeking input from multiple mentors can help to augment the influence of any single mentor on a mentee's potential next steps.

Differences in the mentoring and advising processes arise for students seeking admission to medical school or graduate school. For medicine, there is a structured process with a single gate-keeping admissions system administered by the Association of American Medical Colleges and the American Association of Colleges of Osteopathic Medicine, and a strong emphasis on academic performance and scores on the Medical College Admission Test. The requirements for entry into medical school and progression through training are codified and made highly visible.²² A similarly structured process exists for applying to and being admitted to dental schools.²³ Colleges or universities with a significant number of pre-medical students usually offer formal advising systems, supported by the National Association of Advisors for the Health Professions,²⁴ with those advisors having their own strong professional identities. Thus, advising by dedicated professionals, rather than relying on the variable knowledge of faculty mentors, plays a prominent role in helping undergraduates navigate this system.

A review of several university websites indicates an intent to design mentoring programs specifically for pre-medical students. However, the committee could not find reports of studies of any particular style or design of mentorship in this context. There are also postbaccalaureate programs established for students who cannot gain admission to medical school after completing their baccalaureate degree that provide varying levels and types of classes for credit, as well as mentoring experiences in program contexts.²⁵

Medical School

The role of mentorship during medical school is peripheral to the primary education and training designs, which makes it distinctive from graduate education and training. The intense focus on mastery of a large body of core knowledge during the first two preclinical years has kept medical schools focused primarily on advances in teaching, learning systems, and pedagogy. The required passage of standardized exams for licensure and residency positions also precludes substantial reliance on mentorship as a system to prepare for these standardized expectations. However, functions of mentorship, such as advising, do play a role in medical school training.

One of the most systematic and relevant reviews of mentorship in medical school covered PubMed-indexed publications from 2000 to 2008 (Frei et al., 2010). Of 438 publications identified worldwide, only 25 met their selection criteria related to mentorship purpose and design and mentorship being provided by an experienced medical professional. Of these, 14 reported on a mentorship program with some survey or other measure of mentored experiences. Eleven papers reported on the value of mentorship for medical students in general. Overall, the authors found that mentorship in medical school, when it occurred, was designed to “provide career counseling, develop professionalism, increase students' interest in research, and support them in their personal growth” (p. 1). However, their review did lead them to reach several conclusions for students interested in careers in academia: a traditional dyadic mentorship configuration proves most effective,²⁶ the mentor must serve as a role model both professionally and personally, and career counseling of students by mentors results in students making earlier decisions about specialty and career.

Research has also examined the effects of mentorship in medical school. One study of mentorship groups in medical school designed to increase students' reflectivity for professional development found mixed results,²⁷ with some students reporting positive reactions to the groups and others reporting negative reactions to them. Many students did not feel they should be forced to share their personal reflections in groups of other students and faculty (Lutz et al., 2017). Another study looked at mentorship in 14 new medical schools at various stages of achieving accreditation to admit their first class, considering the possibility that new schools might put more weight on mentorship than established ones (Fornari et al., 2014). However, the constructs and roles of mentorship programs in these schools were just as varied as in established schools, constrained by the same limitations of time and resources. Furthermore, mentorship in these new programs was often difficult to distinguish from advising. Finally, a recent study of a supplemental training to a mentored research program for UR medical students indicated an increase in their academic self-efficacy and interest in an academic career (Fernandez et al., 2019). The results of this study support the potential for programmatic interventions through mentored research to increase diversity in academic medicine.²⁸ There are also some examples of the use of online forums and social media to develop supportive, mentorship networks, especially for women and UR physicians.²⁹

Overall, mentoring does play an important role in medical training and development, but it is less focused on actual development of skills and knowledge, as in STEM, and more focused on professional development and role modeling. In addition, it often takes on more of an advising context as opposed to long-lasting individualized mentoring relationships.

Federal Programs

The largest number of programs has been implemented through federal funding agencies such as NIH and NSF.³⁴ Some of the largest and longest standing have come from the National Institute of General Medical Sciences (NIGMS), including the following:

• Maximizing Access to Research Careers Undergraduate Student Training in Academic Research (MARC U-STAR)
• Research Training Initiative for Student Enhancement
• Initiative for Maximizing Student Development
• Postbaccalaureate Research Education Program (PREP)

Examples of NSF-funded programs with a mentorship component include the following:

• Alliances for Graduate Education and the Professoriate (AGEP)³⁵
• Historically Black Colleges and Universities–Undergraduate Program (HBCU-UP)³⁶
• Louis Stokes Alliance for Minority Participation (LSAMP)
• Research Experience and Mentoring (REM)
• Research Experiences for Undergraduates (REU)³⁷
• Significant Opportunities in Atmospheric Research and Science (SOARS)
• Tribal Colleges and Universities Program (TCUP)³⁸

For most of these programs, “success” has been based almost exclusively on a particular outcome variable, most commonly the number or fraction of students who stay in STEMM or progress to the next career stage in STEMM. For example, an educational outcomes study of the NIGMS MARC program, based on students who participated between 1986 and 2013, showed that about 20 percent had gone on to earn a Ph.D. For a subset of most recent participants, 29 percent earned a Ph.D. or M.D.-Ph.D., 17 percent earned a clinical/professional degree including an M.D. or D.O., and 14 percent earned a master's degree (Hall et al., 2016). A primary element of the MARC program is 2 years of mentored research, largely with an individual faculty member, but sometimes with several faculty members. However, systematic attempts to disaggregate the effects of the mentoring relationship(s) on persistence to advanced STEMM degrees remain an open opportunity for scholarship.

A similar study has been conducted for the educational outcomes of the NIGMS PREP program, which is a nondegree, research-intensive program supplemented with a variety of programmatic elements (Hall et al., 2015). For national PREP scholar cohorts between 2001 and 2014, 65 percent matriculated into Ph.D. programs. In the earliest cohorts, 63 percent graduated with a Ph.D., while many in later cohorts were still in training. Of Ph.D. completers, 50 percent were still in postdoctoral training and the rest were largely engaged in research. For PREP, an independent, qualitative study was conducted of a subset drawn from several PREPs across the United States (Gazley et al., 2014). The first analysis studied individuals just as they started PREP, while the second analysis studied them 1 to 2 years later at the end of PREP (Gazley et al., 2014). This analysis showed that trainees entered PREP for reasons that could be clearly linked to either late-developing or nascent identities as scientists and a need to acquire the cultural capital that could enable them to effectively transition into and succeed in a Ph.D. program (Gazley et al., 2014). The subsequent analysis revealed how the time in PREP enabled the participants to grow, enact, and practice their scientific identity. Using in-depth interviews, this growth could be dissected to show separate contributions by their mentored experiences and relationships, as well as conscious design elements of the program (Remich et al., 2016).

The Academy for Future Science Faculty, a program of the NIH Director's Pathfinder Award to Promote Diversity in the Scientific Workforce, uses a group coaching and mentorship approach.³⁹ This approach was tested with both early- and late-stage biomedical Ph.D. students using a randomized controlled trial design. Cohorts of 100 early-stage and 60 late-stage Ph.D. students from around the United States (with equal numbers of control students) were divided into groups of 10 and matched with an experienced faculty member for mentorship that combined annual in-person meetings and virtual meetings in between the annual meetings. Groups were purposely constructed to include equal numbers of men and women and close to equal membership by race and ethnicity. Groups did not contain students from the same Ph.D. programs, coaches were not from the schools of students in their groups, and faculty and students in each group were not from a student's home institution. Group mentorship was designed to complement and/or fill in for whatever other mentorship students were receiving during their Ph.D. program. The program collects annual interview and survey data.

Several reports on the impacts of being part of the Academy have revealed how students benefit as much from peers as from faculty mentors in this constructed group environment (Thakore et al., 2014; Williams et al., 2016a, 2016b). Several students indicated they would have dropped out of graduate school had it not been for their peers or coach (Williams et al., 2016a, 2016b). Benefits align with the psychosocial and career support functions of effective mentorship, but also reveal vicarious learning that does not typically get assessed in mentorship environments. Ongoing analysis reveals the effects of these groups last well past when they have continued meeting, and further analysis can determine if this group mentorship structure influences career outcomes.

In 2014, NIH created the Diversity Program Consortium,⁴⁰ comprising 10 multi-institutional sites around the country focusing on increasing the number of UR undergraduates who persist into STEM graduate programs as well as centralizing resources to dramatically increase the quality and quantity of mentorship and professional development coaching that is available (Hurtado, 2015). The NRMN serves as the element of the consortium focusing on mentorship and professional development.⁴¹ Since NRMN's inception, more than 12,000 individuals have joined the network in various capacities as mentees and mentors. Additionally, more than 540 postdoctoral researchers and early-career faculty have participated in one of four grant-writing coaching group models in which feedback and coaching is provided for 4 to 12 months, throughout the time of writing a research or training proposal. Studies of the effect of these varied mentoring and coaching experiences are under way (Jones et al., 2017; Sorkness et al., 2017).

The LSAMP program aims to increase the participation of individuals from underrepresented racial and ethnic groups in STEM by increasing “the quality and quantity of students of color who earn bachelor's degrees in STEM fields and pursue STEM-related graduate studies in order to increase the number of underrepresented minorities in the STEM workforce” (NASEM, 2019, p. 136). A quantitative assessment conducted by the Urban Institute in 2006 revealed, of 27 alliances surveyed, 82 percent offered mentorship as part of their student professional development and 60 percent cited mentoring as one of their top five components.⁴² The academic success of LSAMP-supported participants was significantly greater than those of non-LSAMP awardees, when measured by grade point average. In addition, LSAMP-supported students were more likely to matriculate into the STEM workforce after undergraduate training or enroll in STEM-discipline graduate programs than their non-LSAMP counterparts (Clewell et al., 2006).⁴³

The REM program aims to provide high school students, STEM teachers, and undergraduate STEM students and faculty with a particular emphasis on UR students and veterans enrolled in postsecondary education, with mentored, hands-on research experiences that may enhance career and academic outcomes among participants who might not otherwise have engaged in a research project. According to NSF, effective REM programs have many of the following characteristics:

• Mentorship training for researchers and affiliated graduate students or postdoctoral researchers
• Well-designed, introductory training for research participants
• Six to 10 weeks of full-time summer research
• Continued mentorship of research participants throughout the academic year
• Participation of research participants in research team meetings and topic-related conferences or workshops
• Guidance for research participants in coauthoring publications and/or posters⁴⁴

An assessment of one REM program at City College of New York concluded that the program provided a “novel and effective platform to allow more underrepresented students in the greater NYC area to participate in our multidisciplinary research” (Zhu et al., 2016). Another evaluation of a REM program at Clemson University found that students felt they were more prepared to conduct research and had acquired better research skills after participating in the program (McCave et al., 2014).

The SOARS program is a 10-week summer research internship, built around research, mentoring, and community, hosted by the National Center for Atmospheric Research or at laboratories of other SOARS sponsors.⁴⁵ SOARS seeks to involve students from groups that are historically underrepresented in the sciences in atmospheric research by offering comprehensive financial support for summer research, as well as undergraduate and graduate school funding. Each protégé, as the participants are called, has a research, writing, computing, and peer mentor, as well as a coach (Windham et al., 2004). Over the more than 20 years that the program has been running, 90 percent of SOARS protégés have gone on to graduate school, and many have entered the workforce with a master's degree or gone on to complete a Ph.D. (Haacker, 2015).

Several evaluations of SOARS have noted that the program's success is exemplified by the success and quality of its protégés and alumni (Melton et al., 2005; Pandya et al., 2007; Windham et al., 2004). Some 80 percent of the protégés, for example, participate in SOARS for 2 years or more, and between 1996 and 2003, none of the diverse group of participants failed to complete an undergraduate STEM degree. From 1998 to 2007, SOARS protégés presented more than 113 posters and 65 oral papers at scientific conferences, and 12 protégé-coauthored papers resulting from summer research projects were published in peer-reviewed journals. As the authors of one evaluation noted, “Quantitative measures (both those SOARS has been tracking over the years and those we contribute in this report) indicate successes in protégé confidence and comfort interacting with scientists and other professionals, enhanced research, writing, and presentation skills, and sense of belonging among a community of peers” (Melton et al., 2005).

Institutional Programs

One well-known institutionally based STEM intervention program is the Meyerhoff Scholars Program (MSP), initiated in 1988 at the University of Maryland, Baltimore County. Since its inception, it has been supported by an array of private, federal, and institutional resources. Extensive research over the years has shown how MSP has a strong beneficial effect on progression and completion of a Ph.D. for UR students in STEM (Maton et al., 2000, 2012, 2016; Stolle-McAllister et al., 2011). This conclusion is supported by comparing the trajectories and outcomes of participating students with a group of students who were admitted to MSP but chose to go elsewhere. By following the progress of both groups, this naturalistic experiment allows for some level of control of confounding factors, such as self-selection into the program by participants. Despite its value, this kind of comparison is uncommon in the study of academic degrees or programming. Program elements and perceived value by students are closely tied to a sense of community, science identity, and research self-efficacy (Maton et al., 2016).

Preliminary data on efforts to expand MSP to two other campuses indicate some success (Mervis, 2019; Santo Domingo et al., 2019). The University of North Carolina at Chapel Hill and Pennsylvania State University, University Park, have established the Millennium Scholars (MLN) Program and the Chancellor's Science Scholars (CSS) Program, respectively, with both programs designed to emulate elements of MSP. Data on student outcomes from the first four MLN and CSS cohorts, as compared with institutionally identified nonparticipating students, show improvements in STEM retention and grade point average. However, MLN and CSS required “sufficient and sustained administrative support,” full-time dedicated program staff, and participation and support from faculty leadership across campus, among other factors.

Another extensively studied program is the Biology Undergraduate Scholars Program at the University of California, Davis (Barlow and Villarejo, 2004; Gast et al., 2010; Sweeney and Villarejo, 2013). By constructing comparison groups at the university, program investigators were able to show the influence of being in the program on science grades and persistence into STEMM graduate degrees. Their data suggested that a good portion of the effect resulted from mentored research, but they could not completely separate mentorship from the research experience or other program elements.

Probably the most extensive national study of program-based students is The Science Study based at California State University San Marcos (Estrada et al., 2011, 2018; Hernandez et al., 2016, 2018; Schultz et al., 2011). This study used a robust propensity score matched comparison cohort of students at the same institution, and a recruitment strategy and longitudinal design that achieved an exceptionally high retention rate (Hernandez et al., 2013). The study was able to demonstrate increased persistence by UR students into and within STEM Ph.D. programs, as well as the critical importance of developing a science identity in the decision to pursue and persist in the Ph.D.

Mentorship has been shown to be a strategy to promote student success at minority-serving institutions, particularly through the intentional creation of a culture to promote the success and well-being of its students. For example, the Peer Mentoring Program at Xavier University of Louisiana matches incoming freshmen with upperclassmen and student mentors and mentees with faculty advisors (NASEM, 2019). Similarly, Xavier University of Louisiana also houses the Center for the Advancement of Teaching and Faculty Development, which provides faculty mentorship education by hosting faculty workshops geared toward philosophies of mentoring, communication between mentors and mentees, setting goals and expectations, the concepts of stereotype threat and implicit bias,⁴⁶ successfully identifying issues and resolutions, and effective mentorship practices (NASEM, 2019). These programs, however, have not been formally evaluated.

One conceptual example of an institutional program that might encourage mentorship was proposed during the first of three workshops the committee held to gather input for this report. The development of a supportive or “holding environment” might provide interesting scaffolding on which programs could be built (see Box 4-5).

BOX 4-5

The Value of a Holding Environment.

Conference-Based Programs

Several professional development and affinity-based STEM programs are rooted in providing validation of students' diverse identities for the purpose of recruiting and retaining them through degree completion and into the workforce. Research on these initiatives is limited, however, and only a few long-standing national and localized programs have cited mentorship as a strong component of their programming for undergraduate or graduate students.⁴⁷ While mentorship at conferences often takes the form of single, one-off events, the conferences highlighted here were described in the committee's listening sessions and in other forums as developing cultures of mentorship for the duration of the conference. In addition, they often create ongoing structures to facilitate communication as well as career and psychosocial support in between conferences.

The Southern Regional Education Board hosts the Institute on Teaching and Mentoring (ITM), one of the largest gatherings of minority doctoral students in the country. The program is more than 25 years old, and the annual gathering brings together doctoral students with faculty and mentors to receive training and professional development. The primary goal of ITM is to provide scholars with information, knowledge, and skills necessary to navigate graduate school and ultimately become professors as a means of preparing the scholars for mentorship and teaching in the academy. An evaluation of ITM for 2011–2016 found better alignment of the conference's stated aims with outcomes for both women alumni and black alumni, namely, employment status compared to data from the nationwide NSF Survey of Doctorate Recipients. ITM also encourages the scholars to become mentors and develop mentorship programs themselves (SREB, 2018a, 2018b, 2018c).

SACNAS holds an annual national conference that focuses on cultural capital and validates the identities of UR students in STEM with activities such as a powwow, conversations in Spanish, and a Native American blessing. The conference's mentorship components include times when faculty from any university can meet to talk about research and professional development. SACNAS includes orientation sessions for students and mentors, features mentorship workshops, and places specific emphasis on the importance of mentors for scientists of color (Arnette, 2003; Chemers et al., 2011; Collins, 2002; Hurle, 2003).

Similarly, the National Society of Black Engineers (NSBE) (Dickerson and Zephirin, 2017; Ross and McGrade, 2016) and the Annual Biomedical Research Conference for Minority Students (Butts et al., 2016; Casad et al., 2016; Hulede, 2018) bring young and aspiring UR scientists and engineers together, providing both mentors and resources of a mentorship network design. For example, NSBE has implemented an intentional, nationwide, multilayer structure based on ecological systems theory to form a cascade mentoring structure.⁴⁸ Participants at the committee's listening session noted that this type of mentoring was facilitated between the national and regional NSBE meetings through the local chapter structure.

The mission of the Earth Science Women's Network (ESWN), a grassroots, nonprofit, member-driven organization formed in 2002, is to “promote career development, build community, provide opportunities for informal mentoring and support and facilitate professional collaborations” and to “build a resilient community that lifts all women and moves the geosciences forward.”⁴⁹ ESWN offers online mentoring and professional development workshops and networking opportunities at major conferences worldwide and at ESWN-hosted workshops throughout the United States. For example, in October 2018, ESWN hosted two workshops in Boulder, Colorado, on building leadership skills for success in the scientific workforce. Other workshop topics include defining one's research identity, mentor mapping, doing fieldwork, and workforce climate training. Evaluations of ESWN's programming (Adams et al., 2016; Archie and Laursen, 2013; Kogan and Laursen, 2011) show that members report “gains in areas that are often considered barriers to career advancement, including recognition that they are not alone, new understanding of obstacles faced by women in science, and access to professional resources” (Adams et al., 2016).

Capacity-Focused Programs

Several programs have focused on institutional change or even statewide change. The NSF Alliances for Graduate Education and the Professoriate (AGEP) (George et al., 2010), for example, has focused on expanding institutional capacity and approaches to promote STEM diversity.⁵⁰ One study of the state of Maryland's AGEP program, known as PROMISE, used a case study approach to examine how STEM graduate students of color gained access to support through mentorship and developmental networks, including how the PROMISE program influenced their experience of being mentored (Griffin et al., 2018). The investigators interviewed 16 graduate students spanning STEM disciplines, institution types, and years in their graduate program about the relationships that they found important to their development and learning and how the PROMISE program related to these relationships. All participants in this study reported receiving support from multiple sources, including their research advisors, peers, program administrators, professionals outside of their institution, and friends and family. The students also reported that different individuals offered different forms of support, from career guidance to research advice to emotional support. Participants who had less favorable relationships with their research advisors reported drawing more heavily on support from other sources. Finally, the participants felt that the PROMISE program helped them to cultivate and maintain their developmental networks, providing them access to more potential mentors. This research illustrates the value of having access to multiple mentors and how a program can facilitate access to developmental networks (Griffin et al., 2018; Tull et al., 2017).