Can Coaching Bridge the Gap for Incoming Latinx Graduate Students?

Abstract

Attrition rates in the first two years of graduate school are high and ~50% higher for underrepresented students. Here we evaluate an online group coaching intervention using a controlled and an experimental group to determine its impact on skills that are likely predictors of student success, namely the Hallmark of Success indices developed by the National Institutes of Health (NIH). Our intervention includes eight online coaching sessions (four prior to the first year of graduate school and four during the first year of graduate school). Coaching group sessions, led by certified coaches, address Resilience, Risk and Protective Factors, Accountability, Leadership, Teamwork, Professional Networking, and Self-Assessment. We evaluated the change in Hallmark of Success indices with a pre- and post-questionnaire of 44 items used to assess students’ self-evaluation in nine areas. Students in the experimental coaching group showed a significant increase in the Hallmark of Success indices in seven out of nine areas of student performance between the pre- and post-questionnaires compared to the null model of no changes between time periods. In addition, in all of the nine indices of success, we noted significant differences between the pre- and post-surveys for the experimental group. Our experimental design shows that our cost-effective coaching model improves student performance as perceived by the students.

Introduction

Despite significant advances in diversifying the cultural background of the scientific workforce (Johnson et al., 2021; Norris et al., 2020), the proportion of PhD degrees awarded to Latinxs is still 7%. Only 48% of those Latinxs who enroll in a doctorate program complete the degree in 7 years (NSF, 2018; Sowell et al., 2015). Institutional inequalities, ethnocentrisms, unequal access to research opportunities, disconnection with trainees’ cultural backgrounds, and ignoring the holistic and personal needs of trainees are still barriers to supporting all underrepresented students (URP) in biomedical research careers (Carlone & Johnson, 2007; McLachlan, 2012; Pappamihiel & Moreno, 2011).

The rate of attrition to complete a doctoral degree is at least 50% for URP students (Wilson et al., 2018). At the doctoral level, the median time to attrition has been identified as 23 months, with particular challenges in the first year (Kerby, 2015), suggesting interventions would be most effective at this stage (Ampaw & Jaeger, 2012; Sowell et al., 2015). Studies on the well-being of students point to isolation, deficient mentoring, cultural invalidation, and lack of life strategies for overall wellness as key challenges, particularly during the first year of graduate school (Bettinger & Baker, 2014; Deiorio et al., 2016; Williams et al., 2017). Flaster and Gonzalez (2019) found that mental health, self-efficacy, intrinsic motivation, disciplinary identity, and intention to pursue tenure-track positions all decrease after the first year of graduate school. Most of these soft skills are part of the “Hallmarks of Success” indices of the Diversity Program Consortium of the National Institutes of Health, which are identified as key elements for the development of a successful biomedical career (NIH, 2023).

Considering these challenges begs the question of which training strategies would work best to ensure students learn the skills needed to succeed in biomedical careers. Gaining undergraduate research experience (UGR) has demonstrated to be particularly beneficial, especially for underrepresented students (Collins et al., 2017; Haeger and Fresquez, 2016; Schultz et al., 2011). However, success in science requires more than research skills and not all of the aforementioned skills can be obtained with UGR.

Could a cost-effective online coaching program improve those hallmarks for success for Latinx students that have undergraduate research experience (UGR)? This is our research question. Evidence from the literature suggests that coaching can provide students with some of the aforementioned critical life skills to succeed academically by bolstering their potential, resilience, support networks, and sense of self-efficacy, as shown by nationwide research in the US (Bettinger & Baker, 2014; Flaster & Gonzalez, 2019; Bandura, 1997; Bandura & Locke, 2003). Other researchers noted similar findings among members associated with the Society for the Advancement of Chicanos and Native Americans in Science (Chemers et al., 2011) and in a case study conducted at the University of Texas at El Paso (Collins et al., 2017). Coaching can also provide students with a strong supportive peer learning community and professional networks (Montgomery, 2017). Such benefits have been documented for Puerto Rican students in science in the US and in Puerto Rico (Guerrero-Medina et al., 2013). Other documented benefits include overcoming isolation by increasing students’ sense of belonging within the research community in nationwide studies focused on minorities in US professional associations (Chemers et al., 2011); at the university level in 50 universities in the US (Estrada et al., 2011), and in a case study in California State University in San Marcos (Schultz et al., 2011). Coaching has also been found to help students achieve personal goals by promoting their self-care and overall wellness (Schmidt & Hansson, 2018).

Some institutions in the US offer personal life counseling to students, often individually, to help them set goals and develop their scientific identity and overall personal wellness (Deiorio et al., 2016; Pechac & Slantcheva-Durst, 2021; Tull et al., 2012; Williams et al., 2017). However, those one-to-one approaches can be limited in scope, costly, and counselors are moreover rarely certified life skills coaches. With support from the Diversity Program Consortium of the National Institutes of Health (NIH-DPC) program, we developed an innovative group coaching model that was: 1) led by institutional counselors certified by the International Coach Federation as Professional Coaches; 2) conducted in groups of up to approximately 10 students for a broader cost-effective reach; 3) enhanced by advanced senior graduate and postdoc students to encourage peer mentoring and learning communities, and 4) offered according to a structured curriculum and culturally appropriate topic guide. Online coaching activities dealt with cultural barriers, family pressures, building scientific community and networks, mentoring, improving time management, and study skills. We evaluate the impact of our group coaching sessions on undergraduate students transitioning to their first year of graduate school who are pursuing degrees in social or natural sciences, which are areas where Hispanics have even lower representation at the nationwide level. For instance, Hispanics had 8.8% of the total doctoral degrees conferred in science, technology, engineering, and mathematics in 2021 in the US, which is lower than the 9.7% of doctoral degrees conferred to Hispanics in all disciplines (NCES, 2024).

Because previous research has shown that undergraduate research experience is the most significant predictor for successfully completing undergraduate studies (Caraballo-Cueto et al., 2023; Daniels et al., 2016), we randomly assigned students to a control and an experimental group based on their participation in undergraduate research, their department, and their participation in previous mentoring programs. This approach allowed us to adequately assess the particular benefits of coaching, while also considering the impact of the undergraduate research experience (UGR), addressing a gap in the existing literature. Studies with robust evidence of the most effective and sustainable training efforts for underrepresented students transitioning from undergraduate to graduate school are scarce. Too often, studies rely on participant satisfaction and correlation analysis. These do not account for self-selection bias (i.e., better prepared students choose to participate in coaching programs), and rarely use experimental designs with a control group to test their hypotheses or do not have enough URP students to power their analysis. Our experimental design allowed us to test our hypothesis that online coaching can contribute to the success of students in their first year of graduate school, as measured by the Hallmarks of Success indices.

Our group coaching model is also innovative, as coaching is often done in a one-to-one modality. In addition, our interventions were led by a team of certified trained coaches who followed a structured curriculum (available in Spanish at the CienciaPR webpage https://www.cienciapr.org/en/talleres-de-coaching-de-resiliencia). We only found one recent experimental design study where academic-based group coaching took place with certified trained coaches (Fainstad et al., 2022). Nevertheless, that program lacked a specific curriculum or a topic guide, making it challenging to reproduce. In Fainstad et al 2022 study participants were informed about the randomized control trial prior to being placed in the group and knew about the specific aims of the coaching program. Coaches were faculty members as well, who were not blinded, and participants were students of the same coaches who knew their assigned group. These conditions “introduced potential social desirability and selection biases because some participants may have opted to enroll owing to prior experience due to their relationship with the coaches” (Fainstad et al. 2022, p. 10). In our model, we applied a single blind random allocation design to avoid selection biases. We randomly allocated students to the experimental and control groups without their knowledge of the assigned condition.

Our intervention was implemented with student alumni from the University of Puerto Rico at Cayey (UPR-C), who were enrolled in graduate programs locally in Puerto Rico, in the continental US, and in international graduate programs. This span of graduate experiences allowed us to evaluate the role that cultural barriers play in the acquisition of soft skills. We did so by comparing how students in a local graduate program fare in the hallmarks of success vis-a-vis students enrolled in graduate programs outside of Puerto Rico. Given that Hispanics already make up 17% of the United States (US) population, with an expected increase to 29% by 2060 (NIGMS, 2011; US Census Bureau, 2015), additional evidence on cultural barriers is urgently needed to identify the most effective training efforts to diversify the future health research workforce so that it truly represents the general population.

In the next sections we introduce our coaching model, outline our research design, and describe the coaching sessions, including the methodology used to evaluate the intervention and its results. We conclude with a discussion and recommendations for practice.

Our coaching model

Stowers and Barker (2010) define coaching as a “collaborative relationship in which the person being coached, or ‘Coachee’, is coached by an experienced person who is an active inquirer and instrument for change” (p. 365). There are a variety of coaching models that can be adapted to diverse contexts and fields (Deiorio et al., 2022). Our model is reminiscent with recent approaches to coaching in academia, mainly with underrepresented students in the US (Cruz et al., 2021; Fainstad et al., 2022; Klug, 2016) and follows the International Coach Federation (ICF) core competencies as a framework to analyze the transition from undergraduate studies to graduate school. Specifically, we uphold the ICF core competency objectives between participants which are to: a) establish the foundation of an ethical practice, b) co-create the relationship by establishing agreements, c) communicate effectively through active listening, and d) cultivate learning and growth from the participant’s experiences (ICF, 2019). As other academic coaching models, we also encourage a relational approach between coaches and coachees to reflect on strengths and resilience-related skills to achieve academic and personal goals.

Our model intertwines coaching with resilience to promote that students envision taking each step-in life as an opportunity. Experts on psychosocial adversity have expanded the concept of resilience to distinguish those people who, despite being born and living in high-risk situations, achieve a healthy and successful psychological development (Rutter, 1993). Although there are various definitions, we chose Vanistendael’s (1995) definition, who conceptualizes resilience as

“the individual’s ability to do things well despite adverse circumstances. This implies a capacity for resistance and a faculty of positive construction and identifies two important components in this concept: a) resilience in the face of destruction, such as the ability to protect one’s integrity under pressure and b) the capacity to construct positive life behaviors despite difficult circumstances”

(p. 13).

Our coaching for resilience training model integrates these two concepts by taking advantage of coaching as a support and empowerment strategy, and by promoting resilience as a resource for student’s growth and capacity for developing a personal shield in the face of adversity.

Our model is also collective and interdisciplinary. Most academic coaching takes place on an individual basis (Deiorio et al., 2016; Pechac & Slantcheva-Durst, 2021; Sullivan & Polyzoi, 2021). It is common to find individual academic coaching programs throughout the literature (Deiorio et al., 2016; Capstick, 2018) or peer-mentoring programs taking place at the undergraduate or the graduate school level (Bettinger & Baker, 2014; Cruz et al., 2121; Guerrero et al., 2013). However, our innovative group coaching model motivates students from diverse academic stages and fields to reflect on their personal strengths and challenges from a collective perspective that facilitates vicarious learning. Aside from being more cost-effective than individual interventions, some advantages of group coaching when applied to university settings include increased motivation, an increase in self-efficacy and confidence in goal attainment (McDowall & Butterworth, 2014), and reduced procrastination (Mühlberger & Traut-Mattausch, 2015). Enlisting the participation of advanced senior graduate and postdoc students from different scientific fields in behavioral and biomedical sciences who provided peer mentoring to students in their transition to graduate school was instrumental for achieving these goals. These graduate students had been previously trained and were recruited from the Yale Ciencia Academy for Professional Development program in Ciencia Puerto Rico to participate as peer-mentors. In each coach-led session, we provided time for mentors to meet with their mentees to identify needs and practice specific coaching skills. This virtual space provided a noncompetitive environment and community of support. Students were able to strengthen their science identity, learn from one another’s experiences, and discuss challenging situations as well as strategies to overcome them, related to their graduate school, lab, research, or career plans. Previous research findings demonstrate that students’ embracing their professional identities remains a key factor related to biomedical science career persistence among undergraduates and graduate students (Villarejo and Barlow, 2008; Hurtado and Carter 1997; Chemers et al., 2011). Our innovative online group coaching intervention combined students from both stages and from different fields to facilitate setting academic and personal goals (Brown & Grant, 2010; Torbrand & Ellam-Dyson, 2015), supporting senior Latinx students’ transition from undergraduate studies to graduate level enrollment and leading to the successful completion of their first year of a graduate biomedical and behavioral research program.

Coaching curriculum

Coaching in a student’s educational career can lead to engagement, learning, retention, and an increased probability of completing a degree (Bettinger & Baker, 2014; Capstick, 2018; Pechac & Slantcheva-Durst, 2021). However, such transformations are not achieved in a vacuum. The coaching curriculum we developed places the student at the center of different interrelated social systems, considering networks at the family, peer, and professional levels. The theoretical framework that guides this intervention comes from Bronfenbrenner’s ecological systems theory (Bronfenbrenner, 1979) which conceptualized how personal and social contexts interact in facilitating or creating challenges towards developmental outcomes.

Specifically, the curriculum we used to guide the group coaching intervention focuses on the following three key components: 1) promoting teamwork and self-efficacy through various sources of support (Byars-Winston et al., 2016; Chemers et al., 2011); 2) providing support for peer learning and identifying professional networks (Sorkness et al., 2017); and 3) facilitating a structure of belonging within the research community (Chemers et al., 2011; Estrada et al., 2011; Guerrero-Medina et al., 2013; Vieno et al., 2007). Our coaching curriculum guideline also focuses on 1) establishing relationship agreements, 2) sharing learner assessments, 3) developing and implementing an action plan based on their Professional Development Plan (PDP or IDP), and 4) assessing results and revising plans. Hallmarks of success are integrated as “goals” to achieve or move toward over the course of the program. For instance, self-efficacy has been shown to consistently predict behavioral outcomes and changes in individual functioning over time (Bandura & Locke, 2003). Also, networking and mentoring skills in this academic coaching model seek to strengthen scientific identity, a sense of belonging within the research community, and strong academic and professional networks (Guerrero-Medina et al., 2013). These components gave structure to our tailor-made curriculum, providing tools for each session. Since we designed the curriculum and published in Cienciapr.org, it is replicable.

Furthermore, the curriculum guideline for this innovative model is appropriate for underrepresented students and responds to challenges found in under resourced or minority serving institutions like the UPR-C. Besides benefiting from existing coaching literature and NIH hallmarks of success (see Table A2), its design responded to the needs of the UPR-C student population as documented in a 2019 needs assessment survey of UPR-C alumni. More than half (59%) of students with more than two years of graduate school, who had previous UGR experience (n=165) identified the first year as one of the most challenging periods. Alumni overwhelmingly selected “organization and time management skills” (82%), “stress management, self-care and wellness” (78%), “oral communication skills” (76%) and “teamwork” (66%) as the most important skills they believed can better prepare undergraduate students for graduate school. Schramm-Possinger and Powers (2015) found similar responses.

This assessment motivated the thematic topics of the coaching workshops and our preparation of the curricular guide. In the guide, we addressed specific student concerns such as time and stress management, study skills, family pressures to pursue specific careers or remain in Puerto Rico to pursue graduate studies, availability of research opportunities, their level of competence in English skills or in preparedness of a PDP. A collaboration between the Interdisciplinary Center for Student Development (CEDE) and the Institute of Interdisciplinary Research at the UPR-C made possible the development and publication of the Facilitation Guide on Coaching for Resilience Workshops (Reyes-López, 2020) to support students in their personal, academic, and professional career.

This culturally-grounded approach was extended to the Professional Development Plan (PDP, which can be found in https://www.cienciapr.org/en/talleres-de-coaching-de-resiliencia). In 2017, more than 50 higher education faculty researchers and mentors from the UPR system designed this plan in collaboration with faculty from other institutions in Puerto Rico to provide a culturally appropriate PDP in Spanish, that included the needs of our student population, and that could be used by students in academic fields from all disciplines. The PDP serves as a useful tool to: (a) define expectations and goals, (b) organize and prioritize tasks, (c) improve time management, (d) boost productivity, (e) stay focused on tasks, (f) stay motivated, and (g) evaluate the results by reflecting on the strengths and barriers to achieving those goals. We adapted this PDP to support students in their transition from undergraduate to graduate school. Coaches as well as peer-mentors integrated it throughout all coaching sessions. Its design is also meant to help students prepare to face the workforce in their professional life, providing them with a structure to face challenges, solve problems, or achieve a personal goal. Through its use, mentors and students were also able to fosters a relationship of support, learning, and shared experiences.

In summary, we designed group coaching workshops to activate student’s resilience and their ability to transform each adverse situation into an opportunity. By valuing significant people and activating protective factors to create a shield of strength we sought to equip students to achieve their goals, become professionals in their field and, above all, healthy human beings. In accordance with our theoretical model (Bronfenbrenner, 1979) and in dialogue with other successful coaching interventions (Byars-Winston et al., 2016; Chemers et al., 2011; Estrada et al., 2011; Guerrero-Medina et al., 2013; Sorkness et al., 2017; Vieno et al., 2007), we prioritized concepts of connection, interdependence, tribe, and community. This was done throughout our workshops, providing, as we discuss in the next sections, an effective model for bridging support to Latinx students pursuing biomedical and behavioral science careers.

Research Design

Below we describe the institution selected for our study and the student sample selection process. We then discuss the intervention and our evaluation instrument.

Setting

We selected the UPR-C campus to implement our intervention because its student population is circa 100% Latinx, who qualify as “underrepresented populations” in the NIH (2024) definition, where representation is measured with respect to the national level.¹

UPR-C ranks 14^th nationally (within the USA sphere) among source institutions for female Hispanic PhDs in natural and social sciences (NSF, 2017). In order to graduate, all UPR-C students are required to engage in research, creation, or community service, and faculty are required to mentor students accordingly to obtain tenure (Academic Senate Cert. #49 2002–03, #24 2015–16). The UPR-C offers 25 academic programs. Most UPR-C alumni are from the natural sciences (45%) and pursue studies in Science, Technology, Engineering, Mathematics (STEM) or health-related fields.

Sample selection

To be part of the sampling frame (i.e. a list from where we randomly select our sample), a student had to fulfill three criteria. The first was being a candidate for graduation in natural or social sciences, which was determined by the Registrar’s Office. The second criterion was prior participation in a mentored UGR experience, specifically the number of research courses in which a student was enrolled in UPR-C. Students with no undergraduate research courses were excluded. The third criterion was application and acceptance to a biomedical or behavioral graduate program in the fall.

Based on UPR-C interviews with the graduating classes in previous years, we estimated that 80 students would meet the above criteria in a given year, allowing us to adequately power our analysis. This sample size is appropriate considering the number of groups, type-one error, the expected differences among groups (control and experimental), and the number of variables included in the model (Hickey et al., 2018; Jung et al., 2007).

For the first cohort (2020), we contacted 607 students from the behavioral and natural sciences who had applied for graduation in the period December 2020 –April 2021. A total of 321 students answered, 163 of whom had prior undergraduate research experience, as explained next. In this group, 50 students indicated that they were applying for a graduate program in the biomedical area (Figure 1), attended the virtual orientation, and all signed an agreement to participate. Five students of this cohort were not admitted to graduate programs.

Figure 1.

First Cohort: Recruitment Timeline, Sample Sizes, and Experimental/Control Groups

Note: UGR stands for undergraduate research; METAS+ is the name of our intervention.

For the second cohort (2021), we identified 229 students who had graduated from the social and natural sciences (Figure 2). A total of 138 students answered our initial approach, out of which 103 had prior undergraduate research experience and 85 had also applied for a biomedical graduate program. In the second cohort, the application for virtual orientation was answered by 61students and 40 signed an agreement to participate in the study.

Figure 2.

Second Cohort: Recruitment Timeline, Sample Sizes, and Experimental/Control Groups

Note: UGR stands for undergraduate research; METAS+ is the name of our intervention.

Selection and assignment criteria: Once we identified all qualifying students who agreed to participate, we randomly allocated participants to a control and an experimental group. To minimize sampling errors, we followed a stratifying sampling method, which is more efficient than simple sampling (Staffa & Zurakowski, 2018). The first stratum (categories in the sampling frame from where we selected the student) was the number of undergraduate research courses that each student was enrolled in. In these courses the student should have actively participated in a research project, had a professor from UPR-C or another accredited university acting as a mentor, and had tangible evidence of their contributions to the research project (e.g., report, poster, and publication, among others). Four courses were identified as meeting this criterion: one offered by the Honors program (PREH4990), one by the Biology Department (BIOL4990), one by the Chemistry Department (QUIM4999), and one course offered by the Institute of Interdisciplinary Research (INTD4116). The second stratum or category was the academic department from which the student had graduated. Those departments were Biology, Chemistry, Natural Science, and Social Science. The third stratum was if the student had participated in a previous research mentoring program.

In the first virtual orientation, we did not tell the students they would be divided into two groups or that there would be a coaching intervention for a subset of students. Allocating them into the experimental group based on that interest could have introduced a self-selection bias. To eliminate the possibility that only the experimental group would have interested participants, we randomly allocated the interested participants into both the control and experimental groups. Once we allocated participants to the control or experimental group, we offered a second orientation designed separately for each group. Students who agreed to participate submitted an IRB-approved consent form and evidence of having applied to graduate school.

Overall, we randomly allocated 25 students to the experimental group and 25 students to the control group for the first cohort, and 20 students to the experimental group and 20 students to the control group for the second cohort. Most of these students had been accepted to graduate programs in Puerto Rico. Others were accepted at Penn State University, University of Florida College of Pharmacy, University of Illinois Urbana-Champaign, University of Minnesota-Twin Cities, Texas A&M University, Lake Erie College of Osteopathic Medicine (LECOM), Temple University, Tulane University School of Medicine, University of Michigan, Autonomous University of Guadalajara, American University of Antigua, University of Virginia, Notre Dame of Maryland University, Temple University, University of Pennsylvania, Washington University in St. Louis, and University of Buffalo, among others.

The participation of female students in both cohorts was 71% and 60% in the control group and the experimental groups, respectively (Table 1). These percentages approximate the female representation in the general student population at UPR-C. Characteristics related to their undergraduate Grade Point Average (GPA), socioeconomic status, and the first generation in university were distributed similarly between the control and experimental groups (Table 1).

Table 1.

Socioeconomic Characteristics of the Participants of Cohort 1 and Cohort 2

	Control	Experimental
Total number of recruits
Cohort 1	25	25
Cohort 2	20	20
Students excluded in
Cohort 1.	3	3
Cohort 2****	0	2
Total students included in the analyses and accepted to graduate school in
Cohort 1	22	22
Cohort 2	20	18
Number of females
Cohort 1	16	15
Cohort 2	16	10
Number of males
Cohort 1	6	7
Cohort 2	4	8
Mean undergraduate GPA (SD)
Cohort 1
- Female	3.48 (0.23)	3.56 (0.29)
- Male	3.50 (0.46)	3.59 (0.27)
Cohort 2
- Female	3.65 (0.25)	3.74 (0.16)
- Male	3.50 (0.25)	3.67 (0.33)
Mean number of undergraduate research semesters (SD)
Cohort 1
- Female	3.0 (1.6)	2.9 (1.7)
- Male	4.3 (1.5)	2.9 (1.9)
Cohort 2
- Female	1.6 (0.9)	2.1 (0.8)
- Male	1.5 (0.6)	2.4 (0.7)
Mean (SD) number of coaching sessions attended, experimental only
Cohort 1: (after 8 coaching sessions)
- Female	NA	6.9 (1.2)
- Male	NA	6.3 (1.3)
Cohort 2 in progress (after 6 coaching seasons)
- Female	NA	5.8 (0.6)
- Male	NA	6.0 (0.0)
Number of participants that answered the questionnaires: Note the sample size for the analysis for evaluating the impact of coaching are those students that answered both questionnaires.
Cohort 1	22	22
- pre-survey	18	17
- post-survey	18	17
- both surveys
Cohort 2
- pre-survey	20	20
- post-survey	20	18
- both surveys	20	18
Household income distribution while an undergraduate:
Cohort 1
- Same or less than $14,999	5	4
- $15,000–$49,999	9	11
- $50,000 or more	8	5
- No data	0	2
Cohort 2
- $7,500–$14,999	3	3
- $15,000–$49,999	10	14
- $50,000 or more	7	3
- No data	0	0
The number of students whose parents have a university degree:
Cohort 1
- Yes	20	20
- No	2	1
- No data	0	1
Cohort 2
- Yes	19	16
- No	1	4
Socioeconomic status: Students who received Pell Grant:
- Yes	15	13
- No	7	9
Cohort 2
- Yes	13	16
- No	7	4

Coaching implementation

In our group coaching model, online coaching sessions were led by a professional coach. Each Coach responsible for implementing a uniform detailed agenda with the objective of strengthening complementary life skills for a group of 10 students beginning their first year of graduate studies in biomedical and behavioral sciences. A total of eight senior Latinx alumni from CienciaPR also participated actively as graduate peer mentors in each online coaching session. They took two training sections prior to the coaching activities on topics such as how to use the PDP, professional networking and sense of belonging, and how to choose a lab or research practice scenario, among others. These senior alumni were recruited from Yale Ciencia Academy, an NIH-sponsored program of the Yale-sponsored Ciencia Puerto Rico network. The UPR-C coaching team facilitated online chat groups following CienciaPR Yale Academy and National Research Mentoring Network (NRMN) best practices (Sorkness et al., 2017).

In each two-hour coaching session, the Coaches, Coachees, and peer mentors from CienciaPR participated online. During the first hour, the Coaches guided each Coachee towards a desired goal through powerful questions. During the second hour, graduate student mentors were paired with one or two Coachees in breakout rooms to work together on the coaching session’s competencies on a one-to-one basis, with specific instructions from the Coach.

To ensure uniformity in purpose and sequence, the coaches used the facilitation guide developed by the UPR-C team. Each skill addresses and builds on self- and situational awareness, resilience, and proactive approaches to professional development.

Four experimental groups (two groups per cohort composed of approximately 10 students each) and 32 Latinx alumni participated in online coaching sessions each semester for a period of approximately 10 months. The UPR-C ICF-certified coaches offered four online sessions before and after students entered graduate school. In the coaching sessions, students discussed how to transition to grad-school, language and cultural barriers, family pressures, integrating within a scientific community and professional networks, mentoring, time management, and study skills. On a deeper level, the Coachees shared experiences that described the sessions as “within a safe space” or “oasis” to identify opportunities and overcome adversities.

Following Villarejo et al. (2008), who found that developing a Professional Development Plan (PDP) was a key tool for promoting students’ success in their transition to graduate school, the coaches integrated the previously discussed culturally relevant PDP as a key tool for promoting URP students’ success in their transition to graduate school. Aside from the PDP, we were persuaded by the related literature to add other relevant tools for a successful graduate career: the Eisenhower Matrix for time management (Kennedy & Porter, 2022), the Vanistendael paradigm of resilience (Vanistendael et al., 2009), and the personal and academic leadership wheel (Sullivan & Polyzoi, 2021). Life coaching provided undergraduate students with these tools and skills by bolstering their potential, resilience, and support networks (Bettinger & Baker, 2014; Deiorio et al., 2016; Williams et al., 2017) to motivate them to achieve personal goals, self-care, and wellness in their transition to graduate school (Patel, 2017; Schmidt & Hansson, 2018).

After analyzing the related literature and student needs as documented in the institutional survey previously described, we designed the following topic guidelines for structuring the online coaching group discussions:

1. Resilience / The Pillars of My Life

2. Goals / Connecting with the Goal

3. Risk and Protective Factors / Experiences of Overcoming Adversity

4. Accountability / 1,2,3 Priorities

5. Leadership / Living my Leadership

6. Teamwork / One’s Goal is Everyone’s Goal

7. Professional Networking / We are all One

8. Self-Assessment / Do Your Best / Connection to Personal Power

A total of 22 students from Cohort 1 and 19 students from Cohort 2 participated in at least five of the eight coaching sessions offered, which ended in February in each year.

Pre-/post-survey evaluation

To evaluate the effectiveness of our coaching strategy, we conducted a pre- and post-survey evaluation. We grouped survey questions according to nine areas, eight of which are aligned with the student/training Hallmarks of Success (STU) indices from NIH (https://nigms.nih.gov/training/dpc/Pages/success.aspx) and one (Emotional Maturity) was developed by us. The indices included High self-efficacy as a researcher (STU-2); Satisfaction with quality of mentorship (STU-4); Perceived sense of belonging within the research community (STU-6); Intent to continue (or pursue) a career in biomedical research (STU-7); and Participation in academic or professional organization related to biomedical disciplines (STU-13). Students answered the questionnaire before and after the eight coaching sessions.

We developed sets of survey questions to assess each student’s status in relationship to the different Hallmark of Success indices. For example, the general concept of self-efficacy as a researcher refers to how much the student believes in his/her ability to carry out or collaborate in tasks associated with research (Bishop & Bieschke, 1998). Its measurement here was constructed from four basic sub concepts: Affective Emotional Arousal, which refers to the state of being activated and reactive to stimuli (Niven & Miles, 2013) (two questions); Self Perspective as a Researcher (nine questions); Scientific Performance (three questions); and Efficacy as a Researcher (nine questions).

We also evaluated Vicarious Learning, which is the ability of students to use another’s experience to learn, (STU-4; three questions; Roberts, 2010), Sense of Belonging within a scientific community (STU-6; six questions), Scientific identity as Social Persuasion (STU-7, two questions), Peer Networking (STU-13, five questions), and a new hallmark termed Emotional Maturity (five questions). Our empirical analysis is shown in the next section. We analyzed the hallmarks as a whole in a composite index (all items with equal weight) and individually in indices for each STU index. The numbers of questions used to develop the different indices varied and are similar to those used by Byars-Winston et al. (2016).

The associated hallmark of success, the scale, and the references used to construct the different indices are shown in Table 2. We used a 10-point scale to better observe differences among students’ perceptions, which could be concealed in a shorter scale such as that used by Byars-Winston et al. (2016), who used a 5-point scale. All of the questions had a scale from 1 to 10. However, we inverted the scale for the following statements with negative connotations: It is difficult for me to view myself as a future independent researcher; Thinking about my ability to conduct research makes me anxious; I often feel isolated from my academic environment; I feel that there are only a few fellow students with whom I can share my research interests; I hardly know anyone who works in my field. We recalibrated the inverted items to the 1:10 scale prior to calculating the mean indices and subsequent analyses.

Table 2.

Taxonomy of the Pre- and Post-Questionnaire

Index	Sub-group	Hallmark Code	Item Text	Scale (1–10)	Reference
Research Self-Efficacy	Related to coaching	STU-2 High self-efficacy as a researcher	I have been able to identify research opportunities within my areas of interest. I can explain the research that I have conducted with other researchers. I think that research presents an important opportunity to develop my commitment to academic goals. I feel prepared to make decisions about my academic future. It is difficult for me to view myself as a future independent researcher. (inverted scale) I feel prepared to transition into a graduate program. I feel I can succeed in a graduate program.	Normal	Byars-Winston et al., 2016
			I feel prepared to face any difficulties that I may experience during research. I feel comfortable discussing research topics related to my area with other researchers.	Normal	Chemers et al., 2011
Source of Self-Efficacy	Vicarious Learning	STU-4 Satisfac-tion with quality mentor-ship	I have been able to identify at least one research mentor in my area of interest. I have been able to identify at least one person who can advise me on my application or my integration into graduate school. Having multiple mentors helps you overcome challenges and be effective in the field of research.	Normal	Byars-Winston et al., 2016
Source of Self-Efficacy	Social Persuasion	STU-7 Intent to pursue a career in biomedical research	I have developed a professional development plan with the help of another person. Students who are more advanced than I am have motivated me to explore or to pursue a career as a researcher.	Normal	This study Byars-Winston et al., 2016
Source of Self-Efficacy	Affective Emotional Arousal	STU-2 High self-efficacy as a researcher	Thinking about my ability to conduct research makes me anxious. I am capable of succeeding in any research activity that I aim for.	Inverted	Byars-Winston et al., 2016 Chemers, 2011
Source of Self-Belonging	To a Scientific Community	STU-6 Perceived sense of belonging within the research community	I have a strong sense of belonging to a research community. My relationship with my research mentor has been productive. Being a researcher is an important part of my self-image.	Normal	Chemers et al., 2011
			I am confident that in the near future I will be, or continue to be, part of a community of researchers. I feel that there are only a few fellow students with whom I can share my research interests. (inverted scale)	Normal	This study
			I often feel isolated from my academic environment.	Inverted	Hurtado & Carter, 1997
Scientific Identity	Peer Networks	STU-13 Participa-tion in academic or professional organiza-tion related to biomedi-cal career pathway	I have been in contact with researchers outside of my university. I am an active member of at least one professional research organization. I know many fellow students and professors who are involved in research.	Normal	Byars-Winston et al., 2016
			I hardly know anyone who works in my field.	Inverted	This study
			I participate in social events with faculty, fellow students, or other people interested in research.	Normal	Chemers et al., 2011
Life Skills	Emotional	NA	I feel that I have the skills to manage my own time effectively. I feel that I have the skills to handle adverse situations in my academic life. I feel emotionally ready to achieve my career goals. I feel that I can handle any family pressure effectively and instead focus on my graduate studies. I am capable of dealing with the characteristic stress brought about by studying.	Normal	This study
Research Self-Efficacy	Research	STU-2 High self-efficacy as a researcher	I feel that I can make important contributions to a research team. I am going to pursue a career in which I conduct one or more research projects.	Normal	This study
			I feel comfortable working as a researcher on a project belonging to me or to another scientist.	Normal	Byars-Winston et al., 2016
			I feel comfortable with the skills I have in the application of research tools and techniques. I feel very confident in employing the specialized language and terminology used in my discipline. I feel very confident in developing a strategy to collect data. I feel very confident in analyzing data and observations. I feel very confident interpreting the results of a study. I feel confident in making an oral research presentation or presenting a research poster.	Normal	Chemers et al., 2011
Research Self-Efficacy	Performance	STU-2 High self-efficacy as a researcher	I am prepared to work independently on an experiment or research project. I am prepared to write a report of my research. I am qualified to prepare a poster or oral presentation in my research area.	Normal	Byars-Winston et al., 2016

Survey question consistency

We developed our survey questionnaire in English, and then translated it to Spanish for the benefit of participants. To determine if each item was a faithful rendition of the English original, we piloted a preliminary version with four current undergraduate students who fit the study sample demographics. These seniors had applied to a graduate program for the next fall and were not part of the control nor the experimental group. Immediately after administering the pilot survey, we interviewed each student separately and confronted their answers by discussing them out loud. We solicited clarifications when questions did not bring to mind the desired concept expected in the English original. This feedback led to rewording of specific questions. During the fourth (and last) iteration, no changes were suggested. Prior to administering the survey, we also incorporated input from voluntary NIH expert staff which improved the accuracy of the questionnaire’s wording. We administered the final version of the survey to all cohorts. Approximately a month after the intervention, we asked student participants if they had doubts or lack of understanding of the questionnaire in a focus group. No problems or issues of response bias were detected.

We compared pre- and post-evaluations to assess changes in students’ attitudes in the items and Hallmark of Success categories. With those responses, we created indices of the differences between the post and pre survey responses. Indices were calculated for each question for each student. If a student did not respond to one question, the answer was imputed by using the median answer by the group (low rates of answers were left blank <.01%). A difference of zero between post- and pre-evaluation suggests no change in the evaluated item or hallmark of success category. A negative value suggests a negative impact and a positive value a positive impact between the post- and pre-evaluation. A positive value between the post- and pre-evaluation in the control group would suggest that mere participation in graduate school has a positive impact, regardless of the coaching intervention.

A total of 35 alumni from Cohort 1 completed the pre- and post-survey tests (18 from the control and 17 from the experimental group), while a total of 38 alumni from Cohort 2 completed the pre-survey in July 2022 (20 control and 18 experimental). Both groups completed the post-test after the coaching intervention was finalized.

Methodology

We performed all analyses of changes in the Hallmark of Success indices between the pre- and post-surveys using R (version 4.3.1) and RStudio (version 2023.06.1+524). Data wrangling, preparation, and visualization were performed using the tidyverse set of packages (Wickham et al., 2019). We estimated all the central tendencies (median) and dispersion (95% quantiles) using Singh’s modification (1998) of the winsorizing approach prior to bootstrap. This method limits the impact of outliers (skewed data) on the central tendencies and dispersion by decreasing the weight of extreme values and increasing the weight of values at the center of the distribution. We implemented the trimpb function estimators as described by Wilcox (2017, R function trimpb, p. 129), which is more robust than traditional methods when the data do not comply with normal distribution and have outliers. The trimpb function estimates the central tendency and its dispersion by trimming extreme data (in this case 5%) and returning the median and the 95% quantile intervals of the bootstrapped distribution. This function is appropriate for repeated measures. Using the same function, we tested if the changes in Hallmarks of Success were significantly different from zero (post- minus pre-questionnaire response with an alpha of 10%).

To test differences among groups (experimental vs. control) we used Yuen’s test (Yuen, 1974) for two independent groups, such as

$$t=\frac{{\widehat{\mu }}_{t1}-{\widehat{\mu }}_{t2}}{\sqrt{d_1+d_2}}$$ (1)

with

$$d_j=\frac{(n_j-1)s_{wj}^2}{{\widehat{n}}_j({\widehat{n}}_j-1)}$$ (2)

where ${\widehat{\mu }}_{tj}$ is the trimmed mean for the jth group, $s_{wj}^2$ is the variance obtained after the winsorization process, ${\widehat{n}}_j$ is the sample size after trimming.

We applied the function yuenbt (Wilcox, 2017, p. 173) to this test. We then used the yuenbt function with an equal-tailed confidence interval (CI) of the bootstrap intervals with trimming of 5% of outliers and 10% statistical significance. The bootstrap approach for the t-test version was derived from Guo and Luh (2000). Both Wilcox’s approaches used here are Monte Carlo-type simulations with trimming of 5% to reduce the impact of outliers on the location (mean and median) and quantile dispersion (95% CI). The type-1 error was set at 10% for all tests. All the simulations procedures were of 10,000, from which we calculated the 95% quantiles. We performed the analyses on the mean differences of the STU indices.

To determine if there was a differential effect of coaching between students enrolled in a university in Puerto Rico and students enrolled elsewhere, and the interaction among these factors, we performed a linear regression with Gaussian response variables. In this case the response variables are the hallmarks of success. We use the function “glmrob” from the robust base package (Maechler et al., 2023). This function considers outliers, as in the previous tests. We also evaluated the interaction effect, which is the interplay between the experimental/control and if the students were studying in PR or elsewhere. The approach and explanation of robust linear generalized models can be found in Maronna et al. (2019).

Results

Difference between pre- and post-survey periods

The composite index (all items combined) developed for our study measures the general effect of our intervention in the nine hallmarks under evaluation. Our analysis revealed that the composite index of the experimental group showed a significant change after our online coaching intervention. However, the control group did not show a statistically significant change in this composite index, revealing that the result of the intervention/experimental group effect was not by chance. Post-hoc tests for multiple comparisons followed suggestions by Wilcox (2017, p. 346).

In all of the nine indices of success, we noted significant differences between the pre- and post-surveys for the experimental group (note that the mean and CI do not overlap the null model H₀=0); in the Singh’s test; the black horizontal line. Moreover, we found statistically significant differences between the pre- and post-survey in the experimental control group in the following indices: Emotional Maturity, Scientific Community, Affective Emotional Arousal, Self-Perception as a Researcher, Scientific Performance, Efficacy as a Researcher, Social Persuasion, Vicarious Learning, and Peer Networking. This is in spite of the fact that they may overstate their knowledge in the pre-survey (given before entering their graduate program) via the Dunning-Krueger effect (Dunning, 2011). The Dunning-Krueger effect states that when one has little knowledge about a topic one tends to overstate that amount of knowledge. Thus, our results are probably conservative.

On the other hand, those who did not participate in coaching showed no effect of a significant increase in most indices. Note that students in the control group also had significant change between the pre- and post-survey for Social Persuasion and Vicarious Learning. (The Singh’s test estimator location (median) and CI do not overlap the null model H₀=0, p <0.05, Figure 1.)

Differences between experimental and control groups

There are differences between the control and experimental groups in the composite index, as shown in Figure 3. The change in the compound index in the experimental group was statistically significant, while in the control group there were no discernible changes. This supports the notion that our intervention caused an improvement in the nine hallmarks in general.

Figure 3.

Differences in Change in Indices in 8 Hallmarks of Success and Emotional Maturity and Composite Index (All Items) of Students Who Participated in Coaching (orange) and Control Group (black; no intervention)

Notes: The median (point) and 95% quantiles (range, estimated from bootstrap simulations) are shown. The black horizontal line represents the null model of no change between pre- and post-test questionnaire results. Indices that do not overlap the black line (H_o= 0) are significantly different from zero (p <0.05). The values at the top of the figure represent the statistical significance resulting from the test in changes in survey responses comparing control with the experimental group of the students using Yuen’s test. Ns= not significant, * p <0.05, ** p <0.01.

Some hallmarks showed a larger effect than others. Seven of the nine hallmark indices for the experimental group show a marked positive improvement as compared to the control group. No evidence of differences was noted between the control and experimental group for the indices of Vicarious Learning (STU-4) and Peer Networking (STU-7). All other comparisons were statistically significant and showed a positive impact in the experimental coaching group (Yuen’s test p <0.05). In all significant cases, the experimental group had higher post-evaluation scores in the survey as compared to the pre-evaluation survey.

Differences among students who studied in Puerto Rico or elsewhere and interaction effect

The sample size for this analysis included the experimental group (who studied at institutions in Puerto Rico = 28, elsewhere = 10) and the control group (who studied at institutions in Puerto Rico = 20, elsewhere = 15).

We noted significant differences in six of the nine analyses of the Hallmark of Success indices (Figure 4 above and Table A1 in Appendix 1), between the experimental and control group, with the experimental group having higher coefficients. Comparison of the pre- and post-differences between students who studied in Puerto Rico and those who studied elsewhere only showed significant differences in one of the Hallmark of Success indices (Efficacy as a Researcher). An interaction effect can be described as the contrary effect when considering both experimental/control factors and if the students studied in Puerto Rico or not. We found three indices where coaching participation had an effect and in which studying locally had an interaction effect. The indices were Social Persuasion, Emotional Maturity and Scientific Performance. The experimental group had higher differential scores in Emotional Maturity between pre- and post- when studying outside PR, while the students in the experimental group had higher scores for the indices of Social Persuasion and Scientific Performance if studying in PR.

Figure 4.

Interaction plots between experimental/control factors and university location

Notes: Tests and figures were performed using robust generalized linear model (see methods for explanation). Only three of the indices showed significant results and the figures are shown below. Change in Scientific Performance and Social Persuasion students in the Experimental group studying on the island did significantly better than other groups, while students in the experimental group that studied off island had higher emotional maturity.

Discussion

Success in science requires more than research skills. Soft skills are fundamental for a successful research career (Deiorio et al., 2016). Our innovative coaching intervention improved a range of life skills with results that warrant attention in light of the recent scholarship.

In the related literature we identified coaching interventions given to graduate students, but their evaluations were qualitative (Williams et al. 2017) or correlational (Cruz et al. 2021; Hall et al. 2016; Estrada et al. 2019). Experimental designs that measure the impact of coaching are mainly focused on undergraduate students (Bettinger & Baker 2014) and on early-career scientists (Weber-Main et al. 2022). The only recent experimental design we identified that applies academic-based group coaching at the graduate level probably suffered from selection bias by not implementing a blind allocation of participants (Fainstad et al. 2022). Our research broadens the scope of previous studies by applying 1. an experimental approach of students transitioning into graduate programs into social and natural sciences; 2. a single blind random allocation design to avoid selection biases; 3. a coaching intervention provided by certified coaches unrelated to the participants; and 4. a robust empirical comparison between students facing cultural barriers vis-à-vis students not facing cultural barriers in Puerto Rico. We measured this last aspect by comparing students who enrolled in graduate programs in Puerto Rico with their pairs in the U.S.

Specifically, we show that coaching works to acquire the skills identified in the Hallmarks of Success established by the NIH. This is especially true for Affective Emotional Arousal and Social Persuasion, two hallmarks where we observed the largest differences between the control and treatment group. Thus, our model is consistent with recent academic group coaching interventions that address skills related to leadership development (Okpala, et a., 2021) and the imposter phenomenon (Siddiqui, et al., 2024). These factors can also impact underrepresented students’ feeling of belonging and cultural inclusion in biomedical science careers (Weber-Main, et al., 2022). According to Estrada and collaborators (2011) a sense of belonging to the scientific community is important for academic persistence, especially for underrepresented students. We found statistically significant improvements on this index as well, both by comparing students before and after our coaching intervention and by contrasting the results between the control and experimental groups.

The lack of statistically significant differences between the control and experimental groups in the hallmarks Vicarious Learning and Peer Networking deserves discussion. Possible reasons for this lack of significance was that, even though students in the treatment group exhibited larger gains than the control group, these were not enough to mark a difference. Perhaps there are other sources from which students obtained the soft skills linked to the Vicarious Learning and Peer Networking in their first year of graduate school. Williams et al. (2016) did a randomized intervention with a small student sample, stating that “Peers (including fellow graduate students and postdoctoral scientists) are also important potential sources of vicarious learning” (p. 17). A similar analysis can be made with respect to Peer Networking, which involves skills that can be obtained from other experiences, though the coaching enhances such experiences as well.

Finally, even though all of our participants are underrepresented in their academic careers at the nationwide level, we also evaluated the role that cultural differences play in the acquisition of soft skills. Coaching was more effective for participants who were pursuing a graduate degree at the local level (i.e. Puerto Rico) for the hallmarks Social Persuasion and Scientific Performance, than for students who entered a graduate program outside Puerto Rico. Our experimental design is consistent with qualitative studies that found that cultural shocks have an impact on the adaptation to a successful graduate career for international students (Hartshorne and Baucom, 2007), who are affected by differences in teaching philosophies and cultural insecurities (Abdulai et al., 2021). However, coaching was more effective in helping students outside their home country in reaching Emotional Maturity. These opposing effects of coaching on students in and outside their home country appear to average out, so that there were no statistically significant differences between them in the composite hallmark index (which measures overall soft skills).

Conclusions

Underrepresented populations are still struggling to successfully finish their graduate program, especially during their first year of study. The NIH developed a number of Hallmarks of Success that are likely predictors of student success. With an experimental design, we were able to rigorously evaluate the effect of an innovative coaching model on the development of soft skills of first-year graduate students, beyond mere correlations. Based on our robust design, we draw the following conclusions from our data analyses. First, our intervention positively impacts underrepresented students, as measured by the Hallmark of Success indices, in their first year in graduate programs. Overall, we found a statistically significant effects from our intervention in the nine hallmarks under evaluation. Given that we conducted an experimental design, and the statistical analysis is based on robust statistics, we can conclude that this effect was not by chance.

Second, our findings suggest that our coaching intervention is particularly effective for helping students manage their feelings and interact with research organizations. Emotional Maturity, Self-Perception as a Researcher, Scientific Performance, Efficacy as a Researcher, and Social Persuasion are the Hallmarks of Success that were significantly improved by just eight coaching sessions when compared to the control group. The effect on Social Persuasion and Emotional Maturity were the highest relative to other STUs.

Third, our intervention was not effective for some NIH hallmarks, at least in the short term (we only measured the effect of coaching immediately after finishing the sessions). Specifically, Affective Emotional Arousal and Peer Networking showed the lowest differences between the control and experimental groups. This finding may be simply due to the fact that graduate programs typically have near-peer programs in place upon entry. Since we randomly selected students to either group, this was a factor that was more dependent on the graduate program itself rather than the coaching intervention. However, our experimental design did not evaluate the impact of coaching on student’s sense of confidence in their research or in their networking process in the longer term. It is possible that the results of the nonsignificant indices may require longer time periods to manifest.

Our research carries implications for practice as well. First, the proposed online group coaching is a sustainable and cost-effective intervention that can be implemented relatively easily in other universities, especially at under-resourced or minority-serving institutions. The cost is relatively low for several reasons: the small number of sessions; the lower average cost per student compared to individual coaching approaches; and because part of the work is done by graduate students, who are more affordable than a coach. Second, this intervention can be done either by the universities that receive the students or by the sending institutions. Graduate schools interested in improving students’ integration into their graduate program should consider implementing a coaching intervention for their incoming students and offer online coaching sessions both prior to and during their first year of graduate school. Graduate schools that support their future students are likely to see benefits, in that students gain a better sense of belonging to a scientific community, learn to “navigate” the research culture in grad school, have more skills to overcome the imposter syndrome (Bravata et al., 2020; Chakraverty, 2019), prevent mental health problems, and deal with economic challenges. Third, since this cost-effective coaching works for students pursuing a career in medicine as well as for those within social and natural science, several graduate programs can establish their own coaching interventions. Fourth, our culturally relevant curriculum is also freely available on the CienciaPR website. Fifth, since our approach works for students in the U.S. as well as in Puerto Rico, universities located anywhere can seek to implement a coaching intervention that increases their academic success.

A potential limitation of the study is that we do not know what effect our intervention would have on students who did not have previous undergraduate research experience. We recommend that future studies analyze the intervention’s effect on that set of students. In addition, researchers can evaluate whether coaching has other benefits. For instance, our intervention may also help students to finish their second year of graduate school and may impact indices within other stages, such as the time to graduation. Such approaches exceed the scope of this paper.

Appendix 1.

Table A1.

Results of tests for determining if the effect of coaching and location of studies (within PR or outside PR) and their interaction is significantly different. Test performed with the function “glmrob” from the robustbase package. Interaction effect: is the interaction between the experimental/control and if the students were studying within PR or elsewhere, Coef= Estimated Coefficient.

Variables	Coef	Coaching p-value	Coef	Vicarious Learning p-value	Coef	Social Persuasion p-value
Experimental vs Control	0.944	0.024*	0.122	0.771	0.775	0.064
Studied in PR or elsewhere	0.062	0.870	−0.123	0.774	−0.333	0.377
Interaction effect	0.143	0.781	0.687	0.183	1.735	<0.001**
	Coef	Affective Emotional Arousal p-value	Coef	Scientific Community p-value	Coef	Peer Networking p-value
Experimental vs Control	1.422	<0.001**	1.011	0.016*	0.853	0.042*
Studied in PR or elsewhere	0.189	0.617	0.213	0.573	0.539	0.154
Interaction effect	−0.002	0.997	0.313	0.543	−0.362	0.483
	Coef	Emotional Maturity p-value	Coef	Efficacy as a Researcher p-value	Coef	Scientific Performance p-value
Experimental vs Control	2.007	<0.001**	1.256	0.003*	0.311	0.458
Studied in PR or elsewhere	0.311	0.410	1.039	0.006*	−0.110	0.772
Interaction effect	−1.338	0.009*	−0.250	0.628	1.698	<0.001**
	Coef	Mean index p-value
Experimental vs Control	1.067	0.011*
Studied in PR or elsewhere	0.328	0.385
Interaction effect	0.078	0.879

Table A2.

Coaching activities and skills related to NIH Hallmarks of Success indices

Coaching activity	Coaching skills							Hallmarks of Success
	PDP / IDP	Time management	Stress management	Family pressure	Preparing for exams	Studying skills	Self- confidence
Resilience / The Pillars of My Life	X			X	X		X	Emotional Maturity Self-efficacy as a researcher (STU-2)
Goals / Connecting with the Goal	X							Intent to pursue a career in biomedical research (STU-7)
Risk and Protective Factors / Experiences of Overcoming Adversity	X		X	X	X	X	X	Emotional Maturity Quality of mentorship (STU-4)
Teamwork / One’s Goal is Everyone’s Goal	X	X						Sense of belonging within the research community (STU-6) Self-efficacy as a researcher (STU-2)
Accountability / 1,2,3 Setting Priorities	X	X				X		Emotional Maturity
Leadership / Living my Leadership	X	X		X			X	Self-efficacy as a researcher (STU-2)
Professional Networking / We are all One	X	X					X	Participation in academic or professional organization related to biomedical disciplines (STU-13) Quality of mentorship (STU-4) Sense of belonging within the research community (STU-6)
Self-Assessment / Do Your Best / Connection to Personal Power	X		X	X				Self-efficacy as a researcher (STU-2)