Agentic Actions and Agentic Perspectives Among Fellowship-Funded Engineering Doctoral Students

Abstract

In the US and Europe, institutions, foundations and governments invest significant financial resources in doctoral fellowships. Unlike other graduate funding mechanisms, fellowships are typically not tied to specific projects or job responsibilities and thus may afford more agency to students. We examined how fellowship funding contributes to or undermines agency of doctoral student recipients. We interviewed 23 US engineering doctoral students primarily funded on a fellowship for at least one semester. We qualitatively analyzed the interviews, using inductive and deductive methods of coding. Participants described increased flexibility with their projects, advisor, and personal life; additional access to physical resources, people and networks, and research experiences; and feelings of internal validation and external recognition from fellowship awards. Contexts of advising, timing of fellowship, source of fellowship, financial circumstances, and fellowship structure influenced their experiences. Agentic perspectives and actions included choice of advisor and research projects, switching advisors if necessary, completing internships and visiting other labs, and enjoying a higher standard of living. Advisor support is a necessity for students funded on fellowships. Multi-year fellowships from external sources, in comparison to internal sources, more often supported agency. We make recommendations for institutions to structure and administer fellowships to better support students.

1. Introduction

Graduate student funding impacts important outcomes of doctoral training such as choice of dissertation topic (Isaac et al., 1989), skill development (Grote et al., 2021), retention (Ampaw & Jaeger, 2012; Mendoza et al., 2014), and time-to-degree (Horta et al., 2019). In some countries, such as the Netherlands, Australia, China, Kazakhstan, and Chile, domestic students are funded through a uniform mechanism (Torka et al., 2022; Van Rooij et al., 2021). In other systems such as Europe and the US, doctoral students are funded through a variety of sources including project work associated with a research grant, employment in teaching roles, and stipends awarded directly to students (Torka et al., 2022). In such systems, educators and policymakers are interested in questions of how funding mechanisms impact doctoral student outcomes. For example, in Portugal, Horta et al. (2018) found that recipients of competitive individual research grants had higher research productivity as compared with those on research assistantships.

In the United States, in which our study is situated, engineering doctoral programs typically begin with two years of coursework. In many disciplines, students earn their master’s degree as part of their progression (Lange, 2006). Following coursework, doctoral students typically take a qualifying examination they must pass in order to continue with their degree program. Next is the dissertation proposal, after which the student begins their dissertation research as a PhD candidate (Council of Graduate Schools, 2005). Time-to-degree varies from four to six years for engineering doctoral students, with a quicker timeline for students with a previously earned master’s degree (Sowell et al., 2008).

In most science, technology, engineering, and mathematics (STEM) disciplines in the US, doctoral students are funded through three main mechanisms:

• Fellowships are competitively funded by sources internal and external to the institution, including government agencies and nonprofits. Fellowships typically provide for students’ financial needs without requiring additional work tasks (Bowen & Rudenstine, 1992; Lovitts, 2002; Mendoza et al., 2014). Thus, fellowships are often considered by students and their advisors to be the most desirable form of funding because students can focus on their coursework and dissertation research without additional teaching or research responsibilities. The open-ended nature of fellowships affords students much greater autonomy and agency (Graddy-Reed et al., 2021; Herzig, 2004; Szelényi, 2013) compared to other funding.
• Research assistantships are often funded through research grants awarded to faculty members; work responsibilities are motivated by the goals of the grant, funding agency, advisor, and any other collaborators on the research grant (Niemczyk, 2015). Research assistantships provide structured research training and opportunities to interact with faculty and peers (Millett & Nettles, 2006) as opportunities for socialization into the department and discipline. Within engineering, research assistantships are typically funded through either government entities or industry.
• Teaching assistantships are typically funded by academic departments to help with the teaching responsibilities of a class, such as grading, holding office hours, and leading recitation or lab sections (Prieto & Meyers, 2001). Teaching assistantships are perceived as the least desirable funding because they distract from research and degree progress (Borrego et al., 2021). Some prior studies found that teaching assistantships extend STEM doctoral students’ time-to-degree (Knight et al., 2017) while others found no difference (Nettles & Millett, 2006).

While our present study focuses specifically on engineering doctoral students, we include literature on funding mechanisms from a broader STEM perspective. Research assistantships are the dominant funding mechanism in the US, serving as the primary funding source for as many as 56% of STEM doctoral candidates; however, in many biological sciences and biomedical engineering, up to 47% of doctoral candidates are funded primarily by fellowships (Knight et al., 2024). Overall, life sciences doctoral students tend to be funded by fellowships at higher rates compared to engineering and physical sciences doctoral students (Knight et al., 2018), which has implications for whether students are funded at all and what proportion gain teaching and research experience through assistantships.

Fellowships are also the most under-researched of the three funding mechanisms listed above and are worthy of study, given potential policy implications. Fellowships are unique in that organizations can be strategic in their policies, procedures, and selection criteria, potentially influencing graduate education. Fellowships are typically awarded after a competitive application process, which may or may not be tied to admittance to a specific graduate program. While individual faculty members award research assistantships through their research grants, fellowships tend to be controlled centrally (i.e., at the institution, college, or department organizational levels) and by external sources (e.g., foundations, government entities). These entities may influence graduate education in unique ways by applying research findings to update their fellowship policies.

Although fellowships are viewed favorably, the limited prior research has identified both benefits and drawbacks to fellowship funding as compared to other types of funding. In one study, STEM doctoral students primarily funded by external fellowships reported significantly lower skill development in research, communication, peer mentoring, teamwork, and project management when compared to students funded via research assistantships (Grote et al., 2021). The common requirement for research assistants to participate in a research group could explain this finding, as prior work found that research groups facilitate the development of learning competencies for engineering graduate students (Burt, 2017). Fellowships also relate to retention for doctoral students, although with mixed findings. Mendoza et al. (2014) found positive relationships with retention for doctoral students funded on fellowships, although the sample may have been biased by competitive fellowship selection criteria. Ampaw and Jaeger (2012) found that doctoral students funded on fellowships had increased odds of reaching candidacy but decreased odds of completing their degrees. Fellowship funding has also been found to relate to career outcomes. Biomedical sciences PhDs funded through fellowships were less likely to select research-focused jobs for their post-graduation employment than PhDs funded through research assistantships (Blume-Kohout & Adhikari, 2016). In another study, STEM doctoral students funded primarily on fellowships were found to be less likely to graduate with no job prospects than those funded primarily on research assistantships (Kinoshita et al., 2020).

More detailed research is needed into how financial circumstances and funding impact graduate students (George & Newhouse, 2024). Specific to fellowships, research indicates both positive and negative outcomes, even with mixed findings related to the same outcome, i.e., retention. This raises the question of what additional, unexplored circumstances lead to these varied outcomes and their implications for how to best support graduate students through fellowship funding. Our study examines the role fellowship funding plays in the agency of engineering doctoral student recipients, addressing the following research questions:

• How does fellowship funding contribute to or undermine agency of graduate student recipients?
• What fellowship circumstances, such as timing and internal vs. external source, determine the extent to which a fellowship will impact doctoral student agency?

Graduate student agency is a central concept in recent graduate education studies. As it relates to higher education, agency has been defined as assuming strategic perspectives or taking strategic actions towards goals that matter (O’Meara, 2013, p. 2). In taking agentic perspectives and agentic actions, students become “proactive agents in their own development” (Maher et al., 2020, p. 4), taking control of their doctoral training and at times pushing back against department culture to make change (Feldon et al., 2022; Rodriguez et al., 2022). For example, students may exhibit agency in selecting an advisor or a laboratory rotation (Wofford & Blaney, 2021), switching advisors to move into a more supportive relationship (Feldon et al., 2022), strategically deciding when to time teaching experience with respect to dissertation progress (Borrego et al., 2021), or seeking internships and information on future careers (Griffin et al., 2023; Huang & Jung, 2024; Thiry et al., 2015).

Agency is a useful lens for studying fellowships specifically, given the flexibility and autonomy of fellowships described in prior work (Graddy-Reed et al., 2021; Herzig, 2004; Szelényi, 2013). Agency must be considered in context (Griffin et al., 2023; O’Meara, 2013). For doctoral students, there are institutional, economic, cultural, and disciplinary forces influencing which pathways are viable (Feldon et al., 2022). Agency is concerned with whether an individual thinks they have influence over these forces (O’Meara, 2013). The current study does not seek to attribute doctoral student experiences entirely to their funding, but we are concerned with how agency changes when fellowship funding is involved.

2. Literature Review

2.1. Agency in Graduate Education

Theoretical understanding of agency as it relates to graduate education has been advanced through a number of studies on career choice (e.g., Griffin et al., 2023; Huang & Jung, 2024; Thiry et al., 2015) and the increasing recognition of the role of agency in socialization. Jaeger et al. (2017) and Portnoi et al. (2015) are credited with the earliest critiques of socialization as a passive process that does not recognize the agency of students in influencing their own experience or changing the culture of the departments and disciplines to which they are being socialized. In their analysis of graduate student agency in pathways to the professoriate, Feldon et al. (2022) take an institutionalism perspective which centers graduate student agency within a system of institutional, economic, cultural, and disciplinary structural influences. Prior work specifically related to agency and funding mechanism examines agency in the context of overall funding stability and career planning, with limited focus on fellowships (Borrego et al., 2021).

According to the view of agency adopted in doctoral education research, “individuals are embedded in social contexts that deeply shape the range of agency they may experience at any given time” without determining their fate (O’Meara, 2013, p. 2). Agency is context specific, and in a given context “individuals with higher levels of agency understand that despite constraints, they have a role to play in their ability to make successful progress toward their goals” (Griffin et al., 2023, p. 100; McAlpine, 2012; McAlpine & Amundsen, 2011; O’Meara, 2013). Further, an individual’s sense of agency is influenced by their identities and how those interact with local contexts (Jaeger et al., 2017; O’Meara, 2013; Rodriguez et al., 2022).

In this conceptualization, there are two forms of agency: students’ agentic perspectives and agentic actions (O’Meara et al., 2014). Agentic perspectives are “a form of self-talk, a set of perspectives or views taken when experiencing opportunities and constraints in pursuit of a goal” (O’Meara et al., 2014, p. 157). Graduate students can assume agentic perspectives by seeing situations as possible to overcome, recognizing that they can get certain types of support from their advisor and other needed support elsewhere, feeling in control of their career goals, and viewing critical feedback as a way to learn and grow (O’Meara et al., 2014; O’Meara et al., 2013). Examples of agentic perspectives include graduate students viewing themselves as in control of their dissertation success and seeing the challenges and setbacks of graduate study as building their resilience (Nguyet Nguyen & Robertson, 2022).

Agentic actions “are strategic and are enacted with self-awareness of goals and contexts” (O’Meara et al., 2014, p. 158). As noted in O’Meara et al. (2014), “agentic action is discrete from, but often closely follows, and is related to, agentic perspective” (p. 158). Graduate students can take agentic actions by being strategic in managing their relationship with their advisor, taking strategic actions to get needed support from people other than their advisor (e.g., peers, other faculty, Tran et al., 2016), asking for help when needed, and taking steps to obtain skills or knowledge to advance their career goals (Griffin et al., 2023; O’Meara et al., 2014; O’Meara et al., 2013; Thiry et al., 2015). Agentic actions include international students proactively attending workshops and classes to develop second language skills and ending a “dysfunctional relationship” with their advisor by finding another advisor (Nguyet Nguyen & Robertson, 2022, p. 821). Students can derive pleasure from exercising their agency (McAlpine & Amundsen, 2009). Yet students do not always have immediate positive outcomes; taking agentic action can lead to disagreements or strained relationships with peers or advisors, even though these agentic actions can lead to longer-term positive outcomes (O’Meara et al., 2013), for example in the upheaval associated with switching advisors or graduate programs or leaving graduate study altogether.

In this study, we are interested in the role fellowships play in doctoral student experiences. The literature suggests that fellowship funding affords a certain amount of desirable freedom and autonomy as compared to other funding mechanisms (Graddy-Reed et al., 2021; Herzig, 2004; Szelényi, 2013), which may support student agency. Yet there are also mixed results suggesting too much autonomy can delay progress and cause frustration (e.g., Ampaw & Jaeger, 2012). We use agency as a lens to interpret how the lack of uniform structure in fellowships influences graduate student agency in both positive and negative ways. Specifically, we coded for experiences of fellowship-funded engineering doctoral students deriving from their funding, then examined whether and how these experiences led to further agentic perspectives and actions.

2.2. The Context for Doctoral Student Agency

Agency is contextual, and as such, doctoral student agency is never studied without consideration of the graduate education context. While a complete description of the many levels of influence are beyond the current scope, we describe three aspects known to influence doctoral student agency: relationship with advisor (O’Meara et al., 2013; Paglis et al., 2006), academic discipline, and financial concerns (Borrego et al., 2021).

One reason for mixed results in prior studies focused on graduate student funding in general and fellowships specifically may be variation in advising and mentoring received1. A student’s advisor plays an undisputedly important role in their graduate school experience and future career (Lee, 2008; Shanachilubwa et al., 2025). In fact, one review concluded supervision is the single most influential factor in doctoral student completion, achievement, and wellbeing (Sverdlik et al., 2018). Poor mentorship by faculty advisors can negatively impact graduate student mental health (Bork & Mondisa, 2022; Charles et al., 2021; Mackie & Bates, 2019; Sallai et al., 2023). Multiple studies on engineering doctoral student attrition illustrate how the advisor contributes to student decisions to leave their graduate programs, although reasons are complex and involve multiple factors (Bahnson et al., 2023; Berdanier et al., 2020; Zerbe et al., 2022). However, an encouraging advisor can help students cope with negative graduate school experiences and enable them to envision futures in their fields (Amon, 2017).

Another important context influencing graduate student experiences, funding, and agency is the discipline. Multiple graduate education scholars cite the department as the primary site of graduate student socialization (Gardner, 2007; Golde, 1998), due to their strong control over funding, degree requirements, and admission, influenced by disciplinary norms (Golde, 2005). In engineering, most doctoral students are trained as part of a research lab, or group of students and postdocs all working under the same advisor (Burt, 2017; Crede & Borrego, 2012). Such labs tend to be well funded, and most students can expect to be supported through some combination of assistantships and fellowships (Nettles & Millett, 2006). In terms of distribution of fellowships (19%), research assistantships (56%), and teaching assistantships (8%), the four disciplines in the current study align with most engineering disciplines (Knight et al., 2024). A high proportion of engineering PhDs find employment in private industry, and partnerships such as sponsored industry research help align engineering doctoral programs with preparation for industry careers (Germain-Alamartine & Moghadam-Saman, 2020).

Finally, it is reasonable to expect that fellowship funding would allay some of the stress stemming from doctoral students’ financial uncertainty. Financial stress influences graduate student mental health (Charles et al., 2021; Mackie & Bates, 2019). Graduate students with greater financial concerns reported greater symptoms of depression, although this effect could be lessened with greater social support, a supportive department, and optimism about job prospects (Charles et al., 2021). Finances can also be a stressor due to competition for funding opportunities and the potential for funding instability (Grady et al., 2014). Additionally, low stipends and insufficient funding lead many students to seek additional avenues of income in addition to their roles as graduate research assistants, leading to stress about managing several responsibilities at once (Grady et al., 2014). International students, who comprise the majority of graduate students in engineering (National Science Foundation, 2024a), were more likely to report financial hardships than their domestic peers (Hyun et al., 2007)—these students reside within a federal policy environment that limits their work-for-pay opportunities.

2.3. Types of Fellowships

As additional background for the current study, we describe the landscape of fellowship funding in the US. A wide variety of fellowships fund engineering doctoral students at US institutions. Fellowships can either be funded internally through the doctoral students’ institution or through an external source. Common external sources for engineering students include the US government (e.g., the Department of Energy, the National Science Foundation), for-profit companies (e.g., Microsoft), and foundations (e.g., Hertz Foundation). The financial award of a fellowship also varies from a small scholarship to a fully funded annual stipend. For example, National Science Foundation’s (2024b) Graduate Fellowship Program provides a USD 37,000 annual stipend. External fellowships with fully funded stipends often pay doctoral students at a higher rate than they would receive with an RA or TA assignment, an incentive for students to apply for fellowships.

The awarding of fellowships typically occurs after a competitive application process in which doctoral students may need to submit a research proposal, personal statement, and professional references. Institutions also award internal fellowships to students with their admission to the university, often as a recruitment tool (Haines et al., 2023). Historically in the US, most external fellowships have eligibility requirements of being a US citizen or permanent resident, with fewer opportunities available for non-US engineering doctoral students. In our study, external fellowships tended to offer funding for a duration of three to four years, whereas internal fellowships were likely to be for only one or two years. Fellowships can be awarded at any year of study in graduate school, although specific fellowships often have eligibility requirements for the student’s year of study. There may be requirements as part of the fellowship award, such as presentation of research results, attendance at a conference, or work on a specific research project. In other instances, fellowship recipients provide an annual one-page summary or have no reporting requirements. Several well-funded US universities have comprehensive guides on the different external fellowships available to doctoral students; we refer readers to specific examples (MIT Office of Graduate Education, n.d.; NC State University, n.d.; Northwestern University, n.d.).

3. Methods

Given the limited and contradictory research on graduate fellowships, we focus the current study on fellowship funding. The use of qualitative interview methods allowed us to engage with participants and listen to their stories as told in their own words (Creswell & Poth, 2018), while the open-ended nature of qualitative research is well-suited for emergent findings on understudied topics. We intentionally designed the study as an exploratory study of fellowship funding. Our goal was to understand the range of experiences of students primarily funded through fellowships, given the limited and contradictory findings related to how fellowships function as a funding mechanism.

3.1. Participants

This study took place at a US research-intensive public institution well known for high research activity, highly ranked programs, and a high success rate for external graduate fellowships. We recruited participants from four engineering departments through an email sent by their respective graduate program coordinators. Students interested in study participation were directed to complete a Qualtrics screening survey prior to receiving an interview invitation. Following screening, twenty-three doctoral student interviewees met our qualifying criterion of being primarily funded on a fellowship (i.e., greater than 50% of their funding) for at least one semester during graduate study. Given our goal to understand the range of experiences of fellowship-funded students, we sampled students with both internal and external fellowships.

Our sample comprises 19 women and 4 men doctoral students. Of the men, 3 identified as Black and one as Latino. Of the women, 12 identified as white, 4 identified as Asian, 1 identified as Black, and 2 identified as multi-racial (Latina and white; Asian/Pacific Islander and white). Three shared their status as international students. Participants were from chemical, civil, electrical, and mechanical engineering. The year in program varied from Year 1 to Year 7. Internal fellowships primarily funded 8 participants, while external fellowships primarily funded 18 participants; 3 participants had both internal and external fellowships. Internal fellowships include recruitment (i.e., first year) and continuing (later year) fellowships. The graduate school, college of engineering, and engineering departments funded internal fellowships. A variety of US government agencies, non-US governments, and foundations funded the external fellowships.

3.2. Data Collection

We conducted and audio recorded interviews with the 23 participants during fall 2019, in person in a secure location, except for one Zoom interview. The first author, a graduate student at the time, interviewed all participants. Interviews lasted between 31 and 77 min, with most interviews taking about one hour. We used a semi-structured interview protocol (Miles et al., 2014), created by the first and fourth authors. The interview protocol included questions about the fellowship application process, fellowship requirements, their work while on fellowship funding, benefits and drawbacks of fellowship funding, and comparison between their experiences on fellowships and any other types of funding (i.e., research and teaching assistantships). Our framing of this study as exploratory informed the creation of our interview protocol. Questions on the protocol focused broadly on a variety of aspects related to fellowship funding and the experiences participants had while being funded through fellowships. Agency emerged in participant responses during the interviews.

3.3. Data Analysis

We transcribed the interviews and removed identifying information. We employed constant comparative analysis methods of grounded theory (Charmaz, 2006) to uncover patterns and relationships, to better understand agency in the context of different fellowship circumstances.

During the first round of coding, we focused on the experiences of participants while primarily funded on their fellowship(s). Initial inductive codes that emerged (n = 26) from this line-by-line initial coding included advisor interactions, project choice, lifestyle choices, and access to resources. Then, we identified fellowship circumstances for each participant; see

Table 1. Abbreviated codebook.

Agency Codes
Agentic perspectives—perspectives or views taken when experiencing opportunities and constraints, e.g., seeing situations as possible to overcome, recognizing that they can get certain types of support, feeling in control of their career goals, and viewing critical feedback as a way to learn and grow		Agentic actions—strategic and enacted with self-awareness of goals and contexts, e.g., managing advisor relationship, getting needed support from people other than advisor, asking for help when needed, and taking steps to obtain skills or knowledge to advance their career goals
Inductive Codes (to describe experiences)
Flexibility—flexibility around project(s), advisor(s), or personal life arising from fellowship conditions	Access—access to physical resources, people, and networks, learning new skills, or research experiences as direct or indirect result of fellowship funding		Validation—internal or external perceptions of the student from having received a competitive or prestigious fellowship
Fellowship Circumstance Codes
Fellowship timing—when fellowship funding starts with respect to year of study in graduate school and the duration of the funding	Fellowship source—internal fellowships are from the participant’s university; external fellowships originate from organizations other than their own	Fellowship structure—non-monetary aspects, including fellowship requirements and optional opportunities for involvement	Financial—monetary compensation associated with the fellowship, typically in the form of a salary or stipend, including benefits

for finalized code definitions (fellowship timing, fellowship source, fellowship structure, and financial). We additionally coded for advisor attitudes and practices (as a fellowship circumstances code). Next, we completed a third round of axial coding (for the experience codes) to collapse these findings into three main inductive code themes (flexibility, access, and validation); Charmaz (2006) labels this focused coding. Across these first three rounds of coding, we discussed via analytic memo the relationships between the emergent codes and themes and the lens of agency.

We transitioned to theoretical coding around the central theme of agency when we identified and labeled interview passages that were related to either agentic actions or agentic perspectives as defined by O’Meara et al. (2014). In this culminating round of coding, we first assigned codes for “agentic actions” and “agentic perspectives” to the full data set. We then analyzed the overall relationships between agentic perspectives, agentic actions, fellowship circumstances, and experience related inductive codes, searching for patterns. Process codes were developed to elucidate these relationships. Figure 1 illustrates the resulting relationships identified among the levels of coding.

Charmaz’s (2006) stance on use of prior literature is that it should sensitize the researcher but not dictate the emergent theory. Given that agency has been previously conceptualized as having only two components (agentic perspectives and agentic actions), it is appropriate that our interpretive and cyclical engagement with the data and existing literature expands these components substantively.

3.4. Quality

Several aspects of our process address trustworthiness (Lincoln & Guba, 1986). The first two authors met weekly to define and assign codes, using analytic memos to capture the evolving codebook, with definitions and examples, and writing summaries of the emerging findings (process reliability, Walther et al., 2013). We present the codebook in Table 1 and the final relationships between codes in Figure 1 (Walther’s communicative validation). The two coders met with the fourth author frequently and at least once between each round of coding to discuss findings (Walther’s procedural validation) and with the entire author team to ground findings in the graduate education literature (Walther’s communicative validation).

3.5. Author Positionality

As qualitative research is highly interpretive through the perspectives of the researchers, it is important to acknowledge the experiences and lenses we bring to this study (Charmaz, 2006). The first two authors were graduate students at the time of analysis and writing, both funded at various points primarily by internal fellowships, which granted us unique insight into the experiences of our participants and informed our analysis. We counteracted potential over-interpretations of the findings by coding the interviews together and discussing all findings with the overall project team. All authors have graduate degrees in science, engineering, or education disciplines, bringing an interdisciplinary perspective to our work. Two of the authors have held administrative positions with responsibility for managing graduate education portfolios, which brings institutional understanding of how the US graduate education system tends to work.

4. Findings

We organize our findings into two main sections: (1) agentic perspectives among fellowship recipients, and (2) agentic actions among fellowship recipients. Within each section we discuss the contexts and fellowship circumstances that allowed students to take agentic actions or perspectives. We also discuss when certain fellowship circumstances interacted with context to limit agency. For reference,

Table 2. Fellowship characteristics, agentic perspectives and agentic actions by participant.

Pseudonym	Fellowship Source	Timing of Fellowship Start	Fellowship Duration	Agentic Perspectives						Agentic Actions
				PF	AF	APN	ALNS	IV	ER	PF	AF	PFQL	APRC	ARE
Kalia	Internal	Year 1	1 year						+	−		−
Priya	Internal and External	Year 1 (I) Year 3 (E)	1 year 3 years	+, −	+			+	+			+
Malik	External (2)	Year 1 (E1) Year 4 (E2)	5 years 3 years	+	+	+		+	+	+	+	+, −		+
Caitlin	External (2)	Year 3–4 (E1) Year 5 (E2)	1 year 7 mos					+				+	+
Nadia	External	Year 1	5 years	−							+	−		+, −
Jamal	External	Year 1	5 years			+				+		+, −		+
Sophia	External	Year 3	3 years	+	+, −		+		+	−	+	−
Elaine	Internal	Year 1	2 years	+				+
Allison	External	Year 1	3 years		+				+	−		+, −
Taylor	Internal	Year 1	3 years	+								+		+
Sydney	Internal and External	Year 1 (I) Year 3 (E)	1 year 3 years				+		+			+
Caleb	External	Year 3	3 years										+
Carlos	External	Year 2	3 years	+	+	+		+, −	+	+		+
Hannah	External	Year 2	4 years	+	+	+		−	+		+	+, −	+
Avery	External	Year 2	3 years		+		+	+	+	+	+	+
Emma	External	Year 3	3 years	+	+		+		+	−		+
Jenny	External	Year 1	3 years	−	+					+, −
Maeve	External	Year 3	3 years	+			+				+
Keysha	External	Year 1	4 years	−	+	+	+			+		+, −
Mei	Internal	Year 1	1 sem		+		+
Rose	External	Year 1	3 years		+		+
Charlotte	Internal	Year 1	5 years		+			+
Melissa	Internal and External	Year 2 (E) Year 5 (I)	3 years 1 year		+			−	+	−	+	+, −

PF = project flexibility; AF = advisor flexibility; APN = access to people and networks; ALNS = access to learning new skills; IV = internal validation; ER = external recognition; PFQL = personal flexibility and quality of life; APRC = access to physical resources and collaborators; ARE = access to research experiences.

lists summary fellowship characteristics for each participant as well as the agentic perspectives and agentic actions discussed in their interviews. We saw the most examples of agentic perspectives leading to agentic actions for experiences aligned with project or advisor flexibility.

4.1. Agentic Perspectives

4.1.1. Flexibility Influencing Agentic Perspectives

Participants described agentic perspectives due to multiple types of flexibility associated with their fellowship. One primary example is in Project Flexibility. While some participants took agentic action in choosing their projects, Hannah enjoyed the agentic perspective of project flexibility without acting upon it: “I also feel like I can just go down some weird research path because I wanted to. Not to stick to my advisor’s plan. Even though I rarely do deviate.” A related agentic perspective was the awareness that students in other situations might be forced into multiple projects. Sophia “fear[ed] that I would have had a lot of other smaller projects bogging me down if I didn’t have the fellowship, other smaller projects that my advisor has promised on some grant application.” Participants discussed project flexibility much less frequently in the context of internal fellowships. However, Elaine and Taylor surmised that their internal fellowships resulted in less pressure to produce during their first two years as compared to grant project-funded research assistantships.

Not all agentic perspectives related to project flexibility were positive. Some participants shared that lack of guidance made making decisions about their future and trajectory difficult. Jenny negatively experienced increased project flexibility with her multi-year external fellowship, attributing her lack of guidance to her advisors’ reasoning of “You’re free. Well, I’m not going to tell you what to do because I don’t have to pay for you.” Jenny worked on multiple discarded projects, saying in frustration, as if to her advisor, “Okay, I guess maybe I learned something, but I kind of burned my time right there, on the [fellowship]” struggling with the “tiny bits of guidance you gave me.” Jenny was frustrated, with a less agentic perspective toward research deriving from the flexibility of her fellowship because her advisor did not provide needed guidance. In her fourth year, a co-advisor intervened to resolve the situation, but it wasn’t clear to Jenny what agentic actions would have been possible for her to take. Her experience illustrates how advisor support and project goals are essential for forming positive agentic perspectives for students entering graduate study with fellowships.

Even with a supportive advisor, the lack of formal deliverables and external collaborators can be difficult to navigate. Keysha felt that lack of deliverables negatively impacted her degree progress “because this was not industry-sponsored, so there was no pressure to deliver.” Alternatively, Priya explained that working on a grant-funded project while on fellowship supported her progress, saying “I think the fact that there is grant money attached to him being a PI actually holds him accountable, because I have a very scatterbrained advisor.”

Participants also discussed Advisor Flexibility as contributing to agentic perspectives. Malik, Jenny, Keysha, Rose, and Charlotte commented on how their multi-year external fellowships at the start of graduate school expanded their advisor selection options. Choosing an advisor is not in itself an agentic action, since all students must eventually do so, but recognizing expanded options is an agentic perspective toward the advisor selection process. Keysha stated, “when I came into my first year and people had to decide which professor to work with, I kind of had the luxury of choosing whichever professor because I already had my funding”—likely an overestimated perspective given the need for advisors to agree to take on new students. Elaine, Taylor, Sydney, and Charlotte believed that having a previously established relationship with an advisor prior to graduate school led to first year internal fellowship funding for each of them.

Beyond initial advisor selection, fellowships also gave Priya, Malik, and Hannah the agentic perspective that they could leave their advisor if needed. Hannah described how her funder was

pretty adamant about the money follows you, so had I decided to switch universities or finish my PhD at a national lab that the money would stay with me. So, it’s not really contingent on you sticking with your advisor if something goes wrong.

(Hannah)

Again, while others took the agentic action of changing advisors, these students had the agentic perspective that switching was an option if their situation deteriorated.

Some also shared that perceived leverage with advisors through financial freedom shaped their agentic perspectives. While most of these also led to agentic actions, the students described their perspectives in detail. For example, Sophia’s explained, “I feel [my advisor’s] more willing to pay for [international conference travel and bioinformatician consulting hours]. She’s not paying my tuition, she’s not paying my salary, so I have more resources and access to that.” Regarding internships, Melissa said, “So if I’m like, ‘Hey I’m going to do this thing this summer,’ which [my advisor’s] not really thrilled about. But it’s like he can’t tell me not to because he doesn’t pay for me anyway.” She described this changed dynamic as, “the increase in power or academic freedom.” Emma felt positive psychological impacts: “I feel less personal pressure on myself […] I just don’t really feel like [my advisor’s] ever going to get upset at me if things go wrong.”

4.1.2. Access Influencing Agentic Perspectives

Participants also viewed Access to People and Networks as contributing to agentic perspectives. These connections to influential individuals helped shape fellowship recipients’ career goals and ways they could envision themselves in engineering. Several participants gained access to people and networks through their external fellowship funding, including fellowship recipients and alumni and research collaborators, whom they accessed through conferences and email listservs. Malik, Jamal, Carlos, Hannah, and Keysha felt a sense of support and community through their fellowships’ peer and alumni networks, most often interacting during conferences. Jamal found value in building relationships with fellows from different institutions by attending a conference: “I’m still in touch with some of those people.” Carlos found that attending a conference with peers, postdocs, and faculty fellows helped him through a rough patch with his research. He said, “And they were [a] very, very supportive community. I feel like just being there really energized [me].”

Online platforms provided by certain fellowships also influenced these agentive perspectives towards career development. Malik and Carlos spoke positively of their fellowship-wide email listserv, where alumni and other individuals shared open faculty positions and funding opportunities. Carlos noted that people also used the listservs to receive advice, such as how to negotiate during the hiring process. The presence of the email listservs contributed to a feeling of support for Malik, as he explained, “it also makes me feel comfortable, because I’m, like, if I ever need someone at this university, there’s probably a [fellowship] professor there who I can reach out to and can help me.” Keysha also enjoyed her fellowship’s alumni updates, saying that, “it’s also inspiring to see just different women from different countries doing inspiring things.” In general, participants highly valued the peer and alumni networks associated with their fellowship funding, as Malik described, “sometimes more valuable than the money.” Participants didn’t describe specific agentic actions taken as a result of these interactions, but their agentic perspectives toward graduate study and eventual job searching were supported.

Fellowship recipient Access to Learning New Skills and academic norms also influenced agentic perspectives. Participants recognized how their communication skills developed, most often during the fellowship application process. Seven participants (Sophia, Sydney, Avery, Emma, Maeve, Keysha, and Rose) noted the writing skills they gained from writing a research proposal for their fellowship applications, viewing it as useful preparation for future grant proposals. In this way the process of applying for and receiving a fellowship helped influence confidence related to career trajectory. For many, it was their first time applying for funding. Sophia described how she learned to appeal to the interests of the funding agency and adjust the technical content depending on her audience, stating, “I learned a lot about telling a story.” Rose said, “I learned so much from identifying a knowledge gap” and “thinking about how my research could fit there. And it was just such a helpful way of looking at how to plan a research project.” Mei talked about the development of presentation skills for required presentations to industry fellowship sponsors. At this point, students had not taken specific agentic actions, but they had confident and optimistic perspectives about expanded future options.

4.1.3. Validation Influencing Agentic Perspectives

Participants felt a sense of Internal Validation from their fellowship awards, such as feelings of respect, appreciation, and pride. These feelings contributed to their agentic perspectives in how they thought about their future capabilities and career choices. Malik pointed out the difference in getting validation from an external source: “you sometimes feel like your professor is sort of like your mom, who’s always going to say that you’re good.” Priya discussed imposterism, saying “Someone else believes that I can do this even though I might not. […] Honestly, it’s one of the reasons I’m still here.” Avery described how this validation (perspective) allowed her to change advisors (action): “It’s somebody else having faith in funding me as a scientist. That was like, ‘I don’t deserve to be treated like this.’” In the context of graduate imposterism, she said the fellowship “gives you a little bit more dignity to treat yourself with a little bit more respect. I think that’s definitely one of the factors that got me out of a bad situation.” Elaine saw the money connected to her internal fellowship as a sign of the institution’s interest in her, saying, “It definitely made me feel like the school valued whether I was there or not.”

Participants also acknowledged the External Recognition associated with the prestige of winning a competitive award. Ten participants (Kalia, Priya, Malik, Sophia, Allison, Sydney, Carlos, Hannah, Emma, and Melissa) stressed how their fellowship will help with future job prospects, bolstering their resume or CV. From their perspective, receiving the fellowship as an award signals their competence and abilities to others including potential employers. Priya noted that “you have a tendency to be rewarded” after receiving a fellowship. Allison echoed, “That definitely is going to be helpful for whatever is next […] That will help people recognize, ‘Okay, you do good stuff.’” While a few participants alluded to slightly increased pressure to perform because of their fellowship, recognition was predominately a positive aspect shaping agentic perspectives related to career trajectory.

4.2. Agentic Actions

4.2.1. Flexibility Influencing Agentic Actions

Participants discussed how flexibility in their fellowship experience influenced agentic actions related to trajectory. Some expressed that Project Flexibility was a key factor in what actions they pursued. For example, Malik had a positive experience in designing his own dissertation research project. He described his advisor’s enthusiastic response when he approached her with his idea and stressed that his advisor guided him through project development and execution, with weekly check-in meetings. Jamal spoke positively of the greater ability to choose his projects, stating that his funding allowed him to choose between many project options his advisor presented him during his first year.

Other students felt forced by the conditions of their fellowships to take actions with less agency. Participants who received a multi-year, external fellowship in their second year or later described limited project flexibility gained from fellowship funding. In most cases, they continued working on the same projects. As Allison described, “it’s the same things no matter how I was funded. I show up, I do the same things […] My fellowship or financial situation never dictated that.” Emma similarly reflected, “certainly not, [fellowship] has really made no difference in the direction that my projects have gone in, which projects I’ve chosen to work on.” These examples illustrate how fellowship circumstances (year awarded, or timing) specific to an individual’s situation influence the degree to which they experienced project flexibility (i.e., no additional flexibility), which subsequently did not lead to more opportunities for agentic actions (see Figure 1).

Two participants (Nadia and Avery) on multi-year external fellowships used the Flexibility provided by fellowships to leave their advisors. Avery found herself in a toxic work environment and needed to leave her lab and advisor: “I couldn’t ignore it anymore. It affected my mental health and physical health”, which was a major source of “stress and anxiety.” She expanded on the harmful actions of her advisor, explaining, “when you’re working 60 h a week and then you’re told, ‘oh, you’re not working enough,’ or, ‘you’re not trying hard enough,’ those things get to you after a while.” Her advisor had “never graduated a female PhD student,” and she watched another female student leave early with a master’s degree. Avery’s fellowship helped her transition between advisors, since “It was just attached to me.” When Avery informed her advisor that she was leaving, he told her to relinquish her funding, since, “He believed he was responsible for my fellowship.” However, she refused “because I earned that.”

Fellowship funding also gave participants additional leverage in their interactions with their advisor, redirecting financial support from advisors for agentic actions such as attending more conferences (Sophia and Maeve), completing internships (Hannah and Melissa) and working remotely and taking vacation days (Malik). Another student (Maeve) shared that one of the major benefits of her fellowship was that she did not need to continually submit more applications to secure her own funding. These purposeful choices highlight the agentic actions related to advisor flexibility that fellowship recipients were able to make.

Finally, fellowships allowed these graduate students to take agentic actions related to Personal Flexibility and Quality of Life. Many participants mentioned how their fellowship funding allowed them to live a comfortable graduate student lifestyle due to higher stipend rates as compared with assistantships. Emma was able to live alone, moving to a nicer apartment closer to campus upon receiving her second fellowship. Two participants mentioned they were able to eat better food; Allison explained, “I wasn’t eating SpaghettiOs. I wasn’t eating ramen. If I wanted to have shrimp that week, I went and bought shrimp that week.” Taylor used the increased salary to purchase a gym membership and equipment for hobbies. Participants also discussed the ability to save money for unexpected expenses, stop carefully tracking their monthly budget, invest their money and pay off loans. The stability of multi-year funding allowed recipients to plan and take agentic actions with their finances.

However, not all fellowships have equal benefits, and one participant, Nadia, needed to take on additional employment to pay for her childcare costs. Other participants experienced complications with their fellowship salary in terms of lack of consistent pay rates, issues with payment processing and fellowship funding ending before they graduated, which negatively impacted their personal flexibility. For example, the foundation that funded Malik granted cost-of-attendance-based fellowships. When he received a second fellowship, he encountered institutional barriers:

I have to use them simultaneously because I need [first fellowship] to pay for my tuition, and I need the [second fellowship] to pay for my rent. But then, the Financial Aid office will say, “Well, now you have $[X] fellowship, so we’re going to deduct that from your other funding” […] I’ve had to write letters and convince Financial Aid that I need this money.

(Malik)

Institutional complications also occurred for Hannah while completing an unpaid TA assignment required by her department. The US government agency funding her fellowship deducted a 20 h TA salary from her stipend, and she had to escalate the issue to the Dean of the Graduate School when her department refused to pay her for the TA assignment. Allison, Keysha, and Melissa had decreased salaries in their fifth year compared to previous years since their fellowships only covered the first four years of graduate study. Their reactions varied from minor inconvenience to intense frustration that less experienced graduate students were receiving higher salaries. Timing and duration of fellowship funding contributed to the decreased pay, since most multi-year fellowships were for three or four years. Sophia expressed her displeasure about a supplemental fellowship stating, “It was four years, which is kind of a bummer because I think that’s misleading. Because most people don’t graduate in four years.” These fluctuations in expectations, constraints, and institutional navigation elicited the need for students on fellowships to take actions related to their career trajectories that conformed to the context of their fellowship rather than that of their educational program.

4.2.2. Access Influencing Agentic Actions

Participants discussed how Access to Physical Resources and Collaborators influenced their agentic actions while on fellowship. Several participants’ external fellowships granted them access to equipment, services, research spaces, and collaborators. As Caitlin summarized: “But really, it is the research opportunities. It’s being able to have access to more expertise and more technology […] I’ve done better research while on fellowship” and “I couldn’t have done anything like this back at [institution].” Caitlin and Hannah’s US government fellowships required them to work in national laboratories, using their laboratory space and equipment while making professional connections with others within their field. Hannah also received a small stipend each year for supplies and conference travel, and had additional course requirements for her fellowship. Caleb worked with research collaborators at another university for the requirements of his foundation-funded fellowship, where he spent two weeks learning in their research laboratories. Of note, multi-year, external fellowships funded all participants who discussed access to physical resources and collaborators.

However, some fellowships provided access to physical resources which participants did not use, most concerningly when advisors limited students’ agentive actions. This situation prevented participants from taking on internships as part of their funding. A participant explained how engaging in specific programs tied to her fellowship funder was not an option: “Not that I have applied to them, because for instance, my advisor does not like us going to another lab for a few months…my advisor does not allow that.” In other cases, students did not access internships because “none of the institutions unfortunately were in my area of interest.” This serves as an example of fellowship circumstances (advisor) specific to an individual’s situation influencing the degree to which they experienced access to physical resources (i.e., that it was limited), which did not lead to an increased opportunity for agentic actions (see Figure 1).

Finally, fellowships influenced agentic actions through access to research experiences and to graduate study itself. Malik and Jamal attended graduate school only because they received a foundation-sponsored multi-year external fellowship prior to graduate school. Jamal acknowledged the financial complications of graduate school, explaining “for a person like me, money is always an issue” and “I would have not gone to graduate school if I would not have had this fellowship.” Malik highlighted the financial complications with continuing his education, saying, “a big one for me, why I wanted to work so quickly, was your family’s sort of depending on you or waiting for you to be somebody to help them out and have an income.” He went on to say, “But knowing that I had this funding, I think, sort of opened that door for me.” He was not aware of who funds graduate students, coming from a primarily undergraduate institution. Nadia, who was funded for up to five years by a non-US government, would not have been able to attend graduate school in the US without her external fellowship: “So for being international and then getting funding is hard. So, I was like, ‘If I have scholarship or internship, then I can have this opportunity and then come and study here.’”

5. Discussion

We explored how fellowship funding promotes or inhibits agency for engineering doctoral student recipients. Our findings add to the literature by providing further insight into the benefits and drawbacks related to fellowship funding, with an explicit focus on what circumstances lead to different experiences and enactment of agency. Prior research can be summarized as “fellowships seem beneficial for employment (Blume-Kohout & Adhikari, 2016; Kinoshita et al., 2020), detrimental to skill development (Grote et al., 2021), and mixed for retention (Mendoza et al., 2014; Ampaw & Jaeger, 2012)” with little to no explanation of the mechanism or contextual circumstances that may also influence these outcomes. This prior work is predominately situated in the overall landscape of STEM doctoral education and often with a broad focus on all funding mechanisms (fellowship, research assistantship, teaching assistantships). The fellowship-specific findings in prior work are typically presented in contrast with other funding mechanisms. The present study focuses solely on fellowships and extends prior work by illustrating and providing examples of the varied experiences of fellowship-funded engineering doctoral students. We provide insight into specific experiences (i.e., project flexibility, advisor flexibility, etc.) and what fellowship circumstances influence certain experiences and how (i.e., timing of fellowship, fellowship source, etc.). As should be expected, the agency of fellowship-funded doctoral students is complicated by context, and while we can describe the influential circumstances, we cannot draw oversimplified conclusions such as the best form of funding or set of circumstances ideal for all engineering doctoral students.

In sum, we found that fellowships have the potential to provide flexibility, resources, and recognition—essentially, to provide options. Fellowship recipients take agentic perspectives when they realize they have additional options or can avoid less-desirable alternatives, as compared with students in other funding situations. Agentic actions are taken when students decide among these options, selecting from a variety of projects, designing their own dissertation project, switching advisors, doing an internship or research at a different site, or traveling to certain conferences. Although the benefits are often financial in nature, both agentic perspectives and agentic actions augment the graduate school experience of fellowship recipients through reduced stress and unique opportunities for growth. In short, fellowship funding can disrupt power dynamics between student and advisor, giving the student more agency and control over their experience. In our data set, fellowships awarded by entities external to the institution and from the beginning of graduate study offered the most flexibility and agency to recipients.

However, fellowship awards do not uniformly lead to positive experiences. Some of the most detrimental effects of fellowships we saw occurred when advisors were unavailable or unwilling to provide scaffolding for students to become independent researchers—the goal of most doctoral programs. Such scaffolding is more readily available in research and teaching assistant positions (Graddy-Reed et al., 2021; Herzig, 2004; Szelényi, 2013), as well as some fellowships tied to project grants or with reporting requirements. There were also several issues with the timing and financial structures of specific fellowships, the onus of which fell on the students and limited their agentic perspectives and actions. Thus, agency is a useful lens for understanding the nuances of how fellowship funding can support engineering doctoral students. We address our research questions below.

5.1. RQ1 How Does Fellowship Funding Contribute to or Undermine Agency of Graduate Student Recipients?

Fellowship funding has the potential to change the power dynamic between doctoral student and advisor. With a fellowship, a student may be less dependent on their advisor for employment and other resources, and thus feel less obligated to spend their time exactly as dictated by the advisor. Melissa described it as “the increase in power or academic freedom,” while Emma felt “less personal pressure on myself” to avoid disappointing her advisor. This extends to work–life balance considerations such as vacation time and disposable income for personal travel and hobbies, which some participants explicitly noted as a higher standard of living than their peers. These benefits may also contribute to students’ wellbeing and the ability to cope with the pressures of doctoral study.

A faculty member’s supervision style may be magnified with fellowship-funded advisees. For example, a “hands-off” advisor, which students in prior studies have described as leading to more independence and agency (Maher et al., 2020), may become too hands off and fail to provide sufficient guidance to a fellowship student. In the current study, Jenny perceived “burned” or lost time resulting from this type of relationship with her advisor, struggling to make progress with only “tiny bits of guidance.” Keysha also described her progress as impeded by “no pressure to deliver” on her research project. An advisor who is more “hands-on” or focused on enculturation to a workaholic academic culture (Lee, 2008) may find that they can’t control a fellowship student as expected because the student now has power to refuse. Melissa explained that if her advisor is “not really thrilled about” her summer plans, “he can’t tell me not to because he doesn’t pay for me anyway.” The advisors most lauded by our fellowship recipient participants had a mix of emancipation and critical thinking goals as defined by Lee (2008) coupled with a functional focus on project management to scaffold students’ growth. For example, Malik designed his own dissertation research, supported by weekly meetings with his advisor.

This finding aligns with prior work highlighting structural considerations for fellowships to be beneficial for students (e.g., Herzig, 2004), particularly with respect to relationships with faculty to support student progress. Advisors also seemed to limit students’ agentic actions in accessing optional benefits, such as internships and fellowship-related conferences, that would entail time away from campus (e.g., “my advisor does not allow that”). By not requiring certain kinds of engagement, students lost out on agentic actions because of their advisors’ emphasis on time in their lab.

To summarize, our participants described several different agentic perspectives. First, these fellowship recipients recognized that they had more options in selecting an advisor, particularly when they began graduate school with the fellowship in hand. Keysha “had the luxury of choosing whichever professor” as her advisor. As Artiles et al. (2023) found, engineering doctoral students have little flexibility and few information sources available when selecting an advisor and often must have an advisor tied to their assistantship funding mechanism upon matriculation. Given the importance of the advisor for graduate experiences and success (Bork & Mondisa, 2022; Burt et al., 2016; Charles et al., 2021; Lee, 2008; Mackie & Bates, 2019; Perkins et al., 2020), feeling greater agency in making—or if ever needed, changing—this critical decision is an important affordance granted by a fellowship. As Hannah explained “the money follows you” and is therefore “not really contingent on you sticking with your advisor if something goes wrong.”

Second, interview participants highlighted that having a fellowship meant feeling less pressure to make progress on research. Hannah said, with her fellowship she “can just go down some weird research path…Not to stick to my advisor’s plan. Even though I rarely do deviate.” Third, interview participants demonstrated awareness that a community of fellowship recipients and alumni could be advantageous to their careers. Some participants noted access to diverse peers and collaborators through this network as particularly comforting, which could be a useful support mechanism for students from marginalized groups in particular. Malik described his fellowship alumni network as “sometimes more valuable than the money.”

The final two agentic perspectives we uncovered related to the confidence that the fellowship helped spark. Students noted confidence in proposal writing and public speaking built through their fellowships, both in terms of preparing the application materials and completing required presentations. Given mixed results of the development of these kinds of skills of students funded via fellowships relative to students funded via research assistantships (Grote et al., 2021), it is important to consider how fellowship administrators can ensure structures facilitate such confidence-building opportunities for all recipients. Students also described the validation associated with being awarded a prestigious fellowship. This kind of external recognition by a selection committee was helpful for students’ agentic perspective that they were capable of being successful in graduate study. Priya cited her fellowship award as evidence that “Someone else believes that I can do this even though I might not.” These fellowship recipients also recognized how the prestige of the fellowship could pay dividends in the future, effectively describing the Matthew Effect (Merton, 1968), whereby having the fellowship on their CV would act as a positive feedback mechanism to enable other opportunities in the future.

We also saw several instances of agentic actions. First, two participants described how having an external fellowship was the sole reason they decided to attend graduate school rather than working full time. Jamal directly stated, “I would have not gone to graduate school if I would not have had this fellowship.” Haines et al. (2023) showed that funding stability is an important consideration for students from marginalized groups as they are considering graduate school, such as Malik’s reference to family expectations to start working and contributing financially. Second, Avery changed advisors when she did not feel supported, and Jenny relied on her co-advisor. Since their funding was not beholden to any particular advisor, the students had more agency in determining who could best support their research development. Avery’s fellowship helped her realize “I don’t deserve to be treated like this,” transferring her fellowship to work with a different advisor, much to the improvement of her “mental health and physical health.” These agentic actions afforded by fellowships are a crucial benefit, given the negative influence of financial concerns on graduate students’ mental health (Charles et al., 2021; Grady et al., 2014; Mackie & Bates, 2019). We essentially observed a reduced interdependency between the advisor and graduate student, just like resource dependency theory would predict at the organization level (Pfeffer & Salancik, 2003). At the individual level, the faculty member wields less power over the graduate student, which seemed to outweigh any negative short-term implications for the student of a disagreement with their advisor, as has been shown by O’Meara et al. (2013).

Finally, we saw evidence of agentic actions with respect to research projects and networking. Some students selected their own projects because of the autonomy afforded by their fellowship funding. They also acted on opportunities to access different facilities and collaborator networks, such as through national labs. We also saw evidence that since advisors did not have to pay fellowship students’ stipends and tuition, students could use project funds for networking and development opportunities such as conference travel. While our study is focused on fellowships, it is important to note that opportunities such as networking would be beneficial for all graduate students, regardless of primary funding mechanism (research assistantship, teaching assistantship, etc.).

5.2. RQ2 What Fellowship Circumstances, Such as Timing and Funder, Determine the Extent to Which a Fellowship Will Impact Doctoral Student Agency?

Although prior research has described benefits of fellowships in aggregate (e.g., Szelényi, 2013), this study suggests there are important considerations around the timing, financial structures, and nature of the fellowship. If a student received a fellowship after their first year, they did not experience the same agentic perspectives or take agentic actions regarding project flexibility and advisor flexibility as described above. At that stage, students felt less flexibility around their choice of advisor or research topic, although other benefits were noted. Work habits and expectations seemed to be established by that time with little opportunity for renegotiation. Allison described it as “it’s the same things no matter how I was funded. I show up, I do the same things,” and Emma’s fellowship “made no difference in the direction that my projects have gone in.”

Internal fellowships also tended to limit agency, at least partially due to their shorter duration. In these cases, the advantages seemed to be more for the advisor (e.g., saving money) rather than to fundamentally change the nature or power dynamics of the advising relationship to the students’ benefit. The advisor might play a larger role in nominating the student or supporting a nomination, and it’s less clear to students whether the fellowship is transferable to another advisor. For some students, asking such questions could risk falling out of favor with an advisor who controls their academic future. Students also described the length of the fellowship as potentially problematic. Students focused on the effective reduction in their stipend when these funds ran out, since fellowships typically pay more than the research and teaching assistantships these students took on to fill the gap to graduation. That shift added financial strain and sent a problematic signal when students’ pay rates declined despite increased research skills.

Finally, whenever there were administrative challenges in processing awards or institutional policies to navigate with external fellowships, the fellowship recipients—and not the institution—bore the negative experience or gap in pay. Hannah and Malik both had to petition university level administrators to resolve issues of severely reduced pay due to fellowship policies meant to prevent duplicate payments. These situations were challenging for students and negatively affected their agentic perspectives and limited the possibility of agentic actions.

5.3. Contributions to Theory

As compared with other studies that have coded for agentic perspective and agentic actions taken by STEM doctoral students (e.g., Borrego et al., 2021; Griffin et al., 2023; Thiry et al., 2015), we found stronger agency among our participants. However, these prior studies tended to focus on the general lack of agentic perspectives among graduate students related to career planning, particularly for nonacademic careers. Given their sampling focused on STEM doctoral students, where careers trend towards academia, the limited agentic perspectives found in prior work are perhaps unsurprising. Our work adds to the current understanding of agency in graduate school by illustrating examples of increased agentic perspectives and actions in an engineering context, which may be more focused on industrial employment.

It is important to consider that this study exclusively sampled students with fellowships, and that applying for competitive funding may be considered an agentic action in itself. Further, this group was successful in their fellowship applications and had developed agentic perspectives about their graduate training experiences and how fellowship funding augmented those experiences. The question remains as to whether having fellowship funding helps students develop agentic perspectives or students who already have agentic perspectives are more successful at obtaining fellowship funding. In our sample, participants generally worked in well-funded research labs that also had high success rates of students obtaining fellowship funding. They noted a general expectation of applying for certain fellowship opportunities but that funding through their advisor’s research grants would be available if they had not been successful. However, a few participants were required to apply for fellowships because their advisor did not have funds for them or proactively sought out fellowship opportunities.

The value of agency, then, is in the broader context of a graduate student’s experience. Agency cannot be separated—nor can funding—from the interactions, dependencies, and pressures on doctoral students (Griffin et al., 2023; Jaeger et al., 2017). The current study’s contribution is to bring subtleties of funding into the conversation about graduate student agency in context. Prior research on graduate funding has tended to compare major categories of funding, sometimes atheoretically, so it’s no surprise that definitive conclusions about the impacts of funding in isolation can barely be drawn—no single funding mechanism can ensure a positive graduate student experience. The current analysis explains how and when fellowship funding changes the nature of the student–-advisor relationship and graduate student agency.

5.4. Implications

Our findings have a number of implications for academic leaders. Graduate programs should be more proactive in matching fellowship recipients with advisors who are most likely to provide needed support, for example by requiring that a mentoring plan is included with fellowship nominations. Internal fellowship administrators can stipulate that deliverables, presentations, and progress reports or individual development plans be put into place if faculty wish to advise such students. Programs and institutions should have funds on which they can draw to support students as they navigate administrative issues in processing fellowships (which occurred for a few of our study participants, leaving them without pay). Institutions, rather than students, should absorb the financial strain, since the implications for students can be catastrophic. Institutions, particularly universities, should consider how additional networking opportunities could be incorporated into awarded fellowships. Given the benefits of access to people and networks as highlighted in our study, it also may be beneficial to set up such opportunities for all doctoral students, regardless of funding mechanism.

Faculty advisors should consider how to best provide the needed structure for fellowship students who are not involved in externally funded research projects. Our findings indicate that fellowship students struggled with too much freedom, particularly if this occurred earlier in their degree program. To help provide structure, faculty advisors should incorporate fellowship students into their research groups and actively facilitate connections with other group members. For example, fellowship students could be assigned a senior mentor in the research group who has expertise in a relevant topic. Fellowship students would likely benefit from a similar structure to students funded through research assistantships, with schedules for research deliverables and regular meetings with their faculty advisor where they receive actionable project feedback. Students funded on research assistantships regularly work towards dissemination through conference and journal publications. Faculty should implement similar goalposts for fellowship students, given the integral skill development and project management skills involved with writing. In short, faculty advisors should be cognizant that too much freedom can be harmful for fellowship students if they do not receive proper support.

Faculty members should also recognize and allow students to access the added benefits of fellowships such as internships and networking conferences. Although such experiences may temporarily delay research progress, they also have the potential to increase motivation and lead to full-time employment offers, easing stress during the dissertation writing phase. Graduate students with external fellowships should negotiate with potential advisors to plan research activities and deliverables that will engage them in mentored research experiences. As multi-year fellowships approach their end, students should work with advisors early to identify other funding opportunities.

Our results have several implications for entities that fund external fellowships. Students pointed to the range of benefits associated with networking opportunities, professional development, and accessing resources and facilities—our work supports continuing and expanding those efforts. External funding entities could require advising supports, progress reports, and other structures to ensure fellowship recipients receive proper scaffolding. Many European national fellowships (e.g., the Luxembourg National Research Fund, Marie Skłodowska-Curie Doctoral Networks) already have such supports embedded within their fellowship programs, requiring fellowship recipients to complete a Personal Career Development Plan (PCDP) that is regularly updated (Luxembourg National Research Fund, n.d.; European Commission, n.d.). Similar models may be beneficial to adopt within the US to provide additional supports for fellowship recipients. In addition, it could be helpful to require some of the supplemental opportunities, such as internships and opportunities for training or collaboration with other institutions, following the model for European national fellowships.

Finally, our results demonstrate how having an external fellowship prior to matriculation at a graduate institution might make attending graduate school feasible and give students the agency to choose a supportive advisor. External entities should carefully consider the range of institutions from which they recruit applicants. Elite, private institutions tend to have strong access to external fellowship dollars (Knight et al., 2018), so building partnerships to support fellowship proposal submission from students who attend a broader range of institutions could be a more effective strategy. Further, there is substantial variation across engineering disciplines in the proportion of doctoral students funded by fellowships (Knight et al., 2024), which raises additional concerns.

5.5. Limitations and Future Work

There are several caveats to interpreting our findings. First, our study took place at a single, selective, research-intensive US institution prior to the COVID-19 pandemic. Findings translate most readily to similar institutions. Only engineering students were included in our sample, and doctoral study in engineering tends to differ significantly from that in other disciplines (humanities, biological sciences, etc.). As this was an exploratory study, we did not include a comparative group primarily funded through research or teaching assistantships. It is possible that some of the participants’ negative experiences would have been exacerbated if they were funded through a different mechanism or that their fellowship funding gave them more agency to resolve potential issues that arose. Future work should examine agency within the context of different funding mechanisms.

Fellowship selection criteria are changing rapidly, and future work should explore the implications of such policy changes. Our sample includes students who have successfully persisted in their graduate programs and who agreed to interviews. Thus, it is possible our results are biased to positive experiences. We did not interview advisors, so interpretations of incidents are based entirely on student recall. This in-depth exploration of fellowship funding uncovered several circumstances of fellowship funding that impact students’ agentic perspective and actions and ultimately the quality of their graduate experiences. Such circumstances need to be considered by asking for more detailed data from participants in future studies of graduate students exploring the impacts of funding sources.

6. Conclusions

There are many benefits to graduate fellowship recipients, including flexibility with their projects, advisor, and personal life and feelings of validation. The students we interviewed appreciated the networking, mentoring, and collaboration opportunities built into their fellowships. The stability and independence of multi-year fellowship funding that follows a student affords them more agency to take advantage of such opportunities during their graduate study. To realize the full benefits of fellowship funding, careful research structuring and planning with the advisor are important. Those responsible for allocating fellowship funds should consider the implications of fellowship timing and duration for decisions between graduate study and full-time work, research direction, and student agency.