To investigate how advisor relationships influence early departure and research lab change considerations among doctoral engineering students, our study implements the Graduate Attrition Decisions (GrAD) model (Berdanier et al., 2020). Although originally developed to explain attrition, the GrAD model is well suited to investigating laboratory switching due to the similarity in departure from a program and departure from a lab influences. Further, both types of departure decisions should be seen as a continuum of decision-making rather than a binary or one-time decision. The GrAD model provides a structured approach to understand how various factors influencing graduate students’ decisions are interrelated, enriching the discourse surrounding the retention of engineering students. Berdanier’s formulation of the GrAD model was derived from an extensive qualitative analysis of graduate student feedback collected from social media platforms. The GrAD model identifies six interrelated themes (Figure 1), four of which have been explored in engineering and higher education scholarship concerning graduate attrition: advisor relationships, support networks, and cost (Berdanier et al., 2020). Among these, advisor relationships served as a central axis around which many of the other themes revolved. Advisors shape doctoral students’ experiences on multiple fronts (Zhao et al., 2007; Berdanier et al., 2020; Schlosser et al., 2003). Through the socialization they facilitate, advisors can strengthen or erode students’ academic and social support networks (Hocker et al., 2019; Weidman et al., 2001). Similarly, advisor dynamics can shape a student’s experience of cost, not just in financial terms, but also in emotional, research, and professional development aspects, as well as opportunity costs (Brawner et al., n.d.). The GrAD model explains how the advisor relationship intersects with and amplifies other naturally occurring factors, shaping graduate students’ experiences of stress, persistence, and decision-making within the broader doctoral ecosystem. Building upon the GrAD model, we seek to understand how the advisor theme of graduate attrition contributes to departure or lab change considerations while assessing the interaction of how students feel about assessing their progress. Figure 1 illustrates the GrAD model and emphasizes the intersection of advisor relations with student progress evaluation, departure decisions, and lab change choices.

Advisors serve as the central architects of doctoral success, with their multifaceted roles spanning three primary domains of influence. First, advisors support academic success by laying the foundation for degree progress and completion (Bair and Haworth, 2005) and guide students’ development throughout the program (Burt et al., 2019). Second, advisors contribute to personal growth by providing mentorship (Saddler, 2009; Artiles et al., 2023, 2025) and support socialization, helping students shape their scholarly identities (Sweitzer, 2008). Third, advisors shape the professional trajectory through the overall graduate experience (Lee, 2008; Shanachilubwa et al., 2023), facilitate professional development (Simon et al., 2015), and influence students’ future careers (Denton et al., 2025; Choudhary and Jesiek, 2015). However, advisor roles are not just faculty obligations; they are relational cues that doctoral students use to assess the quality of their relationships with their advisors. For instance, when advisors provide (or fail to provide) research guidance or psychosocial support (Sweitzer, 2008), students use these enacted roles as the basis for evaluating the quality of the relationship. One study indicated that the difficulty of research tasks intersects with the quality of the advisor relationship (Parker et al., 2019).

Scholars define advisor relationships using descriptive labels, which are based on how students describe their experiences or how researchers interpret those stories. According to McCain et al. (2024), student narratives reveal four primary relationship dynamics. Strained relationships are defined by conflict and negative emotion, whereas evolving relationships focus on growth and overcoming challenges. On the positive end, Supportive relationships provide professional support and satisfaction, while Egalitarian relationships elevate the student to a colleague, prioritizing collaboration and peer status. Distinct from student narratives, researchers rely on broad categorical labels to classify advisor relationships. They characterize problematic relationships as poor, negative, or weak, while describing supportive relationships as positive or supportive.

In some cases, these labels represent a distinction without a substantive difference. For example, Mackie and Bates (2019) argue that “poor” advisor relationships negatively influence mental health, while Barnes et al. (2010) reached a similar conclusion using the term “negative” advisor relationships. Moving beyond broad descriptors, Levecque et al. (2017) categorized advisor relationships into three distinct leadership styles: autocratic, inspiring, and laissez-faire. They found that the laissez-faire style is statistically linked to negative mental health outcomes.

In this study, we define a negative advisor relationship as one characterized by tension, conflict, or behaviors that undermine a student’s sense of support and belonging, such as dismissiveness, poor communication, or overtly critical interactions (McCain et al., 2024; Barnes et al., 2010). While similar, a poor advisor relationship, by contrast, reflects insufficiency or absence, where the advisor fails to fulfill expected roles or responsibilities, such as providing timely feedback, guidance, or resources, without necessarily displaying hostility (Mackie and Bates, 2019). While these categories are conceptually distinct in the literature, survey instruments tend to capture advisor relationship quality along a single continuous dimension. Finally, a positive advisor relationship involves constructive engagement, mutual respect, and supportive behaviors that foster student growth, well-being, and academic persistence (McCain et al., 2024; Mackie and Bates, 2019; Barnes et al., 2010; Levecque et al., 2017). These styles can be understood to reflect varying levels of autonomy, support and involvement as measured by the scale of the advisor relationship used in this study. Making these distinctions allows us to move beyond interchangeable labels and clarify the functional differences between relationships that are actively harmful, those that are inadequate, and those that are affirming and their implications for persistence and departure consideration.

Departure consideration refers to the evolving process by which doctoral engineering students begin to question their continuation in the program, ranging from initial doubts and intentions to leave, to active planning and concrete steps toward withdrawal (Bahnson and Berdanier, 2023). This perspective places attrition as a gradual and ongoing process, rather than a sudden event (Bahnson and Berdanier, 2023; Lovitts, 2001; Saddler, 2009; Artiles et al., 2023). Related terms used in the literature to discuss students’ graduate departure from their programs include intention to depart, consideration to leave, departure, or attrition consideration (Weidman et al., 2001; Artiles et al., 2025; Sweitzer, 2008; Lee, 2008; Shanachilubwa et al., 2023; Simon et al., 2015). Several factors influence departure considerations, with personal and program-related challenges often serving as critical events that prompt serious considerations of leaving the doctoral program (Artiles et al., 2023; Lee, 2008; Shanachilubwa et al., 2023). Among program-related factors, poor or negative advisor relationships have been strongly linked to departure considerations, with a growing body of literature identifying them as a primary influence on the departure considerations of engineering doctoral students (Berdanier et al., n.d.; Zhao et al., 2007; Donkor and Harshman, 2023; Artiles et al., 2023; Sweitzer, 2008; Denton et al., 2025; Choudhary and Jesiek, 2015; Parker et al., 2019).

Nevertheless, students with strained advising relationships often refrain from placing full blame on their advisors. Instead, they tend to internalize these difficulties, attributing them to personal shortcomings or perceived failures to meet the inherent rigor and demands of doctoral study (Simon et al., 2015; Denton et al., 2025; Mackie and Bates, 2019). This tendency reflects a broader pattern in the literature, where students are responsible for their own persistence or attrition, even when programmatic factors play a significant role (Lovitts, 2001; Artiles et al., 2023; Barnes et al., 2010; Levecque et al., 2017). Such internalization of blame may further complicate students’ ability to accurately assess their degree progress, reinforcing uncertainty and departure considerations. However, when self-blame fails to alleviate stress, emotional exhaustion, or mental burnout (Denton et al., 2025; Golde, 2005), students reach a tipping point and begin to seriously consider leaving their programs (Simon et al., 2015; Denton et al., 2025; Mackie and Bates, 2019; Zerbe et al., 2022). However, the departure decision is rarely impulsive. The decision is often preceded by a period of deliberate reflection, during which students weigh several critical factors: the severity of breakdown in the advisor–student relationship, the availability of external support or mediation, and the degree to which their academic path aligns with long-term career goals (Bahnson and Berdanier, 2023; Shanachilubwa et al., 2023; Denton et al., 2025; Zerbe et al., 2022; Bahnson et al., 2023). While these considerations are common, the decision-making process is far from uniform. It varies widely across individuals, shaped by each student’s unique personal circumstances, academic expectations, and coping capacities (Sallai et al., 2022) and can be strongly influenced by a student’s self-assessed progress (Borrego et al., 2017; George et al., 2018; Ruud et al., 2018; Sallai et al., 2023).

Progress in doctoral education rests on two pillars: completing coursework and producing original research, culminating in the dissertation (Borrego et al., 2017; George et al., 2018; Ruud et al., 2018; Sallai et al., 2023). Foundational coursework provides students with the theoretical grounding, methodological repertoire, and technical skills necessary for lab-based inquiry (Lovitts, 2005). Yet coursework alone cannot signal real progress, because doctoral study ultimately hinges on research productivity. Traditional indicators, such as GPA, rarely predict whether a student will complete the PhD (Lovitts, 2005). Rather, successful completion is more closely tied to the student’s capacity to evolve into an independent researcher (George et al., 2018; Ruud et al., 2018; Sallai et al., 2023). Advisors, therefore, play a pivotal role in this transition (Borrego et al., 2017; Bahnson et al., 2024). In this study, progress assessment is operationalized using a single self-report item capturing perceived clarity, rather than milestone completion due to the continuing research and writing time following completion of most degree milestones. They shape research direction, provide financial support, and connect students to professional networks that accelerate learning and degree completion (Borrego et al., 2017; Crede and Borrego, 2012).

In engineering programs, research is often carried out in alignment with advisors’ funded projects, offering students structured opportunities for hands-on experience and scholarly development (Borrego et al., 2017; Crede and Borrego, 2012). However, tension can emerge when advisors and students pursue divergent goals or operate with unclear expectations (Devos et al., 2016). Such misalignment can delay key milestones and lead to confusion about how progress is assessed (Sverdlik et al., 2018; Hoey, 2008). For instance, students may develop their own perceptions of progress that do not always align with those of their advisors. A healthy and constructive advisor relationship, marked by clear and collaborative communication, is particularly important to assessing research progress that is inherently context-dependent (Roy et al., 2023). When expectations are discussed openly and regularly, both students and advisors are better able to interpret milestones, assess competencies, and develop a shared understanding of academic progress (George et al., 2018; Ruud et al., 2018; Sallai et al., 2023; Bahnson et al., 2024; Zerbe et al., 2023). Positive advising relationships help bridge gaps in understanding and reduce the risks posed by mismatched research goals or ambiguous evaluation criteria (Roy et al., 2023). In such environments, students are more likely to see themselves as active collaborators, which fosters mutual trust, open dialogue, and confidence in navigating the challenges (Berdanier et al., n.d.; Ruud et al., 2018; Roy et al., 2023). However, for wide variety of reasons students still may find departure from a lab to be the best, only, or least disruptive option to persist in the doctoral endeavor.

Attrition is not always a binary process of staying or leaving; sometimes, students switch labs as an alternative to leaving doctoral education altogether. Lab change can restore a sense of support and mitigate the risk of attrition among doctoral engineering students (Bahnson and Berdanier, 2023). In preparation for changing labs, speaking with a potential advisor is often the ‘recommended’ next step. Yet, this process requires explaining to a potential advisor the reasons for the transition, which can raise concerns about retaliation if the change is unsuccessful (Borrego et al., 2017; Crede and Borrego, 2012; McAlpine and Amundsen, 2011). Faculty members may also be reluctant to accept a student from a colleague due to the potential strain on professional relationships (Matthews, 2019). These relationships are typically well-established and may extend beyond the student’s time in the program, making faculty cautious about disrupting long-standing academic networks. Further, universities have policies on academic progress, advisor relationships, and funding that may inadvertently create hurdles for students seeking to leave or raise concerns about their experiences (Lovitts, 2001; McAlpine and Amundsen, 2011).

Therefore, doctoral students who consider switching labs often evaluate how the decision might affect their academic success, departmental relationships, the availability of a suitable new mentor, and the administrative hurdles involved including funding and cost structures (Bahnson and Berdanier, 2023; Zerbe et al., 2022). To navigate these complexities, they often take several preparatory steps. For instance, they may search for information on their own, consult with peers or social media communities (Berdanier et al., 2020), review program-specific guidelines or graduate college resources, and seek advice from program administrators (Bahnson and Berdanier, 2023; Berdanier et al., 2020). Moreover, the strategies students choose can also vary by nationality and gender. International doctoral engineering students, whose visa status is closely tied to their funding, tend to proceed more cautiously than domestic students (Gao, 2021; Ledford and Basilio, 2025). Gendered communication norms matter as well: women are more likely to use indirect channels and may feel less comfortable initiating difficult conversations with advisors or other authority figures (Wang and Houdyshell, 2021). In engineering’s male-dominated culture, this reluctance can arise from a desire to avoid confrontation or critical feedback (Holloway et al., 2020; Lovitts, 2005). While not directly tested in the analyses presented here, these dynamics can create a feedback loop in which some students feel supported and move forward with a lab change, while others hesitate because of perceived barriers or fear of retaliation.

This research is part of a larger mixed-methods project that examines the persistence pathways of doctoral engineering students as they change research labs to continue their graduate education. The goal of this analysis is to investigate the impact of advisor relationships on student considerations of departing their program without a degree or with a master’s degree; the roll uncertainty in progress assessment in the program in departure considerations; and the association of advisor relationship with actions to change research labs. The following are the research questions for this study:

As authors, we draw on our disparate training to investigate how advisor relationships influence doctoral engineering students’ progress assessment and departure considerations while acknowledging the importance of our personal experiences. The first author and principal investigator identifies as a White, queer man with expertise in engineering education and social psychology research. The second author identifies as a Black African graduate student with a background in engineering and education. His experience as an international graduate student, coupled with navigating lab changes, has provided him with valuable insights into the challenges faced by doctoral engineering students, particularly those who are international or from underrepresented groups. Our training from social sciences, education, engineering, and psychology provide multidisciplinary perspectives to support this research. With a balance of perspectives, we are able to look beyond individual disciplines to identify trends and themes. Our collective expertise positions us to extend the impact of our work beyond our individual disciplines. However, our lived experience has provided different views of changing labs and research advisor relationships. Indeed, power dynamics between the authors may have influenced research practice and interpretation reflecting trainee and supervisor roles under investigation in this manuscript. While not apparent in our collaboration, we are mindful of our interactions as reflecting the research topic under investigation. Simultaneously, we share common goals and a commitment to improving the educational experience of graduate engineering students.

The 2022 edition of Engineering and Engineering Technology by the Numbers (Profiles of Engineering and Engineering Technology, 2022) from the American Society for Engineering Education (ASEE) was referenced to identify the top 50 institutions awarding engineering doctoral degrees. Invitations were sent via email to 448 engineering graduate program directors, coordinators, or department heads, requesting that they share details about the study with their doctoral students. Doctoral students voluntarily participated in the study by completing a survey hosted on the Qualtrics online platform. Participants were given the opportunity to enter a drawing for a $10 Amazon gift card, with ten recipients chosen and notified via email, representing about 4% of the participants. The authors obtained approval from the institutional review board (IRB) for all data collection procedures.

Engineering doctoral students from 26 top-ranked universities in the United States and 17 doctoral engineering disciplines responded to an online survey on departure considerations, advisor relationships, and lab change actions. In our sample (n = 269), women (n = 108) are underrepresented compared to men (n = 121) with gender non-conforming individuals comprising a smaller proportion (n = 12), and 28 participants not self-identifying a gender. Among racial and ethnic groups, White students comprised a smaller portion of the sample (n = 102), whereas Asian students are well-represented (n = 105) with some representation from of Black/African American students (n = 13), Middle Eastern or North African students (n = 19), and Hispanic, Latino/Latina/Latinx, or Spanish origin students (n = 21).

Participants answered a series of questions on departure considerations, advisor relationships, and lab change actions as part of the online survey.

Departure Considerations were measured with two items: I am considering leaving my current program without a degree; and I am considering leaving my program with a Master’s. Participants responded to each item by selecting one of the following: Strongly disagree (1), Somewhat disagree (2), Neither agree nor disagree (3), Somewhat agree (4), or Strongly disagree (5). These items were analyzed separately because they represent conceptually distinct outcomes rather than indicators of a single latent construct.

Advisor relationship was measured with nine items measuring a single factor (Bahnson et al., 2024) with a stem (My Advisor…) asking students to rate items such as values my work; is easy to approach. Participants responded to each item by selecting one of the following: Strongly disagree (1), Somewhat disagree (2), Neither agree nor disagree (3), Somewhat agree (4), or Strongly disagree (5). The mean of the nine items is used as the advisor relationship variable. The scale has strong internal reliability (Cronbach’s α = 0.95) similar to existing research (Bahnson et al., 2023). The items developed in past research represent common advisor relationship needs in engineering doctoral education (Cass et al., 2017). Although the high internal reliability may indicate some redundancy in items, it does reflect strong coherence among the indicators for advisor relationship quality.

Progress Assessment was measured with one item: I find it difficult to evaluate my degree progress. Participants responded by selecting one of the following: Strongly disagree (1), Somewhat disagree (2), Neither agree nor disagree (3), Somewhat agree (4), or Strongly disagree (5). As a single-item measure, this construct should be interpreted as indicative rather than comprehensive.

Lab Change Actions were measured with six items. Students were asked to respond Yes (1) or No (0) to the following items: Discussed changing with your current advisor; Discussed changing with a program administrator (graduate program director, department chair, or similar); Searched for program information about changing research labs; Searched for graduate college information about changing research labs; and Discussed changing with potential new advisor(s). The last item was an optional item with a write-in space: Other actions in preparation for changing research labs or universities. These items were generated based on existing qualitative research on lab change in doctoral engineering programs (Bahnson et al., 2019; Bahnson and Abane, 2025).

Initial analysis of the relationship between advisor relationship and departure consideration used linear regression. A significant regression was found (F (1, 267) = 36.85, p < 0.001). The R² was 0.12 indicating that advisor relationship explained approximately 12% of the variance in departure consideration without a degree. While modest, this level of explained variance is meaningful in doctoral education research in which persistence decisions reflect multiple interacting factors. Similarly, a significant regression was found (F (1, 267) = 35.66, p < 0.001). The R² was 0.12 indicating that advisor relationship explained approximately 12% of the variance in departure consideration with a master’s degree. Again, the modest level of explained variance represents a meaningful amount of variation given the multiple interacting factors in doctoral persistence considerations. With both linear regressions, the advisor relationship is an important predictor of departure consideration.

The mediation analysis demonstrated advisor relationships strongly associated with degree progress assessment and progress assessment strongly associated with departure consideration for both departure without and with a Master’s degree. These relationships establish the importance of measuring the statistical mediation of progress assessment of the linear regression relationship between advisor relationship and departure consideration. Mediation analysis demonstrated progress assessment significantly, partially mediated the relationship between advisor relationship and departure consideration without and with a master’s degree (Table 2). The finding supported importance of a student’s ability to assess their progress in their departure considerations.

Advisor relationship significantly negatively predicts student’s engagement in several common steps to changing research labs (Table 3). Doctoral engineering students who consider changing labs due to advisor relationships do not make this decision lightly; they engage in various lab change actions, as reported in Table 3. The significant negative coefficients across different actions, such as discussing lab changes with advisors, program administrators, and potential new advisors, suggest that students experiencing poor advisor relationships may hesitate to initiate conversations about lab changes with their current advisor. Moreover, the negative association between advisor relationships and discussing lab changes with program administrators highlights how negative advising experiences may discourage students from seeking institutional support. Similarly, students in strained relationships may be less likely to search for program-specific resources or graduate college information, suggesting a reluctance to engage in formal decision-making processes related to lab switching. It is also possible that institutional policies or funding constraints, independent of advisor relationships, influence students’ willingness to pursue lab changes.

Our analysis indicates that advisor relationships are strongly associated with progress assessments and departure considerations. Mediation analysis demonstrated that progress assessment significantly and partially mediated the relationship between advisor relationships and departure considerations, both with and without a master’s degree. These results are consistent with existing scholarship, the quality of advisor relationships are strongly associated with doctoral engineering student success, satisfaction, research growth, and commitment to completing a doctoral degree (Academies, 2018; Hocker et al., 2019; Bahnson et al., 2024; Breen et al., 2024), and the characteristics defining both positive and negative advising experiences in doctoral engineering education (Barnes et al., 2010; Breen et al., 2024; Zerbe et al., 2023). Doctoral engineering students who experience negative advisor relationships may perceive their academic progress less favorably, which could increase uncertainty about degree completion. This aligns with prior research indicating that positive advisor relationships foster academic growth, self-efficacy, and retention in doctoral programs (Hocker et al., 2019).

Part of the negative effect of advisor relationships was mediated through progress assessment. This mediation reflects statistical association rather than causal sequencing. Lower advisor relationship scores were associated with greater difficulty in progress assessment, which could contribute to increased departure considerations with or without a master’s degree. This may be because advisors provide research direction, financial resources, and establish support networks that are vital for effective learning and progression (Crede and Borrego, 2012). A poor advisor relationship can leave doctoral students with ambiguity, uncertainty, and a diminished sense of support and guidance. This exacerbates their difficulty in navigating the complexities of their doctoral studies and increases the difficulties in achieving critical degree milestones, and conflicts in assessing degree progress (Bahnson et al., 2024; Sverdlik et al., 2018; Hoey, 2008), which could contribute to departure considerations. Nonetheless, this difficulty can be mitigated by a positive advisor relationship. For instance, doctoral students in a supportive and positive advisor relationship often feel like collegial partners working alongside their advisors, tend to communicate more openly about difficulties, seek guidance on complex research topics, and obtain critical support essential for degree progress (Bahnson et al., 2024; Sherman et al., 2021).

While some doctoral engineering students make departure considerations either with or without a master’s, others may consider changing labs as an alternative to find a more supportive environment. Reasons for doctoral students taking these actions may differ but based on findings in the literature, some factors may account for the lab change actions. For instance, doctoral students in academic environments where perceived loyalty and commitment to an advisor or research group are highly valued may worry that changing labs might harm their relationships with current advisors or negatively affect their reputation within their academic community (Zerbe et al., 2022; Crede and Borrego, 2012). Moreover, depending on the size and nature of the department, as well as how long students have been in the doctoral program, some may face increased scrutiny, disruptions to their graduation timeline, and jeopardized funding commitments (Dinsmore and Roksa, 2023). In other instances, the competitive atmosphere in academic settings, especially in engineering doctoral programs, can hinder open communication between students and advisors (Wofford and Blaney, 2021).

The importance of and significance of progress assessment demonstrates a need to include student progress assessment into the GrAD model of attrition. While course grades are included in the model, these represent a limited portion of the doctoral experience. Further, instructional courses are usually completed early in doctoral education and represent one of the first milestones achieved. An added model node of progress assessment could add meaningful connections to existing theme nodes such as Advisor, Support Network, and Goals and sub-theme nodes such as Coursework, and Certainty of Goals, and Change of Goals.

The GrAD model focuses on attrition as an outcome of the connected themes and sub-themes. However, our work demonstrates the possibility for a potential off-ramp from attrition consideration to changing research labs. Articulating changing lab investigation pathways for students through exploration of GrAD themes and sub-themes beyond the advisor could further benefit students and institutions seeking to continue in doctoral engineering programs.