Doctoral training is often judged through visible, linear indicators such as publication count, time to degree completion, and technical productivity, yet these measures do not explain why students with comparable preparedness, effort, and ability diverge sharply in their long-term trajectories. We argue that doctoral development is better understood as a nonlinear interaction between accumulated effort (a summation function), accelerated efficiency (a multiplicative function), and strategic direction (a power law). Effort builds capacity, efficiency converts accumulated work into momentum, and strategic direction determines whether that momentum is amplified, maintained, or constrained. To clarify the developmental structure of doctoral training, we frame it through three hierarchical stages (alignment, productive, and independence), which can be delineated by the three progressive functions, respectively. This framework synthesizes scholarship on doctoral socialization, mentoring, and independence with the growing reality that AI can substantially amplify efficiency without replacing the human judgment required to identify consequential questions, recognize significance, and exercise strategic judgment under uncertainty. We anticipate that this hierarchical, non-linear model can not only describe the doctoral process but also offer a broader way to think about development processes in individuals and societies.
This study identifies the key factors influencing higher education faculty members’ Sense of Teaching Agency (SoTA) among faculty at U.S. four-year research-oriented institutions. Sense of Teaching Agency refers to a faculty member’s perceived capacity to make intentional instructional decisions that influence student learning within the opportunities and constraints of their teaching context. Drawing on a three‑round Delphi process with a panel of 30 U.S.-based faculty, educational developers, and academic leaders, the study achieved expert consensus on eight core factors: time and workload management, institutional and/or departmental support, autonomy, self‑efficacy, content expertise, pedagogical expertise, technological expertise, and an individual’s resistance to change. Internal factors reached consensus more readily than external ones, suggesting that SoTA is largely grounded in faculty capacities and perceptions but moderated by structural and contextual conditions. These findings establish SoTA as a distinct dimension of faculty agency and offer an innovative framework for redesigning faculty development, workload expectations, and evaluation systems to support instructional autonomy, pedagogical innovation, and equity in higher education.
Student persistence and academic performance are central indicators of student success in higher education, yet they are often modeled using similar analytic assumptions despite reflecting distinct underlying processes. Guided by theories of student engagement and persistence, this study examines whether student persistence intention and college grade point average (GPA) differ in their structural complexity and whether analytic approaches of varying complexity offer differential explanatory value. Using data from 23,215 students in the 2024 Community College Survey of Student Engagement (CCSSE) in the United States, we compare pre-specified additive regression models, regularized regression models, and algorithmically flexible tree-based models to predict persistence intention (self-reported likelihood of reenrollment) and GPA using the same theory-aligned predictors. Results show that regression-based approaches performed as well as algorithmically flexible models for persistence intention and better than them for GPA, and that no model exceeded the majority-class baseline in classification accuracy for persistence intention. Theory-informed interaction terms suggested by the tree-based models did not improve fit or prediction when specified directly within regression, and a multiple-imputation sensitivity analysis indicated that substantive estimates were robust to the missing data produced by listwise deletion. These findings indicate that analytic complexity offered no predictive gains for either outcome, and that apparent context-dependent relationships require explicit specification and testing rather than being inferred from algorithmic flexibility alone. By comparing analytic approaches across two theoretically distinct outcomes, this study highlights the importance of aligning methodological choices with outcome structure and contributes to theory-informed research on student success in community colleges.
The rapid integration of Artificial Intelligence (AI) has compelled many higher education institutions (HEIs) to implement strategies that promote the use of AI-enhanced technologies in online learning environments to serve the needs of all students. However, there remains limited research on how online leaders examine the transformational impact of AI integration in online learning in higher education. Therefore, the study explored the perspectives of online leaders on the transformational impact of AI regarding how these leaders lead and manage online programs in the era of AI-driven online learning. It also examined challenges online leaders face in leading and managing online programs in the era of AI. Grounded in a transformational leadership theory through semi-structured interviews of 60 online learning leaders (vice provost of online learning) from 52 public and eight (8) private universities in the United States. Data analysis revealed the following themes: leadership initiatives on AI policy, guidance, and governance; leadership efforts: providing AI resources; AI as a transformational agent in administration, teaching, and learning (AI is changing online teaching and learning, Reimagining the potential of the Learning Management System (LMS); and challenges. The paper discusses the role of online leaders in navigating AI as well as guiding and managing the impact of AI on online teaching and learning.
