Trauma-Informed Design and Guided GenAI Use May Help Online Students Persist, Study Finds

A new peer-reviewed study from researchers at the University of Phoenix argues that student persistence in online higher education is not simply a matter of grit or motivation, but the product of learning environments that either amplify or ease the cognitive strain carried by non-traditional learners. The study, published July 15, 2026, in the journal Innovative Higher Education, examines how stress, self-regulation, resilience, belonging, psychological safety and generative artificial intelligence interact in the daily experiences of working adult students pursuing degrees online. Its central conclusion is striking: trauma-related regulatory strain can intensify the mental demands of navigating digital courses, and when institutions design for that reality, persistence becomes something an institution helps build rather than something a student must achieve alone.

The research was conducted by fellows and scholars affiliated with the University’s Center for Educational and Instructional Technology Research, known as CEITR, which is housed within the University of Phoenix College of Doctoral Studies. The author team includes Melinda Kulick, Jessica Sylvester, Chunfu Cheng, Christina Bergren and Jim Croushore, each of whom brings a practitioner background in assessment, higher education operations, research methodology, counseling and school leadership. Their article, titled “Designing for Persistence in Online Higher Education: A Trauma-Informed, GenAI-Integrated Model for Non-Traditional Learners,” introduces a conceptual framework intended to translate qualitative findings into actionable institutional design principles.

Methodologically, the study is a descriptive phenomenological investigation of 45 undergraduate students enrolled in six bachelor’s programs within the College of Social and Behavioral Sciences at a large U.S. online university. Participants ranged in age from 18 to 51 and older, with most between 36 and 50, and represented the varied employment, caregiving and prior educational experiences commonly associated with non-traditional online learners. The researchers used purposive sampling to recruit participants and collected data through an anonymous online questionnaire with open-ended prompts exploring stress, self-regulation, resilience, belonging, psychological safety and GenAI use. Respondents could answer in writing or by audio, an accommodation that itself reflects attention to accessibility.

Analysis followed Colaizzi’s descriptive phenomenological method, a structured qualitative approach in which researchers identify significant statements, extract their meanings and organize them into thematic clusters. Five themes emerged from the data, and these themes informed the development of the study’s central contribution, the Trauma-Informed AI Persistence Model, abbreviated as TIAIP. Importantly, the study did not assess specific trauma histories or clinical diagnoses. Instead, it examined participant experiences involving stresses such as financial pressure, caregiving responsibilities, employment instability and competing life demands, treating trauma-informed practice as a design lens rather than a clinical category.

The findings paint a vivid picture of how chronic stress reshapes cognitive capacity. Participants described how ongoing stress and competing responsibilities eroded concentration, problem-solving and their ability to engage fully in academic work. One participant explained that stress creates significant brain fog, noting that when stressed, the ability to problem solve diminishes incredibly. Another described the strain of managing school, work and a household simultaneously, saying it detracts from the ability to fully engage in academic work. These accounts align with established research on cognitive load, which holds that working memory is a limited resource and that regulatory demands from chronic stress consume capacity that would otherwise support learning tasks.

Course design emerged as a decisive factor in whether that limited capacity was spent learning or merely coping. Participants reported that unclear instructions, inconsistent course organization and difficulty knowing where to begin increased cognitive strain and made task initiation harder. Conversely, small acts of instructional acknowledgment carried outsized weight. One participant recalled that when an instructor acknowledged how hard it is to juggle everything, they felt seen, and it made them want to stay engaged. The researchers found that trauma-related regulatory strain intensified the cognitive demands associated with course navigation, pacing expectations and limited instructional presence, while belonging and psychological safety consistently supported engagement. In other words, the emotional climate of a course and its structural clarity are not separate concerns but intertwined determinants of whether students persist.

Generative artificial intelligence occupied a nuanced position in the findings. Participants described GenAI as a situational support that could reduce overwhelm and help them initiate academic tasks during periods of acute stress. One student said that when their brain is overloaded, AI helps them get started and takes away the panic of a blank page. Another explained that when feeling overwhelmed by an assignment or unsure where to start, AI helps break things down into smaller, manageable steps, providing structure that reduces stress and yields a clearer plan of action. These accounts position GenAI as a kind of cognitive scaffold for task initiation, the phase of academic work that participants identified as most vulnerable to regulatory strain.

Yet the same participants raised pointed concerns about AI dependency and unclear ethical boundaries around its use for learning. The study does not present GenAI as a replacement for human support or student thinking. As co-author Jessica Sylvester, a senior manager of College Operations, associate faculty and CEITR research fellow, put it, the more useful question is how institutions can create the conditions in which technology serves as an intentional scaffold alongside clear expectations, thoughtful course design and meaningful human connection. That, she noted, requires helping students understand both what these tools can support and where their limitations and ethical boundaries lie. The framing treats GenAI as a conditional support whose benefits depend on institutional guidance rather than an autonomous solution.

Building on the five themes, the authors introduce the TIAIP Model as a conceptual framework integrating trauma science, online learning design and guided GenAI use. The model positions student persistence in online higher education as an institutional outcome shaped by the interaction of trauma-related demands, digital course architecture and structured AI integration. It addresses how regulatory strain can intensify the cognitive demands of course navigation, pacing and limited instructional presence, while belonging and psychological safety influence engagement. In practice, TIAIP encourages institutions to reduce unnecessary barriers to persistence through clearer course structures, supportive instructional presence, deliberate attention to belonging and psychological safety, and purposeful guidance for GenAI use. Kulick, an assessment manager, associate faculty in the College of Doctoral Studies and CEITR research fellow, summarized the implication: for students managing work, family responsibilities and other pressures alongside their education, persistence cannot be understood simply as a matter of motivation or individual resilience. The findings, she said, point to the importance of designing online learning environments that reduce unnecessary cognitive demands, strengthen students’ sense of connection and give learners appropriate support for navigating moments when demands become especially high.

The authors are candid about the study’s limits. The findings provide context-specific qualitative insights from students at one online university and should not be generalized to all higher education populations. The research relied on self-reported experiences, and the open-ended questionnaire offered less opportunity for follow-up probing than interviews would have. Findings were not analyzed by intersecting demographic identities, and the TIAIP Model is a preliminary conceptual framework derived from the qualitative findings that requires further study and validation across institutions and learner populations. Even so, the study arrives at a moment when universities are simultaneously confronting rising numbers of non-traditional learners and the rapid, often unregulated arrival of generative AI in classrooms. By connecting trauma-informed pedagogy, digital course architecture and guided AI use within a single model, the researchers offer institutions a testable hypothesis: that persistence is designed, not merely demanded, and that technology, when bounded by clear ethical and instructional guardrails, can help students begin the work that stress once made impossible.

Subject of Research: Trauma-informed online learning design and generative AI support for non-traditional student persistence

Article Title: Study on trauma-informed online learning and GenAI support for student persistence published by University of Phoenix researchers

Article References: Study on trauma-informed online learning and GenAI support for student persistence published by University of Phoenix researchers. (n.d.). Original publication

Image Credits: AI Generated

DOI: Not provided

Keywords: online learning, student persistence, trauma-informed pedagogy, generative AI, cognitive load, psychological safety, belonging, non-traditional learners, higher education, qualitative research, instructional design, self-regulation