This case study examines how formative assessment performance, student engagement traces, and instructor feedback language co-vary with performance on a final integrative assessment. The study was conducted in a 12-week, second-year Framework for Higher Education Qualifications (FHEQ) Level 6) undergraduate course on empirical analysis and the scientific method (N = 49). We triangulate low-stakes assessments, platform-generated engagement indicators (talktime, hand raises, votes, reactions, and chat messages), and Linguistic Inquiry and Word Count 2022 (LIWC-22) features of instructor feedback. Psychological safety serves as a theoretical lens for interpreting students’ contributions, responses to feedback, and continued engagement after lower early formative scores; the empirical analysis rests entirely on observable assessment, participation, and feedback data.Classroom interaction data make some aspects of participation visible, but cannot, by themselves, reveal the learning conditions students experienced. We therefore treat behavioural and linguistic traces as descriptive features of instructional interaction. In-class verbal assessments were more strongly associated with integrative performance than group-based tasks, and students who maintained or increased participation following lower early scores showed upward formative trajectories. Instructor feedback was consistently characterised by high clout, positive tone, and analytic clarity, which we interpret as text-level features of the feedback environment. Grounded in educational psychology and learning analytics, the study offers a practical way to examine how assessment, participation, and feedback align with active learning instruction, using descriptive correlations, bootstrap confidence intervals, within-family false discovery rate adjustments, and an exploratory Classification and Regression Tree (CART) visualisation. The findings are specific to a single-course, small-cohort case based on one instructor’s assessments and synchronous records.
Situated against the backdrop of an evolving regulatory landscape, this study investigates the perceptions of United Kingdom university educators regarding the use of generative artificial intelligence in summative assessment. Through a mixed-methods survey of 330 academic staff, the research identifies a significant gap between high levels of digital literacy and low evaluative trust. While respondents report high familiarity with AI tools, confidence in the ability of artificial intelligence to provide fair or nuanced marks remains low. Qualitative analysis of open text questions reveals a nuance crisis, where educators perceive these systems as incapable of comprehending complex student submissions. The findings reveal significant gender disparities in technical proficiency, however, a consistent pattern of low evaluative trust persists across teaching experience levels, with no statistically significant differences detected based on professional seniority, suggesting a potential need to consider alternative pedagogical frameworks, such as performative and synchronous assessments, to safeguard academic integrity.
Artificial intelligence (AI) chatbots have rapidly transformed academic research practices amongst Nigerian university students. This study investigates the dual impact – beneficial and detrimental – of AI chatbot usage in research at Delta State University, Abraka, 2020–2024. Using a mixed-methods approach, quantitative data were collected via structured surveys from 320 students, whilst qualitative insights were gathered through semi-structured interviews with 27 faculty members and administrators. Secondary data included institutional documents and peer-reviewed literature, with coverage extending to 2025 for theoretical framing. Findings reveal that 78% of students use AI chatbots to brainstorm, summarise literature and structure assignments – enhancing productivity and access to global knowledge. However, over 60% admitted to submitting AI-generated content without verification or attribution, raising serious concerns about plagiarism and intellectual dishonesty. Additionally, Western-centric AI design limits cultural relevance in Nigerian contexts, potentially marginalising indigenous knowledge systems. This study concludes that AI chatbots are valuable academic aids when used ethically and critically but pose significant risks when misused. Recommendations include institutional AI literacy programs, transparent usage policies, localised AI development and alternative assessment strategies to uphold academic integrity. AI should complement – not replace – human reasoning in scholarly inquiry.
This study investigates the comparative efficacy of physical and online learning modalities in management education, focusing on student satisfaction and learning outcomes. It examines how improvements in Management Basics (MB), Management Organisation (MO), Management Techniques (MT), Problem-Based Learning (PBL), and Task-Based Learning (TBL) affect satisfaction across both environments. A quantitative research design involved 128 first-year management students in Taiwan across two semesters. Data from online and physical learning contexts were analysed using correlation and multiple regression to assess relationships between learning improvements and satisfaction. Regression analysis indicates that PBL exerts the strongest influence on satisfaction in both modalities, followed by TBL in physical learning. In online learning, MB and MO are key predictors, while MT and TBL show limited impact. The findings suggest that educators should prioritise PBL across both environments to enhance student satisfaction. This study contributes to management education literature by integrating course-unit and pedagogical dimensions into a unified framework for evaluating satisfaction across modalities. It highlights the differential impact of PBL and TBL in online versus physical settings.
What drives Generation Z (Gen Z) university students in Vietnam to adopt and keep using Mobile English Learning Apps: content quality or something else? This study proposes a Social Override effect, that is, a pattern in which peer validation and hedonic engagement, rather than content quality, appear to drive actual use even when information quality is adequate. Using an explanatory sequential mixed-methods design, this study analysed survey data from 150 Vietnamese university students alongside 10 semi-structured interviews. Information Quality did not significantly predict actual use (β = 0.147, p = 0.103); Hedonic Motivation did (β = 0.401, p < 0.001). This pattern is consistent with a hygiene-factor interpretation of content quality: its absence may generate dissatisfaction, but its presence does not, on its own, increase usage. The interviews help explain why. In this setting, peer recommendation and group-based app use create a buffer that lets students overlook technical limitations and advertising friction. Among these Gen Z students, trust was expressed through interface aesthetics and social visibility more than through content accuracy. Whether the same pattern holds beyond Vietnamese university students is a question this single study cannot settle.
