Student Perceptions of Automated Feedback in Undergraduate Mathematics (STACK, Kenya)

Online assessment systems that deliver automated feedback offer a scalable means of supporting mathematics learning, yet most research on their use has been conducted in WEIRD (Western, Educated, Industrialised, Rich, and Democratic) contexts. This study examined participants' perceptions of automated feedback within a year-long undergraduate mathematics course at a Kenyan university, where the STACK system was implemented through a five-quiz model. A mixed-format questionnaire was completed by 146 participants, of whom 125 provided open-text responses. Closed questions were analysed using descriptive statistics, and open-text responses were coded using Lipnevich and Smith's (2022) framework of cognitive ('Understand'), affective ('Feel'), and behavioural ('Do') processing, with two independent coders. Participants reported that STACK-based homework was useful for learning and expressed a clear preference for it over paper-based assessment. A majority (71.2%) favoured structured, step-by-step question formats, valuing them for clarity, fairness, and the opportunity to demonstrate reasoning. Worked solutions were the most frequently valued feedback feature, while guiding hints were appreciated for supporting independent thinking. Coded responses were dominated by the 'Understand' function, indicating that participants largely interpreted feedback as clarifying their understanding. Participants also reported infrastructural and design challenges, notably workload, unreliable internet and device access, and difficulty entering mathematical expressions on mobile devices. The findings suggest that while core perceptions of automated feedback are broadly shared across contexts, aspects of the feedback experience are shaped by local infrastructure and educational context, underscoring the value of context-sensitive assessment design.© the author