2026· MATEC Web of Conferences· 0 citations· 10 references
TL;DR
The results indicate that what students do with feedback matters more than the source from which it originates, and that self-efficacy is the pivotal mechanism translating feedback experience into self-regulated learning.
Abstract
Feedback is a core mechanism of formative assessment, yet its effect on learning depends on how students conceive of and act upon it. This study tests an AI-enhanced student conceptions of feedback model in English as a Foreign Language (EFL) writing, relating six feedback conceptions to academic self-efficacy and self-regulation. Survey responses from 300 undergraduate students at a state university in Indonesia were analysed with partial least squares structural equation modelling (PLS-SEM) using 5,000 bootstrap subsamples. The measurement model met every reliability and validity criterion: outer loadings ranged from 0.743 to 0.888, composite reliability from 0.857 to 0.931 and average variance extracted from 0.609 to 0.750, while the Fornell-Larcker criterion, cross-loadings and HTMT ratios all confirmed discriminant validity. All seven structural paths were significant. Active use of AI feedback was the strongest antecedent of self-efficacy, followed by enjoyment, whereas ignoring feedback exerted a significant negative effect. Self-efficacy in turn strongly predicted self-regulation and accounted for 43.5% of its variance. The results indicate that what students do with feedback matters more than the source from which it originates, and that self-efficacy is the pivotal mechanism translating feedback experience into self-regulated learning.
With growing pressure in today’s world, students are facing burnout from meeting academic demands. This may influence students’ motivation and academic engagement, which causes growing concern in higher education. Therefore, this study is conducted to examine the influence of burnout on students’ motivation by integrating the Job Demands–Resources Model and Expectancy–Value Theory. These relationships are then tested using Partial Least Squares Structural Equation Modeling (PLS-SEM). A questionnaire consisting of three components including value, expectancy, and affective was distributed to 125 undergraduate students. The results were then analysed using SmartPLS, following a two-stage procedure involving measurement and structural model assessment. The findings revealed that the measurement model exhibited satisfactory reliability and validity across all constructs. Structural model results revealed that disengagement exerted significant effects on all three motivational components (value, expectancy, and affective), indicating its strong and consistent influence on students’ motivational beliefs and emotional responses toward learning. Exhaustion, however, showed a more selective impact, significantly influencing expectancy but not value or affective components. Overall, the results suggest that motivational withdrawal affects students' motivation greater than emotional fatigue. These findings contribute to a more detailed understanding of how multiple aspects of burnout may affect students' motivation. The study also highlighted that PLS-SEM is useful in studying the complex relationship between psychological constructs. The results of this study contribute to addressing students’ loss of interest and highlight the role of educators in supporting student motivation during lessons.
Saidah Ismail, Haslinda Md. Isa, Noor Hanim binti Rahmat et al.· International journal of res...· 0 citations
Academic underachievement in Indonesian vocational high schools (SMK) is often attributed to cognitive ability alone, yet non-cognitive factors may matter. Drawing on Bandura's social cognitive theory and Astin's involvement theory, this study tested whether student engagement mediates the associations of self-efficacy and learning motivation with examination performance. A cross-sectional survey was administered to 282 students in four state vocational high schools in Pesisir Selatan Regency, West Sumatra, sampled proportionately by school. Constructs were modelled as reflective-reflective higher-order constructs and estimated with the disjoint two-stage approach in PLS-SEM; performance was a single summative examination score drawn from school records and not equated across the four schools. After item purification the engagement construct retained only its behavioural and emotional dimensions, so H5 to H7 test a narrower construct than the introduction describes. Discriminant validity was not established for the self-efficacy and engagement pair (HTMT = .908, above the pre-specified .90). Self-efficacy (β = .640, 95% CI [.537, .743]) and learning motivation (β = .141, 95% CI [.027, .255]) were associated with engagement, explaining 55.7% of its variance. No path reached examination performance: the model explained essentially none of its variance (R² = .005, adjusted R² = -.006), and all three paths were statistically equivalent to zero within ±.20. Five hypotheses were not supported. The null result is consistent with criterion misalignment, but the design did not test that explanation.
Generative artificial intelligence (GenAI) has shown potential in supporting academic writing, yet limited research addressed the intention and actual use of this technology by foreign language students. To address this gap, the study employs an extended Unified Theory of Acceptance and Use of Technology 2 (UTAUT2) model by integrating AI self-efficacy and AI trust to investigate the factors influencing EFL students’ intention and actual use of GenAI in academic writing. An online survey using a cross-sectional design involving 481 purposively selected EFL students from Indonesian universities was analyzed using partial least squares structural equation modeling (PLS-SEM) with SmartPLS 4. The results revealed that performance expectancy, effort expectancy, social influence, habit, and AI self-efficacy significantly shaped EFL students’ intentions to use GenAI for academic writing, while behavioral intention, habit, and AI self-efficacy significantly influence actual use. In contrast, AI trust, facilitating conditions, and hedonic motivation show no significant effect on either intention or actual use. The model demonstrated strong explanatory power, with R² values of 0.806 for behavioral intention and 0.617 for actual use. The study enhances the explanatory power of UTAUT2 in GenAI-supported academic writing and offers practical insights for pedagogy and policy to promote GenAI literacy, critical evaluation skills, and responsible use in academic writing.
This study aims to analyze the influence of teachers’ teaching styles and social support on the self-efficacy and learning motivation of 11th-grade students at SMK Negeri 44 Jakarta. The method used was a quantitative approach with a population of 210 students; the research sample consisted of 136 students. Data collection was conducted via an online questionnaire using Google Forms. Data processing and analysis were performed using Partial Least Squares (PLS)-based Structural Equation Modeling (SEM) with SmartPLS 4.0 software. The results of the outer model test showed that all indicators met the criteria for convergent validity, discriminant validity (HTMT < 0.85), and reliability, with composite reliability and Cronbach’s alpha values above 0.70. Meanwhile, the inner model evaluation yielded an R-Square value of 0.413 for the self-efficacy variable and 0.457 for the learning motivation variable. Hypothesis testing showed that teaching style (β = 0.397; t-value = 3.908; p-value = 0.000) and social support (β = 0.318; t-value = 3.069; p-value = 0.002) had a positive and significant effect on self-efficacy. In addition, social support (β = 0.367; t-value = 3.040; p-value = 0.002) and self-efficacy (β = 0.281; t-value = 2.295; p-value = 0.022) had positive and significant effects on learning motivation. Conversely, the teacher’s teaching style did not have a significant effect on learning motivation (β = 0.151; t-value = 1.311; p-value = 0.190). The implications of this study indicate that students have high confidence in achieving their learning goals; future aspirations are the primary factor driving learning motivation; teachers are the source of social support most felt by students; and teachers’ nonverbal communication has supported the creation of engaging learning experiences, although the pace of speech still needs improvement.
Destria Amara, Dita Puruwita, Rizka Zakiah· TOFEDU: The Future of Educat...· 0 citations
This study investigates how AI-generated writing feedback influences English writing self-efficacy among Chinese university students and examines the dual role of achievement goal orientation as both a mediator and a moderator in this relationship. A quasi-experimental pretest-posttest design was implemented with 148 s-year English majors over a 12-week semester. Participants in the experimental group received feedback from a GPT-4-based automated evaluation platform, while the control group received conventional teacher feedback. Writing self-efficacy, achievement goal orientation, and AI feedback perception were measured at both time points. Results indicated that AI feedback significantly enhanced writing self-efficacy (d = 0.94), with the largest gains in ideation and self-regulation. Mediation analysis revealed that mastery-approach goal orientation partially mediated the feedback–efficacy relationship, accounting for 33.4% of the total effect, whereas performance-approach orientation did not function as a significant mediator. Moderation analysis showed that mastery-approach orientation amplified the positive effect of AI feedback on self-efficacy, while performance-approach orientation attenuated it. A moderated mediation model confirmed that the indirect effect through mastery-approach goals varied across learner motivational profiles. These findings extend social cognitive theory into AI-mediated pedagogical contexts and suggest that instructors should incorporate goal-framing activities to maximize the efficacy-building potential of AI writing tools.
Xiangyu Kan, Lin Li· Frontiers in Psychology· 0 citations
This research compares AI-generated and teacher feedback on student reflections in Japanese K-12 classrooms that adopt self-paced learning, focusing on feedback characteristics and learner receptivity. Two AI conditions were designed based on Hattie and Timperley’s four-level feedback model: Theory-AI, employing GPT-4.1-mini with this framework alone, and Context-AI, employing GPT-5.1 with an Intelligent Tutoring System (ITS) architecture that integrates learner and domain models. These were compared with teacher feedback (Teacher) in a within-subject design involving 110 students across four schools and five grade levels. Feedback characteristics (RQ1) were analyzed along three dimensions—volume, generation time, and content composition across the four levels (task, process, self-regulation, and self)—and their uniformity both between and within class groups. Learner receptivity (RQ2) was assessed via clarity, specificity, and empathy on a 5-point scale. For RQ1, both AI conditions exhibited higher proportions of process- and self-regulation-level feedback than teachers, whereas Theory-AI produced excessive self-level feedback. AI feedback was generated approximately ten times faster and exhibited substantially greater uniformity in both volume and content composition, whereas teacher feedback varied considerably between class groups (e.g., self-level inclusion: Cramér’s $V = .468$ ) and within the same class group (character-count SD: 9.4–37.1). For RQ2, no differences were observed in clarity; specificity ratings followed Context-AI > Theory-AI > Teacher, and empathy ratings followed Theory-AI > Context-AI > Teacher, with all pairwise differences significant. These findings indicate that ITS-integrated AI feedback offers an efficient, theoretically grounded complement to teacher feedback in K-12 education.
