Superficial feedback by peers diminishes the efficacy of peer assessment. AI-assisted peer assessment might address this challenge, but its effect on vocabulary acquisition and academic resilience remains underexplored. Therefore, we examined how AI-assisted peer assessment contributes to EFL learners’ vocabulary acquisition and academic resilience. Applying a mixed-methods design, three intact classes were randomly divided into three conditions: an AI-assisted peer assessment condition, a conventional peer assessment group, and a control group. A researcher-made test measured learners’ baseline and post-intervention vocabulary knowledge. Secondly, narrative frames and semi-structured interviews explored learners’ perceptions and experiences of academic resilience. The results demonstrated that the AI-assisted peer assessment condition achieved the highest posttest vocabulary scores, followed by the conventional peer assessment group and the control group. Thematic analyses of qualitative data revealed that AI-assisted peer assessment contributed to academic resilience by promoting self-regulated persistence, assisting learners in managing uncertainty through conditional trust in AI mediation.
Cross-sectional research on academic resilience often identifies “low-SES but high-performing” students, offering limited insight into changes in high-performance status. Using two-wave longitudinal data from the China Education Panel Survey (2013–2015), this study classified 2867 low-SES junior high school students into four pathways: persistent non-high performance, resilience growth, sustained high performance, and fragile high performance. Fragile high performance denotes the observed transition from W1 high to W2 non-high performance under the adopted percentile-based definition. Low-SES students had lower entry and maintenance rates and a higher exit rate than middle- and high-SES peers. Machine-learning analyses showed that the incremental predictive value of baseline information differed across transitions, with the largest improvement for resilience growth. Baseline achievement position provided substantial predictive information, but full models outperformed percentile-position benchmarks. These findings show that equity-oriented educational evaluation should consider both disadvantaged students’ access to high performance and the stability of that status over time.
The association between teaching quality and socioeconomic disparities in student achievement remains underexplored, with existing studies offering competing perspectives. Two main hypotheses guide this debate. The differential effectiveness hypothesis suggests that certain teaching practices disproportionately benefit students depending on their socioeconomic background. In contrast, the inequality of educational opportunity hypothesis focuses on disparities in access to high-quality teaching across schools serving different populations. Drawing on TIMSS 2019 data from 46 education systems, we tested these hypotheses using mediation and moderation analyses. Our findings provide support for the inequality of opportunity perspective: in several countries, higher classroom discipline and instructional clarity partly mediate the association between classroom socioeconomic composition and achievement. By contrast, we find limited evidence for differential effectiveness, as neither instructional clarity nor classroom discipline consistently moderated the relationship between individual SES and achievement.
Tutoring is among the most effective learning interventions, yet it remains costly and difficult to implement at scale. Peer tutoring offers a promising alternative, but schools face barriers related to training, pairing, assessment, and classroom management. This study evaluates a technology-mediated peer tutoring system designed to address these barriers. Using a mixed methods design-based research approach with 56 middle school students in India, we examined learning outcomes and implementation processes. The system generated evaluative data streams and produced learning gains across 208 tutoring sessions, with students improving from 40% to 65% proficiency (estimated gain = 1.25/5, 95% CI [0.96, 1.55], d=.95). Qualitative findings revealed shifts toward learner-centered tutoring practices and highlighted compatibility-based matching in fostering comfort, risk-taking, and engagement. These findings demonstrate how technology can sustain the behavioral, motivational, and organizational conditions needed for peer tutoring to scale in real schools, while simultaneously generating continuous evaluative data to guide improvement."The happiness you get from teaching is just not comparable to anything else."- 7th grade student, Pune, India
Early engagement with formative assessment (FA) in pre-service teacher education is crucial, as it shapes how pre-service teachers (PSTs) conceptualise and enact assessment. Conducted in a four-year English language teacher education programme at a state university in Türkiye, this mixed-methods study examines how pre-service teachers’ FA literacy, beliefs, and practices evolve through a designed intervention across coursework and practicum. Results indicate that FA development is a dynamic, contextually shaped process that supports student-centred teaching and student engagement yet remains complex and contingent on personal and contextual factors. Sustained support from instructors and mentors enhanced pre-service teachers’ awareness, confidence, and responsiveness to student learning. The study highlights the importance of scaffolded practicum experiences that integrate reflective dialogue and progressively increased responsibility for FA.
