Categories
Higher Education Language Development

Task Offloading to GenAI in the Writing Feedback Process and Its Effects on Writing Development

Lai and colleagues used a quasi-experimental design to investigate whether offloading different tasks to GenAI in the “draft–assess–revise” workflow affects English-as-a-foreign-language (EFL) students’ writing development. Conducted over seven weeks, the study involved 101 Chinese university students and compared three conditions: (1) GenAI drafting, student assessment, student revision; (2) student drafting, GenAI assessment, student revision; and (3) student drafting, student self-assessment, student revision.

Results showed that the GenAI drafting group performed best in writing ability (M = 18.21, SD = 2.21), significantly higher than both the GenAI assessment group (M = 16.30, SD = 1.70) and the no GenAI group (M = 15.15, SD = 1.79). This indicates that including GenAI in the feedback workflow generally benefits writing, but offloading the drafting task rather than the assessment task produces the strongest effects.

Regarding cognitive processes, the GenAI drafting group showed significantly higher cognitive engagement, utilizing more total prompts (median 20.50 vs. 9.00, U = 195, p < .001) and more learning-oriented prompts (median 10.50 vs. 8.00, U = 398.50, p < .05) than the assessment group. They also outperformed the other two groups in metacognition in problem identification (M = 4.41), argumentation (M = 2.82), and providing constructive feedback (M = 4.97) during peer review.

In writing self-efficacy, the GenAI drafting group demonstrated the largest improvement (M = 4.74, SD = .55), significantly higher than the other groups, whereas the GenAI assessment group (M = 4.42, SD = .53) and no GenAI group (M = 4.45, SD = .78) did not differ significantly. Interviews revealed these students experienced a stronger sense of control and competence because they could critically evaluate and revise AI-generated drafts. In contrast, some students in the GenAI assessment group worried about overreliance on AI feedback, leading to “false confidence.”

Overall, the study highlights that the educational value of GenAI in writing feedback depends on which tasks are offloaded to AI. Offloading the drafting task to GenAI while having students handle assessment and revision promotes deeper cognitive engagement, stronger metacognitive understanding, and higher self-efficacy. The authors recommend that writing instruction should prioritize AI-generated drafts paired with student evaluation and revision, serving as a catalyst for critique and reflection rather than replacing student thinking.

Source (Open Access): Lai, C., Pan, M., Guo, K., & Cui, Y. (2026). What task to offload to GenAI in the writing feedback process? – Effects of task-offloading approaches on EFL learners’ writing skill development. Computers & Education, 252, 105675.

https://doi.org/10.1016/j.compedu.2026.105675… Read the rest

Categories
Achievement Higher Education

Student agency in Colombian higher education: a dual-pathway model of mediation and moderation between student engagement and academic achievement

A recent quantitative study by Torres Castro and Pineda-Baéz examines the relationship between student engagement and academic achievement, with particular attention to how student agency functions as both a mediating and a moderating mechanism in this relationship.

The study surveyed 1,713 final-year students from six accredited private universities across five regions in Colombia. Grounded in social cognitive theory and the ecological model of student engagement, the study conceptualizes student engagement as a multidimensional construct. It encompasses ten indicators across four domains: academic challenge, learning with peers, experiences with faculty, and campus environment. Student engagement is assessed through the National Survey of Student Engagement, with scores standardized on a 0 to 60 scale. Student agency is operationalized as agentic engagement through the five-item Agentic Engagement Scale. This scale captures proactive behaviors such as expressing preferences, asking questions, and suggesting adjustments to instruction. Academic achievement was measured by self-reported cumulative grade point average on Colombia’s standardized 0.0 to 5.0 scale. The study employed hierarchical regression with robust standard errors, structural equation modeling, and bootstrap-based mediation analysis using R version 4.3.1.

The findings reveal that among the ten engagement indicators, only collaborative learning and student-faculty interaction are positively and independently associated with academic achievement. Among the five agentic engagement behaviors, only expressive voice emerged as a significant mediator, channeling approximately six percent of the total effect of collaborative learning on grade point average. Expressive voice also moderated the relationship between student-faculty interaction and grade point average in a compensatory pattern. The association between faculty interaction and achievement was stronger among students with lower levels of expressive voice, whereas it was attenuated yet remained positive among those with higher levels. Although effect sizes were modest, the findings demonstrate that student agency operates as a context-dependent dual mechanism.

The study further suggests that instructional practices should combine structured guidance with opportunities for student-initiated contribution, and that support systems should prioritize students with lower levels of agentic engagement.

Source (Open Access): Torres Castro, U. E., & Pineda-Baéz, C. (2026). Student agency in Colombian higher education: a dual-pathway model of mediation and moderation between student engagement and academic achievement. Higher Education, 1-20.

https://doi.org/10.1007/s10734-026-01669-3… Read the rest

Categories
K-12 Education Programme Evaluation

The Effects of Integrated STEM Education on K12 Students’ Achievements: A Meta-Analysis

Integrated STEM education refers to T&L of scientific, technological, engineering, and mathematical knowledge and skills in integrative ways, emphasizing the connection between abstract knowledge and real-world problems. Integrated STEM education is characterized by four core features: multidisciplinary integration, real-world application, authentic inquiry or design-based practice, and active student learning. Based on 124 extracted and coded studies (2010-2022), Chen et al.’s (2025) meta-analysis reports on the effects of integrated STEM education based on three main types of interventions: (1) adopting integrated STEM education, (2) using extra teaching and learning strategies to enhance integrated STEM education, and (3) using specific learning technologies to support integrated STEM education.

All three types of interventions yielded a medium effect on knowledge acquisition and a small effect on student perceptions. Besides, adopting integrated STEM education had a large effect on cognitive skills; using extra teaching and learning strategies in integrated STEM programs produced a medium effect on cognitive skills and problem-solving task performance; using specific learning technologies had a small effect on problem-solving task performance. Some factors, such as task type (inquiry or design-based task) and program duration, may influence STEM learning outcomes.

To maximize the efficacy of integrated STEM education, practitioners should embed its four core characteristics into curriculum design while favoring short-to-medium duration programs (one month to a semester). Educators must carefully balance hands-on design and minds-on inquiry tasks by providing necessary scaffolding tailored to students’ prior knowledge. Furthermore, deploying targeted instructional strategies and learning technologies can enhance engagement with complex, real-world problems. Ultimately, evaluating these programs requires a multidimensional approach that prioritizes skill development and practical problem-solving performance alongside traditional knowledge acquisition.

Source (Open Access): Chen, B., Chen, J., Wang, M., Tsai, C. C., & Kirschner, P. A. (2026). The effects of integrated STEM education on K12 students’ achievements: A meta-analysis. Review of Educational Research, 96(2), 619-668.

https://doi.org/10.3102/00346543251318297… Read the rest

Categories
Language Development Primary School Education Secondary School Education

The Effects of Interventions for Students With Reading Difficulties in Grades 4–12

Killingly and colleagues conducted a systematic review and meta-analysis of school-based interventions for students with reading difficulties in Grades 4–12 published between 2011 and 2023. The study examined the overall effectiveness of these interventions and further tested whether study characteristics, sample characteristics, and intervention characteristics moderated their effects. A total of 104 publications and 586 effect sizes were included, representing 97,114 participants. Methodologically, the authors used a Correlated and Hierarchical Effects model combined with robust variance estimation to address dependency among multiple outcomes and effect sizes within the same study, while also estimating effects across overall reading performance and specific reading domains.

The results showed that, overall, reading interventions had a small but significant positive effect for students with reading difficulties in Grades 4–12, with an overall effect size of g = 0.212 (95% CI [0.163, 0.261], p < .001). This suggests that although the gains were not large, these interventions did produce reliable improvements in students’ reading performance. Across specific reading domains, the strongest effects were found for vocabulary (g = 0.422), followed by decoding/word recognition (g = 0.199) and reading comprehension (g = 0.187). Fluency showed only a very small but significant effect (g = 0.080), spelling was not significant (g = 0.015), and phonological processing, although showing a larger effect size on the surface (g = 0.531), did not reach significance and was therefore considered unstable. Overall heterogeneity was very high (I² = 89.71%), indicating that differences in study design and sample characteristics had a substantial influence on intervention effectiveness. GRADE assessment further suggested that the overall quality of evidence ranged from moderate to high, with the strongest evidence for fluency, moderate evidence for vocabulary, and moderate-to-low evidence for phonological processing.

Moderator analyses showed that intervention effects varied according to both study and sample conditions. Overall, more recently published studies showed stronger effects (β = 0.015), and journal articles produced significantly larger effects (g = 0.268) than research reports (g = 0.062). In terms of sample characteristics, low socioeconomic status was not significantly related to overall effects, but a higher proportion of students with learning disabilities was associated with slightly stronger effects (β = 0.006). For students from a language background other than English, overall differences were not significant, but in the vocabulary domain, a greater proportion of such students was associated with stronger effects (β = 0.016), suggesting that vocabulary instruction may be particularly important for this group. Regarding intervention design, intervention focus, duration, and measurement type were all significant moderators. Comprehension-focused interventions showed relatively strong overall effects (g = 0.313), multicomponent interventions showed stable effects (g = 0.178), and word study interventions had smaller effects (g = 0.096), whereas vocabulary-focused interventions, though fewer in number, showed the largest effect (g = 0.716). Shorter interventions were actually associated with stronger effects, with effect sizes of g = 0.405 for 0–5 hours and g = 0.409 for 6–15 hours. In addition, researcher-developed measures yielded significantly larger effects (g = 0.542) than standardized measures (g = 0.127). Although there was no significant overall difference between interventions delivered by teachers and those delivered by researchers, in vocabulary interventions teacher-led delivery produced stronger effects (g = 0.733) than researcher-led delivery (g = 0.249), suggesting that classroom teachers may hold particular advantages in providing vocabulary support.

Overall, this study shows that reading interventions for older students with reading difficulties are indeed effective, although the magnitude of their effects depends on the reading domain being targeted and on the design of the intervention. Vocabulary and reading comprehension appear to be the most promising focuses, while multicomponent interventions also demonstrate stable benefits. By … Read the rest

Categories
K-12 Education Social and Motivational Outcomes

Social and Emotional Learning Programs and Students’ Prosocial Behavior: A Meta-Analysis

A recent meta-analysis conducted by Hung and colleagues examined the effectiveness of school-based social and emotional learning (SEL) programs for K–12 students’ prosocial behavior. Prosocial behavior is conceptualized as any voluntary behavior intended to benefit others, such as helping, sharing, comforting, and defending others. The researchers analyzed 66 studies and 157 effect sizes involving 52,914 youth.

Effect sizes were calculated using Hedges’ g, which includes a small sample bias correction to the effect size estimate to account for small studies. Because most studies contributed multiple effect sizes, the authors used a correlated effects (CE) model with robust variance estimation (RVE) to account for within-study dependence among effect sizes. Approach refers to whether an SEL program takes a curricular, interactional, structural, or combined approach. To examine whether the effect of SEL programs on prosocial behavior was moderated by sample, program, methodological, and publication characteristics, the authors conducted a mixed-effects meta-regression analysis with all moderators added to the model simultaneously. In total, they investigated 14 moderating variables such as approach, school level, urbanicity, and dosage.

The remaining moderating variables yielded no statistically significant differences. Results indicated that effects for rural, suburban, and combination areas were not statistically different from samples from urban areas. Results also indicated that effects were not statistically significantly different between samples with higher versus lower proportions of students qualifying for free or reduced-price lunch. Effects of curricular and curricular combined with structural or interactional approaches were not significantly different from SEL programs that only used an interactional approach. Findings indicated that effects of studies delivered at Tier 2 were not statistically different from Tier 1 studies. Studies that used a quasi-experimental design and single-group pre–post design yielded similar effects on prosocial behavior compared to studies that used a randomized controlled trial. Effects were similar across different types of prosocial behavior measures. Effects from studies that did not meet baseline equivalence were not statistically significantly different from studies that met baseline equivalence, and effects from studies that did not report implementation fidelity were not statistically significantly different from studies that reported fidelity of implementation. Results indicated that effects from studies conducted across earlier decades were not statistically significantly different from studies conducted more recently, and effects were similar for peer-reviewed and non-peer-reviewed studies.

The authors further noted that most studies were conducted with elementary school children (56%), the majority implemented universal Tier 1 interventions (89%), and a curricular approach was the most common (77%). Additionally, a considerable proportion of studies did not report key demographic data, with 71% failing to report free or reduced-price lunch rates.

A key implication for practice from this meta-analytic review is that school-based SEL programs are effective in promoting K–12 students’ prosocial behavior, and that “more is not necessarily better” — a moderate dosage and moderate duration may be most ideal. Future policy and practice should take into account this “less is more” finding. At the same time, more research is needed involving secondary schools, rural schools, non-curricular approaches, and diverse student populations in order to fully understand the effectiveness of SEL programs.

Source (Open Access): Hung, C., Brass, N. R., Brockmeier, L., Bergin, C., Imler, M., & Luper, S. B. (2026). Social and Emotional Learning Programs and Students’ Prosocial Behavior: A Meta-Analysis. Review of Educational Research, 00346543261438462.

https://doi.org/10.3102/00346543261438462… Read the rest

Categories
Effective Teaching Approach Secondary School Education

Creative Visual Programming for Secondary Students: Enjoyment, Self-Efficacy, and Gender Differences

Smit et al. (2025) examine how students’ enjoyment during visual programming tasks relates to their self-efficacy beliefs and gender differences in programming confidence. Grounded in Pekrun’s control-value theory of achievement emotions, the study focuses on whether positive emotional experiences in programming can strengthen students’ beliefs in their ability to program. The research was conducted in a daylong visual programming workshop titled “Creativity in Science and Technology-Smart Textiles”, where secondary school students programmed LED matrices connected to micro:bit devices (small programmable computers commonly used in STEM and coding education) and applied them to creative, real-world tasks such as smart shirts and bicycle shirts.

The study involved 269 lower-secondary students from 16 Swiss classes in Grades 7 to 9. Students completed pre- and post-questionnaires measuring self-efficacy for visual programming, while their momentary enjoyment was measured four times during the workshop through experience sampling. The course was structured to move from more guided tasks in the morning, including Morse code and debugging activities, to more open and creative tasks in the afternoon, such as designing smart textile applications. Structural equation modelling, including latent state-trait theory and latent growth curve models, was used to examine changes in enjoyment and self-efficacy over the day.

Results show that students’ enjoyment remained relatively stable across individual tasks and was largely shaped by their general enjoyment of programming rather than by specific task situations. However, students with lower initial enjoyment showed stronger increases during later, more creative tasks. Girls reported lower enjoyment than boys at the beginning of the workshop, but their enjoyment increased more strongly over time, narrowing the gender gap. Both girls and boys showed increased self-efficacy for visual programming by the end of the course. Although girls initially reported substantially lower self-efficacy than boys, the gender difference was no longer significant in the final model after the workshop.

Overall, the findings suggest that application-oriented and creative visual programming activities can foster students’ confidence in programming, especially among girls. The combination of smart textiles, visual coding, debugging practice, and open-ended design tasks appeared to create a motivating learning environment that supported positive emotions and self-efficacy development. The study highlights the importance of designing programming instruction around authentic, creative, and personally meaningful tasks, rather than treating programming as an abstract or purely technical activity.

Source (Open Access): Smit, R., Schmid, R., & Robin, N. (2025). Experiencing enjoyment in visual programming tasks promotes self‐efficacy and reduces the gender gap. British Journal of Educational Technology, 56(3), 1231-1247.

https://doi.org/10.1111/bjet.13523… Read the rest