Open research questions in Innovative Teaching and Learning Methods

What the literature leaves open

• Future research should examine emotion regulation strategies as they unfold in interaction (Jones et al. However, a key limitation of qualitative observation is that it does not allow for generalizable conclusions about how emotional challenges typically occur or how groups engage in regulation.

  Outspoken emotions – Emotion regulation during emotional challenges in collaborative learning · 2026 · DOI

• Constructive alignment is thereby re-specified as a partial organizing principle whose explanatory scope is delimited by the expansion of design dimensions in EFL contexts.

  RECONFIGURING INSTRUCTIONAL COHERENCE: A FUZZY DELPHI VALIDATION OF A MULTI-DIMENSIONAL DESIGN FOR BIG IDEAS–ORIENTED EFL INSTRUCTION · 2026 · DOI

• Digital mathematics instruction demands robust problem-solving skills, yet the predictive relationship between Task-Technology Fit (TTF) and Metacognitive Self-Regulation (MSR) remains under-researched.

  <b>Mapping the Digital Math Classroom: A Path Analysis of Task-Technology Fit, Metacognitive Self-Regulation, and Problem-Solving Proficiency </b> · 2026 · DOI

• Future research should explore Human–AI co-regulation Page 16 of 20 F1000Research 2026, 15:1234 Last updated: 29 JUL 2026 across diverse educational settings, including K–12, vocational education, and professional learning environments. Future studies may investigate comparative effectiveness across different generative AI architectures, multimodal AI systems, or emotionally respon- sive AI agents.

  Human–AI Co-Regulation in Adaptive Learning: Developing GPT-Supported Self-Regulated Learning Models · 2026 · DOI

• transparency mainly in the system itself, in its accuracy, explainability, and oversight, and in the institutional safeguards around it. The selfregulatory paradox points to a condition these framings tend to leave implicit. A system may be technically transparent and still be experienced as opaque by a student who does not regard herself as entitled to read and act on her own assessment data. Transparency, on this view, is partly relational: it depends on the position the learner occupies within the system, not on the interface alone. The highest-rated item in our data makes this concrete. That students expect staff, rather than themselves, to act when analytics flag a risk of failure is itself a stance towards an algorithmic assessment function. It can serve equity, by prompting timely support, while also concentrating interpretive authority in the institution and leaving the classifying criteria out of students’ view—a cost borne most heavily by those least placed to contest a misclassification. Accessibility does not resolve this. Opening a dashboard to students is necessary but not sufficient literacy and interpretive scaffolding, access does not become the capacity to question an algorithmic output. The point extends to validity. Beyond whether a system measures what it claims to, its consequential validity (Messick, 1995) turns on how the outputs are used, and by whom; analytics that learners hand over wholesale cannot support the self-regulatory uses often invoked to justify them, which weakens the case for deploying them as learner-facing tools at all. Student expectations thus belong responsible design of automated the among to assessment, alongside its technical and policy dimensions. Fairness and transparency are not only engineered into systems; they are co-produced by the learners who have to recognize themselves as agents within them. transparency; without data inputs for 5.5 Limitations Several limitations should inform the interpretation of these findings. First, the study draws on a single institution, and the sample was a large but self-selected convenience sample recruited through a voluntary survey; women and the health and an rather system towards imagined social sciences are over-represented relative to the wider student body, so the descriptive proportions should not be read as population estimates, and generalization to other contexts cannot be assumed.

  The self-regulatory paradox of learning analytics: student expectations and the conditions for fair algorithmic assessment in higher education · 2026 · DOI

• The findings indicate that AI-enhanced listening, when deliberately integrated into a coherent curric- ulum and accompanied by transparent assessment practices, produces clear and meaningful improve- ments in listening comprehension in the present sample. The pattern of results is held under alterna- tive analytic specifications, suggesting that the advantage reflects substantive learning-related change rather than an idiosyncrasy of a single analytic choice. These outcomes align with contemporary syn- theses indicating that systems that offer adaptive practice, targeted feedback, and varied speech input are well-positioned to accelerate L2 listening development. Several qualifications circumscribe the breadth of inference. One limitation concerns the imbalance in engagement between groups. The experimental condition introduced multiple novels, high-interest tools (viz., chatbots, gamification, and VR), while the control group relied solely on textbook activi- ties. This disparity introduces the possibility of a novelty effect, as increased motivation and curiosity, rather than the AI tools themselves, may have contributed to the observed performance gains. Alt- hough typical in early-stage intervention research, this design asymmetry limits inferences about whether the advantage stems from AI-mediated scaffolding or from heightened learner engagement associated with new technologies.

  Tuning in With Technology: AI-Enhanced Listening Instruction in the Jordanian EFL Classroom · 2026 · DOI

• 42.39, p < .001, partial η² = .49. Adjusted means favored the experimental group (M = 16.56, SE = .29) over the control group (M = 13.94, SE = .29), in- dicating a substantial advantage for learners receiving coordinated, multi-tool AI instruction. These gains were observed within the B1 range of the assessment. Sequence and blend multiple affordable AI tools (e.g., chatbots, LingQ, auto- mated transcription, short VR scenarios) in short in-class rotations, and use platform analytics to tailor difficulty and feedback rather than relying on a single app or textbook. Replicate and scale the design across multiple schools and larger samples; use mixed methods to isolate which tool components drive gains; examine modera- tors (proficiency, digital literacy, teacher training); and, when possible, apply for- mal standard-setting to link outcomes to CEFR levels. Effective, cost-sensitive AI scaffolding can broaden access to international me- dia, higher education, and employment for secondary learners in under-re- sourced MENA settings, thereby supporting educational inclusion and social mobility. Pursue longitudinal, multi-site trials to test durability and transfer to spontane- ous spoken interaction, compare single-tool versus multi-tool architectures, evaluate teacher professional development needs, and assess cost-effectiveness across diverse Jordanian and regional contexts.

  Tuning in With Technology: AI-Enhanced Listening Instruction in the Jordanian EFL Classroom · 2026 · DOI

• For mathematics teachers, it is recommended to explicitly teach AI-supported learning routines by establishing classroom norms such as requiring students to attempt problems first before consulting AI, explaining solutions in their own words, and verifying AI-generated answers using notes, teacher-provided examples, or alternative solution methods. Teachers should also strengthen students’ self-directed learning skills through structured scaffolds such as goal-setting templates, weekly learning plans, reflection logs, and self-check checklists aligned with key SRSSDL dimensions like awareness, learning strategies, and evaluation. In addition, classroom assessments and learning tasks should be designed to emphasize reasoning rather than copying by incorporating open-ended problems, error analysis activities, and solution-justification tasks that discourage simply pasting AI outputs. Finally, verification and evaluation should be taught as core mathematical habits by training students to check AI solutions for accuracy in units, logical steps, computations, and final answers, and to identify common reasoning errors that may occur in AI-generated explanations. For school leaders and program coordinators, developing school-based AI literacy guidelines is essential to promote responsible AI use, protect academic integrity, and ensure compliance with the Data Privacy Act of 2012 (RA 10173). Schools should also provide teacher capacity-building opportunities through training on AI-integrated lesson design, formative assessment strategies, and practical methods for guiding learners to use AI as a scaffold rather than a shortcut. Moreover, schools are encouraged to establish monitoring and support systems that allow teachers to track AI-related learning behaviors and provide targeted interventions, especially for students with low SDL readiness who may be more vulnerable to overreliance or passive use of AI tools. For students, ChatGPT should be used as a learning tutor rather than a solver by prioritizing requests for explanations, examples, and hints instead of complete final answers. Learners are encouraged to practice a “verify and reflect” habit by checking AI responses using notes or alternate methods and writing brief reflections on what they learned and how their understanding improved. To build independence gradually, students should reduce reliance over time by practicing similar problems without AI support after receiving initial guidance, ensuring that learning becomes internalized rather than tool-dependent. For future researchers, it is recommended to measure the quality of AI use rather than focusing only on frequency by including indicators such as verification behaviors, prompting strategies, and critical evaluation of AI outputs. Researchers may also test more explanatory models by examining mediators and moderators for example, whether SDL mediates the relationship between ChatGPT use and performance, or whether self-regulation moderates the benefits and risks of AI use. Longitudinal or experimental designs are encouraged to track changes over time and strengthen causal interpretation, and further studies should compare groups and contexts across grade levels, achievement levels, and access conditions like internet or device availability to explore equity implications and identify which learners benefit most from AI-supported learning.

  AI-Facilitated Self-Directed Learning and Mathematics Performance: A Mixed-Methods Study on ChatGPT Use among Generation Z Students · 2026 · DOI

• This contribution describes a work-in-progress project that is currently situated at an early stage of implementation. Our goal is to further disseminate this current work to teachers and educational developers, while at the same time refining and expanding. Ultimately, the success of any educational tool depends on its connection to practice. By designing with teachers rather than for them, we shift from prescription to partnership. Co-creation is not just a method; it is a mindset that institutions must embrace to truly support self-regulated learning.

  A plea for the structural anchorage of co-creation: empowering educators to foster self-regulated learning. · 2026 · DOI

• Effective group work is central to Problem-Based Learning (PBL) in higher education, yet the optimal strategy for forming student groups remains unclear.

  A Comparative Study of MBTI and Learning Style- Based Grouping for Enhancing Group Effectiveness and Balance in a Pedagogical Setting · 2026 · DOI

• To enhance generalizability, future research should consider exploring diverse academic disciplines (e. Third, while our findings show that GA support differentially impacted students depend- ing on their SRL levels in terms of certain dimensions, the underlying mechanisms of this differential effect remain unclear.

  Exploring the combined effects of group awareness support and students’ self-regulated learning levels on socially shared regulation of learning and learning outcomes in CSCL · 2026 · DOI

• However, most existing studies have merely applied scripted roles in a single collaborative environment, with limited research exploring their effectiveness in promoting CKC across diverse environments.

  The impact of scripted roles on students’ viewpoint depth and interaction pattern in collaborative knowledge construction: comparing online and offline collaborative learning · 2026 · DOI

• This study provides empirical insights into how task complexity and metacognitive scaf- folding interact to influence different learning outcomes in students’ collaborative pro- gramming. The findings indicate that task complexity and metacognitive scaffolding have significant interaction effects on computational thinking tendency and learning motivation, but not on metacognitive awareness or programming learning achievement. However, the study focused solely on middle school students, which may limit the generalizability of the findings to other educational levels. Future research could explore the effects of task complexity and metacognitive scaffolding across different K–12 seg- ments to examine potential developmental differences. Additionally, this study employed a purely quantitative methodology without incorporating qualitative data. A mixed-methods approach in future research could provide a more comprehensive understanding of how task complexity and metacognitive scaffolding shape students’ collaborative programming learning, offering richer insights into their cognitive processes, problem-solving strategies, and learning experiences. Another limitation concerns the fixed sequence of programming tasks in the present study. Both groups completed tasks in the same order (low, medium, and high complexity), which may introduce potential order effects and cumulative learning advantages. Because the intervention group received metacognitive scaffolding in earlier tasks, their improved performance in later, more complex tasks may partly reflect cumulative preparation or learning advantages rather than solely the immediate effects of scaffolding under higher task complexity. Future studies may adopt counterbalanced or randomized task sequences to better disentangle immediate scaffolding effects from cumulative advantages across tasks of increasing complexity. 1 3Exploring the interactive effects of task complexity and metacognitive… 27 Page 22 of 25 Funding This study was supported in part by the National Science and Technology Council of the Republic of China [NSTC 112-2410-H-216-002, NSTC 113-2410-H-216-001-MY2] and the Ministry of Education of Humanities and Social Science Project of the People’s Republic of China [24YJA880096]. Data availability The data in this study can be accessed by sending request e-mails to the corresponding author.

  Exploring the interactive effects of task complexity and metacognitive scaffolding on students’ performance in collaborative programming · 2026 · DOI

• Future research should investigate how synthesis interventions impact stu- dents’ individual growth—such as their sense-making and writing skills—and examine how these approaches function across a wider range of contexts, including disciplinary writing, interdisciplinary collaboration, and large-scale online learning environments.

  Advancing collaborative discourse through knowledge synthesis · 2026 · DOI

• This study applied the Apriori algorithm to interaction logs from a mathematics tutoring system to examine behavioral patterns related to LH across differences in LH level, intervention condition, and problem-solving outcome. The analysis found that not using available hints was the most frequent pattern linked to unsolved problems, while persistence behaviors, such as continuing with a problem, appeared less often. Low-LH students showed stronger links between persistence and solved problems, as well as positive associations between hint use and correct solutions. High-LH students displayed more avoidance-oriented patterns, particularly skipping behaviors tied to unsolved results. Students without system-based intervention exhibited stronger persistence- success associations, while those with intervention tended to show more skipping behaviors. Across all groups, continued engagement with problems was consistently related to solved outcomes, whereas avoidance without hint use was more often connected to unsolved ones. There are several limitations in this study. The dataset did not include demographic or contextual variables that might influence engagement patterns. The LH classification was derived from a Random Forest model trained on behavioral features and grounded in Yates’ (2009) validated teacher-rated scale; nonetheless, the use of a model-derived binary label as a proxy for a psychological construct introduces some degree of construct validity limitation in how the low and high LH groupings should be interpreted. The intervention and non-intervention groups were drawn from different schools without randomization, so differences in behavioral patterns between groups cannot be attributed to the intervention and should be interpreted as associations rather than causal effects. The study focused on Grade 8 learners in the Philippines, so patterns may differ in other educational levels or contexts. The session-level unit of analysis does not account for within-student dependencies, as individual students contributed multiple sessions to the dataset. This means that behavioral tendencies of high- frequency contributors are reflected more prominently in the reported patterns, and the findings should be interpreted as characterizing session-level co-occurrences rather than consistent behavioral tendencies within individual learners. Future studies employing student-level aggregation or multilevel modeling approaches could more precisely examine whether the reported associations generalize across individual students. Because behavioral indicators were aggregated at the session level, within-session event sequences were not modeled, www.ejel.org 154 ©The Authors John Paul P. Miranda and future studies with timestamped action-level logs are encouraged to apply other pattern mining alongside association rule mining. A sensitivity check confirmed that the primary avoidance-related patterns were stable across varied threshold combinations, though future work may explore a wider range of values to further establish robustness. The analysis also relied solely on log data, which captures observable behaviors but not the underlying cognitive or emotional processes that drive them. Despite these constraints, the observed patterns suggest several opportunities for improving ITS. Adaptive features could be designed to detect early signs of avoidance, such as repeated skipping or low hint use, and to provide timely prompts that encourage persistence and improve help-seeking. For students with higher LH, strategies may involve guiding them toward productive hint use and reinforcing persistence after mistakes. For students with lower LH, approaches could aim to sustain engagement and gradually increase problem difficulty to strengthen resilience. Educators might also use these patterns to identify students who could benefit from targeted support, combining system data with classroom observations for a more complete understanding of learning behaviors. Future enhancements to the AES platform may include real-time pattern detection, dynamic adjustment of problem difficulty, and context-sensitive feedback. Additional studies that integrate behavioral logs with self-reported or interview data could help explain the motivational and emotional factors driving these behaviors. Ethical Statement: This study used student data collected by the author under ethics clearance from the UE Ethics Review Committee, with informed consent obtained from all participants and full compliance with the Data Privacy Act of 2012. AI Ethics Statement: The author confirm that they did not use AI when writing this study.

  Apriori-based Analysis of Learned Helplessness in Mathematics Tutoring: Behavioral Patterns by Level, Intervention, and Outcome · 2026 · DOI

• Based on the findings, several practical recommendations can be derived for FC implementation. First, the results demonstrate a clear pathway from SRL capacity through engagement and satisfaction to PLE. Accordingly, fostering SRL should be treated as an explicit instructional objective rather than an assumed learner characteristic. Instructors can embed structured goal-setting prompts at the beginning of each preclass module, provide guided preparation checklists, and incorporate brief metacognitive reflection tasks that require students to monitor their learning progress. Low-stakes formative quizzes administered before class can further support students in evaluating understanding and adjusting strategies accordingly. Second, engagement should be intentionally designed rather than left to spontaneous interaction. To convert SRL into observable ENG, in-class time should prioritize collaborative problem-solving, case-based discussions, peer instruction, and real-time polling activities. Clearly defined group roles and tasks requiring application and analysis of pre-class content can strengthen behavioral and cognitive engagement. Third, SAT can be enhanced through instructional coherence and feedback quality. Alignment between pre-class materials, in-class activities, and assessment criteria is essential. Timely, specific, and constructive feedback reinforces students’ sense of competence and progress, thereby strengthening the affective foundation of perceived effectiveness. Opportunities for student reflection and voice can further consolidate positive learning experiences. Moreover, FC design should follow a developmental sequence rather than isolated interventions: pre-class scaffolding that strengthens SRL, in-class structures that activate ENG, and feedback mechanisms that consolidate SAT. Instructional sophistication should prioritize developmental alignment over technological complexity. Investments in digital tools may yield limited benefits if regulatory competencies and engagement structures are insufficiently developed. Strengthening SRL strategies is therefore likely to generate cascading effects across engagement and satisfaction, ultimately amplifying PLE.

  Mechanisms linking self-regulated learning competence to perceived learning effectiveness among pre-service teachers in flipped classrooms · 2026 · DOI

• Future research should explore the underlying mechanisms that drive the differential effects observed across various group com- positions. As students self-formed their groups, it is assumed these prior experiences were generally positive or neutral; however, future research should investigate how the valence of prior collaborations might influence subsequent group dynamics. However, a generalised approach to facilitation is insufficient.

  Examining group dynamics and composition characteristics with HLM in online collaborative instructional planning among pre-service teachers · 2026 · DOI

• Future studies could consider increasing the number of participants, which could strengthen the robustness of the findings, for example, a larger number of students might provide similar responses for the indicators of interest. Future research could consider the aspect of collaboration when investigating interest in the classroom. Future research could address this by including such measures to more rigorously distinguish the effects of IDC-aligned instructional strategies from baseline motivation. Additionally, future research could examine how individual motivation and the nature of problems in project-based learning courses interact to shape learners’ interest over time. Future research could investigate how educational and cultural factors shape learners’ interest. The third limitation is that we did not account for learners' initial intrinsic motivation or the types of problems presented in the ISD course, both of which may have influenced how their interest was sustained.

  Sustaining learners’ interest: Applying IDC theory in a semester-long project-based learning course · 2026 · DOI

• challenge to reconsider what they do know and to seek additional information them SAM model and also lesson the plan. The Umeed workshop helped me to empathise issues with the facing they are rather just someone reading else’s story.” than The data was coded by two trained educational technology researchers, and inter-rater reliability was assessed using Cohen’s kappa (κ). Both coders were well-versed in IDC theory and project-based learning courses. In the first round, both researchers independently coded 11% of the total data (i.e., interviews and survey responses from two participants, and 11% of the in-class discussion responses from the TPACK activity and reflection journals). Discrepancies were then discussed and resolved through consensus in the second round. The coding of indicators of trigger, immerse, and extend phases of the interest cycle demonstrated high agreement between coders, with κ = 0.80 for in-class discussion responses (TPACK episode), κ = 0.90 for interview and survey responses, and κ = 0.77 for reflection journal responses. These results indicate reliable coding procedures, with overall kappa values reflecting substantial inter-rater reliability (Halpin, 2024).

  Sustaining learners’ interest: Applying IDC theory in a semester-long project-based learning course · 2026 · DOI

• 1. Incorporate Multiple Intelligences in Curriculum Design Individual is a social being and one of the most important features that distinguish him from other living things is his ability to learn. This situation ensures that the children and young people who are growing up 36 Vision International Scientific Journal, Volume 10, Issue 2, September 2025 Assoc. Prof. Dr. Arafat Useini, Assist. Prof. Dr. Şehida Rizvançe Matsani are in harmony with the society and age in which they live in a healthy and productive way (Rizvançe Matsani & Koca, 2023). Educators should design curricula that address all dimensions of intelligence, allowing students to engage in activities that match their cognitive strengths (Bümen, 2005; Demirel, 1999). This could include project-based learning, hands-on experiments, collaborative exercises, and arts integration.

  EXPANDING LEARNING HORIZONS: EXPLORING GARDNER’S MULTIPLE INTELLIGENCES AND THEIR CONTRIBUTION TO EDUCATION · 2026 · DOI

• Future research should consider controlling for tutors’ procedural and conceptual knowledge dur- ing experimental design to better understand the underlying mechanisms driving the relationship. This points to an additional medi- Behaviormetrika1 3 ating factor—potentially motivational or metacognitive—that warrants further investigation.

  Beyond prior knowledge: the predictive role of knowledge-building in tutor learning · 2026 · DOI

• The findings of this study should be interpreted in light of several limitations. First, the relatively small sample size (n = 46) from a single online course limits both statistical power and generalisability. Accordingly, the results should be interpreted as context-spe- cific, and replication with larger and more diverse samples is needed. Second, the study primarily focused on perception and behavioural outcomes; accordingly, broader rela- tional aspects of feedback quality, such as the extent to which feedback accurately and comprehensively addressed the targeted needs, were beyond its current scope. Third, the first author reviewed the GenAI-generated feedback prior to delivery to ensure compli- ance with ethical requirements and contextual appropriateness. Although the cognitive demands of reviewing text are not necessarily equivalent to those of composing feedback from scratch (Xavier et al., 2026), this reviewing process was not formally measured or accounted for as a workload factor in the present study. Fourth, the tutor's awareness of the GenAI feedback condition may have influenced their approach to feedback drafting in unquantifiable ways, leading them to engage in greater deliberation or inadvertently alter their usual practice.

  Supporting self-regulated learning through generative AI feedback in online higher education: the importance of student perceptions of the source of feedback · 2026 · DOI

• Given the promising patterns observed here and the relatively small, single-course cohort on which they are based, future research should investigate whether these findings per- sist across diverse educational settings and larger, more diverse student populations. Future studies should also compare GenAI and tutor feedback with a no-feedback con- dition to provide more robust insights into the longitudinal impact of different feedback Yilmaz et al. International Journal of Educational Technology in Higher Education (2026) 23:16 Page 20 of 25 regimes on learning progress. Given that the tutor in this study was aware of the GenAI feedback condition, future studies should consider withholding this information until all interventions are completed, to ensure that tutor-generated feedback remains consis- tent with usual practice and is not influenced by awareness of the experimental condi- tions. Future studies should also consider tracking the same students' SRLs performance in contexts without feedback, enabling a more robust assessment of whether observed changes persist and translate into sustainable skill development beyond the intervention period. In addition, future studies should examine both feedback perception and uptake, encompassing processes such as understanding, accepting, elaborating on, and critically engaging with feedback (Meyer et al., 2025), to develop a more nuanced understanding of how feedback interventions shape learning. This distinction matters because higher perception scores alone should not be interpreted as evidence that feedback functions effectively in all respects. To enable more rigorous comparisons of feedback quality across conditions and their association with observed outcomes, future studies should consider content-based evaluation techniques, such as precision, recall, and F1-based analyses of how accurately feedback addresses targeted learning needs, as well as den- sity-based metrics, such as the proportion of targeted needs addressed per unit of the text, and their association with learning outcomes. Finally, future studies should inves- tigate whether GenAI-generated feedback is equally effective for students from diverse backgrounds and learning profiles, attending to questions of educational equity (Stein- bach et al., 2025). How individual characteristics such as prior AI experience and cur- rent self-regulation level interact with source awareness to shape feedback engagement remains an important question.

  Supporting self-regulated learning through generative AI feedback in online higher education: the importance of student perceptions of the source of feedback · 2026 · DOI

• One limitation of the present study is that not all students from the participating school classes took part. Some did not have parental consent, so the social network data only reflect the networks among those with consent, rather than the full class networks. Additionally, in line with ethical guidelines, students were informed before the study began that participation was voluntary, that task performance would not count for school grades, and that they could withdraw at any time. As a result, some students who completed the social network measures in Session 1 chose not to participate in the concept task in Session 2. This means our data may not fully represent natural classroom populations; in particular, lower-performing or less motivated students may have been more likely to opt out. A strength of the study is that testing was teacher-led and took place in regular classrooms, and the task enabled self-regulated learning, as students could allocate their time freely and choose whether and what to restudy. At the same time, the task was researcher- ADOLESCENTS’ METACOGNITION, LEARNING, AND SOCIAL NETWORKS 34 designed and highly structured and standardized: concepts were presented one per page during learning, the order could not be changed, and restudy was allowed only once. Completing the task for the first time may have reduced strategic regulation, as students were possibly uncertain about the benefits of restudying. Repeated task experience appears to increase strategic restudy (Van Loon & Laninga-Wijnen, 2025). In typical school settings, where students return to tasks over time, they may make more effective decisions. Moreover, the task was framed as a pointsbased competition to motivate engagement and prevent dropout, since ethical constraints prevented us from counting task performance toward schoolwork. This framing may have prevented drop-out, but could also have reduced perceived personal or academic relevance. Overall, the controlled task effectively assessed monitoring, study time, persistence, restudy, decision making, and performance. However, regulation of learning may differ in naturalistic learning contexts. Future research should examine whether our findings generalize to everyday learning practice in the classroom and when working on homework. A further limitation is that we only assessed who students collaborated with on school tasks and did not obtain detailed insights into how these collaboration choices were made or how students actually worked together. As shown in Table 1, classes differed in the extent to which collaborations involved friends: in some classes, students collaborated (almost) exclusively with friends, whereas in others, they also worked with peers they did not identify as friends. We accounted for class-level effects in the multilevel analyses of social networks, but this does not explain why such differences exist.

  Linking Adolescents’ Monitoring and Regulation Processes to Learning and Social Classroom Networks · 2026 · DOI

• Based on the conclusions, future research may include additional variables not examined in this study to account for the remaining 39% of the variance in the affective outcomes of learners with disabilities. Exploratory studies may also be conducted to generate themes that can serve as potential variables, with emerging sub-themes functioning as corresponding indicators. Furthermore, educational leaders may initiate targeted training programs and allocate resources to enhance teacher implementation of differentiated instruction, as it demonstrated a stronger influence on learner affective outcomes.

  Influence of Cooperative Learning and Differentiated Instruction on the Perceived Affective Outcomes of Learners with Disabilities · 2026 · DOI

Most-cited papers in Innovative Teaching and Learning Methods

• Not Another Inventory, Rather a Catalyst for Reflection · To improve the academy · 1992 · 596 citations
• A Systematic Review of Research on Personalized Learning: Personalized by Whom, to What, How, and for What Purpose(s)? · Educational Psychology Review · 2021 · 384 citations
• Self-regulated learning training programs enhance university students’ academic performance, self-regulated learning strategies, and motivation: A meta-analysis · Contemporary Educational Psychology · 2021 · 345 citations
• The Role of Direct Strategy Instruction and Indirect Activation of Self-Regulated Learning—Evidence from Classroom Observation Studies · Educational Psychology Review · 2020 · 296 citations
• Student Interaction and Learning in Small Groups · Review of Educational Research · 1982 · 279 citations
• Supporting students’ self-regulated learning in online learning using artificial intelligence applications · International Journal of Educational Technology in Higher Education · 2023 · 237 citations
• Peer Instruction versus Class-wide Discussion in Large Classes: A comparison of two interaction methods in the wired classroom · Studies in Higher Education · 2003 · 205 citations
• Examining the key influencing factors on college students’ higher-order thinking skills in the smart classroom environment · International Journal of Educational Technology in Higher Education · 2021 · 196 citations
• Fostering Metacognition to Support Student Learning and Performance · CBE—Life Sciences Education · 2021 · 195 citations
• Empowering ChatGPT with guidance mechanism in blended learning: effect of self-regulated learning, higher-order thinking skills, and knowledge construction · International Journal of Educational Technology in Higher Education · 2024 · 193 citations

Most recent work

• The role of metacognitive awareness as mediator and moderator in the nexus between self-regulated learning and mathematics achievement · Cogent Education · 2026
• Inquiry-based learning patterns in large language model-driven learning environments: An exploratory study from Bloom’s perspective · Australasian Journal of Educational Technology · 2026
• An integrative systematic review analysis of research on technology-enabled personalized learning and self-regulated learning · Smart Learning Environments · 2026
• Transforming Self-regulated Learning – Multimodal Insights and Future Directions · Educational Psychology Review · 2026
• Advancing collaborative discourse through knowledge synthesis · International Journal of Computer-Supported Collaborative Learning · 2026
• Could Instructor Talk Drive CURE Effectiveness? A Comparative Study of Instructor Talk in Introductory Lab Courses · CBE—Life Sciences Education · 2026
• Artificial Intelligence–mediated peer collaboration in EFL argumentative writing: Exploring learners’ inferential reasoning development · Technology in Language Teaching & Learning · 2026
• Student-AI interaction patterns in collaborative problem solving using ordered network analysis · Interactive Learning Environments · 2026
• Whose visions for what learning? Perspectives, policies and practices in private supplementary tutoring · International Review of Education · 2026
• Addressing sociocultural gaps through feedback: Culturally Responsive Instructional Supervision for equitable outcomes · Theory Into Practice · 2026

Related topics in Psychology