Olga Chernikova

Collaborative learning is widely recognized in educational research as an effective pedagogical approach for facilitating the development of a broad range of skills. However, this approach also faces several challenges in ensuring optimal learning outcomes. Adaptive instructional support has emerged as a promising solution for addressing these problems by enhancing both individual- and group-level learning. This meta-analysis of 83 empirical studies builds on an integrated theoretical framework to examine the effects of adaptive instructional support on different learning dimensions as well as the moderating role of design features, learner characteristics, and contextual conditions in collaborative learning. Overall, adaptive instructional support exhibited a positive medium effect ( g = .73, 95% CI [.56, .90], p < .001) on collaborative learning with varying effects across learning dimensions. Design features proved the most influential moderator: Adaptations based on cognitive variables, implemented through cognitive models or data-science approaches, targeted at collaborative activities, and applied over short-term durations yielded the strongest effects; however, the role of teacher and learner involvement in decision-making around adaptivity requires more careful consideration. Learners’ prior knowledge, domains, and learning context also significantly moderated the effectiveness of adaptive support.

Teachers work in socially and emotionally complex environments that require rapid professional decision making under conditions of uncertainty. Classroom simulations provide opportunities for preservice and in-service teachers to practice, receive feedback, and reflect without placing students or teachers at unnecessary risk. Despite their growing use, the field lacks a contemporary and inclusive definition of classroom simulation. Existing frameworks often categorize simulations by technology or immersion level, overlooking important differences in agency, responsiveness, adaptation, and instructional design. These limitations have become more significant with the emergence of generative artificial intelligence (AI) and AI agents. This conceptual paper defines classroom simulations as deliberately designed approximations of classroom practice embedded within a structured learning process that includes preparation, enactment, and reflection. We propose a four-layer framework encompassing the professional practice represented, the simulated situation, the distribution of agency and adaptation, and the orchestration of learning. The framework is used to examine how simulations can support social-emotional and inclusive teaching competencies and how AI may influence scenario generation, student behaviour enactment, personalization, and feedback. Effective implementation requires theoretically grounded representations of learners and classroom interactions, alignment with learning goals, meaningful human oversight, and protections for agency, inclusion, privacy, and emotional safety.

In psychological assessment, gauging the impact of personality traits on academic outcomes is vital. Many studies explore the relation between academic achievement and traits like conscientiousness but prioritize description over prediction. Addressing this gap by focusing on actual prediction can refine assessment methodologies and deepen theoretical understanding. Our study focuses on predicting the influence of conscientiousness facets on standardized test scores using various machine learning strategies. Data from N = 7,949 Luxembourgish Grade 9 students showed a gradient boosting model with item-level predictors outperformed traditional linear regression ( R² = .123 vs. R² = .077). This model revealed both linear and nonlinear ties between conscientiousness facets and achievement. Our findings accentuate conscientiousness’s underestimated predictive power for academic success and advocate for machine learning as a pivotal tool in psychological testing, particularly for outcome prediction.

We investigated the effect of cognitive training of executive functions on children’s cognitive outcomes. To address this issue, a systematic meta-analysis of published research articles on cognitive training interventions was performed considering children’s age, training duration, -procedure, and -technology in moderator analyses. The results ( N = 57) of a random-effects-model showed that cognitive training was effective with a total effect size of g = 23 The training was more effective for younger compared to older children. Training benefits were found for near- and far-transfer tasks. The largest gains for a near-transfer skill were found for working memory. Both numeracy and literacy skills profited from training. Computer training was very effective, however, only a few studies used mobile technology. Non-adaptive training was associated with greater effect sizes and both group and individual training were similarly effective. Verbal feedback was important for younger children. School was an effective context for training, however, only a few studies were conducted at home or at the lab. The findings are discussed and advocate an early start of cognitive training interventions.

This meta-analysis builds on and extends Chernikova, Heitzmann, Stadler, et al.’s (2020) meta-analysis on simulation-based learning in higher education by summarizing the findings of 214 empirical studies. The main focus of the current meta-analysis was to investigate the role of prior knowledge and design features, namely (1) the authenticity of simulated scenarios (as both functional correspondence with a real task and a physical resemblance of the environment) and (2) the salience of relevant information in moderating the effects of simulation-based learning on facilitating complex skills in higher education. The analysis identified that the salience of relevant information contributes significantly to explaining the variation in the effectiveness of simulation-based learning environments. We conclude that authenticity and salience can be assessed independently of each other, and both can be considered important design features in creating effective simulation-based learning environments to meet the needs of learners with different levels of prior knowledge.

This meta-analysis includes 217 empirical studies and investigates the effectiveness facilitating learners' agency as a strategy to adapt and individualize instructional support to learners' prior knowledge. We conclude that (1) facilitating learners' agency might have different effects on learners with different levels of prior knowledge; (2) allowing learners to control simulation flow enhances effects of simulations.