Methodology gaps in Education

Representative open questions

Showing 30 of 796 — one per source paper, highest-quality first.

• Power-Optimized AI-Enhanced Telepresence Robots: A Validated Multi-Modal Framework for Sustainable Remote Learning in Higher Education (2026) · doi

  Battery endurance is currently unspecified in tested configurations; expansion to 8-10 hour operational periods using 6000 mAh high-capacity cells requires hardware redesign and power-consumption validation to confirm the 35% energy reduction advantage is maintained.

• El desarrollo de la comprensión del aspecto matemático del logaritmo mediante la intuición y la axiomatización en los estudiantes de Licenciatura en Educación Secundaria con especialidad en Matemáticas al término de la aplicación de su programa educativo. (2026) · doi

  Students demonstrate persistent difficulties when performing logarithmic subtraction operations involving negative characteristics and mantissas, particularly when the subtrahend exceeds the minuend; the paper identifies confusion about characteristic value changes and position shifts in decimal notation but does not propose or test specific didactic sequences to remediate these conceptual errors in logarithmic arithmetic.

• e-DigCompEdu: Digital Competences for Online Higher Education Framework (2026) · doi

  The Scientific Digital Literacy section (7) identifies the need to integrate advanced digital tools and AI techniques for data-driven research and robust data management, but provides no concrete specifications for which AI tools, data management frameworks, or integration protocols should be implemented across different research methodologies (quantitative vs. qualitative).

• A systematic review of generative artificial intelligence in education: Pedagogical impacts, ethical risks, and future directions (2026) · doi

  While the paper discusses integration of AI-generated content with human oversight in curriculum development, it does not address specific quality assurance protocols, validation criteria, or measurable benchmarks for determining when AI-generated lesson plans, quizzes, and case studies meet pedagogical correctness standards for multilingual and competency-based learning contexts.

• Ethical challenges of artificial intelligence in education: A systematic literature review on bias, privacy, and academic integrity (2026) · doi

  Current plagiarism detection tools have become obsolete against generative AI outputs that appear original despite lacking authentic authorship. The paper identifies an active arms race between generative AI tools and AI detection systems with mixed success, but does not specify what multimodal detection techniques (combining text analysis, metadata analysis, and behavioral patterns) should be developed or tested to reliably distinguish AI-generated academic work from human-authored content.

• Error Patterns and Predictors in Solving Algebraic Word Problems: Evidence from Secondary School Students (2026) · doi

  The study identifies that comprehension errors occur in 40-49.5% of students solving linear equation word problems, yet does not empirically test whether explicit instruction in algebraic term interpretation (e.g., distinguishing variables as placeholders vs. fixed numbers) reduces comprehension error rates compared to standard procedural instruction.

• ChatGPT in computer programming education: A review of current literature and applications (2026) · doi

  Standardised research designs and evaluation frameworks for assessing ChatGPT's role in programming education are lacking, reducing consistency and comparability across published studies. Development of validated assessment instruments and methodological protocols would enable meta-analysis and evidence synthesis on AI integration effectiveness.

• Coding Science: The Role of Block-Based Programming Activities in Enhancing Computational Thinking and Science Achievement (2026) · doi

  Longitudinal studies must investigate sleeper effects in block-based programming integration with science instruction by tracking whether science achievement gains in the experimental group emerge after the initial cognitive load of coding diminishes over extended time periods.

• When teachers learn by observation: Vicarious reinforcement and student response systems in teacher education (2026) · doi

  The study documents that student response systems (AhaSlides/Mentimeter) enhanced participation and content understanding in teacher education, but does not systematically compare vicarious reinforcement effects across different question formats (multiple-choice vs. short-answer) or investigate which specific types of peer answer visibility most effectively motivate shy or introverted student teachers to participate.

• A(n) (re)awakening: preservice teachers connecting their culture to science teaching along the US–México border (2026) · doi

  Translation methodology in qualitative research integrating Spanish-English bilingual data requires refinement to capture full nuance of colloquialisms and dialects across different border regions, as current multi-reviewer approaches may still miss contextual meaning in translated testimonios and family narratives.

• Digital Platforms and Relational Practice: A Netnographic Study of Educator-Family Communication (2026) · doi

  The paper identifies that explicit educator prompts and invitation questions significantly increase family engagement and willingness to contribute to digital platform discourse, but does not systematically test different prompt typologies. Future work should experimentally manipulate the linguistic framing, question type, and invitation specificity of educator posts to quantify which exact communicative cues most effectively shift families from passive recipients to active collaborators in educator-family digital communication.

• When faces trigger feelings: multi-stakeholder experiences, use, and expectations of facial-emotion learning technologies and materials for autistic people (2026) · doi

  The paper reveals that current facial-emotion apps present static images or rapid video transitions that do not match how autistic users experience emotional state changes individually. There is a need to develop facial-emotion learning tools that visualize gradual emotional transitions (e.g., neutral to happy) rather than discrete emotional states to align with how some autistic individuals process emotional progression.

• Segmental and suprasegmental pronunciation training via AI: An exploration of university students’ perceptions and attitudes (2026) · doi

  The paper does not systematically compare AI speech recognition accuracy rates across different accent varieties or non-native speaker input types when evaluating segmental pronunciation features, limiting generalizability of findings to diverse learner populations.

• The current landscape of teachers artificial intelligence acceptance: Relationships with TPACK and technostress (2026) · doi

  Professional development initiatives focusing on pedagogical integration of AI into subject-specific teaching practices require empirical evaluation; no studies have measured the effectiveness of collaborative teacher communities and peer/reverse mentoring models in reducing technostress while enhancing AI-supported instructional strategies adoption.

• Effect of Van Hiele group guided-discovery instructional approach on student engagement in learning plane geometry (2026) · doi

  The paper notes that assessment methods traditionally focused on testing rather than understanding exacerbate low engagement, but the intervention's specific formative assessment strategies within VHGGDIA are not detailed. Research should compare how different assessment approaches (diagnostic, formative, performance-based) embedded in Van Hiele guided-discovery instruction differentially impact plane geometry engagement.

• On the development and potential influential aspects of prospective science teachers’ competencies for the use of explainer videos for learning in physics education (2026) · doi

  Open-ended response formats from a sub-sample will be analyzed in subsequent work to examine differences in how prospective teachers' professional vision can be promoted during subject-specific teacher training in physics, but this qualitative analysis has not yet been completed or reported.

• Producing concept-motivated signs supports learning of STEM in American Sign Language (2026) · doi

  Study 2 employed pre- and post-tests assessing only the first three levels of Bloom's taxonomy (remember, understand, apply) for STEM learning outcomes in ASL. Higher-order cognitive assessment levels (analyze, evaluate, create) were not measured, limiting the ability to determine whether concept-motivated versus engagement-motivated ASL signing differentially impacts deeper learning and transfer of STEM concepts.

• Pluriliteracies for global citizenship: the 4Rs framework for deeper learning in the modern language(s) classroom (2026) · doi

  The Deeper Learning Episode (DLE) framework presented in Table 4 provides guiding questions for four phases (Activation, Surface, Consolidation, Transfer) with success indicators (Thinking, Feeling, Saying, Doing, Having, Being), but the paper lacks specification of how teachers assess or operationalize these shifts, particularly the distinction between lateral reading and critical ignoring as threshold practices between Surface and Consolidation phases.

• Adequacy and Utilization of Physical Resources: Implications for Effective Teaching and Learning (2026) · doi

  The study found significant disparities in physical resource adequacy between urban and rural schools in Delta State, but did not investigate the specific mechanisms driving these disparities across different school types (secondary vs. primary) or administrative structures. Future research should systematically compare resource allocation patterns, infrastructure investment cycles, and maintenance protocols between urban and rural schools to identify actionable policy interventions.

• Digital Disruption of Academic Integrity (2026) · doi

  The partial mediation model demonstrates that IT facilitates plagiarism through AI paraphrasing tools and essay mills while simultaneously enabling detection through software monitoring, but the competing mechanisms and their relative effectiveness across different AI tool types (generative AI, paraphrasing engines, essay mills) lack empirical comparison within postgraduate populations.

• The Role of Artificial Intelligence in the Career Expectations of Ukrainian Students: Implications for Higher Education (2026) · doi

  Complete-case analysis was implemented with excluded records not preserved separately, preventing systematic evaluation of missingness patterns and creating potential selection bias. Future studies must implement reproducible preprocessing with retained excluded cases to facilitate missingness diagnostics and sensitivity analyses for the composite indices measuring AI perceptions.

• The Effects of Handwriting and Typing on Chinese Character Learning in the Digital Age: Evidence from Arab Adolescent Learners (2026) · doi

  The study compared handwriting versus typing effects on Chinese character learning in Arab adolescents but did not investigate whether these effects vary across different proficiency levels (beginner, intermediate, advanced). The orthographic-semantic mapping findings may differ significantly depending on learners' existing character knowledge and stage of language development.

• No one-size-fits-all: a study of prompt techniques and large language models to enhance AI’s mathematics educational quality (2026) · doi

  The operationalization of mathematics educational quality through binary coding obscures nuances of pedagogical effectiveness, and binary logistic regression showed considerable model misfit. Future research should develop multi-dimensional quality metrics and explore non-binary coding schemes that better capture the complexity of pedagogical effectiveness in LLM-generated mathematical explanations.

• Cluster Pattern Analysis of Students Stress using Machine Learning Algorithms with Feature Engineering (2026) · doi

  Only four supervised learning classifiers (Decision Tree, Random Forest, SVM, and KNN) were evaluated for stress cluster prediction. Alternative supervised learning techniques such as gradient boosting, neural networks, or ensemble methods combining different feature engineering approaches beyond Chi-square selection should be tested to determine if superior classification accuracy can be achieved for student stress prediction.

• Multimodal AI in education: an avatar-based intelligent learning system for the Kazakh language (2026) · doi

  The sentiment processing component uses lexicon-based fallback mechanisms for specialized terminology, but the paper does not specify which Kazakh educational domains (e.g., STEM, humanities, technical vocabulary) have limited lexical coverage or how sentiment quantization thresholds should be calibrated for different subject matter in language learning contexts.

• The use of large language models to solve mathematical modeling problems: preservice mathematics teachers’ use practices, perceived affordances and challenges, and trustworthiness judgments of AI-generated outputs (2026) · doi

  The study identifies that validation and interpretation phases of the modeling cycle were the least supported by LLMs and were often omitted in submitted group work, but does not investigate what specific instructional interventions or prompt design strategies could redirect PSTs' LLM use toward these critical evaluative components during mathematical modeling.

• Pre-service Teachers' Self-Directed Learning in a Blended Learning Environment: A Study on Scale Development and Affecting Factors (2026) · doi

  The study employed a cross-sectional correlational design that cannot establish causal relationships between identified SDL predictors (student awareness, technological ability, self-monitoring, autonomy) and SDL outcomes; longitudinal studies tracking pre-service teachers over multiple semesters are needed to confirm whether interventions like Project-Based Learning and Problem-Based Learning actually increase SDL.

• Development of User Story and Design Thinking Integration Teaching Model for Software Engineering Education (2026) · doi

  The US-DT teaching model's Perceived Ease of Use (PEOU) factor showed no significant improvement, with students struggling specifically with new tools and terminology during Design Thinking phases and constrained by strict session time limits. Future research must systematically investigate which specific DT tools and terminology create the highest cognitive load and test targeted interventions like preparatory workshops to address this gap.

• Does “Maniplo-Spatial Arty” Really Work? A Study of the Effectiveness of “Maniplo-Spatial Arty” as a Teaching and Learning in Science Pedagogy (2026) · doi

  The study design permitted observation of only single lessons containing maniplo-spatial tasks, preventing assessment of learners' understanding at Level 2 (medium to long-term retention) through follow-up diagnostic assessments either shortly after or weeks following the observed lesson. Creating task-specific diagnostic apparatuses for each maniplo-spatial task observed would be needed to measure delayed retention of ideas and observables.

• Quality Evaluation of Indonesian Student-generated User Stories: Insight from Human and ChatGPT Evaluation (2026) · doi

  Semantically and context-dependent QUS criteria (Conceptually Sound, Estimable, Independent, Complete) demonstrated higher disagreement and larger effect sizes in binary scoring; the paper suggests but does not implement or validate whether rubric-based evaluation guidance with partial fulfillment scoring would reduce disagreement and improve criterion reliability for these semantic criteria.