Archive

Published JEES: Journal of Education and Educational Sciences • 2026

The Effect of the Snowball Drilling-Based Jigsaw Model on Learning Independence and Mathematical Spatial Ability of Elementary School Students

Ike Kurniawati, Herlin Kurniasari

This study aims to determine the influence of variables in the snowball drilling-based Jigsaw learning model. The research method employed a pre-experimental design using a one-group pretest–posttest design. The research subjects were school students with a saturated sampling technique and a sample size of 60 respondents. The research instruments consisted of a cognitive test of mathematical spatial ability and a questionnaire on students' mathematical learning independence as well as a questionnaire on responses to the Jigsaw learning model. Data analysis techniques used multiple linear regression tests to determine the effect of the snowball drilling-based Jigsaw model on students' learning independence and mathematical spatial ability. Path analysis was used to test the relationship between aspects of independence and mathematical spatial ability. The results showed that the application of the snowball drilling-based Jigsaw model had a significant effect (p < 0.05) on improving elementary school students' mathematical learning independence and mathematical spatial ability. This study contributes novelty by integrating the Jigsaw cooperative model with Snowball Drilling to simultaneously strengthen students’ learning independence and mathematical spatial ability at the elementary level, an area that has been rarely examined in previous studies. In addition, there is a positive relationship between aspects of the variables of learning independence and mathematical spatial ability, as can be seen from the p-value < 0.05. The Jigsaw model can encourage students to increase their learning independence through responsibility and develop mathematical spatial abilities through understanding relationships, perception, visualization, rotation, and spatial orientation. Thus, this study can be used as a reference in similar studies.

Published: 01 May 2026 View Details

Published JEES: Journal of Education and Educational Sciences • 2026

The Effect of the Project-Based Learning Model in Terms of Learning Styles on the Literacy and Numeracy Skills of Elementary School Students

Irma Sofiasyari, Rizky Arianty

This study aims to determine the effect of the project-based learning model in terms of learning styles on the literacy and numeracy skills of fourth-grade elementary school students in the Jamblang sub-district in 1 cluster, namely SDN 1 Wangunharja, SDN 2 Wangunharja, SDN 1 Sitiwinangun, SDN 2 Sitiwinangun, SDN 3 Sitiwinangun, SDN 1 Bojong Wetan, SDN 2 Bojong Wetan, and SDN 3 Bojong Wetan. This study used a quantitative method with a one-group pretest-posttest design. The sample in this study consisted of all 240 fourth-grade elementary school students. The research instruments consisted of a learning style questionnaire, literacy and numeracy tests with multiple choice questions for the pretest and posttest. Data collection was conducted before and after the implementation of the PjBL learning model. Data analysis included validity, reliability, normality, homogeneity, simple linear regression analysis, and paired sample t-test. The results show that each data point in the variables also meets the assumptions of normality and homogeneity. From the descriptive statistics calculation results, the hypothesis test of simple linear regression analysis and the paired sample t-test show a significant difference between the pretest and posttest scores for students' literacy and numeracy abilities. with the students' literacy ability level at the time of the pretest averaging 42.88 and the posttest score averaging 76.33, and the students' numeracy ability level at the time of the pretest averaging 43.46 and the posttest score averaging 71.21. Based on the hypothesis test, the significance value of each variable was 0.000 (

Published: 01 May 2026 View Details

Under Review MIREJ: Multidisciplinary Innovation Research Journal

A Comparative Evaluation of Large Language Models for Enterprise Deployment: Performance, Safety, Cost, and Scalability Using a Multi-Criteria Decision Framework

Jurgen Mecaj, Erarda Vuka

The proliferation of large language models (LLMs) across enterprise, research, and public-sector applications has created an urgent need for rigorous, multi-dimensional evaluation frameworks. This paper presents a comprehensive comparative analysis of seven state-of-the-art LLMs — GPT-4o, Claude 3.5 Sonnet, Gemini 1.5 Pro and Flash, LLaMA 3 70B, Mistral Large, and Claude 3 Haiku — across eight evaluation dimensions: benchmark accuracy, safety alignment, cost efficiency, inference latency, context handling, deployment flexibility, multilingual capability, and scalability. A weighted Multi-Criteria Decision Analysis (MCDA) framework is applied to produce transparent composite rankings from empirical benchmark data using five standardized benchmarks (MMLU, HumanEval, HellaSwag, GSM8K, MATH). Results indicate that Claude 3.5 Sonnet achieves the highest MCDA composite score (0.801), driven by accuracy (90.4% MMLU, 92.0% HumanEval) and safety alignment (4.9/5). Gemini 1.5 Flash emerges as optimal for cost-sensitive deployments ($0.075/1M tokens; 210 tok/s). The paper analyzes architectural trade-offs between dense transformers and Mixture-of-Experts designs, provides a deployment recommendation matrix, and contributes an extensible, evidence-based decision framework for enterprise AI practitioners.

View Details

Declined MIREJ: Multidisciplinary Innovation Research Journal

A Comparative Evaluation of Large Language Models for Enterprise Deployment: Performance, Safety, Cost, and Scalability Using a Multi-Criteria Decision Framework

Jurgen Mecaj

The proliferation of large language models (LLMs) across enterprise, research, and public-sector applications has created an urgent need for rigorous, multi-dimensional evaluation frameworks capable of guiding model selection beyond single-metric leaderboards. This paper presents a comprehensive comparative analysis of seven state-of-the-art LLMs — GPT-4o (OpenAI), Claude 3.5 Sonnet (Anthropic), Gemini 1.5 Pro and Flash (Google DeepMind), LLaMA 3 70B (Meta AI), Mistral Large (Mistral AI), and Claude 3 Haiku (Anthropic) — across eight evaluation dimensions: benchmark accuracy, safety alignment, cost efficiency, inference latency, context handling, deployment flexibility, multilingual capability, and scalability. The study applies a weighted Multi-Criteria Decision Analysis (MCDA) framework to produce transparent composite rankings from empirical benchmark data. Evaluation leverages five standardized benchmarks (MMLU, HumanEval, HellaSwag, GSM8K, MATH), official API pricing data, throughput metrics, and curated safety evaluation datasets. Results indicate that Claude 3.5 Sonnet achieves the highest MCDA composite score (0.801), driven by combined strengths in accuracy (90.4% MMLU, 92.0% HumanEval) and safety alignment (4.9/5). Gemini 1.5 Flash emerges as the optimal choice for cost-sensitive, high-throughput deployments ($0.075/1M tokens; 210 tok/s). The paper additionally analyzes architectural trade-offs between dense transformers and Mixture-of-Experts (MoE) designs, scaling law evidence, safety evaluation profiles, and provides a nine-row deployment recommendation matrix. This work contributes an extensible, evidence-based decision framework with practical guidance for practitioners, researchers, and enterprise decision-makers navigating the rapidly evolving LLM ecosystem.

View Details

Published DUTIES: Education and Humanities International Journal • 2026

Deciding When and How to Use AI in EFL Speaking Instruction: Evidence from Surveys and Teacher Interviews

Samikshya Bidari, Muhammad Aulia Taufiqi

The integration of Artificial Intelligence (AI) in English as a Foreign Language (EFL) instruction has expanded rapidly, yet little is known about how teachers make pedagogical decisions regarding AI use in speaking classrooms. This study investigates how EFL teachers explain and justify their decisions about when and how to use AI tools in speaking instruction, employing a convergent mixed-methods design. Quantitative data were collected through a survey of 24 teachers, analyzed using descriptive statistics, while qualitative insights were obtained from semi-structured interviews with 12 teachers, analyzed thematically. Findings reveal that teachers adopt a selective, context-sensitive approach, prioritizing AI for pronunciation practice and fluency exercises, where immediate feedback and structured practice are most effective. Teachers exercise professional judgment to balance AI affordances with pedagogical objectives, contextual constraints, and ethical considerations, ensuring that AI supplements rather than replaces human interaction. The study highlights the multidimensional nature of teacher agency in AI-supported speaking instruction and provides practical implications for professional development and curriculum design. Future research could examine AI integration across other language skills, diverse educational contexts, and longitudinal impacts on learners’ speaking proficiency and autonomy.

Published: 01 Mar 2026 View Details

Published DUTIES: Education and Humanities International Journal • 2026

When Feedback Must Be Human: Pedagogical Resistance to AI in EFL Speaking Classrooms

Ega Nur Fadillah, Uwaimir Ahad

The rapid advancement of artificial intelligence (AI) has intensified debates about its role in language education, particularly in providing automated feedback. While existing research has largely focused on teachers’ acceptance and use of AI tools, limited attention has been given to teachers’ deliberate decisions not to use AI in specific pedagogical contexts. This qualitative study investigates EFL teachers’ pedagogical resistance to AI-mediated oral feedback in speaking classrooms. Drawing on in-depth semi-structured interviews and reflective accounts from EFL teachers, the study employs thematic analysis to explore how teachers explain their resistance and the pedagogical values underlying their decisions. The findings reveal that resistance is grounded in teachers’ concerns about interactional immediacy, learner affect, dialogic engagement, and ethical responsibility. Oral feedback is viewed as a relational practice that requires human sensitivity to timing, tone, and emotional cues, which teachers perceive as inadequately addressed by current AI technologies. Rather than signaling technological reluctance, pedagogical resistance emerges as an enactment of teacher agency and professional judgment. The study contributes to critical discussions on AI integration in education by reframing non-use as a principled pedagogical choice and highlighting the need for context-sensitive, human-centered approaches to AI use in EFL speaking instruction.

Published: 01 Mar 2026 View Details

Published DUTIES: Education and Humanities International Journal • 2026

AI Game-Based Learning in Low-Resource Classrooms: Teachers’ Innovation Under Constraint

Mochamad Rizqi Adhi Pratama, Nur Karmila Maisara

Abstract: The integration of artificial intelligence into game-based learning has been widely promoted as a means of enhancing engagement and personalization in education. However, existing research largely assumes well-resourced classroom conditions, offering limited insight into how such approaches are enacted in constrained contexts. This qualitative multiple case study explores how teachers design and implement AI-supported game-based learning in low-resource classrooms and how contextual constraints shape their pedagogical reasoning and innovative practices. Drawing on interviews, classroom observations, teaching artifacts, and stimulated recall, the study foregrounds teachers’ agency in mediating AI use under conditions of limited infrastructure, device access, and institutional support. The findings reveal that teachers adopt selective and hybrid approaches to AI game-based learning, combining AI-generated content with offline activities and teacher-led scaffolding. Contextual constraints function not merely as barriers but as catalysts for reflection, improvisation, and ethical decision-making related to fairness and inclusivity. This study contributes to AI in education research by reframing innovation as a situated, teacher-driven process and by highlighting the importance of context-sensitive approaches to AI-supported pedagogies, particularly in underrepresented low-resource educational settings.

Published: 01 Mar 2026 View Details

Published DUTIES: Education and Humanities International Journal • 2026

AI as a Metacognitive Mirror: How Students Use AI to Monitor and Repair Reading Comprehension Breakdowns

Admiral Indra Supardan, Ma. Wilma Capati

The increasing availability of artificial intelligence (AI) tools has transformed how EFL students engage with academic reading, yet little is known about how AI shapes learners’ metacognitive processes during reading. This qualitative study conceptualizes AI as a metacognitive mirror and investigates how EFL students use AI to monitor and repair reading comprehension breakdowns. Data were collected from undergraduate EFL students at a public university through academic reading tasks, screen recordings, think-aloud protocols, AI interaction logs, and stimulated recall interviews. Thematic analysis revealed that students used AI to externalize comprehension monitoring by confirming interpretations and articulating sources of confusion. AI also supported comprehension repair through strategy-specific and iterative regulation, enabling learners to request paraphrases, examples, and simplified explanations in response to perceived difficulties. However, the findings also indicate tensions between productive metacognitive support and uncritical reliance on AI, particularly when learners accepted AI-generated explanations without verification. The study contributes to AI-assisted reading research by shifting attention from learning outcomes to metacognitive processes and learner agency. Pedagogical implications highlight the importance of guiding students toward reflective and responsible AI use to support academic reading comprehension.

Published: 01 Mar 2026 View Details

Published DUTIES: Education and Humanities International Journal • 2026

Teachers’ Beliefs and Decisions Regarding Artificial Intelligence Use in Education

Muhammad Abdul Azis, Muhammad Numan

The increasing integration of artificial intelligence (AI) into educational settings has raised important pedagogical and ethical questions, particularly regarding how teachers understand and decide to use AI in their instructional practices. This qualitative study explores teachers’ beliefs about AI use in education and examines how these beliefs shape their instructional decisions. Drawing on semi-structured interviews and thematic analysis, the study reveals that teachers hold nuanced and evaluative beliefs about AI, viewing it as a supportive pedagogical tool while expressing concerns about overreliance, learning quality, and professional responsibility. Rather than adopting AI uncritically, teachers exercise agency through selective integration and pedagogical regulation of AI use, particularly in relation to assessment and student accountability. Teachers’ decisions are shown to be context-sensitive and grounded in humanistic values that emphasize ethical judgment and meaningful teacher–student interaction. The findings suggest that AI use in education is best understood as a belief-driven and value-laden practice rather than a purely technical innovation. This study contributes to educational and humanities-oriented discussions by foregrounding teachers’ professional judgment in shaping responsible and pedagogically sound AI integration.

Published: 01 Mar 2026 View Details

Published GENFABET: Generasi Pendidikan Dasar • 2026

Peningkatan Pembelajaran Aktif melalui Pengembangan Paket Media di Kelas V SD Swasta Pardamean Medan

Kania Tiur Adelia Siringoringo, Romaito Pardosi, Christi Pelita Hutagaol, Sryance Romaster Simarmata, Lidwina Dedea Ginting

This study was conducted at Pardamean Private Elementary School in Medan to analyze the learning process in fifth grade, particularly focusing on the level of student participation during classroom activities. This research employed a descriptive qualitative approach, with data collected through classroom observations and interviews with the fifth-grade teacher, Mrs. Situmorang. The observation results indicated that most students tended to be passive, relying heavily on teacher instructions and showing limited initiative, which led to a predominantly one-way learning process. Although the teacher had made efforts to deliver explanations optimally, student-centered activities were not fully implemented due to limited learning media, time constraints, and variations in student abilities. The learning media used were primarily textbooks and the blackboard. Based on these findings, the study formulated an initial concept of simple learning media, including flash cards, question cards, student discussion sheets, and learning modules, designed to support more interactive classroom learning. In conclusion, the study highlights the need for practical and accessible learning media to encourage active student participation and reduce teacher-centered learning in elementary school classrooms.

Published: 27 Feb 2026 View Details

Published GENFABET: Generasi Pendidikan Dasar • 2026

The Effect of Metacognitive-Based Deep Learning Models in Mathematics Learning on the Academic Resilience of Elementary School Students

Muhammad Bayu, Shinta Mutiara Dewi

Mathematics learning at the elementary level is often associated with anxiety and low academic resilience, which may hinder students’ persistence and performance. Strengthening resilience through innovative instructional models is therefore essential. This study aims to examine the effect of a Metacognitive-Based Deep Learning model on the academic resilience of elementary school students in mathematics. A quantitative true experimental design with a posttest-only control group was employed. The sample consisted of 120 fourth- and fifth-grade students divided equally into experimental and control groups. Data were collected using a 20-item mathematical resilience questionnaire measuring value, struggle, growth, and perseverance. Confirmatory Factor Analysis using PLS-SEM confirmed the validity and reliability of the instrument. Data were analyzed through an independent samples t-test, Cohen’s d effect size, and two-way ANOVA. The results revealed a highly significant difference between groups (p < .001), with a mean difference of 24.233 points and a very large effect size (d = 3.46). The findings indicate that the intervention consistently improved students’ academic resilience across grade levels without significant interaction effects. This study contributes theoretically by reinforcing the role of deep learning integrated with metacognitive strategies in developing non-cognitive competencies in elementary education. Practically, it offers an evidence-based instructional alternative to enhance students’ resilience and reduce mathematics anxiety.

Published: 27 Feb 2026 View Details

Published GENFABET: Generasi Pendidikan Dasar • 2026

Effect of the Realistic Mathematics Education (RME) Learning Model Assisted by Stick Media on Understanding the Concept of Multiplication

Azzahra Septia Hanami, Neha Nurabkafiya

This study is grounded in the need to enhance students’ conceptual understanding of multiplication, which often remains procedural and lacks meaningful comprehension at the elementary level. The research aimed to examine the effect of the Realistic Mathematics Education (RME) model assisted by stick media on students’ understanding of multiplication concepts. A quantitative approach with a quasi-experimental design was employed, involving 64 elementary school students divided into an experimental group and a control group. Data were collected using a validated post-test instrument measuring conceptual understanding. Data analysis techniques included descriptive statistics, Pearson correlation, and simple linear regression to determine the strength and significance of the relationship and predictive contribution of the independent variable. The findings indicate that students taught using the RME model assisted by stick media achieved higher post-test scores compared to those receiving conventional instruction. The regression analysis demonstrates a strong and significant positive effect of the RME model on conceptual understanding, with a substantial proportion of variance explained by the model. These results confirm that contextual learning integrated with concrete manipulatives enhances students’ ability to construct mathematical meaning. The study contributes empirically to the development of mathematics learning strategies by demonstrating the effectiveness of combining RME principles with simple, low-cost media. Its added value lies in providing a practical and theoretically grounded instructional alternative that can be readily implemented to strengthen conceptual learning in elementary mathematics.

Published: 27 Feb 2026 View Details