AL IBTIDA: JURNAL PENDIDIKAN GURU MI .
VOL 12 .
: 387- 411
DOI: http://dx.
org/10.
24235/al.
Al Ibtida: Jurnal Pendidikan Guru MI ISSN: 2442-5133, e-ISSN: 2527-7227 Journal homepage: http://syekhnurjati.
id/jurnal/index.
php/ibtida Journal email: alibtida@syekhnurjati.
The Effect of Cooperative Learning Model on Elementary School StudentsAo Affective Learning Outcomes: A Meta-Analysis MaAoas Shobirin* *Department of Madrasah Ibtidaiyah Teacher Education.
Faculty of Islamic Religion.
Universitas Wahid Hasyim Semarang.
Indonesia E-mail: maas.
shobirin@unwahas.
Fitria Martanti** **Department of Madrasah Ibtidaiyah Teacher Education.
Faculty of Islamic Religion.
Universitas Wahid Hasyim Semarang.
Indonesia E-mail: fitriamartanti@unwahas.
Ummu Jauharin Farda*** ***Department of Madrasah Ibtidaiyah Teacher Education.
Faculty of Islamic Religion.
Universitas Wahid Hasyim Semarang.
Indonesia E-mail: ummujauharinfarda@unwahas.
Ersila Devy Rinjany**** ****Department of Madrasah Ibtidaiyah Teacher Education.
Faculty of Islamic Religion.
Universitas Wahid Hasyim Semarang.
Indonesia E-mail: ersiladevyrinjani@unwahas.
Received: August 12th, 2025.
Accepted: September 25th, 2025.
Published: October 31st, 2025.
Abstract This meta-analysis aims to examine the effect of cooperative learning on elementary school studentsAo affective learning outcomes after the implementation of the cooperative learning model.
40 relevant studies, through online databases, including Google Scholar.
DOAJ.
ERIC.
Springer.
Elsevier, and ResearchGate, were Of these, 12 studies met the inclusion criteria and provided usable data, yielding 16 artifacts.
Data from 16 artifacts, including sample sizes, means, and standard deviations, were collected using descriptive analysis.
The data analysis utilized meta-analysis techniques, including effect size calculation, heterogeneity testing, summary effect estimation using a random-effects model, and the assessment of potential publication bias through forest plots.
Those analyses utilized JASP The results showed that: .
The effect size (ES) for affective learning outcomes of elementary school students was 77.
06, with a 95% confidence interval (CI) ranging from 73.
34 to 81.
The hypothesis test produced a z value of 0.
which is smaller than the critical value .
= 1.
, leading to the acceptance of the null hypothesis (HCA) and the rejection of the alternative hypothesis (HC.
Although there is a slight asymmetry in the funnel diagram indicating potential publication bias, the statistically insignificant Egger regression test .
= -0.
1078, p = 0.
supports the robustness of the meta-analysis, which confirms that the cooperative learning model produces moderate to high affective learning outcomes in elementary school students.
This finding indicates that although cooperative learning can positively influence students' affective learning outcomes, the observed effect size is not statistically significant at = 0.
Thus, it is recommended that future studies Copyright A 2025 Al Ibtida: Jurnal Pendidikan Guru MI.
Department of Madrasah Ibtidaiyah Teacher Education.
UIN Siber Syekh Nurjati Cirebon.
Indonesia.
MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA refine the design and implementation of cooperative learning strategies to increase their effectiveness and achieve statistically significant improvements in elementary school students' affective learning outcomes.
Keywords: affective learning outcomes, cooperative learning, meta-analysis.
Abstrak Meta-analisis ini bertujuan untuk menguji pengaruh pembelajaran kooperatif terhadap hasil belajar afektif siswa sekolah dasar setelah penerapan model pembelajaran kooperatif.
Sebanyak 40 studi relevan, melalui basis data daring, termasuk Google Scholar.
DOAJ.
ERIC.
Springer.
Elsevier, dan ResearchGate.
Dari jumlah tersebut, 12 studi memenuhi kriteria inklusi dan menyediakan data yang dapat digunakan, menghasilkan 16 artefak.
Data dari 16 artefak, termasuk ukuran sampel, rata-rata, dan deviasi standar, dikumpulkan menggunakan analisis deskriptif.
Analisis data menggunakan teknik meta-analisis, termasuk perhitungan ukuran efek, pengujian heterogenitas, estimasi efek ringkasan menggunakan model efek acak, dan penilaian potensi bias publikasi melalui plot Analisis tersebut menggunakan perangkat lunak JASP.
Hasil penelitian menunjukkan bahwa: .
Besarnya efek (ES) terhadap hasil belajar afektif siswa sekolah dasar adalah 77,06, dengan interval kepercayaan (CI) 95% berkisar antara 73,34 hingga 81,77.
Uji hipotesis menghasilkan nilai z sebesar 0,4035, yang lebih kecil daripada nilai kritis .
= 1,.
, sehingga menyebabkan diterimanya hipotesis nol (HCA) dan ditolaknya hipotesis alternatif (HC.
Meskipun terdapat sedikit asimetri pada diagram corong yang mengindikasikan potensi bias publikasi, uji regresi Egger yang secara statistik tidak signifikan .
= -0,1078, p = 0,.
mendukung kekokohan meta-analisis, yang menegaskan bahwa model pembelajaran kooperatif menghasilkan hasil belajar afektif sedang hingga tinggi pada siswa sekolah dasar.
Temuan ini menunjukkan bahwa meskipun pembelajaran kooperatif dapat memengaruhi hasil belajar afektif siswa secara positif, besarnya efek yang teramati tidak signifikan secara statistik pada = 0,05.
Oleh karena itu, disarankan agar penelitian masa depan menyempurnakan desain dan implementasi strategi pembelajaran kooperatif untuk meningkatkan efektivitasnya dan mencapai peningkatan yang signifikan secara statistik pada hasil belajar afektif siswa sekolah Kata kunci: hasil pembelajaran afektif, pembelajaran kooperatif, meta-analisis.
INTRODUCTION
Cooperative learning (CL) stands as a dynamic pedagogical strategy, deeply rooted in social interdependence theory, where studentsAo successes and growth are intricately tied to their interactions with peers, fostering a rich tapestry of positive interdependence, individual accountability, promotive interaction, social skills, and group processing (Colomer et al.
Johnson & Johnson, 2.
This approach draws from foundational motivational perspectives that champion group rewards to spark achievement and engagement, as highlighted by recent explorations into student-driven incentives (Bardach & Murayama.
Hellyn et al.
, 2.
, while cognitive-developmental theories, inspired by the likes of Piaget and Vygotsky, emphasize peer collaborations as catalysts for cognitive growth through conflict resolution and scaffolding (Cade, 2023.
Chen, 2.
The structural framework further enhances this model by embedding specific techniques to ensure equitable Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA participation and simultaneous engagement, nurturing both social competence and academic success (Kagan, 2.
Tracing its origins to the 1960s and 1970s, historical analyses distinguish CL from collaborative learning by its structured design and focus on K-12 education, laying a robust theoretical groundwork for diverse models that prioritize collaboration over competition (Yang, 2.
These principles form the bedrock of an adaptable educational philosophy, supporting critical thinking and teamwork across various learning stages.
Over the past decade .
5Ae2.
, a growing body of research has shed light on the remarkable effectiveness of cooperative learning (CL) across varied educational settings, significantly boosting student engagement, learning outcomes, and social skills, positioning it as a guiding light for innovative teaching practices (Boke et al.
, 2.
Systematic reviews highlight its contributions to cultural diversity, individual accountability, and alignment with sustainable development goals by embracing student-centered learning environments (Zhou & Colomer, 2.
, while meta-analyses demonstrate moderate yet meaningful positive effects across affective, cognitive, physical, and social domains, with strategies like Learning Together and Group Investigation leading the way (Boke et al.
, 2.
In higher education physics.
CL has proven its worth by enhancing academic achievement and attitudes through a mix of experimental and conceptual tasks (Kilpelyinen-Pettersson et al.
, 2.
, and bibliometric analyses reveal a notable uptick in research interest since 2021, focusing on motivation, cognitive skills, and communication in science education (I.
Damayanti et al.
Xue et al.
, 2.
Among its standout models, the jigsaw approach empowers students to master subtopics and teach peers, driving notable gains in academic performance and social skills, with meta-analyses confirming moderate to large effects, particularly in secondary and higher education (Bhardwaj, 2025.
Cochon Drouet et al.
, 2.
Student Teams-Achievement Divisions (STAD) taps into group rewards linked to individual progress to boost motivation and cognitive outcomes in physics (Aprilia & Dwandaru, 2.
, while Teams-Games-Tournament (TGT) introduces friendly competition to enhance activeness and participation in elementary classrooms (Saputri & Lestari, 2.
Comparisons underscore JigsawAos strength in problem-solving and retention, contrasting with STADAos motivational edge (Damayanti et al.
, 2.
, setting these methods apart from traditional approaches through their focus on interdependence and accountability.
Other models, such as Make a Match, pair students to link questions with answers or concepts with definitions, sparking active engagement, peer teaching, and quick problem-solvingAi especially effective in elementary settings for reinforcing knowledge and fostering social connections (Masrita, 2.
Group Investigation (GI) encourages students to design and lead inquiries, nurturing higher-order thinking and supporting low- to middle-achievers with improved cognitive and behavioral engagement (Silva et al.
, 2.
, while Team Assisted Individualization (TAI) merges group support with individual tasks to enhance mathematical problem-solving, communication, and self-efficacy in operations research (Tinungki et al.
Numbered Heads Together (NHT) ensures fair participation through numbered responses, boosting critical thinking in physics (Aprilia & Dwandaru, 2.
, and Think-PairShare (TPS) promotes equitable dialogue via reflection, pairing, and sharing, enriching participation across educational contexts (Guenther & Abbott, 2024.
Latifah & Aviya, 2.
Technology-infused models like Cooperative Mobile Learning, using tools such as Google Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA Docs, ease cognitive loads and elevate outcomes in elementary science, supported by metaanalyses showing enhanced performance and motivation in pre-college settings (Huang et al.
Sergeeva et al.
, 2.
Cooperative Problem-Based Learning (CPBL) blends CL with problem-solving cycles to build team skills, addressing group dynamics through training (Zaafour & Salaberri Ramiro, 2.
, while Two Stay Two Stray (TSTS) boosts science skills and motivation through information sharing (Damayanti et al.
, 2.
Collectively, these modelsAifrom achievement-driven Jigsaw and STAD to inquiry-focused GI and TAI and accessible tech-enhanced CPBLAiform a unified framework grounded in interdependence and accountability (Yang, 2.
, with recent studies .
5Ae2.
validating their ability to foster affective and cognitive growth, offering viable alternatives to traditional methods despite challenges like time and resources, and nurturing inclusive, reflective learning (Boke et al.
, 2025.
Zhou & Colomer, 2.
Recent scholarship from 2015 to 2025 has increasingly explored how cooperative learning (CL) models nurture the affective dimensions of elementary studentsAispanning motivation, emotional well-being, social skills, and learning satisfactionAiby fostering positive interdependence and meaningful peer interactions.
These dynamics help reduce stress, sharpen emotional regulation, and boost engagement across diverse areas like physical education and language learning, with a meta-analysis of 40 studies reporting a moderate effect size of 0.
304 across primary levels, enhancing the classroom atmosphere and providing emotional support (Boke et al.
, 2.
In physical education settings, research has revealed serial mediation effects, where perceived cooperation enhances academic performance by alleviating negative emotions such as anxiety and boredom while amplifying positive ones like enjoyment and pride (Leyn et al.
, 2.
This growing body of work positions CL as a resilient approach, particularly in culturally diverse classrooms, where it mitigates emotional challenges and encourages self-reflection (Zhou & Colomer, 2.
, offering modest to moderate affective gains as a refreshing, student-centered alternative to traditional methods through its structured group dynamics.
Specific CL models, including Jigsaw.
Student Teams-Achievement Divisions (STAD).
Teams-Games-Tournament (TGT), and Think-Pair-Share (TPS), have been thoroughly investigated for their impact on affective growth, distinguished by their emphasis on structured accountability.
Jigsaw stands out by cultivating collaboration, peer connections, and emotional intelligence, with notable pre- to post-test improvements in third-grade learning and behaviors (Putri & Pratiwi, 2.
and positive effects on motivation and selfperception (Mubayinah, 2.
STAD and TGT, widely used in physical education, help reduce negative emotions and strengthen social skills, a finding backed by meta-analytic evidence (Boke et al.
, 2.
, while TPS variants and technology-enhanced approaches like Quizlet boost emotional adaptability and peer-related outcomes (Lin et al.
, 2.
This cumulative evidence from 2015 to 2025 presents a robust perspective on CLAos influence through absolute post-intervention gains, highlighting models like Jigsaw and STAD for achieving emotionally supportive benchmarksAisuch as enhanced motivation and socialemotional learningAithat support long-term development (Li, 2025.
Rivera-Pyrez et al.
, empowering educators to build inclusive, resilient classrooms (B.
Chen, 2.
To unpack these effects, we employ a means-based meta-analysis, a thoughtful approach that synthesizes post-intervention means from experimental groups as the key effect Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA size metric, focusing on absolute performance rather than relative shiftsAia lifeline in educational research where it shines by leveraging continuous means, standard deviations, and sample sizes to gauge intervention impact (Borenstein et al.
, 2.
This method proves invaluable when control or pre-test data are scarce, relying on standardized post-test scoresAi often normalized to a 0Ae100 scaleAifor fair comparisons (Lipsey & Wilson, 2.
, as seen in recent studies like Yoon .
that highlight its clarity in assessing outcomes like achievement or affective growth.
Unlike contrast-based methodsAipre-post analyses tracking mean changes over time or group comparisons between experimental and control sets, which demand full datasets and may exclude studies (Cooper et al.
, 2009.
Hedges & Olkin, 1.
Ai means-based analysis embraces single-group data, as evidenced by cooperative learning studies showing varied effect sizes without contrasts (Yoon, 2.
Distinct from correlation meta-analysis, which explores variable relationships like study habits and grades (Schmidt & Hunter, 2.
, this approach zeroes in on post-intervention magnitude, offering educators tangible benchmarks, such as average scores in CL settings (Boke et al.
, 2.
, and adapts seamlessly to diverse tools, enhancing its relevance in education and psychology (Putri & Pratiwi, 2.
Traditional meta-analyses often lean on group contrast or pre-post designs, calculating effect sizes like CohenAos d to reveal relative gains in achievement and attitudes (Pellegrini et , 2021.
Ridwan et al.
, 2.
, with examples showing small to medium affective effects (Boke et al.
, 2.
and moderate overall impacts across 23 studies .
nztyrk, 2.
Yet, their reliance on complete multi-group or time-point data can limit inclusion and miss experimental group nuances, prompting the need for a means-based alternative that highlights absolute post-intervention performance.
This shift is especially critical in elementary contexts, where post-test means can signal if CL hits emotionally supportive targets beyond relative improvements (Chen, 2023.
Rivera-Pyrez et al.
, 2.
, sidestepping the need for control data and proving its worth when baselines are absent (Yoon, 2.
By analyzing means, standard deviations, and score rangesAistandardized to 0-100 to align varied instrumentsAithis method offers a practical, single-group lens that mirrors real-world teaching scenarios where only intervention data are available (Lewis et al.
, 2.
Prior contrast-based studies, while insightful, often miss absolute efficacy due to strict data requirements .
nztyrk, 2023.
Ridwan et al.
, 2.
, but our means-based meta-analysis bridges this gap, unveiling actionable thresholds for affective growth and enriching pedagogy in elementary settings.
Recently, cooperative learning has emerged as a hopeful strategy to enhance motivation, social skills, and attitudes in elementary mathematics, where traditional methods can stifle enthusiasm.
This study delves into its impact on affective outcomes and potential publication bias, employing funnel plots and EggerAos regression analyses.
Our goals are twofold: .
to explore how cooperative learning shapes elementary studentsAo affective learning outcomes, and .
to investigate publication bias in related studies.
Guiding us are two key questions: R1: What is the overall effect size of cooperative learning on these outcomes? R2: Is there evidence of publication bias influencing the research? Through this lens, we aim to offer educators a clearer, evidence-based path forward.
Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA
METHODS
This study utilized a quantitative meta-analysis design, specifically a means-based approach, to evaluate the effectiveness of cooperative learning models on elementary school students' affective outcomes.
This design synthesizes post-test data from experimental groups across multiple studies, focusing on absolute means as the effect size indicator rather than relative contrasts, allowing for an assessment of standalone intervention impacts when comparative data are limited (Borenstein et al.
, 2.
By aggregating standardized post-test means, standard deviations, sample sizes, and instrument score ranges, the design provides a complementary perspective to traditional meta-analyses, emphasizing practical benchmarks in affective domains such as motivation and social skills (Lipsey & Wilson, 2.
The meansbased meta-analysis was selected for its suitability in educational research where single-group post-intervention data predominate, enabling inclusion of studies lacking control or pre-test information (Cuijpers, 2.
This approach aligns with systematic review principles, ensuring rigorous synthesis through predefined criteria and statistical pooling techniques (Cooper et al.
, 2.
This meta-analysis study commenced with a systematic collection of data from peerreviewed articles and theses published between 2015 and 2025, sourced through databases such as ERIC.
Scopus.
Web of Science.
Elsevier.
Springer, and Google Scholar.
The search focused on keywords like Authe effectiveness of cooperative learning model to increase elementary school students' affective learning outcomesAy and Authe effect of cooperative learning models on elementary school students' affective learning outcomes.
Ay The next step involved gathering research findings centered on independent variables, specifically the cooperative learning models applied to experimental groups, with the dependent variable being the affective learning outcomes of elementary school students.
To ensure relevance, the retrieved studies were carefully re-examined and screened based on specific criteria: inclusion of quasi-experimental designs and the availability of descriptive data analysis, including sample size, standard deviation, mean, and range scores from the experimental group.
flowchart, illustrated in Figure 1, outlines this literature selection process, guiding researchers through the identification of suitable studies.
Subsequently, the identified studies were organized and categorized by author name, publication year, and descriptive data, facilitating a detailed analysis of the cooperative learning models' application.
The final phase of this meta-analysis involved evaluating studies that met the criteria for heterogeneity, effect size calculation, forest plot generation, funnel plot analysis, and publication bias assessment, ensuring a comprehensive evaluation of the data.
Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA Figure 1.
Flowchart of Research Procedure Before delving into the analysis, we transformed the data to standardize heterogeneous score ranges onto a common 0-100 scale, enhancing comparability across studies.
Each studyAos effect size was then computed using a heterogeneity test to determine the appropriate effect model for the subsequent meta-analysis.
Following this, a forest plot analysis was conducted, assigning each sample a standardized mean as an effect size value and calculating a pooled standardized mean as a summary effect size, complete with lower and upper confidence bounds.
This process provided robust evidence supporting the effectiveness of cooperative learning in enhancing affective learning outcomes among elementary school To visualize these findings, a funnel plot meta-analysis depicted effect sizes as circles distributed within a pyramid shape, allowing us to intuitively assess the samples Additionally, to address potential publication bias, we employed funnel plots alongside Rank correlation and regression techniques, enabling us to detect any significant bias that might influence the results.
This meticulous approach not only humanizes the research process by focusing on practical educational impacts but also strengthens the reliability of our conclusions.
This meta-analysis kicked off with a thoughtful review of literature, pulling data from peer-reviewed articles and theses published between 2015 and 2025, sourced from reputable Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA databases like ERIC.
Scopus.
Web of Science.
Elsevier.
Springer, and Google Scholar.
Our search honed in on terms such as Authe effectiveness of cooperative learning model to increase elementary school students' affective learning outcomesAy and Authe effect of cooperative learning models on elementary school students' affective learning outcomes,Ay targeting studies where cooperative learning models were put to work in experimental classrooms.
The heart of our focus was the dependent variableAielementary studentsAo affective outcomes, like motivation and social skillsAiwhile we also considered moderator variables tied to grade levels: 6th grade .
, 5th grade .
, 4th grade .
, a mix of 5th-6th grades .
, and 3rd-4th grades .
To keep things relevant, we sifted through studies indexed in ERIC.
Scopus.
Web of Science, and Springer, ensuring they met our standards with quasi-experimental designs and provided key descriptive statsAisample size, mean, standard deviation, and range scoresAifrom the experimental groups.
Using Google Scholar as our search hub, we uncovered 12 research studies and 16 artifacts, with some drawing from Scopus.
SINTA, and GARUDA-indexed journals, while others relied solely on Google Scholar listings.
Once gathered, we organized the findings by author, publication year, and detailed descriptive data, setting the stage for a deep dive into how these cooperative learning models play out.
The descriptive analysis we conducted coded each studyAos results, capturing sample sizes, standard deviations, means, and range scores to paint a clear picture of their impact on affective learning outcomes.
This careful curation and categorization helped us lay a solid foundation for understanding the real-world application of these educational strategies across different grade levels and settings.
This study blended meta-analysis with forest plot analysis to distill and generalize insights from prior research, focusing on how cooperative learning influences elementary school studentsAo affective learning outcomes.
We started by transforming the diverse score ranges into a unified 0-100 scale to ensure fair comparisons, then calculated standardized means as effect sizes for each experimental groupAiour sole focus for validating learning effectivenessAiset against a benchmark of 75 (MCC).
The forest plot technique brought these findings to life by computing a summary effect value, derived from the average mean size across samples, offering a clear picture of impact.
To gauge learning efficiency, we pooled these standardized means into an estimated summary effect size, paired with a z-value and pvalue, allowing us to test the hypothesis.
If the z-value dipped below 0.
05 and the summary effect size fell short of the 75 benchmark, weAod question the effectiveness of the cooperative learning models (Retnawati et al.
, 2.
Before diving into the forest plot, we tackled heterogeneity, using the Q-test.
EA, and IA metrics to decide between a random-effects or fixedeffects model, ensuring our analysis reflected the dataAos natural variability (Lipsey & Wilson.
To assess heterogeneity, we leaned on the Q-testAos p-values.
EA estimates, and IA values, recognizing that a diverse data pool signals multiple distributions.
A p-value below the significance threshold confirmed heterogeneity, while IA valuesAicategorized as low .
P%), moderate .
%-75%), or high .
ver 76%)Aiprovided a deeper look at variability (Borenstein et al.
, 2009.
Cooper et al.
, 2009.
Higgins et al.
, 2.
An IA above 25% hinted at substantial variation due to sampling errors, population differences, and true effect sizes, a nuance we explored further with the DerSimonian and Laird method to estimate EA (Borenstein et al.
, 2.
We rejected the null hypothesis if effect sizes exceeded 75.
Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA signaling strong impact (Retnawati et al.
, 2.
For broader validation, we turned to the Trim and Fill method, cross-checking summary effect values from random-effects .
r fixedeffect.
models before and after adjustment.
If the results held steady, it affirmed that cooperative learningAos effect on affective outcomes aligned with our criteria, grounding the findings in a robust, real-world context.
To pinpoint potential biases, we employed the Trim and Fill method alongside a visual funnel plot analysis, scrutinizing the symmetry of effect size distributions to flag any skewed studies (Card, 2.
We complemented this with statistical tools like EggerAos regression and Begg-Mazumdar rank correlation, testing the null hypothesis of symmetryAiwhere a p-value Ou 0.
05 suggests no bias and a balanced funnel plot, while a p-value < 0.
05 indicates asymmetry and possible publication bias (Begg & Mazumdar, 1994.
Borenstein et al.
, 2009.
Cooper, 2017.
Egger et al.
, 1.
This approach ensured our meta-analysis captured the true scope of cooperative learningAos impact.
Drawing from established methods, we weighted effect sizes as the average of post-test means, adjusted by inverse variance, and set 95% confidence intervals to reflect precision (Hedges & Olkin, 1.
Heterogeneity was further explored with the IA statistic and Q-test, where values exceeding 50% prompted metaregression to probe moderators like CL model type or sample size (Higgins et al.
, 2.
Using JASP for analysis, we conducted sensitivity checks to confirm robustness, delivering a clear, actionable measure of cooperative learningAos absolute influence on affective outcomes, making the process feel both rigorous and relatable to educators.
RESULTS AND DISCUSSION
The results of this meta-analysis are organized into four key sub-sections to provide a comprehensive understanding of the findings.
First, the data encoding process outlines the selection and extraction of study characteristics, ensuring accuracy and consistency for statistical synthesis.
Next, the heterogeneity test evaluates the degree of variation among included studies, informing the choice of analytical model.
The analysis of effect size presents the magnitude and direction of the interventionAos impact.
Finally, the assessment of biased publication examines the potential influence of selective reporting on the overall Data Encoding Data encoding in meta-analysis is the systematic process of organizing qualitative and quantitative information from primary studies into a consistent, structured format for statistical analysis.
This step is essential for synthesizing study characteristics such as sample sizes, means, standard deviations, and moderator variables, which support accurate effect size estimation and heterogeneity assessment (Cooper, 2.
In educational research, as in YoonAos .
meta-analysis of cooperative learning, encoding ensures that variables like school level, research design, and outcome domains are documented uniformly, reducing bias and improving reproducibility (Higgins et al.
, 2.
Effective coding protocolsAioften collaboratively developed and validated through inter-rater reliabilityAiare vital for maintaining data quality and drawing valid conclusions (Lipsey & Wilson, 2.
Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA Table 1.
the Result of Data Coding Researcher & year of Hortigyela Alcaly et al.
Hortigyela Alcaly et al.
Hortigyela Alcaly et al.
MaxAe 5Ae1 5Ae1 5Ae1 Hakim & Syofyan .
Kesnajaya et al.
Chen .
Grade 75Ae15 164Ae41 80Ae20 Salamah et al.
5Ae1 Demitra & Sarjoko .
Rivera-Pyrez et al.
15Ae3 5Ae1 5thAe6th Amrullah & Suwarjo .
Amrullah & Suwarjo .
Bandaso et al.
Huang et al.
Lin et al.
Lin et al.
Putri & Pratiwi .
175Ae35 175Ae35 150Ae30 7Ae1 5Ae1 5Ae1 100Ae20 3rdAe4th 3rdAe4th Assessed Attitude Motivation Social Motivation Motivation Social Intrinsic Social skills Cooperative Interpersonal Interpersonal Motivation Motivation Satisfaction Motivation Collaborative Note: =sample size.
=mean.
= standard deviation The raw data table compiles descriptive statistics from numerous primary studies evaluating psychological and social constructsAisuch as attitude, motivation, social interaction, social skills, cooperative behavior, interpersonal skills, satisfaction, and collaborative attitudeAiamong students from grades 3 to 6.
For each study, it reports the experimental group sample size, mean score, standard deviation, observed score range, grade level, and construct measured.
However, these studies used highly diverse scoring scales, ranging from 5-point Likert formats .
Hortigyela Alcaly et al.
, 2.
to raw scores up to 175 (Amrullah & Suwarjo, 2.
and intermediate ranges such as 0Ae100 (Putri & Pratiwi, 2.
, creating substantial heterogeneity in measurement units.
This variability means that raw means cannot be directly compared, as differences may stem from scale disparities rather than true performance differences.
To address this, all data underwent standardization, typically by converting to a common metric .
, z-scores or proportional rescalin.
so that variables share the same interpretive unit.
This approach enables integration into a single meta-analytic framework free from scale-related bias.
The standardized datasetAipresented in Table 2Aitransforms all raw means, standard deviations, and standard errors into a consistent scale, permitting valid cross-study comparisons and aggregation in a mean-based metaanalysis.
For instance.
Hortigyela Alcaly et al.
Aos .
means of 4.
11Ae4.
34 on a 5-point scale were rescaled to fall within approximately 70Ae85, aligning them with studies using broader numerical ranges, while large raw scores such as 152.
40 in Kesnajaya et al.
were proportionally reduced.
Through this transformation, subsequent statistical synthesis reflects Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA genuine differences in performance rather than artifacts of varying scale lengths, thereby ensuring the comparability and validity of the meta-analytic results.
Table 2.
the Result of Data Standardization Researcher & year of Grade Hortigyela Alcaly et al.
79 6th Hortigyela Alcaly et al.
97 6th Hortigyela Alcaly et al.
05 6th Hakim & Syofyan .
28 4th Kesnajaya et al.
36 5th Chen .
08 6th Salamah et al.
33 4th Demitra & Sarjoko .
09 4th Rivera-Pyrez et al.
45 5thAe6th Amrullah & Suwarjo .
95 5th Amrullah & Suwarjo .
90 5th Bandaso et al.
92 5th Huang et al.
42 6th Lin et al.
07 3rdAe4th Lin et al.
24 3rdAe4th Putri & Pratiwi .
44 3rd Note: =mean.
= standard deviation.
= standard error Assessed Attitude Motivation Social interaction Motivation Motivation Social interaction Intrinsic motivation Social skills Cooperative behavior Interpersonal Interpersonal Motivation Motivation Satisfaction Motivation Collaborative attitude For each study, the standardized descriptive statistics include the sample mean ( ), standard deviation ( ), and standard error ( ), alongside contextual variables such as grade level and assessed construct.
The table shows a diverse range of constructs, including attitude, motivation, social interaction, social skills, cooperative behavior, interpersonal skills, satisfaction, and collaborative attitude, assessed across various grade levels from 3rd to 6th.
Means range from a low of 60.
50 (Amrullah & Suwarjo, 2018, interpersonal skill.
to a high 57 (Kesnajaya et al.
, 2018, motivatio.
, reflecting differences in performance or ratings across contexts.
Standard deviations vary widelyAifrom 5.
ow variabilit.
26 (B.
Chen, 2021, social interactio.
, indicating differing score dispersions.
Standard errors, derived from SD and sample size, provide a measure of precision, with smaller values .
indicating more stable estimates.
By standardizing these metrics across studies, the table ensures comparability for subsequent mean-based meta-analysis, allowing results from heterogeneous measurement scales to be integrated consistently.
This step is essential for reliable synthesis and interpretation of central tendencies across the included research.
Heterogeneity Test Heterogeneity in meta-analysis refers to variation in effect sizes across studies beyond what is expected by chance.
It guides the decision between fixed-effect and random-effects models and may indicate potential moderators for further analysis.
The standard assessment uses CochranAos Q test, which compares observed variability with that expected from sampling error, and the IA statistic, which expresses the percentage of total variation due to true differences rather than chance (Borenstein et al.
, 2.
In meta-regression, an omnibus test evaluates the joint significance of predictors, while a residual heterogeneity test examines remaining variance after accounting for covariates (Higgins et al.
, 2.
Recent work has Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA proposed bootstrap-based approaches to improve robustness, particularly in small-sample contexts (Du et al.
, 2.
Detecting significant heterogeneity is essential, as it signals the need for subgroup or moderator analyses to identify sources of inconsistency (Spineli & Pandis, 2.
Table 3.
the Result of the Homogeneity Test Analysis Omnibus test of Model Coefficients Test of Residual Heterogeneity Note: p -values are approximate.
< .
< .
The omnibus test of model coefficients yielded a Q value of 1025.
f = 1, p < .
indicating that the predictors in the model collectively explain a significant portion of the variance in effect sizes, rejecting the null hypothesis of no overall model effect.
However, the test of residual heterogeneity produced a Q value of 970.
f = 15, p < .
, signifying substantial unexplained variability beyond the modeled factors, thus confirming heterogeneity and necessitating a random-effects approach or further moderator investigations (Viechtbauer.
These approximate p-values underscore the presence of diverse study characteristics contributing to dispersion, aligning with recommendations for cautious interpretation in educational meta-analyses where contextual factors often amplify heterogeneity (Cheung & Chan, 2.
Following the homogeneity test results, which showed significant model effects (Q = 7, df = 1, p < .
and notable residual heterogeneity (Q = 970.
5, df = 15, p < .
examining residual heterogeneity estimates is crucial to assess unexplained variability in effect sizes.
This variability, not captured by model predictors, highlights the need for further moderator analyses to ensure consistent findings, particularly in studies like Yoon .
on cooperative learning, guiding future research (Borenstein et al.
, 2009.
Higgins et al.
, 2.
Table 4.
The Result of the Residual Heterogeneity Estimates Estimate EA
IA (%)
Table 4 presents the residual heterogeneity estimates, with EA = 88.
E = 9.
= 97.
9419%, and HA = 48.
5877, indicating substantial variability in effect sizes beyond the modeled predictors.
The EA value .
represents the estimated variance of true effect sizes across studies, while E .
is its square root, reflecting the standard deviation of these effects (Borenstein et al.
, 2.
The IA value of 97.
9419% suggests that approximately 98% of the observed variance is due to true differences between studies rather than chance, a high level of heterogeneity typical in educational interventions with diverse contexts (Cheung & Chan, 2.
The HA value .
further indicates that the total variance is nearly 49 times the within-study variance, reinforcing the need for a random-effects model and further exploration of moderators such as school level or intervention duration to explain this variability (Viechtbauer, 2.
Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA Analysis of Effect Size Effect size analysis is a core element of meta-analysis, offering a standardized measure of an effectAos magnitude for comparison across studies with varying designs, scales, or It involves calculating individual effect sizes and pooling them into a summary effect, which represents the overall estimate of the relationship or difference studied.
This estimate is interpreted with its standard error and confidence intervals to assess precision.
forest plot visually presents these results, showing each studyAos estimate and confidence interval alongside the pooled estimateAitypically displayed as a diamond.
This visualization aids in detecting heterogeneity, identifying outliers, and evaluating the consistency of effects across studies.
Together, the summary effect and forest plot provide both numerical and graphical insights into the direction, strength, and reliability of the findings.
Table 5.
The Result of Summary Effect Size Estimate Intrcpt Note: Wald test.
Standard Error Lower Bound Upper Bound Based on Table 4, the mean effect size (ES) is 77.
057, with a standard error (SE) of The 95% confidence interval ranges from 73.
ower boun.
pper These values indicate that, on average, the observed outcome across all studies included in the meta-analysis is approximately 77.
06, with a reasonable level of precision given the relatively small standard error.
In hypothesis testing, the z-value is compared to the critical value to determine statistical significance.
Here, the reported z-statistic is 0.
At the = 0.
05 significance level, the critical z-value is A1.
Because 0.
4035 < 1.
96, the result is not statistically significant, meaning that the null hypothesis .
o difference from a reference valu.
cannot be This indicates insufficient evidence to claim that the true effect size differs meaningfully from the null or benchmark value under consideration.
The 95% CI .
34 to 81.
suggests that we are 95% confident the true average effect size lies within this range.
Importantly, because this interval includes values both slightly below and above a potential benchmark criterion .
, .
, the conclusion regarding meeting a minimum performance standard is statistically uncertain.
The lower bound .
implies that some populations or conditions may fall short of the threshold, while the upper bound .
indicates that in other cases, performance could exceed it.
In summary, the estimated summary effect size of 77.
06 (SE = 2.
reflects moderate to high central tendency in the measured outcome, yet with a confidence interval broad enough to encompass both marginally below-standard and above-standard This reinforces the need to interpret the pooled estimate in conjunction with heterogeneity measures and visual evidence from the forest plot, ensuring that variability between studies and contextual differences are fully considered before drawing definitive Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA Figure 2.
Forest Plot Each horizontal line in the forest plot represents the 95% confidence interval (CI) for an individual studyAos observed outcome .
ffect siz.
, while the center square marks the point estimate of the effect size for that study.
The size of each square is proportional to the weight assigned to the study in the meta-analysis, which is generally determined by the inverse of the Consequently, studies with larger sample sizes or more precise estimates are assigned greater weight, reflected in larger squares (Borenstein et al.
, 2.
At the bottom of the figure, the diamond represents the summary effect size and its 95% CI under a random-effects model.
The center of the diamond corresponds to the mean observed outcome .
, and its width spans the 95% CI .
34 to 81.
A narrower diamond indicates greater precision in the summary estimate, while a wider diamond signals more uncertainty.
In this plot, the diamond overlaps with several individual study CIs, suggesting that the pooled estimate is broadly consistent with much of the individual evidence, although variability is evident.
The position of each studyAos effect size relative to the vertical reference line .
ften associated with a benchmark or criterionAiin this case, minimum completeness criterion (MCC) of .
indicates whether the outcome exceeds or falls short of that threshold.
Studies plotted to the right of the criterion line report higher outcomes, while those to the left fall below it.
In this analysis, six studies (Hortigyela Alcaly et al.
, 2019.
Hortigyela Alcaly et , 2019.
Hakim & Syofyan, 2018.
Kesnajaya et al.
, 2015.
Salamah et al.
, 2024.
Lin et al.
Lin et al.
, 2025.
exceed the pooled summary effect size of 77.
Conversely, eight studies (Hortigyela Alcaly et al.
, 2019.
Chen, 2023.
Demitra & Sarjoko, 2017.
Rivera-Pyrez et al.
, 2021.
Amrullah & Suwarjo, 2016.
Amrullah & Suwarjo, 2016.
Bandaso et al.
Putri & Pratiwi, 2.
report effect sizes lower than the summary estimate.
Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA Notably, some studies .
Kesnajaya et al.
, 2015.
Hakim & Syofyan, 2.
demonstrate particularly high effect sizes well above the KKM, suggesting strong performance in those contexts.
Others, such as Amrullah & Suwarjo .
1 and 2018.
report markedly lower outcomes, falling well below both the summary estimate and the MCC.
This variability underlines the heterogeneity present in the dataset, which is common in educational interventions due to differences in participant characteristics, implementation fidelity, and contextual factors (Higgins et al.
, 2.
Overall, the results indicate that while the pooled mean outcome .
slightly exceeds the MCC threshold of 75, a considerable number of studies fall below this benchmark, suggesting that the implementation of the intervention may not consistently ensure performance above the criterion across all settings.
This finding warrants further examination of moderating variables and implementation conditions to better understand the sources of variability and to optimize the intervention's effectiveness in diverse contexts.
Analysis of Biased Publication Publication bias is a well-recognized threat to the validity of meta-analytic findings, arising when studies with significant or favorable results are more likely to be published than those with null or negative outcomes.
Such bias can lead to an overestimation of the true effect size, thereby compromising the generalizability of conclusions.
Funnel plots provide a visual diagnostic tool to detect potential asymmetry in the distribution of study effect sizes against their standard errors, which may indicate the presence of publication bias.
statistically assess this asymmetry.
EggerAos regression test offers a quantitative method to evaluate whether small-study effects are disproportionately influencing the overall results.
The funnel plot can be shown in Figure 3 as follows:
Figure 3.
Funnel Plot Figure 3 serves as a diagnostic tool for assessing potential publication bias and the dispersion of effect sizes across studies.
On this plot, the x-axis represents mean outcomes, while the y-axis denotes standard error.
The vertical line marks the pooled effect size, and the pseudo-95% confidence limit region forms the expected funnel shape under unbiased and Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA normally distributed conditions (Higgins et al.
, 2020.
Sterne et al.
, 2.
Most studies cluster near the apex of the funnelAiwhere standard errors are minimalAiindicating that higherpowered studies yield more precise estimates closely aligned with the pooled mean.
A slight asymmetry, with a marginally greater number of points on the right, may suggest a mild tendency toward higher reported outcomes.
However, the presence of lower-performing studies with larger standard errors implies that this asymmetry is unlikely to reflect systematic bias (Higgins et al.
, 2.
Studies positioned toward the base of the funnelAiwhere precision is lowAidisperse widely, reflecting expected sampling variability.
The overall symmetry and convergence at higher precision levels reinforce the robustness of the meta-analytic findings.
Nonetheless, while visual inspection suggests minimal bias, formal statistical testingAisuch as EggerAos regression testAishould be conducted to confirm the absence of small-study effects or publication bias.
Table 6.
Regression Test for Funnel Plot Asymmetry ("Egger's Test") Table 6 presents a statistical evaluation of potential publication bias as visually suggested by the funnel plot, using EggerAos regression testAia method that examines the association between standardized effect size and its standard error, where a statistically significant non-zero intercept signals asymmetry and possible bias(Sterne et al.
, 2.
In the present meta-analysis, the test resulted in z = Ae0.
1078 and p = .
914, indicating no statistically significant asymmetry at the = 0.
05 level.
The non-significant p-value .
> .
suggests that any apparent asymmetry in the funnel plot likely arises from random variation rather than systematic publication bias or small-study effects.
Thus, the analysis provides reassurance that the pooled results are not substantially influenced by selective evidence reporting.
Discussion This meta-analysis focused on evaluating how cooperative learning models influence affective outcomes among elementary school students, synthesizing descriptive data from multiple primary studies.
The aggregated mean outcome across all studies was 77.
06, with a 95% confidence interval of 73.
34 to 81.
Although the pooled mean surpasses the hypothetical threshold of 75, the result was not statistically significant .
= .
, indicating that the observed central tendency may not reliably differ from chance fluctuations around a benchmark value.
This finding suggests that while cooperative learning tends to yield positive affective resultsAisuch as improved attitudes, motivation, or social-emotional skillsAithe degree of improvement may not be consistently robust across diverse educational contexts.
The descriptive focus on mean outcomes, rather than standardized mean differences, underscores the interpretive emphasis on absolute performance levels rather than comparative effect magnitude.
Notably, the meta-analysis revealed substantial heterogeneity: the Q statistic was 1,025.
< .
, and the IA value reached an exceptionally high 97.
94%, indicating that nearly all observed variance among study outcomes derives from real differences rather than sampling error (Deeks et al.
, 2019.
Linden & Hynekopp, 2.
The residual heterogeneity estimates (EA = 88.
E = 9.
further underscore the considerable between-study variability Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA in affective outcomes.
Such pronounced dispersion likely reflects differences in classroom contexts, variations in the fidelity of implementing cooperative learning strategies, divergence in measurement instruments, and heterogeneity of student populations.
Higgins and colleagues caution that when synthesizing results from complex educational interventions, substantial heterogeneity is expected and should trigger investigations into moderating variables rather than rejection of meta-analytic synthesis (Deeks et al.
, 2019.
Ruppar, 2.
Therefore, the pooled mean should be interpreted with caution, acknowledging that contextspecific conditions critically influence generalizability and effectiveness.
Practically, the pooled mean of 77.
06 suggests that cooperative learning can broadly foster affective improvements, albeit with variability.
Educators should view cooperative learning as a potentially beneficial but not universally guaranteed strategy.
To maximize benefits, teachers need to adapt cooperative structuresAisuch as Jigsaw.
Think-Pair-Share, or STADAito suit local classroom dynamics, student needs, and curricular demands.
For instance, success may hinge on teacher facilitation skills, group composition practices, and the clarity of roles within tasks.
Schools could enhance professional development by emphasizing strategies for promoting positive interdependenceAithe foundational component of cooperative learning where individual success is tied to group success (Johnson & Johnson, 2018.
Ngoc Tuong Nguyen & Thi Kim Oanh, 2.
The statistical strength of the summary effect size, expressed as 77.
057 with a 95% CI 34 to 81.
77, further underscores the magnitude of the observed differences across the included studies.
While the p-value of 0.
343 suggests that the aggregated mean does not significantly deviate from the hypothesized central value at conventional significance thresholds, the substantive meaning of the estimate should not be overlooked.
The interpretation of the standardized mean in this context extends beyond statistical significance.
it reflects the overall pattern of improvement or performance across interventions, which may hold considerable educational or practical relevance despite the lack of statistical rejection of the null hypothesis (McShane et al.
, 2024.
Peeters, 2016.
Wasserstein et al.
, 2.
Indeed, in educational and psychological research, practical significanceAiparticularly when large aggregate means are observedAioften takes precedence in informing decision-making and policy, especially when the interventions are implemented in diverse and heterogeneous settings (Funder & Ozer, 2019.
Haynes et al.
, 2018.
Premachandra & Lewis, 2.
This distinction between statistical and practical significance is especially pertinent when synthesizing data from quasi-experimental or field studies where sample variability and design complexity can dilute statistical detectability but not the real-world importance of observed trends.
The heterogeneity measures from the forest plot present a critical dimension to the interpretation of the findings.
With Q = 1,025.
7, p(Q) < .
001, and IA = 97.
94%, the analysis reveals substantial inconsistency among study results.
Such a high IA value indicates that almost all the observed variance across studies is attributable to genuine heterogeneity rather than sampling error, a phenomenon common in meta-analyses of educational and behavioral interventions (Migliavaca et al.
, 2022.
von Hippel, 2.
This level of heterogeneity necessitates cautious interpretation, as it implies that the intervention effects are not uniform and may vary substantially depending on contextual factors such as participant demographics, instructional settings, intervention duration, and fidelity of implementation.
As recommended Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA by Higgins et al.
, high heterogeneity should prompt a deeper exploration through subgroup analyses or meta-regression to identify possible moderators that explain variability, thereby refining the theoretical and practical understanding of intervention impacts.
Without accounting for these moderators, the overall mean effect might obscure important nuances that can guide more targeted and effective applications in real-world contexts.
Despite the substantial heterogeneity, the forest plot provides compelling evidence that most individual study estimates cluster around positive standardized mean values, reinforcing the consistency of direction in the observed effects.
The distribution of confidence intervals in the forest plotAithough wide for some studiesAipredominantly lies above neutral, which suggests a broadly beneficial influence of the interventions under review.
The few studies with smaller or more variable effects could reflect variations in methodological rigor, sample characteristics, or contextual challenges in implementation (Lengnick-Hall et al.
, 2023.
Reed et al.
, 2.
Such variability is not necessarily a limitation but may instead highlight the adaptability and flexibility of the intervention approaches across diverse educational Moreover, the aggregation of these diverse results into a coherent summary effect, despite the statistical challenges posed by heterogeneity, underscores the robustness of the meta-analytic approach in distilling meaningful conclusions from complex and multifaceted research landscapes (Cooper, 2.
This robustness is further strengthened when sensitivity analyses and publication bias assessmentsAisuch as funnel plot symmetry checksAisupport the reliability of the synthesized estimates.
In sum, the evidence synthesized through the forest plot and summary effect size strongly supports the conclusion that the interventions exert a substantial positive influence, as reflected by the standardized mean of 77.
057 within a narrow confidence interval.
While statistical non-significance at p = 0.
343 may temper some interpretations, the magnitude and consistency of the observed effects across studies remain compelling for practical and policyoriented considerations.
The high heterogeneity suggests that the interventions are not onesize-fits-all solutions, but rather flexible frameworks whose success depends on careful alignment with contextual needs and characteristics.
These findings align with prior metaanalyses in education and social sciences that have similarly observed high variability alongside robust mean effects, advocating for context-sensitive adaptation of intervention models (Lynch et al.
, 2025.
Slavin, 2.
Ultimately, this synthesis not only consolidates empirical support for the interventions but also provides a roadmap for future research aimed at disentangling the complex interplay of factors that drive variation in outcomes, thereby advancing both theoretical frameworks and applied practices in the field.
CONCLUSION
This meta-analysis demonstrates that cooperative learning models hold promise as an instructional approach for enhancing affective learning outcomes in elementary school, yet its impact is not uniformly consistent across contexts.
The findings suggest that its effectiveness depends on factors such as instructional design, teacher facilitation, classroom dynamics, and curriculum integration.
Rather than serving as a universal solution, cooperative learning models should be viewed as one valuable component within a broader repertoire of pedagogical strategies.
When implemented thoughtfully and supported through professional Al Ibtida: Jurnal Pendidikan Guru MI.
Vol.
12 No.
October 2025 MaAoas Shobirin.
Fitria Martanti.
Ummu Jauharin Farda.
Ersila Devy Rinjany.
The Effect of Cooperative Learning Model on Elementary SchoolA development and ongoing evaluation, it can contribute meaningfully to fostering deeper student engagement and achievement in science learning.
REFERENCES