Augmented reality technology in enhancing learning retention and critical thinking according to STEAM program

The paper addresses how augmented reality (AR) can enhance science learning within a STEAM framework at the intermediate (second grade of middle school) level in Ha'il, KSA. Motivated by rapid technological change, the rise of digitally native learners, and observed shortcomings in students’ understanding of science concepts and STEAM test performance, the study explores AR’s potential to create engaging, learner-centered environments aligned with constructivist and modern communication theories. The research problem stems from insufficient critical thinking skills and inadequate multimedia support in textbooks. The study aims to evaluate AR’s effectiveness on learning retention (LR) and critical thinking (CT) and to examine how AR design (image vs. marker), mental capacity (high vs. low), and gender (male vs. female) interact. The research questions are: (1) How do AR design, mental capacity, and gender interact to affect learning retention? (2) How do these factors influence the development of CT skills? (3) What is the effectiveness of AR in enhancing science learning outcomes according to the STEAM program?

The literature positions AR as a technology blending real and virtual elements with real-time interaction and 3D registration, deployable on common smartphones without specialized equipment. AR content can include text, audio, video, and 2D/3D models, supporting interactive, context-rich learning. Types include location-based and vision-based AR; the latter encompasses marker-based and markerless tracking. In STEAM contexts, AR supports learner-centered, activity-based education, enhancing motivation, collaboration, spatial awareness, and learning gains. AR has been linked to improved handling of complex, abstract concepts, better memory for procedures, and higher-order thinking. Critical thinking (CT) is a multifaceted construct involving analysis, inference, evaluation, and problem-solving. STEAM approaches and AR can foster CT through problem-based, project-oriented learning and visualization of abstract ideas. Mental capacity is framed as the ability to make decisions and sustain performance, influenced by cognitive, emotional, and environmental factors; AR’s kinesthetic, learning-by-doing affordances may build cognitive structures and motivation. Prior work highlights AR’s benefits but also implementation challenges (setup, lighting, device capability, app limitations, curricular constraints).

Design: Quasi-experimental pretest/posttest control-group design with matched groups randomly assigned to experimental and control conditions. In experimental groups, science content was taught with AR; in controls, traditional methods were used.

Participants: 120 second-grade middle school (8th-grade) students from six private schools in Ha'il city (2021–2022); 63 females and 57 males. All owned a smartphone and consented to participate. Mental capacity (MC) was measured using the Juan Pascual-Leone cross-shape test (36 items). Students scoring ≥20 classified as high MC; <20 as low MC (Al-Banna & Al-Banna, 1990).

Grouping: Eight experimental subgroups by 2 (gender: male/female) × 2 (MC: high/low) × 2 (AR design: image/marker). Control group(s) received traditional instruction. Table 3 labels include: G1 male/high/marker; G2 female/high/marker; G3 male/low/marker; G4 female/low/marker; G5 male/high/image; G6 female/high/image; G7 male/low/image; G8 female/low/image.

Instruments: (1) Achievement test (for science topics) used post-intervention and after two weeks to assess learning retention; Aiken’s V validity = 0.78 (5% error rate), 45 valid items, 5 invalid; Cronbach’s alpha = 0.88 (reliable at 0.95 CI threshold r=0.4428). (2) Cornell Critical Thinking Scale (Arabized to Saudi environment) to measure CT (Al-Zahrani, 2017; Ezz Al-Arab & Saad, 2016; Al-Hajri, 2018). Internal consistency supported by point-biserial correlations significant at p ≤ 0.01. Cronbach’s alpha for overall CT = 0.92; subscales ranged 0.75–0.88.

Procedure and materials: AR implemented via HP Reveal on smartphones/tablets using image/marker triggers. Content focused on light waves: characteristics of light, refraction through media and prisms, laws of reflection, uses of mirrors. Educational objects included videos and motion infographics. Students received ready-made AR science cards. AR activities lasted ~120 minutes, with additional face-to-face instruction ~240 minutes over ~3 weeks. Students could also use activity booklets independently at home. Teachers were briefed; researchers observed classes prior to intervention. Ethical procedures aligned with University of Ha'il guidelines (BA-2203), with informed consent.

Analysis: MANOVA examined main and interaction effects of AR design (image vs. marker), mental capacity (high vs. low), and gender on learning retention, CT skills, and learning outcomes. Post hoc comparisons used LSD tests; effect sizes reported where provided.

• Learning retention (LR): Significant effects involving mental capacity and its interactions. From Table 2: MC main effect F=45.463, p<0.001; MC×Gender F=5.105, p=0.008; MC×Gender×AR design p=0.017 with effect size reported as 0.568 (moderate). AR design alone (p=0.934) and Gender alone (p=0.453) were not significant. Group means (Table 3) for LR: Control 26.78; G1 (Male, High, Marker) 35.70 (highest); G5 (Male, High, Image) 34.667; G6 (Female, High, Image) 33.000; G2 (Female, High, Marker) 32.600; G3 (Male, Low, Marker) 32.200; G7 (Male, Low, Image) 31.700; G4 (Female, Low, Marker) 31.300; G8 (Female, Low, Image) 29.909 (lowest). Overall, AR groups outperformed control on LR.
• Critical thinking (CT): Mental capacity significantly affected CT; AR design, gender, and interactions did not. From Table 4: MC F=29.174, p<0.001, effect size 0.936 (high); Gender p=0.170; AR design p=0.767; interactions ns. Post hoc (Table 5): High MC mean=69.744 vs Low MC mean=62.939. Thus, CT gains favored high MC students irrespective of AR type or gender.
• Science learning outcomes (STEAM context): From Table 6: MC F=48.084, p<0.001, effect size 0.998 (high); AR design p=0.738; Gender p=0.907; interactions ns. Authors report differences across gender favoring males on overall learning outcomes (Table 7 means: Males 25.074; Females 24.992), and note AR’s alignment with spatial visualization may advantage males. Broadly, AR improved learning motivation and STEAM skills.
• Overall: AR-based instruction improved learning retention and supported CT development; mental capacity was a key moderator, with high-MC students showing stronger CT outcomes. The specific AR design (image vs. marker) showed no differential effect.

Findings indicate AR-enhanced lessons improved learning retention compared with traditional methods, likely due to multimodal, interactive representations that concretize abstract science concepts, increase attention and engagement, and facilitate problem-based, student-centered learning. The highest LR occurred in male high-MC students using marker-based AR (G1), with the lowest in female low-MC image-based AR (G8), highlighting the moderating role of mental capacity and some subgroup differences. For CT, mental capacity was the primary determinant; AR type and gender did not yield significant differences, suggesting that both image- and marker-based AR implementations can effectively support CT when coupled with engaging, constructivist activities. The authors attribute benefits to AR’s ability to visualize spatial relationships and provide immediate, context-rich feedback, consistent with constructivist and sociocultural learning theories and prior literature. While overall gender main effects were not significant in CT, the authors note male-favoring differences in overall learning outcomes within the STEAM context, potentially due to spatial visualization demands. The results reinforce that AR, when embedded in learner-centered pedagogy, can deepen understanding, promote long-term retention, and support higher-order thinking.

Integrating AR into STEAM-oriented science education enhances engagement, motivation, participation, and learning retention, and supports the development of CT and other 21st-century skills. AR enables students to observe and interact with digitized objects in authentic contexts, aligning with constructivist and sociocultural frameworks and encouraging a shift from lecture-based to active, AR-supported learning. The study recommends incorporating AR-based active learning into future curricula to promote conceptual understanding over memorization and to facilitate transdisciplinary learning central to STEAM. Future research should evaluate AR’s educational benefits at larger scales, optimize best practices for curriculum integration, and investigate long-term effects, costs, and implementation guidelines, including the design of effective AR activities that expand students’ CT.

• Small, convenience sample limited to private schools in Ha'il; findings may not generalize to broader or public-school populations.
• Restricted to second preparatory grade and specific science content; short intervention duration (~3 weeks).
• Potential selection bias (all students had smartphones) and novelty effects.
• Limited investigation of diverse learner needs; underserved, at-risk, special education, and low-income groups were not specifically studied.
• Practical constraints and technical challenges (device capabilities, lighting, connectivity) may affect scalability and fidelity.