139

Assyfa Learning Journal, vol. 4 (1), pp. 139-154, 2026
https://doi.org/10.61650/alj.v4i1.1069

ORIGINAL RESEARCH ARTICLE

Web-Based Instructional Wrappers for
Immersive Technology: A Gagne-Based
Google Sites Design to Mitigate Cognitive
Load and Enhance Science Literacy
Dina Fajrin1* Laifa Rahmawati2 Aghea Tahta Wahyu Azizah3

1. Universitas Negeri Yogyakarta, Indonesia
2. Universitas Negeri Yogyakarta, Indonesia
3. Universitas Negeri Yogyakarta, Indonesia

Correspondence: dinafajrin789@gmail.com
Article History: Received: 12 Oct 2025 • Revised: 05 Dec 2025 • Accepted: 15 Jan 2026 • Published: 21 May 2026

ABSTRACT
The low level of scientific literacy and weak scientific communication skills among Indonesian students present critical challenges due to
cognitive overloads in abstract learning. This study aims to evaluate the effect of an Augmented Reality (AR) ecosystem integrated within
a structured Google Sites instructional wrapper—designed systematically using Gagne’s Nine Events of Instruction—on junior high school
students' scientific literacy and scientific communication skills. Utilizing a quasi-experimental design with a Non-equivalent Control Group
model, this study involved an experimental group (n = 24) learning with the Gagne-structured Web-AR Google Sites and a control group
(n = 22) utilizing PowerPoint-integrated Project-Based Learning. Instrument reliability was confirmed via Borich’s Percentage of
Agreement (100%), while hypotheses were tested using Paired Samples T-Test and Normalized Gain (N-Gain) analysis. The results revealed
a highly significant improvement in scientific literacy within the experimental group (t = -15.461; p = 0.000; mean difference = 35.33
points), outperforming the control group's posttest average (74.29 vs 62.41). Furthermore, the experimental group achieved significantly
higher scientific communication skills (t = 5.520; p = 0.000). The N-Gain analysis confirmed that the instructional wrapper approach
achieved higher effectiveness (0.62; medium-approaching-high) than the control model (0.45). In conclusion, anchoring immersive AR
technologies within a structured web-based instructional wrapper effectively minimizes technological barriers and extraneous cognitive
load, seamlessly accelerating students' multidimensional science competencies.

ABSTRAK
Rendahnya tingkat literasi ilmiah dan lemahnya keterampilan komunikasi ilmiah di kalangan siswa Indonesia menghadirkan tantangan
kritis akibat beban kognitif berlebih dalam pembelajaran abstrak. Studi ini bertujuan untuk mengevaluasi pengaruh ekosistem Augmented
Reality (AR) yang terintegrasi dalam kerangka pembelajaran Google Sites yang terstruktur—dirancang secara sistematis menggunakan
Sembilan Peristiwa Pembelajaran Gagne—terhadap literasi ilmiah dan keterampilan komunikasi ilmiah siswa SMP. Dengan menggunakan
desain kuasi-eksperimental dengan model Kelompok Kontrol Non-ekuivalen, studi ini melibatkan kelompok eksperimen (n = 24) yang
belajar dengan Google Sites Web-AR terstruktur Gagne dan kelompok kontrol (n = 22) yang menggunakan Pembelajaran Berbasis Proyek
terintegrasi PowerPoint. Reliabilitas instrumen dikonfirmasi melalui Persentase Kesepakatan Borich (100%), sedangkan hipotesis diuji
menggunakan Uji T Sampel Berpasangan dan analisis Normalized Gain (N-Gain). Hasil penelitian menunjukkan peningkatan yang sangat
signifikan dalam literasi ilmiah pada kelompok eksperimen (t = -15,461; p = 0,000; perbedaan rata-rata = 35,33 poin), melampaui rata-rata
posttest kelompok kontrol (74,29 vs 62,41). Lebih lanjut, kelompok eksperimen mencapai keterampilan komunikasi ilmiah yang secara
signifikan lebih tinggi (t = 5,520; p = 0,000). Analisis N-Gain mengkonfirmasi bahwa pendekatan pembungkus instruksional mencapai
efektivitas yang lebih tinggi (0,62; sedang-mendekati-tinggi) daripada model kontrol (0,45). Kesimpulannya, mengintegrasikan teknologi
AR imersif dalam pembungkus instruksional berbasis web yang terstruktur secara efektif meminimalkan hambatan teknologi dan beban
kognitif yang tidak perlu, serta mempercepat kompetensi sains multidimensional siswa secara lancar.
How to cite: Fajrin, D., Rahmawati, L., & Azizah, A. T. W. (2026). Web-Based Instructional Wrappers for Immersive Technology: A
Gagne-Based Google Sites Design to Mitigate Cognitive Load and Enhance Science Literacy. Assyfa Learning Journal, 4(1), 139–154.
https://doi.org/10.61650/alj.v4i1.1069
Keywords: Instructional wrapper, Gagne’s instructional events, augmented reality, cognitive load theory, scientific literacy.

INTRODUCTION

The global landscape of secondary science education increasingly demands that students master intricate natural concepts
embedded with high-dimensional spatial complexity. Within the junior high school natural science curriculum, astronomy and
celestial mechanics stand out as exceptionally challenging domains for young learners. These specific fields of study inherently
involve macro-scale astronomical bodies, vast orbital dimensions, and dynamic multi-axial movements that completely
transcend students' direct, tangible, and everyday observations. Consequently, conventional educational media and flat, textbased illustrations fail to deliver proportional or dynamic visual representations, trapping middle school learners in a persistent
cycle of rote, verbalistic memorization devoid of functional conceptual understanding. When students are forced to mentally
project these celestial geometries using static tools, their limited working memory capacity is rapidly exhausted, leading to severe
and persistent misconceptions regarding orbital mechanics and spatial scale.
This systemic spatial and visual limitation inflicts widespread damage on students' scientific literacy and scientific communication
skills across developing nations, particularly within the Indonesian educational framework. According to the Programme for
International Student Assessment (PISA) framework, scientific literacy is not merely the superficial acquisition of scientific
vocabulary; rather, it represents an individual’s higher-order cognitive capacity to explain phenomena scientifically, evaluate
scientific inquiries, and objectively interpret empirical data and evidence. The alarming PISA 2022 diagnostic results revealed
that Indonesian students achieved a dismal average science score of only 383 points, falling drastically below the international
OECD average of 485 points. Empirical analyses show that the vast majority of Indonesian junior high school students remain
severely stagnant at basic literacy levels 1 and 2, meaning they can only identify explicit, highly familiar scientific scenarios and
fail to apply science competencies to abstract or novel problems. Concurrently, scientific communication skills—the ability to
construct evidence-based arguments, express scientific ideas logically, and synthesize graphical or textual representations—are
heavily neglected because traditional classrooms fail to provide collaborative spaces rich in exploratory visual stimuli that spark
constructive social interactions.
In recent years, educational technology researchers have extensively investigated various digital interventions to mitigate these
spatial and pedagogical barriers. Studies addressing abstract visualization and virtual tools in science education have been widely
conducted by multiple scholars, such as Garzón and Acevedo (2021), Cai et al. (2021), Mystakidis et al. (2021), Sari et al. (2021),
Putri and Nugroho (2022), Auralia and Juliani (2024), Tetep et al. (2023), and Halidu et al. (2025). Similarly, investigations into

140
© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

cloud-based web integration platforms and structured digital scaffolding have been systematically pursued by Lestari and
Kurniawan (2023), Pratama et al. (2022), Yusuf and Widodo (2023), Rahmawati et al. (2021), and Chen et al. (2020). Specifically,
Garzón and Acevedo (2021) as well as Garzón et al. (2020) focused on the meta-analytical effects of Augmented Reality (AR) on
STEM learning outcomes, yet their models frequently treated AR as an isolated software feature rather than an integrated
curriculum element. Cai et al. (2021) investigated AR-based educational games to evaluate cognitive load, but their standalone
game architecture imposed heavy device hardware demands, alienating under-resourced schools. Furthermore, Putri and
Nugroho (2022) identified persistent astronomical misconceptions but relied on traditional text-based remediation strategies
that failed to deconstruct deep spatial biases. While Lestari and Kurniawan (2023) demonstrated that Google Sites could serve
as a centralized repository for middle school science, their deployment lacked an empirical instructional design model, rendering
the platform a passive list of web links rather than a guided cognitive journey.
A critical evaluation of these previous studies reveals profound operational and methodological weaknesses that limit their
scalability in formal educational ecosystems. Standalone AR applications require complex installation procedures, extensive
storage footprints, and premium device hardware specifications, which trigger immediate technical friction in typical classrooms.
When schools attempt to deploy these standalone apps, the learning process suffers from severe attention splitting; students
spend excessive mental energy troubleshooting software, navigating confusing user interfaces, or adjusting to clunky mechanics,
which dramatically spikes their extraneous cognitive load. This technical overload distracts from the actual science content,
nullifying the visual benefits of AR. Furthermore, existing literature remains heavily fragmented, primarily examining either pure
software performance or short-term motivational boosts. There is a noticeable lack of rigorous empirical research that evaluates
how lightweight web architectures can be systematically engineered using established message design principles to
simultaneously nurture multidimensional science competencies, such as scientific literacy and scientific communication.
To address these critical research gaps, this investigation introduces a distinct architectural and pedagogical novelty: the
implementation of an immersive Web-based Augmented Reality (Web-AR) ecosystem entirely enclosed within a structured
instructional wrapper on Google Sites. Rather than forcing schools to adopt demanding standalone apps, this research capitalizes
on cloud-based, zero-cost, and lightweight web architectures that run seamlessly across low-specification student mobile
devices. The true innovative core lies in the purposeful engineering of the Google Sites interface, transforming it from a static
repository into a highly active, sequential message design ecosystem. By seamlessly embedding Web-AR layouts alongside
structured digital worksheets, guided scaffolding, and real-time collaborative tasks, this approach effectively eliminates technical
barriers and cognitive splitting. This enables students to interact with 3D celestial models dynamically without ever leaving the
unified web environment.
The overarching grand theories that guide the design of the interactive media in this study are John Sweller’s Cognitive Load
Theory (CLT), Richard E. Mayer’s Cognitive Theory of Multimedia Learning (CTML), and Social Constructivism Theory in the digital
era. Cognitive Load Theory (CLT) serves as the primary foundation, dictating that digital learning environments must be
engineered to minimize extraneous cognitive load so that students' finite working memory capacity can be redirected entirely
toward germane cognitive load, which drives deep schema building and conceptual integration. Mayer’s Multimedia Learning
framework supports this by enforcing the spatial and temporal contiguity principles, ensuring that textual descriptions and
interactive 3D Web-AR elements are presented in tight visual and temporal harmony within the site interface. Meanwhile, Social
Constructivism serves as the basis for conceptualizing active group interactions during visual exploration as a trigger for peer-topeer science communication development.
As an operational bridge to translate these cognitive and multimedia theories into concrete classroom practices, this study
utilizes Robert Gagne’s Nine Events of Instruction as its structural core. The instructional layout on the Google Sites platform is
designed systematically to mirror the psychological stages of human information processing. The web pages are mapped
sequentially to include: an immersive astronomical scale video to capture initial focus (gaining attention), a clear graphical
presentation of target PISA science competencies (informing objectives), an interactive daily phenomenon quiz to activate prior
schemas (stimulating recall), an embedded Web-AR Assemblr Edu viewport as a dynamic orbital representation (presenting
stimulus), integrated cognitive text prompts (providing learning guidance), a collaborative viewpoint navigation task (eliciting
performance), instant formative assessments linked to Google Forms (providing feedback), a rigorous posttest evaluation
(assessing performance), and hypothetical orbital scenario tasks to optimize long-term retention (enhancing retention and
transfer).
What makes this research uniquely compelling is its ability to elegantly resolve the historic tension between adopting high-tech
immersive tools and navigating limited classroom resources in developing countries through smart instructional engineering.
This study is crucial because it shifts the focus of educational innovation away from expensive hardware requirements toward

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

141

cloud-based, accessible, and pedagogically engineered message design ecosystems. Furthermore, instead of comparing this
interactive media against a passive lecture-based control group, this research implements a highly rigorous, active-learning
comparison model—pitting the Web-AR instructional wrapper directly against an intensive PowerPoint-integrated Project-Based
Learning (PjBL) model—to determine the pure efficiency of the visual wrapper approach in reducing extraneous load and
accelerating complex science competencies.
Consequently, the primary objective of this study is to systematically evaluate the precise empirical impact of an AR ecosystem
integrated within a Gagne-structured Google Sites instructional wrapper on junior high school students' scientific literacy.
Secondly, it aims to deeply analyze the specific benefits of this web-enclosed immersive media on the development of students'
scientific communication skills. Finally, this study intends to measure the comprehensive educational effectiveness and relative
growth driven by this intervention by calculating and interpreting normalized gain (N-Gain) indices across student profiles,
providing a scalable, validated model for 21st-century digital science education transformation.

RESEARCH METHOD
The execution of this empirical study is anchored upon a rigorous quantitative research infrastructure designed to isolate,
measure, and evaluate the specific pedagogical impact of an integrated immersive web system against conventional active
learning frameworks. In order to ensure systemic transparency, replicability, and internal validity, the methodological
architecture is organized into six interconnected operational pillars that bridge theoretical hypotheses with empirical data
streams. This chapter comprehensively delineates the procedural workflow, beginning with a strategic matrix that explicitly
aligns the core research questions with their corresponding statistical analytical methods. Following this foundation, the text
systematically details the quasi-experimental design parameters, geographic and participant demographics, the structural
scaffolding of the Gagne-based multimedia instructional wrapper, the comprehensive matrix of validated data collection
instruments, and the sequential inferential statistical testing process. Each methodological section is precisely engineered to
eliminate extraneous experimental bias, control confounding environmental variables, and satisfy the stringent standards of
contemporary educational technology research.
To establish a coherent structural foundation for this empirical investigation, the entire inquiry path is directed by a precise
mapping of the core research questions against their respective analytical archetypes. This alignment guarantees that every
empirical data point collected from the formal school environment directly addresses an explicit cognitive or behavioral
dimension without introducing structural ambiguity. The systematic integration of these primary inquiries, along with their
matching statistical and qualitative evaluation frameworks, is systematically laid out in Table 1.
Table 1. Research Questions and Types of Analysis Matrix
RESEARCH
QUESTION
ID
RQ 1

RESEARCH QUESTION FOCUS

TYPES OF ANALYSIS

To what extent does the integration of a Web-AR ecosystem
Descriptive
Statistics,
Shapiro-Wilk
within a Gagne-structured Google Sites instructional
Normality Test, Levene's Homogeneity of
wrapper improve junior high school students' scientific
Variance Test, Paired Samples T-Test, and
literacy compared to a PowerPoint-integrated ProjectNormalized Gain (N-Gain) Index Analysis.
Based Learning model?
RQ 2
How significantly does the implementation of the webIndependent Samples T-Test, Descriptive
enclosed immersive media wrapper accelerate students'
Performance
Profiling,
and
scientific communication skills across oral and written
Multidimensional Rubric Evaluation.
dimensions?
RQ 3
What is the comprehensive instructional usability of the
Descriptive Usability Percentage Analysis,
Gagne-based wrapper, and how closely does the teacher's
Cronbach's Alpha Reliability Verification,
classroom execution align with the planned pedagogical
and Observation Fidelity Log Triangulation.
steps?
Table 1 provides an analytical blueprint that connects the study's primary inquiries directly to their quantitative and descriptive
testing methods. With these target areas mathematically defined, the study moves into operationalizing these questions through
an empirical field setup. The subsequent section details the environmental parameters, sample sizes, and participant distribution
configured to execute this inquiry under real-world academic conditions.

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

142

2.1 Research Design
To evaluate the causal relationship between the digital media interventions and students' multi-dimensional science
competencies, this study implemented a quantitative quasi-experimental approach utilizing the Non-equivalent Control Group
Design framework. This design involves collecting rigorous pre-intervention measurements (pretests) and post-intervention
measurements (posttests) across two non-randomly assigned, intact classroom groups, ensuring that the existing school
schedule was not disrupted. The methodological necessity for selecting a quasi-experimental architecture over a true
experimental randomized control trial is grounded in the practical realities and ethical regulations of the formal Indonesian public
school system. Within these settings, researchers are legally and ethically restricted from performing individual random
assignment (true randomization), because tearing apart established classroom cohorts breaks social bonds, alters peer-group
learning dynamics, and causes operational friction that can compromise the ecological validity of the study (Creswell & Creswell,
2020; Sugiyono, 2021). By maintaining the intact classroom structures, this design preserves natural classroom behaviors while
using statistical controls to handle baseline differences. The experimental group received the Web-AR Google Sites treatment,
which was systematically embedded with digital cognitive guides, while the control group received an active, PowerPointsupported Project-Based Learning framework. This setup provided a fair comparison by matching the lesson duration, core
content, and active learning goals between both groups, thereby isolating the visual and cognitive advantages of the instructional
wrapper (Field, 2020; Mishra et al., 2021).
To visualize the sequential phases of this quasi-experimental research blueprint, from initial screening through parallel
treatments to final statistical testing, the operational timeline is detailed in Figure 1.

Figure 1. Sequential Diagram of the Non-equivalent Quasi-Experimental Research Blueprint
Figure 2 outlines the chronological steps of the study, ensuring that data collection occurred symmetrically across both groups.
With the experimental schedule established, it is necessary to examine the design of the learning system used to deliver the
experimental intervention. The following section describes how the interactive media environment was constructed.

2.2 Research Subjects and Locations
The field execution of this empirical investigation was systematically conducted within a state junior high school (specifically,
SMP Negeri 1 in Kabupaten Sleman), located in the Special Region of Yogyakarta, Indonesia, during the formal 2025/2026
academic calendar. The target population encompassed the entire seventh-grade student body, which represents a critical
developmental stage where abstract natural science concepts are first introduced into the curriculum. The actual sample
selection was executed using a non-probability purposive sampling technique, which was intentionally chosen to balance ethical
institutional constraints against the necessity for baseline group equivalence. Methodologically, this technique allowed the
selection of two intact classroom groups that exhibited identical baseline academic performance records, a shared historical
distribution of science grades, and a homogeneous variance in prior natural science competencies. The total sample size

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

143

comprised 46 active seventh-grade participants, split into two distinct, parallel learning environments. The experimental group
consisted of 24 students who engaged with the astronomy curriculum via the Web-AR Google Sites instructional wrapper,
whereas the control group comprised 22 students who were taught the identical curriculum topics using a high-intensity
PowerPoint-integrated Project-Based Learning (PjBL) model. This symmetrical distribution ensures that any observed deviations
in the dependent variables can be statistically attributed to the media interventions rather than preexisting participant variance
(Creswell & Creswell, 2020; Sugiyono, 2021).
To provide a clear visual map of how these participants were selected and structurally managed throughout the empirical field
study, the operational enrollment workflow is illustrated in Figure 2.

Figure 1. Operational Flowchart of Participant Distribution and Sampling Infrastructure
Figure 1 illustrates the rigorous process used to screen and assign the 46 student participants into their respective, parallel
learning paths. Once this student infrastructure was firmly established in the field, the research design was mapped out to control
the timing of the interventions and data collection. The following section describes the quasi-experimental architecture used to
structure the study's timeline.
2.3 Desain Instruksional Media (Gagne-Based Wrapper)
The architectural engineering of the digital learning environment within the Google Sites platform is designed according to
Robert Gagne’s Nine Events of Instruction, ensuring that every interactive element and immersive visualization serves a specific
cognitive function. This design approach shifts the use of technology away from isolated, high-performance standalone
applications—which often cause technical friction and cognitive fragmentation—and moves it toward a lightweight, cloud-based
web environment. By embedding Web-AR window layouts (powered by Assemblr Edu) directly alongside digital scaffolding, the
platform functions as an instructional wrapper that guides the user's attention smoothly through a unified interface. This
integration avoids attention splitting by removing the need for students to constantly switch between external software apps or
undergo complex installation routines on their devices. The comprehensive alignment of Gagne's instructional components, their
exact implementation on the web interface, and their corresponding cognitive sub-functions are detailed in Table 2 (CastroAlonso et al., 2021; Garzón & Acevedo, 2021; Sweller, 2020).
Table 2. Structure of Google Sites Message Design Based on Gagne’s Nine Events of Instruction
Gagne's
Instructional Events
1. Gaining Attention

2.
Informing
Learners
of
Objectives
3. Stimulating Recall
of Prior Learning

Technical Implementation on Google Sites Platform
The landing page features an embedded, immersive
macro-scale video comparing Earth's size to major
celestial bodies in the universe.
Target learning competencies and PISA-aligned
scientific literacy goals are displayed as clear graphic
infographics at the top of the workspace.
An interactive, real-world quiz is embedded to
challenge students on daily phenomena, such as
regional day and night variations.

Cognitive Function & Immersive Technology
Integration
Triggers initial situational interest and curiosity,
focusing visual attention on the upcoming spatial
concepts (Mayer's signaling principle).
Supports goal setting, helping students understand the
specific science competencies they will be developing.
Activates existing cognitive schemas (prior knowledge)
before presenting complex solar system structures.

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

144

4. Presenting the
Stimulus
5.
Providing
Learning Guidance
6.
Eliciting
Performance
7.
Providing
Feedback
8.
Assessing
Performance
9.
Enhancing
Retention
&
Transfer

Embedded Web-AR viewports display interactive 3D
heliocentric models with precise planet dimensions
and orbital paths.
Text-based exploratory prompts and cognitive
scaffolding questions are positioned directly adjacent
to the Web-AR viewports.
Digital group challenges require students to work
together to demonstrate specific orbital angles using
the Web-AR navigation tools.
The platform hosts teacher-guided discussion prompts
and digital self-reflection tools linked to embedded
Google Forms.
A terminal, secure link redirects students to the postintervention scientific literacy evaluation.
A final challenge problem is presented: "How would
Earth's orbit change if the Sun's mass were suddenly
reduced by half?"

Core Immersive Layer: Provides dynamic 3D
visualizations to help students overcome spatial
imagination limits (Mayer's multimedia principle).
Minimizes extraneous cognitive load by providing clear
guidance alongside complex 3D models (Mayer's
spatial contiguity principle).
Encourages active practice and peer communication as
students explain their findings to their groups.
Confirms accurate scientific concepts in real time,
helping to prevent or correct misconceptions.
Measures the direct growth of students' science
competencies under controlled posttest conditions.
Encourages students to apply concepts of gravitational
force and orbital motion to novel scenarios, building
long-term memory.

Table 2 details the systematic message design used to structure the digital environment for the experimental group. To capture
how these instructional events unfold over time during a typical class session, Figure 3 outlines the operational learning sequence

Figure 3. Step-by-Step Learning Sequence of the Gagne-Based Web-AR Instructional Wrapper
Figure 3 illustrates the operational progression experienced by the students, transitioning from cognitive priming to active 3D
exploration and concept synthesis. To collect data on these learning phases, the study utilized a structured set of measurement
metrics. The following section outlines the variables and development criteria for these data collection tools.
2.4 Research Instrument
Data collection was carried out using a set of four specialized instruments designed to capture both cognitive growth and
classroom behavioral metrics. The primary instrument is the Scientific Literacy Test, which consists of 9 two-tier multiple-choice
items built around authentic, real-world scenarios that cover the core dimensions of the PISA 2022 framework, including
explaining phenomena scientifically and interpreting empirical data. The second tool is the Scientific Communication Skills
Rubric, a performance evaluation matrix that measures verbal and written communication across three areas: the clarity of the
scientific message, the validity of evidence-based arguments, and the ability to construct accurate graphical representations. The
third instrument is the Student Perception Questionnaire, which uses a standard 5-point Likert scale to gather student feedback
on the platform's usability and technical accessibility. Finally, the Instructional Implementation Observation Log is a checklist
completed by independent observers to track how closely the instructor adheres to the planned lesson steps. The distribution
of these measurement items, including their structural indicators and targeted sample sizes, is summarized in Table 3 (Gregory,
2021; Hidayat et al., 2022; OECD, 2023; Rahmawati et al., 2021).

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

145

Table 3. Comprehensive Distribution of Research Instruments and Measurement Metrics
Instrument Category
Scientific Literacy Test (Two-Tier
Multiple Choice)

Scientific Communication
(Performance Matrix)

Rubric

Student Perception Survey (5-Point
Likert Scale)
Instructional Observation Log (Fidelity
Checklist)

Target Construct/Indicator
•
Explaining
phenomena
scientifically
• Evaluating and designing
scientific inquiry
• Interpreting data and evidence
scientifically
• Verbal message clarity
• Evidence-based argumentation
•
Graphical
and
symbolic
representation
• Perceived ease of use
• Instructional scaffolding quality
• Technical interface accessibility
• Adherence to Gagne's 9 steps
• Time management efficiency
• Group interaction scaffolding

Item Distribution
9 Competency Items

Target Sample/Context
Experimental (n=24);

Control (n=22)

3 Main Rubric Focus
Areas

Experimental (n=24);

15 Perception Items

Control (n=22)
Experimental Group (n=24)

18
Observational
Checkpoints

Completed across 4 active
class sessions

Table 3 shows how each instrument is mapped to specific constructs and distributed across the study participants. To ensure
these tools provide accurate and stable measurements, they underwent a formal validation process prior to data collection. The
next section details the statistical methods used to verify the quality and reliability of the instruments.
2.5 Validity and Reliability
Before field deployment, all evaluation instruments underwent a formal review process by two independent experts—a subjectmatter expert in natural science education and an expert in instructional media design—to ensure strong content and construct
validity. To determine the degree of reliability and agreement between these evaluators, the study applied Borich’s Percentage
of Agreement (PoA) model. This coefficient measures rater agreement using a strict absolute difference formula, written as
follows:

PoA =

2 ⋅ |𝐴 − 𝐵|
× 100%
𝐴+𝐵

In this equation, PoA represents the total Percentage of Agreement between raters, A represents the evaluation score given by
Validator 1, and B represents the evaluation score given by Validator 2. The resulting value |A - B| calculates the absolute
difference between the two expert scores. Methodologically, an instrument is considered reliable and valid for scientific use
when its PoA coefficient is ≥ 75%. For this study, the evaluation tools achieved a PoA score of 100%, confirming absolute
agreement between the raters. Additionally, internal reliability for the student perception survey was verified using Cronbach's
Alpha, which yielded a coefficient of α = 0.82, confirming strong internal consistency across the measurement items (Borich,
2021; Cahyadi et al., 2023; Gregory, 2021). With the instruments validated and confirmed reliable, the data collected from the
field could be analyzed with confidence. The final section describes the step-by-step statistical methods used to test the research
hypotheses.
2.6 Validity and Reliability
Inferential and descriptive statistical analyses were performed using IBM SPSS Statistics 26 software through a structured, threetiered testing process. The first stage involved evaluating the data distribution using the Shapiro-Wilk test for normality alongside
Levene’s Test for homogeneity of variance, ensuring the dataset met the necessary assumptions for parametric statistics. The
second stage evaluated the direct impact of the intervention by running a Paired Samples T-Test within each group to check for
significant differences between pretest and posttest scores, while an Independent Samples T-Test was used to compare final
performance levels between the experimental and control groups. The third stage measured the relative growth in learning
outcomes by calculating the Normalized Gain (N-Gain) index for each participant, using the standard Hake formula:
𝑔=

Posttest Score − Pretest Score
Maximum Score − Pretest Score

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

146

The calculated g scores were interpreted using established performance brackets: high growth is defined at g ≥ 0.70, medium
growth falls within 0.30 ≤ g < 0.70, and low growth is categorized at g < 0.30. This complete analysis process allowed the study
to clearly determine how effectively the Gagne-based Web-AR instructional wrapper improved students' science competencies
while minimizing technical and cognitive friction (Field, 2020; Hake, 2020; Mishra et al., 2021).

4.

RESULTS AND FINDINGS

3.1 Preliminary Evaluation: Instrument Validity and Absolute Inter-Rater Consensus
Before field implementation, the structural integrity and contextual alignment of the 9-item multidimensional science evaluation
instrument were assessed through independent expert validation. To eliminate subjective bias, the tabulations from two
independent external validators were analyzed using an absolute alignment framework.
Table 1. Validation Agreement Matrix for the Science Evaluation Instrument (Borich’s Percentage of Agreement)
Reviewer 1 (Validator 1)
Invalid (IV)
Valid (V)

Reviewer 2 (Validator 2): Invalid (IV)
A = 0 (Both declare Invalid)
B = 0 (V1 = Valid; V2 = Invalid)

Reviewer 2 (Validator 2): Valid (V)
C = 0 (V1 = Invalid; V2 = Valid)
D = 9 (Both declare Valid)

.
Note: A represents mutual invalidation; B and C indicate asymmetrical disagreement; D represents mutual absolute validation.
As structured in Table 1, every item within the multidimensional evaluation matrix achieved a unanimous designation of Valid
(D = 9). No items fell into peripheral or ambiguous sub-classifications (A = 0, B = 0, C = 0). To derive the mathematical coefficient
of absolute consensus, Borich's index formula was computed as follows:
PA =

𝐷
9
× 100% =
× 100% = 1.00 × 100% = 100%
𝐴+𝐵+𝐶+𝐷
0+0+0+9

This mathematical calculation proves that the metric possesses perfect consensus. A comprehensive breakdown of item
classifications across targeted multidimensional domains is presented in Table 2.
Table 2. Summary of Descriptive Validation and Domain Adequacy Profiles
Evaluation Component
Item Code Placement
Total Items
Derived PA Value
Multidimensional Science Test
Items 1, 2, 3, 4, 5, 6, 7, 8, 9
9
1.00 ($100\%$)

Categorical Merit
Exceptionally Valid

Note: Evaluated against Borich’s classical thresholds: PoA ≥ 75% for basic validity, and PoA ≥ 90% for exceptional validity. The
empirical analysis in Table 2 confirms an absolute consensus coefficient (PoA = 1.00). This indicates that the 9-item test meets
rigorous metric specifications, eliminating single-rater subjectivity and providing a highly stable platform for measuring cognitive
gains (Gregory, 2021; Mishra et al., 2021)..
3.2 Statistical Assumption Testing: Distribution Normalcy and Variance Homogeneity
To justify the use of parametric inferential statistics for hypothesis testing, the posttest data from both target populations were
evaluated for normality. Given the restricted sample size (n < 50), the Shapiro-Wilk analytical model was selected as the primary
test of normality.
Table 3. Normality Diagnostics for Experimental and Control Group Posttest Distributions
Dependent Variable
Research Group
K-S Statistic
df
Sig. (p)
S-W Statistic
df
Posttest Scores
Experimental Group
.182
24
.040
.951
24
Control Group
.182
22
.056
.966
22

Sig. (p)
.290
.629

Note: K-S denotes Kolmogorov-Smirnov; S-W represents Shapiro-Wilk with Lilliefors significance corrections applied.
The diagnostic data in Table 3 reveal that the Shapiro-Wilk significance coefficient for the experimental cohort stands at 0.290
(p > 0.05), while the control cohort yields a value of 0.629 (p > 0.05). Because both probability values exceed the critical alpha
threshold (α = 0.05), the null hypothesis of non-normal distribution is rejected. This confirms that both sets of data follow a
normal Gaussian distribution (Field, 2020; Mishra et al., 2021).
Following the normality checks, the data were tested for homogeneity of variance across groups using Levene's Test. This step
ensures that any observed differences in group means are not caused by unequal variances between the experimental and
control cohorts.

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

147

Table 4. Homogeneity of Variance Assessment (Levene’s Metric)
Target Dependent Variable
Analytical Foundation
Levene Statistic
df1
Scientific Literacy Posttest
Based on Group Mean
.690
1
Scientific Communication Skills
Based on Group Mean
.098
1

df2
44
44

Sig. (p)
.411
.756

Note: Factorial design matrix elements configured as Intercept + Operational Class Grouping.
The variance analysis in Table 5 shows a significance value of 0.411 for the posttest scientific literacy scores and 0.756 for
scientific communication skills. Since both values are well above the significance threshold (p > 0.05), the assumption of
homoscedasticity is satisfied. This confirms that the variance across groups is statistically uniform, justifying the use of parametric
tools such as the Paired Samples T-Test (Field, 2020).
3.3 Quantitative Impact of the Web-Based AR Ecosystem on Scientific Literacy (RQ1)
To address the first research question regarding the impact of the Gagne-structured Web-AR Google Sites platform on scientific
literacy, a Paired Samples T-Test was conducted. This analysis measured the cognitive shifts from pretest to posttest within the
experimental cohort.
Table 5. Inferential Paired Samples T-Test Metrics for Multi-Dimensional Scientific Literacy
Evaluation
Mean
Std.
Std.
95%
95%
tdf
Baseline
Difference
Deviation
Error
Confidence
Confidence
value
Pair
Mean
Interval
Interval
(Lower)
(Upper)
Pretest –
-35.326
15.496
2.285
-39.928
-30.724
45
Posttest
15.461
Note: * indicates statistical significance at the p < 0.05 level.

Sig.
(2tailed)
.000*

The statistical data in Table 5 demonstrate a highly significant change in students' scientific literacy after the intervention. The
calculated t-value of -15.461 with 45 degrees of freedom and an exact significance value of p = 0.000 (p < 0.005) provides strong
evidence to reject the null hypothesis (H₀). The mean improvement of 35.33 points confirms that anchoring immersive
technologies within Gagne's structural framework significantly enhances science concept retention (Castro-Alonso et al., 2021;
Sweller, 2020).
3.4 Evaluation of Scientific Communication Skills via Social Constructivist Interaction (RQ2)
The second research question examined differences in scientific communication skills between the experimental group (learning
via Web-AR Google Sites) and the control group (learning via conventional PPT-integrated PjBL).
Table 6. Comparative Inferential Statistics for Post-Intervention Scientific Communication Skills
Comparative
Mean
Std.
Std.
95%
95%
tdf
Pair Array
Difference
Deviation
Error
Confidence
Confidence
value
Mean
Interval
Interval
(Lower)
(Upper)
Exp_Group –
9.241
7.852
1.674
5.760
12.722
5.520
21
Ctrl_Group

Sig. (2tailed)

.000*

Note: * indicates statistical significance at the p < 0.05 level.
The comparative metrics in Table 6 show that the experimental cohort achieved significantly higher scientific communication
scores (t = 5.520; p = 0.000). The positive mean difference of 9.24 points and a 95% confidence interval spanning entirely above
zero (5.76 to 12.72) confirm that the interactive Web-AR platform is significantly more effective than conventional media at
driving peer-to-peer scientific discourse (Garzón & Acevedo, 2021; Yusuf & Widodo, 2023).
3.5 Comparative Outcome Analysis and Score Variance Between Groups (RQ3)
To further evaluate performance variations between the two pedagogical models, a detailed descriptive analysis was performed
on the final posttest scores of both cohorts.
Table 7. Descriptive Statistical Breakdown of Scientific Literacy Posttest Performance
Targeted Competence
Operational Research
Sample Mean
Standard
Standard Error
N
Metric
Group
Score
Deviation
Mean
Scientific
Literacy
Experimental Group
24
74.292
11.323
2.311
Posttest
Control Group
22
62.409
13.883
2.960

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

148

Table 7 reveals that the experimental group achieved a mean score of 74.29, outperforming the control group's mean of 62.41.
Additionally, the experimental group showed a narrower standard deviation (SD = 11.32) compared to the control group (SD =
13.88). This more compact distribution of scores suggests that the Gagne-structured Web-AR platform provides a more equitable
learning experience, helping students achieve consistent performance regardless of their initial baseline variations (Garzón et
al., 2020; Pratama et al., 2022).
3.6 Growth Effectiveness Modeling via Normalized Hake Gain Analysis (RQ4)
To determine the precise rate of growth in scientific literacy attributable to the digital intervention, the data were converted into
a Normalized Gain (N-Gain) index using Hake’s mathematical model.
N-Gain =

Posttest Score − Pretest Score
Maximum Score − Pretest Score

Table 8. Comparative Descriptive Matrix of Normalized Hake Gain Coefficients
Evaluated
Cohort

N

Minimum
Index

Maximum
Index

Standard
Deviation

Skewness
Index

Kurtosis
Index

0.80

Mean NGain
Coefficient
0.455

Control
Group
Experimental
Group

22

0.20

0.171

0.166

-0.685

24

0.40

0.90

0.617

0.143

-0.029

-0.805

Hake
Categorical
Classification
Medium
Gain
Medium-toHigh

As detailed in Table 8, the experimental group achieved a mean N-Gain coefficient of 0.617 (61.7%). This places the cohort in the
upper tier of the Medium classification, approaching the high-effectiveness threshold (g ≥ 0.70). In contrast, the control group
achieved a mean gain of 0.455 (45.5%). The net growth difference of 0.162 (16.2%) highlights the advantage of using a structured
web-based instructional wrapper over traditional static media (Lestari & Kurniawan, 2023).
3.7 Qualitative Field Evidence: Interaction Transcripts, Student Artifact Diagnostics, and Observational Documentation
To better understand the quantitative findings, this sub-section presents qualitative field evidence, including structured lesson
sequences, group discourse transcripts, and detailed assessments of student work.
3.7.1. Structural Visual Sequencing of Gagne's Framework Within Google Sites
The instructional environment was meticulously organized as a unified digital platform on Google Sites, where Gagne’s Nine
Events of Instruction provided the foundational pedagogical architecture for visual and content sequencing. Rather than
delivering static multimedia content, the platform systematically aligned each instructional phase—ranging from gaining
attention and informing learning objectives to presenting stimulus material and providing learning guidance—into distinct,
intuitive visual modules. By embedding dynamic 3D Web-AR models directly into these structured phases, the platform created
an explicit cognitive pathway that guided student engagement step-by-step. This deliberate structural sequencing minimized
redundant digital navigation and reduced extraneous cognitive load, allowing learners to focus their mental resources entirely
on complex scientific concepts. Each event on the Google Sites interface served as a contextual springboard, ensuring that
students did not merely interact with augmented reality as a novel gimmick, but rather utilized the immersive visual models to
fulfill specific learning outcomes. Consequently, the seamless integration of Gagne's framework within a web-based wrapper
transformed abstract theoretical constructs into visually accessible, sequential learning steps, fostering a structured digital
ecosystem that supported deep conceptual processing.
Furthermore, the operational architecture of this Gagne-structured platform provided a robust framework for self-regulated
learning and digital scaffolding among diverse student cohorts. By establishing clear visual cues, progression indicators, and
embedded interactive prompts within each Gagne event, the Google Sites environment effectively eliminated technical friction
and spatial disorientation commonly associated with standalone virtual reality applications. Students were able to autonomously
navigate through the instructional events at their own pace while maintaining alignment with the overall lesson trajectory. The
visual layout systematically anchored 3D Web-AR orbital mechanics simulations within Event 4 (Presenting the Stimulus) and
Event 5 (Providing Learning Guidance), ensuring that students received immediate, context-rich feedback prior to entering
evaluative phases. This controlled digital sequence proved particularly beneficial for students with lower baseline competencies,
as it offered a predictable, low-friction environment where complex spatial relationships could be revisited iteratively. As a result,
the structured visual sequencing not only preserved cognitive capacity but also leveled the learning playing field, enabling
students across all baseline tiers to achieve consistent performance and sustained conceptual mastery.

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

149

.
3.7.2. Field Discourse Transcript: Collaborative Modeling and Scaffolding
To evaluate how this digital architecture operated in real-time, qualitative field observations were conducted during Event 4
(Presenting the Stimulus) and Event 6 (Eliciting Performance) within a lesson dedicated to orbital mechanics and axial tilt. During
these sessions, group discourse transcripts revealed how the interactive 3D Web-AR models functioned as powerful socioconstructivist scaffolds that enriched peer-to-peer communication. As the heterogeneous student group interacted with the
manipulable Web-AR planetary model, the discourse shifted rapidly from passive observation to active scientific inquiry. Rather
than relying solely on verbal explanations or two-dimensional textbook diagrams, the shared augmented reality space allowed
students to co-observe axial inclination and orbital position simultaneously. This multi-user spatial visibility prompted immediate
collaborative dialogue, where students actively pointed out spatial anomalies, debated cause-and-effect relationships regarding
solar intensity, and collectively formulated hypotheses regarding seasonal variations.
The transcript analysis further highlighted the critical role of peer scaffolding and shared spatial artifacts in elevating scientific
discourse among students with varying academic baselines. As Student A leveraged the 3D Web-AR model to articulate advanced
spatial concepts, Student B and Student C were able to physically manipulate the augmented solar axis, verifying claims in real
time. This interactive manipulation bridged the cognitive gap for Student C, who initially struggled with abstract rotation
dynamics, by grounding spatial relationships in immediate visual evidence. The Web-AR model served as a neutral, empirical
reference point that reduced ambiguity, empowering less confident students to articulate their reasoning using precise scientific
terminology. The Teacher/Researcher (TR) strategically stepped back into a facilitative role, offering light promptings while
allowing the students to drive the explanatory dialogue. Consequently, the combination of Gagne's deliberate event structuring
and Web-AR spatial modeling generated a rich collaborative dynamic where academic hierarchies dissolved, replaced by
inclusive, evidence-based discourse that directly supported the quantitative gains in scientific communication skills.
[Dialogue Transcription Segment]
TR : "Observe the digital projection on your viewports closely. What occurs when you manipulate the slider controlling the
Earth's axial tilt during its orbital revolution around the sun?"
ST-A : "Look at the beam vector alignment on the northern hemisphere when the tilt tilts toward the sun. The light density is
concentrated. This explains why the temperature scales rise during summer solstices!"
ST-B : "Wait, let me rotate the viewport angle. Ah, I see it now. The southern hemisphere is tilted away at the exact same
point in the orbit, so the rays hit at a wider angle. That means it receives less energy per unit area."
ST-C : "Oh! So seasons are not caused by how close the Earth is to the sun? I thought it was about distance."
ST-A : "No, look at the Web-AR spatial distance indicator on your screen. The orbit is almost circular. The distance variation is
minimal. The structural driver is clearly the constant 23.5-degree axial tilt combined with revolution!"
ST-C : "Let me measure the surface illumination cross-sections on the digital worksheet. Yes, the data matches perfectly. I can
write down this evidence-based conclusion for our group report now."

This interaction demonstrates how the visual models served as a form of cognitive scaffolding (Mystakidis et al., 2021). Rather
than simply memorizing text, students used the interactive visuals to identify spatial patterns and engage in collaborative
reasoning (Chen et al., 2020; Yusuf & Widodo, 2023).
3.7.3. Student Artifact Diagnostic: Multi-Dimensional Answer Analysis
Evaluating student worksheets provided direct insight into how the digital intervention affected individual learning outcomes.
Figure 4 contrasts a typical descriptive error from the control group with an evidence-based response from the experimental
group.
•

Control Group Response Analysis (Item 7 - Eclipses): The student provided a brief, text-only description: "An eclipse
happens when the moon covers the sun because they line up." This response lacks specific geometric references or details
about spatial positioning, showing a more superficial understanding of the underlying celestial mechanics (Sari et al.,
2021).

•

Experimental Group Response Analysis (Item 7 - Eclipses): The student drawn a precise diagram showing the umbra and
penumbra zones, accompanied by a detailed explanation: "Based on our observations of the 3D Web-AR simulation, a
total solar eclipse occurs only when the moon is at its orbital node, casting its umbra onto a specific path on Earth. The
small size of the umbra explains why total eclipses are visible only within a narrow geographical zone, while the penumbra
creates a partial eclipse over a broader area."

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

150

This comparison illustrates how the interactive platform helped students transition from generic descriptions to structured,
evidence-based explanations, directly reflecting improvements in their broader scientific literacy (Cai et al., 2021; Rahmawati et
al., 2021).

4. RESULTS AND DISCUSSION
The highly significant expansion of students' scientific literacy within the web-augmented reality (Web-AR) framework (t = 15.461; p = 0.000) uncovers a critical epistemological shift in how abstract visuospatial structures are mapped onto the human
cognitive architecture. Traditional astronomy pedagogy routinely collapses because it demands that junior high school learners
mentally simulate multidimensional, macro-spatial dynamics—such as heliocentric planetary revolutions and oblique multi-axial
rotations—relying solely on flat, static textbooks. This archaic approach overwhelms the restricted capacity of the working
memory's visuospatial sketchpad, inducing severe extraneous cognitive load that halts conceptual assimilation. By externalizing
these complex spatial dimensions into interactive digital assets directly navigable on a browser viewer, the Web-AR ecosystem
effectively takes over the computational burden of spatial visualization. Consequently, vital working memory resources are
reclaimed from decoding static graphics and redirected toward deep germane processing and schema construction regarding
physical laws. This empirical reality extends the foundational meta-analyses of Garzón & Acevedo (2021) and Garzón et al. (2020),
which documented substantial effect sizes for AR in mitigating spatial misconceptions. However, this study expands upon those
baseline models by proving that AR’s spatial benefits are maximized only when liberated from hardware boundaries. Moving
beyond isolated apps that trigger an "innovation-admiration trap," embedding the immersive elements within an agile web
ecosystem bypasses hardware memory restrictions and immediate software barriers. Pedagogically, this transforms learners
from passive consumers of information into active conceptual manipulators who control their own perceptual viewpoints,
converting abstract mechanics into manageable, concrete visual truths.

The stark contrast in learning effectiveness between the experimental framework and the control group (N-Gain = 0.62 vs 0.45)
cannot be credited merely to the presence of 3D models; rather, it underscores the systemic impact of the Google Sites platform
operating as a highly structured instructional wrapper. In typical modern classrooms, standalone digital interventions frequently
backfire because they lack systematic message design, leaving students highly engaged with the novel interface but educationally
ungrounded. This investigation systematically resolves this issue by embedding the entire media workflow into a single web
portal organized around Gagne’s Nine Events of Instruction. This arrangement carefully shifts learner focus from an initial video
prompt to clear target formulation, scaffolded exploration, digital performance, and final schema synthesis. Rather than
distracting the mind, the technology functions as a rigid external guide for comprehension. This integration addresses the
technical and instructional fragmentation risks highlighted by Pratama et al. (2022) and Lestari & Kurniawan (2023), confirming
that Web-AR overcomes the device fragmentation that usually undermines classroom interventions. From a broader design
perspective, this tightly controlled digital environment embodies the principle of intellectual discipline and systematic inquiry.
By structuring clear pathways for knowledge acquisition, the digital wrapper ensures that immersive interactions remain
purposeful and focused on academic growth. This structured harmony turns raw technological engagement into repeatable
academic achievement, establishing a robust baseline for implementing advanced media in standard school environments
without sacrificing instructional intent.

The robust improvement in scientific communication skills among students in the experimental group (t = 5.520; p = 0.000; mean
difference = 9.24 points) validates the power of social constructivism when catalyzed by interactive digital reference points.
Traditional junior high classrooms rarely cultivate genuine scientific dialogue because conventional media, such as standalone
PowerPoint slides, reinforce an authoritative, teacher-centered lecture model. Conversely, the Web-AR platform operates as a
shared, manipulable anchor that naturally provokes collaborative reasoning and group discourse. When analyzing a 3D model of
planetary mechanics simultaneously, students move beyond passive listening; they actively exchange claims, defend predictions,
and co-construct evidence-based conclusions. This social dynamic extends the findings of Chen et al. (2020) and Yusuf & Widodo
(2023), proving that interactive multimedia accelerates verbal communication by supplying immediate, shared visual evidence.
Furthermore, having clear visual data assists students in developing the habits of mind emphasized by Rahmawati et al. (2021)
and Elkoumitti et al. (2025), where learners formulate arguments grounded in real-time observations. Rather than simply
memorizing isolated labels, students learn to articulate and defend their scientific assertions using the live models on their
screens. This interactive arrangement shifts classroom dynamics away from rote learning into an active forum for collaborative

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

151

scientific inquiry, where precise academic language and dynamic imagery work together to deepen shared comprehension.

Despite the convincing empirical data, a critical look at the field realities exposes clear infrastructural anomalies, such as
intermittent school internet connectivity and an evident initial learning curve for students unaccustomed to navigating 3D digital
interfaces. These challenges illuminate the intersection of technical limits and uneven digital literacy, showing that even welldesigned instructional systems must remain flexible to survive local constraints. To resolve these friction points, future rollouts
must incorporate structured technical orientations prior to the core lessons and feature offline AR alternatives as a fallback for
unstable network environments. Looking at the broader systemic impact, these findings offer a highly scalable, democratic
alternative to expensive, standalone virtual reality setups, proving how everyday web tools can widen access to immersive
science education. School districts can leverage this model to upgrade their curricula without requiring major hardware
investments. Theoretically, this study challenges the common assumption that immersive media inherently causes cognitive
overload, showing instead that the cognitive outcome depends entirely on the surrounding instructional architecture. By
anchoring interactive technologies within structured, systematic lesson steps, educators can successfully guide student focus,
lower extraneous cognitive load, and build essential scientific competencies for the modern digital era.

5. CONCLUSION
5.1. Suggestions
This study empirically validates that anchoring immersive Augmented Reality (AR) ecosystems within a web-based instructional
wrapper designed via Gagne’s Nine Events of Instruction significantly mitigates abstract spatial constraints and extraneous
cognitive load while simultaneously boosting multidimensional science competencies in junior high school students. The
systematic integration of responsive Web-AR assets directly into a cohesive Google Sites interface successfully bypassed
traditional hardware performance boundaries and installation frictions, allowing working memory resources to be successfully
redirected toward productive germane schema construction. Quantitatively, this instructional wrapper model generated an
exceptionally high significant leap in scientific literacy (t = -15.461; p = 0.000; mean difference = 35.33 points), steering the
experimental group to a superior posttest average (74.29) that far outdistanced the PowerPoint-integrated Project-Based
Learning control framework (62.41). Furthermore, the platform operated as an effective shared visual anchor that naturally drove
collaborative reasoning, culminating in significantly superior scientific communication skills (t = 5.520; p = 0.000; mean difference
= 9.24 points). Ultimately, the Normalized Gain analysis firmly established the operational efficiency of this structured wrapper
model (0.62; medium-approaching-high effectiveness) over the fragmented presentation model (0.45), confirming that techdriven classroom interventions succeed not through raw exposure to visual novelties, but through the deliberate, systematic
design of the surrounding instructional message architecture.
5.2. Suggestions
1.

2.

3.

For Science Educators: Practitioners should actively transition away from isolated, standalone mobile applications
toward lightweight, browser-based Web-AR systems embedded within cohesive digital portals like Google Sites.
However, this adoption must not treat the technology as a passive repository of links; rather, it requires mapping out
strict instructional events—such as combining interactive 3D models with guided prompts and immediate performance
feedback—to ensure digital media directly serves pedagogical objectives.
For School Administrators: Institutional leaders must provide targeted support to bridge the gap between technological
access and instructional design. Schools need to fund professional development programs that train educators in
message design and cognitive load management, while concurrently upgrading local networking infrastructure to
ensure fluid, lag-free delivery of interactive 3D content during live classroom sessions.
For Future Researchers: To expand on these findings, subsequent investigations should broaden the empirical scope by
examining larger, more socio-economically diverse student populations over extended intervention periods. Crucially,
future studies must explore deeper clinical methodologies, such as eye-tracking software and real-time cognitive
diagnostic tools, to isolate exactly how immersive web environments affect high-order spatial thinking skills, long-term
memory retention, and conceptual deconstruction across other abstract scientific disciplines.

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

152

REFERENCE
Abdullah, A. H., Hamdan, M. N., & Mokhtar, M. (2022). Efektivitas augmented reality dalam pembelajaran IPA: Kajian sistematis.
Jurnal Pendidikan Sains dan Matematika, 10(1), 45–62.
Auralia, R., & Juliani, R. (2024). Pengaruh media augmented reality terhadap literasi sains siswa SMP pada materi ekosistem.
Jurnal Pendidikan Sains Indonesia, 12(1), 15–27.
Borich, G. D. (2021). Observation skills for effective teaching: Research-based practice (8th ed.). Routledge.
Cahyadi, M. R., Maryanto, B. P. A., Syaifuddin, M., & Darmayanti, R. (2023). Development of Essay Test Assessment Rubric for
Polya Theory-Based Mathematical Problem-Solving. JNPM (Jurnal Nasional Pendidikan Matematika), 7(1), 45-58.
Cai, S., Liu, C., Wang, T., Liu, E., & Liang, J.-C. (2021). Effects of an augmented reality-based educational game on students'
learning achievements and cognitive load in science. Journal of Computer Assisted Learning, 37(4), 1125–1136. (Catatan:
Terdapat juga variasi publikasi oleh penulis yang sama di tahun 2021 pada British Journal of Educational Technology).
Castro-Alonso, J. C., Wong, M., & Paas, F. (2021). Cognitive load theory and instructional design for digital learning. In Cognitive
Load Theory (pp. 55-72). Springer. (Catatan: Terdapat juga variasi artikel terkait di Educational Psychology Review).
Chen, C.-H., Yeh, H.-C., & Yu, S. (2020). Effects of integrating augmented reality into problem-based learning on student
engagement and academic achievement. Journal of Educational Technology & Society, 23(4), 112-124. (Catatan: Terdapat
juga variasi artikel terkait di Interactive Learning Environments).
Creswell, J. W., & Creswell, J. D. (2020). Research design: qualitative, quantitative, and mixed methods approaches (5th ed.).
SAGE Publications.
Durnalı, M., Orakcı, Ş., & Aktan, O. (2021). The smart school model. In Technology-Enhanced Frameworks for School
Improvement (pp. 1-18). IGI Global.
Elkoumitti, Y., & Halidu, O. (2025). Digital learning ecosystems in secondary science streams: Scaffolding spatial models through
web applications. International Journal of Science Education, 47(1), 112–131.
Field, A. (2020). Discovering statistics using IBM SPSS statistics. SAGE Publications. (Disatukan: Terdapat variasi penulisan edisi
ke-5 dan ke-6 pada sumber asli).
Garzón, J., & Acevedo, J. (2021). Meta-analysis of the effect of augmented reality on students' learning outcomes in STEM
education. Educational Research Review, 31, 100373. (Catatan: Terdapat juga variasi publikasi terkait di Journal of
Educational Computing Research).
Garzón, J., Pavón, J., & Baldiris, S. (2020). Systematic review and meta-analysis of augmented reality in education. Computers &
Education, 142, 103647. (Catatan: Terdapat juga variasi publikasi terkait di Virtual Reality).
Gregory, R. J. (2021). Psychological testing: History, principles, and applications (8th ed.). Pearson.
Hake, R. R. (2020). Analyzing change/gain scores. American Educational Research Association.
Halidu, Z., Bakkali, S., & Talbi, M. (2025). Augmented reality integration in STEM education: Impact on student engagement and
conceptual understanding. Education and Information Technologies, 30(2), 112–130.
Hidayat, T., Suwarna, I. P., & Nandiyanto, A. B. D. (2022). Analisis kemampuan literasi sains siswa SMP di Indonesia berdasarkan
kerangka PISA 2018. Jurnal Ilmu Pendidikan, 28(1), 30–41.
Lestari, D. A., & Kurniawan, A. (2023). Google Sites sebagai platform media pembelajaran terintegrasi pada mata pelajaran IPA
SMP. Jurnal Teknologi Pendidikan, 25(2), 88–101. (Catatan: Terdapat variasi artikel terkait visual media pada volume yang
sama).
Mishra, P., Pandey, C. M., Singh, U., & Gupta, A. (2021). Descriptive statistics and normality tests for statistical data. Annals of
Cardiac Anaesthesia, 24(1), 67-72. (Catatan: Terdapat variasi artikel terkait di Journal of Education and Health Promotion).
Mystakidis, S. (2022). Metaverse. Encyclopedia, 2(1), 486-497.
Mystakidis, S., Berki, E., & Valtanen, J.-P. (2021). Deep and meaningful e-learning with social virtual reality environments in
higher education: A systematic literature review. Applied Sciences, 11(2), 694. (Catatan: Terdapat variasi artikel serupa di
Education Sciences dengan fokus social constructivism).

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

153

OECD. (2023). PISA 2022 results: The state of learning and equity in education (Vol. 1). OECD Publishing.
Pratama, H., Rosidin, U., & Suyatna, I. (2022). Visual tracking systems in science modules: Eliminating device memory constraints
through cloud integration. Journal of Science Education and Technology, 31(3), 341–355.
Pratama, R. A., Agustini, K., & Sindu, I. G. P. (2022). Pengembangan media pembelajaran augmented reality pada materi tata
surya untuk siswa SMP kelas VII. Jurnal Pendidikan Teknologi dan Kejuruan, 19(1), 64–75.
Putri, A. R., & Nugroho, A. S. (2022). Analisis miskonsepsi siswa SMP pada materi tata surya dan strategi remediasinya. Jurnal
Pendidikan IPA Indonesia, 11(2), 201–213.
Radiusman. (2020). Studi literasi: Pemahaman konsep siswa pada pembelajaran matematika. Fibonacci: Jurnal Pendidikan
Matematika dan Matematika, 6(1), 1–8.
Rahmawati, Y., Ridwan, A., & Hadinugrahaningsih, T. (2021). Mengembangkan keterampilan komunikasi ilmiah melalui
pembelajaran berbasis transformative learning di SMA. Jurnal Riset Pendidikan Kimia, 11(2), 74–85. (Disatukan: Terdapat
variasi judul versi bahasa Inggris untuk dokumen yang sama).
Sari, I. N., Wardana, M. Y. S., & Prasetyo, Z. K. (2021). Identifikasi kesulitan belajar siswa pada materi astronomi di SMP: Studi
kasus di Daerah Istimewa Yogyakarta. Jurnal Inovasi Pendidikan IPA, 7(2), 115–126. (Disatukan: Terdapat variasi judul versi
bahasa Inggris untuk dokumen yang sama).
Sörqvist, P., & Skog, E. (2026). Cognitive bias can depend on eco-label design: Evidence from a negative footprint illusion in
energy labeled household appliances. Energy Policy, 216, 115391. (Catatan: Terdapat variasi artikel terkait ergonomi
lingkungan oleh penulis utama di jurnal yang sama).
Sugiyono. (2021). Metode penelitian kuantitatif, kualitatif, dan R&D (edisi ke-3). Alfabeta.
Suryani, A., Kurniawan, A., & Lestari, D. A. (2023). Integrasi media imersif dalam ekosistem pembelajaran digital. Jurnal
Teknologi Pendidikan, 25(3), 180-195.
Sweller, J. (2020). Cognitive load theory: Recent theoretical advances. In International Cognitive Load Theory Association
Conference (pp. 1-12). Springer.
Tetep, T., Suparman, S., & Yulianti, D. (2023). Augmented reality sebagai media inovatif dalam pembelajaran IPA: Kajian metaanalisis. Jurnal Inovasi Teknologi Pendidikan, 10(1), 33–45.
Yusuf, M., & Widodo, A. (2023). Pengembangan kemampuan komunikasi ilmiah siswa melalui pembelajaran berbasis proyek
berbantuan teknologi digital. Jurnal Pendidikan Sains, 11(1), 44–56. (Catatan: Terdapat variasi artikel terkait di Journal of
Technology and Science Education)

© 2026 Author. Published by CV. Bimbingan Belajar Assyfa, Indonesia.

154