When Technology Matters Most: Fitness-App-Integrated Problem-Based Learning as Compensatory Scaffolding for Physical Literacy Among Low-Grit Students

2.1 Study design

This study employed a 2 × 2 factorial between-subjects experimental design to examine the main and interaction effects of instructional model and student grit level on physical literacy outcomes. The first factor was instructional model, consisting of problem-based learning integrated with a fitness application (PBL–App) and problem-based learning without application integration (PBL–NonApp). The second factor was grit, categorized into high and low levels based on pre-intervention assessment. The intervention was implemented as a field-based classroom experiment within regular physical education lessons to maintain ecological validity.

2.2 Research setting and participants

The study was conducted in a public vocational secondary school in Indonesia during the regular physical education program. Participants were Grade XI students enrolled in compulsory physical education. A total of 64 students participated in the study (57 males and 7 females), reflecting the natural gender composition of the school’s automotive engineering program. Gender was not treated as an analytical variable. Given the highly unbalanced gender distribution, gender was not included as a factor in the statistical analyses.

The intervention was implemented over approximately 12 weeks, comprising 16 scheduled physical education lessons (intervention lessons). In addition, baseline and post-intervention assessments were conducted over two class sessions each (Section 2.6), scheduled alongside the regular timetable. All participants followed the same curricular content and instructional duration; experimental groups differed only in the mode of activity monitoring and feedback.

2.3 Randomization, grouping, and experimental conditions

This study adopted a randomized classroom trial design. Intact classes were randomly assigned to either the PBL–App or PBL–NonApp condition prior to the start of the intervention. Grit was measured at baseline and classified using a median-split procedure, resulting in high-grit and low-grit groups. Although grit is frequently treated as a continuous variable, a categorical approach was adopted in this study to facilitate clear interpretation of interaction effects within the factorial experimental framework. Preliminary analyses indicated no significant baseline differences in grit scores across instructional conditions.

The factorial design produced four experimental conditions (n = 16 per group): (1) PBL–App with high grit, (2) PBL–App with low grit, (3) PBL–NonApp with high grit, and (4) PBL–NonApp with low grit.

2.4 Intervention procedures

2.4.1 Core problem-based learning structure

Problem-based learning was implemented using a five-phase instructional sequence adapted for physical education: (1) problem orientation, (2) hypothesis formulation, (3) self-directed information gathering, (4) solution presentation, and (5) reflection. Problems were ill-structured and contextually relevant, such as designing weekly fitness plans or analysing barriers to maintaining physical activity outside school. Teachers acted as facilitators, guiding inquiry and reflection without providing prescriptive solutions. Reflection activities included journaling, peer discussion, and review of individual fitness goals. As part of each problem-based learning cycle, students were also assigned structured independent physical activity tasks to be completed outside scheduled lesson time. These tasks were derived from the problem orientation and hypotheses developed in class and were intended to be implemented in students’ home or community environments, before being revisited during subsequent solution presentation and reflection phases.

Teachers received prior orientation and written instructional guidelines to ensure consistent implementation across experimental conditions.

2.4.2 PBL–App condition

In the PBL–App condition, students used the Strava mobile fitness application to monitor daily physical activity. Application features included step tracking, goal setting, and progress visualisation. Activity data collected outside scheduled lessons were reviewed at the beginning of each class and integrated into the PBL cycle to support reflection, discussion, and strategy revision.

2.4.3 PBL–NonApp condition (Pedometer and manual activity log)

In the PBL–NonApp condition, students monitored daily physical activity using a clip-on waist-mounted pedometer (spring-lever mechanical pedometer). This commonly used non-digital device was attached to the waistband or belt during daily activities and recorded steps based on vertical hip movement. The pedometer provided basic step-count feedback without digital connectivity, automated data storage, or real-time feedback features.

Students also completed a daily paper-based activity log. The log sheet required documentation of: (a) the type of physical activity performed, (b) the total number of steps per day, and (c) the duration of physical activity. To support accountability and parental awareness, each daily log included a section for parent or guardian acknowledgement signature. Logs were collected weekly and reviewed and countersigned by the physical education teacher.

2.5 Instruments and measures

2.5.1 Physical literacy

Physical literacy was assessed using the Canadian Assessment of Physical Literacy, Second Edition (CAPL-2), which measures four domains: Physical Competence, Daily Behaviour, Motivation and Confidence, and Knowledge and Understanding. Scoring followed CAPL-2 guidelines to produce an overall physical literacy score on a standardised 100-point scale.

A minor modification was applied to the physical competence component (CAMSA) to accommodate contextual constraints and participant age. Specific movement tasks were adapted while retaining the core constructs of physical competence assessed in the original CAPL-2 framework. Content validity of the adapted tasks was ensured through expert review by physical education specialists. Reliability of questionnaire-based domains was examined using Cronbach’s alpha (α = 0.81), indicating acceptable internal consistency, while reliability of physical competence measures was assessed using intraclass correlation coefficients (ICC > 0.90), reflecting excellent measurement reliability. These procedures are consistent with prior cross-cultural applications of the CAPL-2 instrument.

2.5.2 Grit

Grit was measured using the Physical Education Grit Scale (PE-Grit). The instrument was translated and culturally adapted for Indonesian vocational students through forward–backward translation and expert validation. Content validity was established through expert judgement during the adaptation process. Reliability analysis from pilot testing demonstrated good internal consistency (Cronbach’s α = 0.81). Given that the PE-Grit scale has been previously validated using confirmatory factor analysis by its original developers, the present study focused on cultural adaptation and reliability testing rather than re-establishing construct validity.

2.6 Data collection procedure

Baseline assessments of grit and physical literacy were conducted in Week 1 across two class sessions to ensure complete administration of CAPL-2 domains and the PE-Grit questionnaire within the constraints of lesson time. The instructional intervention itself was implemented over approximately 12 weeks, comprising 16 physical education lessons. Post-test assessments were conducted in Week 13 and were also administered across two class sessions using identical procedures. Importantly, the pre-test and post-test sessions were assessment sessions and were not counted as part of the intervention lessons. Daily physical activity data were collected throughout the intervention using either application logs or pedometer/manual logs, depending on group assignment. In line with CAPL-2 procedures, step-monitoring devices were issued during the baseline assessment period and were used to support Daily Behaviour monitoring during the intervention; devices/logs were checked periodically and final records were collected at post-test. These data represented students’ task-based independent physical activity, which contributed directly to the Daily Behaviour domain of the CAPL-2, while motivation, confidence, and knowledge domains were assessed through questionnaire-based components in accordance with CAPL-2 guidelines.

Treatment fidelity was monitored through periodic classroom observations conducted by the research team and through structured teacher checklists aligned with the PBL implementation guidelines to ensure adherence to instructional protocols.

2.7 Data analysis

Statistical analyses were conducted using SPSS version 27 (IBM Corp.) with a significance level of α = 0.05. Assumptions of normality and homogeneity of variance were examined using the Shapiro–Wilk and Levene’s tests. Descriptive statistics were computed for each experimental group.

Primary analyses employed two-way ANOVA to test the main and interaction effects of instructional model and grit on physical literacy outcomes. Pre-test physical literacy scores were examined and found to be comparable across groups; therefore, two-way ANOVA on post-test scores was considered appropriate for examining group differences and interaction effects. Effect sizes were reported using partial eta squared (ηp²). Where significant interaction effects were identified, post hoc comparisons were conducted using Tukey’s HSD test.

2.8 Ethical approval

Ethical approval was obtained from the Research Ethics Committee of Universitas Pendidikan Indonesia (No. 336/UN40.K/PT.01.01/2025). Written informed consent was obtained from students and their parents or guardians prior to participation. All data were anonymised and stored securely.

2.9 Data and materials availability

The data generated in this study involve secondary school students, including minors, and therefore contain sensitive educational and behavioral information. In accordance with the approval granted by the Research Ethics Committee of Universitas Pendidikan Indonesia (No. 336/UN40.K/PT.01.01/2025), the datasets are not publicly available to protect participant confidentiality and minimize the risk of re-identification.

All data were anonymized prior to analysis and are stored securely by the research team. Researchers who wish to access the anonymized dataset may submit a reasonable request to the corresponding author. Access will be considered on a case-by-case basis and is conditional upon a clear description of the intended academic use, confirmation that the data will be used solely for non-commercial research purposes, and, where required, approval from an appropriate institutional ethics committee. No data containing personally identifiable information will be shared.

Materials associated with the intervention, including lesson plans, problem scenarios, and scoring procedures, are available from the corresponding author upon reasonable request.

2.10 Generative AI statement

Generative AI tools were used solely for language polishing and formatting. No generative AI was used in study design, data collection, data analysis, or interpretation.

3.1 Participant flow and baseline equivalence

A total of 64 Grade XI vocational students were included in the analysis, resulting in four balanced factorial cells (n = 16 per cell): PBL–App/High grit (A1B1), PBL–App/Low grit (A1B2), PBL–NonApp/High grit (A2B1), and PBL–NonApp/Low grit (A2B2).

Baseline equivalence. Baseline (pre-test) CAPL-2 total scores are summarised in Table 1. Consistent with random class assignment and balanced cell sizes, the group means were close in magnitude, suggesting no substantial baseline imbalance. (Because the dissertation chapter reports assumption checks on baseline data but does not report an inferential baseline-equivalence test for CAPL-2 pre-test, the manuscript reports baseline comparability descriptively.)

All participants (N = 64) completed both the pre-test and post-test assessments; therefore, no cases were excluded and no missing outcome data were present.

3.2 Assumption checks

Assumptions for parametric testing were satisfied. Normality screening did not indicate significant deviation from normality (Kolmogorov–Smirnov p = 0.087; Shapiro–Wilk p = 0.071). Homogeneity of variance was also met (Levene’s test: pre-test p = 0.223; post-test p = 0.310). Therefore, post-test CAPL-2 total scores were analysed using a two-way ANOVA.

3.3 Descriptive statistics (Post-test)

Post-test CAPL-2 total scores by factorial cell are summarised in Table 2. The highest post-test mean was observed in the PBL–NonApp/High-grit group (A2B1), whereas the lowest mean was observed in the PBL–NonApp/Low-grit group (A2B2).

3.4 Two-way ANOVA (Post-test CAPL-2 total score)

A two-way ANOVA tested the main effects of instructional model and grit, and their interaction, on post-test CAPL-2 total scores ( Table 3). The main effect of instructional model was not statistically significant, whereas the main effect of grit and the instructional model × grit interaction were statistically significant.

3.5 Follow-up comparisons (Interaction interpretation)

Given the significant interaction ( Table 3), follow-up comparisons were conducted to clarify the pattern of differences. In line with the dissertation analysis, these comparisons focus on the four theoretically relevant contrasts: (i) grit differences within each instructional model and (ii) instructional-model differences within each grit level.

For transparency, Table 4 reports each contrast as a mean difference with the corresponding standard error computed from the pooled ANOVA error term (MS_error = 81.927; df = 60) and equal cell sizes (n = 16). An adjusted significance threshold was applied across the four planned contrasts (Bonferroni α_adj = 0.0125).

3.6 Interaction plot

Figure 1 presents the interaction between instructional model and grit level on post-test physical literacy scores.

Figure 1. Interaction between instructional model (PBL–App vs. PBL–NonApp) and grit level (high vs. low) on post-test CAPL-2 total physical literacy scores.

The interaction plot shows a non-parallel pattern, indicating that the effectiveness of the instructional model varied depending on students’ grit levels. Specifically, the difference between the PBL–App and PBL–NonApp conditions was more pronounced among students with low grit, whereas this difference was substantially reduced among students with high grit. This pattern suggests that see technology-integrated problem-based learning particularly benefits students with lower levels of grit, while students with higher grit demonstrated relatively comparable outcomes across instructional conditions. These findings are consistent with the significant interaction effect observed in the two-way ANOVA, confirming that the impact of instructional model on physical literacy outcomes is conditional upon students’ grit levels.

4.1 Interpretation of main and interaction effects

The positive association between grit and physical literacy is consistent with the view that persistence and sustained effort can support engagement across physically and cognitively demanding learning cycles in physical education. This interpretation aligns with broader evidence linking grit-related self-regulatory tendencies to educational engagement and performance, while also reflecting critiques that grit effects are typically modest and context-dependent and should not be treated as a single, deterministic explanation of learning outcomes (Credé, 2018; Muenks et al., 2017). In the present context, grit may help students maintain effort during repeated phases of goal setting, practice, feedback use, and reflection that are central to PBL.

The absence of a significant main effect of instructional model suggests that technology integration should not be assumed to function as a universal enhancer. Prior work on digitalisation in physical education emphasizes that outcomes depend on pedagogical alignment and implementation quality, not merely the presence of technology (Osmanović et al., 2021). The most important finding is therefore the significant interaction between instructional model and grit. The interaction pattern is best characterized as an ordinal (non-cross-over) interaction, in which the magnitude of the instructional difference changes across grit levels without requiring intersecting lines. Substantively, app-supported PBL appears to provide greater added value for students with lower grit, whereas high-grit students achieve comparable outcomes with or without app support.

4.2 Why grit may matter for physical literacy development

Physical literacy is multidimensional and includes affective and cognitive components alongside physical competence and daily behavior. These domains are likely shaped by sustained engagement, follow-through, and willingness to persist when tasks become challenging. From this perspective, students with higher grit may be better positioned to benefit from PBL even when scaffolds are minimal because they are more likely to continue practicing, reflect on progress, and complete independent tasks. At the same time, the critical literature on grit cautions against overinterpreting grit as a stable, universally beneficial trait, and encourages attention to contextual supports that may amplify or compensate for differences in persistence (Credé, 2018). This balanced view fits the present findings, where grit appears most informative when examined as a moderator of instructional conditions rather than as an isolated predictor.

4.3 Fitness applications as scaffolding for lower-grit learners

The compensatory pattern for lower-grit students can be interpreted through a self-regulated learning lens. PBL requires learners to sustain effort across multiple phases, and students who struggle with perseverance may show inconsistent engagement during iterative cycles, especially when a substantial portion of activity practice occurs outside scheduled lesson time. In contrast, fitness applications can function as external scaffolds by structuring monitoring routines, making progress visible, and providing feedback cues that support adherence to planned activities. Such scaffolding is most likely to matter when technology outputs are not treated as an add-on, but are brought back into classroom inquiry and reflection so that the data become part of the problem-solving process.

In this intervention, app-based monitoring was integrated into the PBL cycle by using students’ activity records as inputs for discussion, evaluation of hypotheses, and revision of weekly strategies. This integration may have reduced the self-regulatory load for lower-grit learners by converting activity goals into concrete, trackable actions and by strengthening the link between independent activity tasks and classroom learning. This interpretation is consistent with scholarship arguing that technology-enhanced instructional models are most effective when tools are tightly aligned with pedagogy and used to support reflective learning processes rather than merely recording behavior (Cojocaru et al., 2022).

4.4 Limited incremental gains among higher-grit students

For higher-grit students, the limited incremental benefit of app integration suggests that these learners may already possess the persistence and self-regulatory capacity needed to profit from PBL without additional digital scaffolding. Evidence from educational psychology indicates that learners with stronger self-regulation sometimes demonstrate smaller marginal gains from external supports because their internal goal-maintenance and monitoring processes are already robust (Muenks et al., 2017). However, this finding should not be interpreted as evidence that fitness applications are irrelevant for high-grit students. Instead, it indicates that the added value of the application was smaller for this subgroup under the current implementation. Different designs, including more advanced goal personalization or tasks that require explicit interpretation of activity data, may yield different results.

4.5 Implications for technology-enhanced PBL in vocational physical education

The vocational context strengthens the relevance of a “for whom and under what conditions” interpretation. Vocational students often display heterogeneous learning readiness and engagement profiles, and instructional designs that rely heavily on sustained self-direction may therefore produce uneven benefits across students. The present findings suggest that integrating app-based monitoring into PBL can serve as an adaptive scaffold, particularly for students who may otherwise disengage during independent activity components. At the same time, this implication depends on implementation quality. Scholarship on digitalisation in physical education underscores that benefits are unlikely when technology is used superficially or without coherent integration into lesson goals (Osmanović et al., 2021). The current integration approach, which used monitoring data as a basis for collaborative reflection and revision, may therefore be a central feature supporting the observed conditional benefits.

4.6 Practical implications

These results suggest that physical education teachers and schools may consider differentiated technology integration rather than uniform requirements for all students. For learners who demonstrate lower persistence or weaker follow-through on independent activity tasks, fitness applications can provide additional structure through monitoring routines and feedback cues, potentially supporting more equitable physical literacy development.

The findings also reinforce that technology use should be pedagogically purposeful. Teachers can connect app outputs, such as step counts and activity summaries, to inquiry questions and reflection prompts so that data are used for learning and decision-making rather than for tracking alone. Implementation support is likely to be important, particularly professional learning that helps teachers design inquiry tasks around monitoring data and manage classroom routines for reviewing and interpreting activity records.

Finally, technology adoption should be accompanied by attention to practical and ethical considerations, including student access to devices, data privacy, and clear policies for appropriate use in school contexts (Bopp & Stellefson, 2020). These issues may be especially salient in vocational settings where access and resources can vary.

4.7 Limitations

Several limitations should be considered when interpreting the findings. The study involved one vocational secondary school and a highly male-dominated program, which may limit generalizability to other vocational majors and more gender-balanced cohorts. Instructional conditions were assigned at the intact-class level. While this strengthens ecological validity, it also introduces the possibility of class-level clustering; future studies can address this through multilevel modeling or by increasing the number of classes per condition.

Grit was categorized into high and low groups to fit the factorial design, which may reduce sensitivity compared with continuous moderation approaches. Future studies could model grit continuously to capture more nuanced conditional effects. In addition, CAPL-2 was adapted to suit older adolescents and contextual constraints. Although expert review supported alignment with the intended constructs, further validation work is needed to strengthen adolescent-appropriate physical literacy measurement models.

Finally, the PBL–NonApp condition relied on pedometers and paper logs, which may be more vulnerable to recording error or non-compliance than app-based monitoring. Although signatures and teacher checks were used to enhance accountability, differential measurement precision across conditions should be acknowledged.

4.8 Directions for future research

Future research can extend this work by testing whether the observed benefits for lower-grit students persist beyond the intervention period using longer follow-up designs. Mechanism-focused studies can examine whether self-monitoring, feedback interpretation, or peer accountability mediates the conditional effect of app-integrated PBL. Studies involving more diverse vocational majors and multiple schools would strengthen external validity and help identify contextual boundary conditions.

Analytically, future work may benefit from regression-based moderation with continuous grit and from multilevel models that account for class-level clustering. Examining CAPL-2 domain-level outcomes could also clarify whether technology-enhanced PBL primarily affects daily behavior, motivation and confidence, knowledge and understanding, or physical competence. Taken together, the findings indicate that integrating fitness applications into PBL can function as an adaptive scaffold for learners who need more support to sustain engagement and self-regulation, highlighting the importance of aligning pedagogy, technology, and learner dispositions in vocational physical education.