Introduction

Fostering lifelong physical activity (PA) and overall well-being in children is a critical priority in both public health and education, which requires a deep understanding of foundational constructs such as motor competence (MC) and physical literacy (PL) [1–7]. MC refers to proficient motor-skill performance, including coordination and motor control [1]. MC is a multidimensional construct encompassing various motor skills and their underlying mechanisms. It plays a pivotal role in influencing various aspects of children’s health, including their physical fitness, psychological well-being and overall health-related quality of life [2–4].

A holistic concept related to individual health is PL, which is understood as an integrated construct spanning physical, psychological, social and cognitive domains [5], rather than a set of isolated components. In this view, fostering PL entails attending simultaneously to physical competence, motivation and confidence, knowledge and understanding, and the value and commitment to engage in lifelong PA [6,7]. Perceived PL (PPL) specifically refers to how children view their own PL [8], such as evaluating their own physical competence, motivation and confidence, and their knowledge and understanding in participating in PA [9]. Indeed, PL is a broader construct than MC covering not only proficiency in motor skills but, as mentioned encompasses various aspects of an individual’s ability to engage in PA [10]. Both PPL and MC can impact children’s participation in PA [11] and improve their overall health outcomes, essential for maintaining a healthy lifestyle [12].

The literature has established a strong link between PA participation and both MC [13,14] and PL [15]. This relationship is often described as dynamic and reciprocal, suggesting that higher levels of PA are related to better children’s MC and PL development [15–18] and is crucial, as MC and PL are determinants of lifelong PA and overall health [15,19].

MC is associated with numerous types of factors such as biological and demographic, behavioral, cognitive, emotional, and psychological, cultural and social factors, creating a system with adaptable components that work together synergistically and cooperatively to generate a diverse array of motor results [20]. The relationship between MC and PA is also complex and influenced by a variety of correlates such as gender, physical fitness, perceived MC and physical environmental factors [13,21,22]. It is now generally accepted that the physical environment plays a crucial role in facilitating both children’s and adolescents’ PA behavior [23–26]. For instance, positive attributes of the home neighborhood, such as the presence of recreational facilities, higher socio-economic status (SES) and a safe transport infrastructure encourage PA, whereas negative factors such as traffic and safety concerns hinder it [23–26].

Recent studies have highlighted the importance of the physical environmental factors in promoting not only PA but also MC [27–29]. Although the school environment is considered an important factor in PA behavior, as children spend a significant amount of time at school [30], little is known about how environmental characteristics at home and school influence this relationship. In particular, the moderating effects of perceived dimensions, such as PPL, remain underexplored. In fact, the relationship between PA and MC can be moderated by a variety of physical environmental factors, including geographical location and the school environment [27–29]. It seems that children from rural areas have better MC and spend more time outdoors playing than those children from urban areas [28], while less densely populated areas may provide more opportunities for PA and MC development. In relation to the school physical environment, the presence of sports and play facilities and well-designed school grounds (including aesthetics) support MC [27,29]. Further research is still needed to deepen the analysis of the relationships between the neighborhood and school characteristics and the development of MC by considering children’s PA behavior. Future research should also examine the role of environmental factors in PL due to the lack of information on this topic [31].

The relationship between PA, MC, and PPL is a critical area of research in education and public health, while understanding how physical environmental factors moderate these relationships can provide valuable insights for developing effective interventions and policies. The aim of this study was thus to examine the relationship between PA, MC and PPL in a school-age population, considering through a model-based tree approach the physical environment in the school and home-neighborhood as factors that moderate these relationships. We hypothesize that PA is positively associated with MC and PPL, but that these relationships are heterogeneous across the sample. Specifically, we expect that environmental factors in the school and home-neighborhood identify subgroups where the impact of PA on MC and PPL is significantly moderated. By examining the study objective, we considered it would be possible not only to determine whether there are subsamples in which the influence of PA on the outcome variables differs, but also to identify the environmental variables that actually affect this relationship. With this information, recommendations can be made to the relevant stakeholders to maximize the impact of PA on MC and PPL while considering the physical environment.

Methods

Results

Descriptive results

Table 1 gives the descriptive statistics of all the variables included in the present study, plus the differences between boys and girls in the variables. Results indicate significant gender differences in MC, PA, PPL, the physical domain of PL, recreational facilities, and park proximity. In all these variables, boys reported higher scores than girls. The magnitude of these differences was small for the majority of variables (0 < |d| < 0.5), except for MC, which yielded a medium effect size (d = 0.57).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Descriptive data and gender comparisons.

https://doi.org/10.1371/journal.pone.0350957.t001

Model-based trees

To address our primary objective of identifying how environmental factors moderate the impact of PA on MC and PPL, we employed model-based recursive partitioning. The model-based tree for MC is presented in Fig 1. Three school variables emerged as potential mediators (i.e., Sports and play facilities, School aesthetics, and PA policies) leading to four distinct subsamples with different slopes for the PA/MC relationship. Among the participants in schools scoring ≤4 in Sports and play facilities (n = 147), the slope was b = 0.86, representing the baseline relationship where PA contributes to MC, but its impact is limited by the lack of adequate infrastructure. For those in schools where Sports and play facilities >4, School aesthetics ≤5, and PA policies ≤4 (n = 63), the slope increased to b = 0.92. Students attending schools with Sports and play facilities >4, School aesthetics ≤5, and PA policies >4 (n = 142) showed a slope of b = 1.17, suggesting that the implementation of formal PA policies further boosts the effectiveness of PA on motor development. Finally, the participants in schools with over 4 points in Sports and play facilities and >5 on School aesthetics (n = 44) exhibited the steepest slope (b = 2.07). In practical terms, this indicates that the impact of PA on motor development is more than doubled in optimal school environments. These results suggest that high-quality facilities and aesthetics act as a catalyst, significantly amplifying the benefits of PA for children’s MC.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Linear regression-based tree for Motor Competence (MC).

The terminal nodes of the tree plot report the intercept and estimate of physical activity (PA), but the models are controlled by gender (G) and socio-economic status (SES) as reported in equations. Variables related to the school environment are shown with a blue background, while those related to the home neighborhood are shown with a yellow background.

https://doi.org/10.1371/journal.pone.0350957.g001

Three subgroups were identified in the recursive partitioning model for PPL (Fig 2), with one school variable (PA policies) and one home-neighborhood variable (Recreational facilities) acting as potential mediators. The first subgroup comprised students in schools with PA policies ≤4 (n = 126), yielding a slope of b = 0.24, representing the weakest association between PA and PPL in contexts with limited institutional support. A second subgroup included those in schools with PA policies >4 and Recreational facilities ≤2.75 (n = 196), where the slope changed to b = 0.28. The participants in schools where PA policies >4 and Recreational facilities >2.75 (n = 74) displayed the most pronounced slope (b = 0.40). This suggests that the synergy between supportive school policies and accessible neighborhood facilities creates a more stimulating environment, significantly strengthening how PA translates into higher PPL. Essentially, when the physical environment provides more opportunities for practice, the positive impact of PA on PPL is maximized.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. Linear regression-based tree for Perceived Physical Literacy (PPL).

The terminal nodes of the tree plot report the intercept and estimate of physical activity (PA), but the models are controlled by gender (G) and socio-economic status (SES) as reported in equations. Variables related to the school environment are shown with a blue background, while those related to the home neighborhood are shown with a yellow background.

https://doi.org/10.1371/journal.pone.0350957.g002

As shown in Table 2, when comparing the multiple linear models estimated for the entire sample with those obtained through recursive partitioning, the latter demonstrated a slightly better fit. For both MC and PPL, the model-based trees obtained higher Pearson correlation coefficients (with an improvement of r = 0.06–0.09) and lower estimation errors (i.e., MSE and MAE) than the multiple linear models. These performance metrics validate our initial hypothesis: the relationship between PA and the outcomes is not linear across all contexts, but is significantly enhanced by specific environmental thresholds. This also showcase that the recursive partitioning approach might provide a more nuanced and accurate understanding of how school and neighborhood characteristics influence children’s motor development.

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. Performance parameters for model-based trees and linear models for motor competence and perceived physical literacy.

https://doi.org/10.1371/journal.pone.0350957.t002

Discussion

Our findings support the study’s central hypothesis: while PA is positively associated with both MC and PPL, these relationships are not uniform across the sample but are significantly moderated by environmental contexts. Specifically, the model-based tree analysis revealed that school facilities, aesthetics, and PA policies—alongside neighborhood recreational facilities—identify distinct subgroups where the impact of PA is amplified or constrained. In particular, the discovery that optimal school environments can more than double the effect of PA on MC, and that the synergy between school and home-neighborhood factors maximizes PPL, confirms that the developmental benefits of movement are conditioned by specific environmental thresholds.

Children motor development is inherently shaped by a complex interplay of individual, social, and environmental factors, so that its research calls for a multidimensional and ecologically-grounded approach [20]. Within this framework, schools and residential neighborhoods emerge as critical environments in which opportunities for movement and perceptions of competence are either nurtured or constrained [27–30]. This study examined the moderating role of the physical environment—both school and residential neighborhood—on the relationships between PA engagement with MC and PPL in school-aged children, using a model-based tree approach. Despite the ever-increasing recognition of the role of the environment in shaping PA behaviors, few studies have examined how specific school and home neighborhood characteristics influence the strength of PA’s relationship with MC and PPL [46], while even less is known about how these contextual factors interact with one another or with sociodemographic variables such as gender and SES. Addressing this gap, our study contributes novel insights into how place-based factors can condition developmental trajectories (e.g., MC, PPL) in childhood.

The findings of the current study demonstrate that specific physical environmental features significantly moderate the strength of these relationships and that boys outperform girls in MC, PPL, particularly in the physical domain, and PA engagement. These findings are consistent with previous research [2,21], in which boys showed significantly higher levels of MC, PA, and PPL than girls. These differences may reflect gendered patterns of participation in PA and socialized beliefs around MC and self-perception [47]. Boys may be afforded more opportunities for active play and structured sports, which in turn can reinforce higher self-perceptions of competence [3,22]. These gender disparities highlight the need for gender-responsive interventions that promote equitable access and engagement in diverse forms of PA from an early age. To effectively reduce these gaps, interventions should move beyond traditional competitive sports models, which often favor male participation, and instead incorporate a broader range of non-competitive, cooperative, and creative physical activities. Furthermore, school-based programs could benefit from pedagogical strategies that challenge gender stereotypes and provide girls with targeted mastery experiences to bolster their self-perception of competence [48]. From an environmental perspective, designing school playgrounds with diverse ‘activity zones’—rather than predominantly sports-centric layouts—could encourage girls to engage more actively in movement without the pressure of male-dominated spaces.

The model-based trees provided novel insights into how specific environmental factors moderate the association between PA and both MC and PPL. For MC, school-based variables—namely sports and play facilities, school aesthetics, and PA policies—emerged as key moderators. Notably, children attending schools with higher-quality sports facilities and aesthetically appealing environments, coupled with comprehensive PA policies, exhibited stronger associations between PA and MC. This suggests that favorable school environments not only facilitate more PA but may also amplify the developmental benefits of such activity on motor skill acquisition [27,30]. The steepest slope observed (b = 2.07) for children in schools with high scores in all three areas underscores the importance of rethinking the synergistic physical environmental constructions such as the nexus of recreational infrastructure, aesthetics and policies in promoting PA engagement and MC.

In the case of PPL, both school and home-neighborhood environments served as moderators. PA policies again emerged as a significant school-level factor, while recreational facilities in the neighborhood context influenced the strength of the PA–PPL relationship. The positive moderation effect of these variables aligns with the existing literature on the ecological model of health behavior [23,37], in which access to diverse recreational spaces and supportive institutional policies may not only increase children’s opportunities for engagement in PA but also foster greater confidence and self-perception regarding their physical capabilities [9]. Interestingly, the combination of strong school policies and rich recreational access yielded the highest slope (b = 0.40). This emphasizes the interplay between institutional and community-level supports. Importantly, this study assesses PPL—children’s self-evaluation—rather than the actual PL as a holistic construct. While MC captures objective performance, PPL reflects the internalization of PL related-elements of movement and its importance. This distinction is vital, as subjective perceptions often drive long-term engagement independently of actual motor skill levels.

Our findings suggest that PPL, as a self-reflective construct, is deeply influenced by the variety and quality of the contexts where movement occurs. Unlike MC, which represents objective skill, PPL is built through the internalization of successful experiences in diverse settings. High-quality neighborhood facilities strengthen the PA-PPL link by providing non-evaluative spaces for autonomous exploration. In these environments, movement can be experienced as a personal achievement rather than a graded task, directly boosting physical self-confidence. The synergy between school policies and neighborhood resources further facilitates the transition from ‘doing’ to ‘believing’; while schools might provide institutional validation, the neighborhood might offer the ‘practice ground’ to internalize these capabilities. Consequently, in resource-limited environments, this relationship is weakened (as shown by our lower slopes) because children lack the diverse success scenarios needed to translate physical effort into a robust sense of personal capability.

When the moderator effect was examined according to the PL domain, particular nuances were obtained. For instance, the closeness to parks fostered the association between PA and PPL in the social domain (see S3 Fig), which in turn highlights the importance of providing opportunities for children to interact with their peers while autonomously actively playing to empower their PA-based social skills [49]. Additionally, school-based facilities such as drinking fountains and wildlife or nature gardens when coupled with aesthetically enhanced school environments were found to strengthen the link between PA and PPL both in the psychological and cognitive domains (see S2 and S4 Figs). These findings emphasize the need to reconsider how both school and neighborhood infrastructure policies are designed, as they collectively influence children’s holistic motor development and PL acquisition [50].

It is important to note that none of the environmental (school and neighborhood) moderators negatively impacted the association between PA with MC and PPL (which would depict a downhill path), rather it was these environment factors that slightly or strongly moderated the aforementioned association. These results are consistent with the hypothesis of this study, as in this region every school is obliged to fulfill its duty in promoting students’ health (Decree 106/2022). Consequently, school-based PA policies in the form of extracurricular PA, movement-based pedagogies, active transport, etc. are applied, which in turn not only contribute to fostering the students’ PA but also to their MC and PPL. This study thus showcases the importance of schools developing and integrating PA policies so that students can benefit by enhancing not only their level of PA engagement but also their actual competence and self-perception of PL, which can be considered a holistic perspective of health [50].

This study has several strengths in that it used a model-based recursive partitioning method, which allowed us to detect heterogeneity in how PA relates to developmental outcomes across different environmental contexts—an approach that goes beyond the limitations of traditional linear regression. Whereas traditional linear modeling often assumes homogeneity across populations, potentially masking meaningful subgroup-specific variations, model-based trees recursive partitioning enables the identification of the distinct environmental and contextual moderators that influence how PA relates to MC and PPL across different settings [44]. This method can reveal differential patterns that would otherwise be obscured in aggregate analyses [27,29]. The use of this analytical strategy is particularly relevant in PA, in which interactions between behavior and environment are rarely uniform. In doing so, it offers a more nuanced and ecologically valid understanding of how environmental factors shape the developmental outcomes related to movement and health in children [37]. The study also controlled for key covariates such as gender and SES, increasing the validity of the observed moderating effects.

However, certain limitations must be acknowledged. The findings should be interpreted with caution, as the study’s cross-sectional design precludes causal inference. Also, although self-reporting measures for PA and PPL are widely used and validated [8,39], they may be subject to recall bias or socially desirable effects. Furthermore, the study’s dropout rate must be considered, as attrition may have influenced the final sample’s representativeness. Regarding the variables, the limited variability observed in certain school factors (e.g., school grounds suitability) might have restricted the model’s ability to detect additional moderation effects. It is also important to acknowledge that using a purely quantitative approach may not fully capture the holistic and multidimensional nature of PL. Future research could benefit from mixed-methods designs to provide a more comprehensive understanding of children’s movement experiences. Moreover, the psychosocial and cultural factors that were not included in the models (e.g., peer norms, parental modeling) could also interact with environmental moderators. Future studies should adopt longitudinal designs and consider integrating psychological and social dimensions to better understand how complex systems interact over time.

From the point of view of practical implications, these findings offer actionable insights for school administrators, urban planners and policy makers. Enhancing the quality and accessibility of school-based physical environments—such as playgrounds, gymnasiums, and outdoor aesthetics—along with implementing supportive PA policies, can help optimize the impact of PA on children’s motor development and PPL. Likewise, investments in neighborhood recreational infrastructure, including parks, can complement school efforts and foster a more active, confident, and physically literate young generation. Importantly, these improvements should be equitably distributed to ensure safe and inclusive access for girls, specifically designing environments that challenge gender stereotypes and avoid exacerbating existing disparities between the socio-demographic groups.

Conclusions

This study underscores the significance of physical environmental factors in moderating the relationship between PA and MC, as well as PPL in children. While the cross-sectional design precludes causal claims, our findings suggest plausible mechanisms through which the environment may amplify the benefits of PA. Using a model-based recursive partitioning approach [44], we identified how school and neighborhood features—such as PA policies, recreational access, and environmental aesthetics—moderate these relationships. This method allowed us to detect subgroup-specific patterns according to specific environmental variables that influence these relationships. The findings provide valuable insights for stakeholders to develop targeted inclusive interventions and policies that enhance opportunities for skill development and perceived competence regarding MC and PPL. In line with the ecological model of health behavior [23], our data indicate that supportive environments are associated not only with higher PA engagement but also with strengthened perceptions of children’s confidence, competence, and social interaction through movement [9,37]. This suggests that the environment might act as a catalyst; for instance, the synergy between school policies and neighborhood recreational facilities might provide the necessary ‘practice grounds’ to internalize movement as a successful experience, thereby reinforcing PPL. Compared to standard linear regression, which assumes a uniform relationship across the entire population, the model-based tree approach is particularly suited for detecting moderation as it automatically identifies complex interactions and non-linear effects. By uncovering specific environmental thresholds, this method allows for the identification of distinct subgroups where the impact of PA is significantly different, a task that would require pre-specified interaction terms in traditional models. The model-based tree approach thus offers a robust framework for exploring complex interactions and detecting heterogeneity in educational and public health research.

Supporting information

Acknowledgments

The authors would like to express their gratitude to the heads, students, parents and teachers of the participating schools for their cooperation.