In contemporary societies, individuals face everyday challenges that demand effective responses and structured solutions. Ortiz et al. [1] argue that children and adolescents must be prepared to confront today’s demands. This calls for the early development of cognitive and socioemotional skills such as critical thinking and problem-solving. The latter is recognized as a key twenty-first-century competency [2] that enables individuals to navigate complex and shifting environments. UNESCO [3] defines problem-solving as the ability to analyze and apply creative and innovative strategies.

Strengthening problem-solving skills promotes autonomy and the capacity to deal with emerging situations. International assessments such as PISA, administered by the Organisation for Economic Co-operation and Development (OECD), show that Colombia lags other countries in core areas like mathematics and problem-solving, with even wider gaps in rural areas [4].

One example is the Francisco de Paula Santander Educational Institution (IEFPS), located in the municipality of Calamar, Bolívar. There, structural deficiencies in technological infrastructure and instructional resources hinder the development of advanced cognitive skills. A longitudinal analysis of Saber 11 test results in Colombia revealed sustained underperformance in problem-solving skills, particularly in mathematics and science. School administrators and teachers expressed concern over the lack of an institutional pedagogical strategy, limited internet access, scarce digital resources, and declining student motivation. Some students can generate promising ideas but often fail to structure them in systematic ways that lead to solutions.

For this reason, the school was chosen as a case study, as it reflects the structural conditions faced by many rural schools across Latin America. In this context, we propose a pedagogical strategy focused on computational thinking as a formative path to promote problem-solving skills. George-Reyes et al. [5] point out that this approach allows educators to address educational challenges through a conceptual framework tailored to real-world contexts. According to Seehorn et al. [6], computational thinking involves processes such as decomposition, abstraction, algorithmic formulation, and structured design to tackle complex problems.

Since its original formulation by Wing [7], computational thinking has been closely linked to problem solving, system design, and the understanding of human behavior. It has since become an interdisciplinary tool for addressing both mathematical and social issues [8]. At the curricular level, it has been promoted by organizations such as the CSTA, which advocate for its integration into the education systems of the United States and Canada. Brennan and Resnick [9] redefined it through the Scratch programming language, identifying three core components: computational concepts, practices, and perspectives aimed at expressing oneself, connecting with others, and engaging in inquiry.

This study adopted three analytical dimensions, computational concepts, required tasks, and evaluative capacities, drawing on the contributions of Román-González [10], Barboza et al. [11], and Cáliz et al. [12]. These dimensions were selected for their relevance to science education in Colombian secondary schools. Problem-solving has been theorized from various perspectives. Pinzón et al. [13] distinguish between a psychological level, when one does not know how to reach a goal, and a cognitive level, when the goal is known but the means are not. For Vásquez et al. [14], it is a competency that involves analyzing processes and achieving specific outcomes. Pi ñeiro [15] describes it as a procedural process that is not necessarily linear, while Olivares [16] emphasizes its importance in tasks that lack clear solution paths.

Classical authors such as Polya [17] proposed heuristic approaches based on strategies like generalization and analogy. Newell and Simon [18] introduced the problem space model as a goal-oriented mental structure. Perales [19] defined a problem as a situation of uncertainty that triggers a search process, and D’Zurilla and Nezu [20] expanded this concept by including emotional and behavioral dimensions. In pedagogical terms, Galleguillos et al. [21] stress that teachers must design learning experiences within their subject areas that enable students to develop these skills.

Pedagogical strategies that focus on enhancing computational thinking and problem-solving are relevant insofar as they support the development of competencies applicable to real-life contexts [22], while also boosting student motivation through experiential methods [23]. Martínez-Daza [24] underscores the importance of strategic design to integrate teaching, learning, and assessment. Still, Picado-Arce et al. [25] warns that merely providing technological resources without sustainable planning or pedagogical support does not guarantee meaningful learning.

Computational thinking strategies go beyond digital tools. They also include resources such as robotics, programming, digital games, and activities that do not rely exclusively on sophisticated technologies [26], [27], [28]. Recent literature reports promising results. Paucar-Curasma et al. [2] found connections between computational thinking and the stages of problem-solving among Peruvian university students. Urquizo et al. [29] used play as a mediating tool, achieving positive outcomes. Caballero-González and García-Valcárcel [30] demonstrated the impact of robotics at the primary school level. Ayuso et al. [31] combined computational thinking and Polya’s method using Scratch in schools in Andalusia. In Colombia, Quiroz-Vallejo et al. [26] documented how these tools support the reformulation and automation of solutions, even under adverse conditions.

Scholarly contributions in this field have reinforced the view of computational thinking as an educational paradigm oriented toward the development of transferable cognitive tools rather than the mere acquisition of technical skills. García-Pe ñalvo [32], for instance, argues that preparing students for digital societies requires moving beyond training them as users of technological tools and instead fostering structured forms of reasoning that support problem solving in evolving technological contexts. Empirical studies further indicate that learning gains in computational thinking are strongly mediated by pedagogical structure and instructional task design, regardless of the specific technologies employed [33], [34].

Despite this growing body of research, there is still a lack of empirical evidence on the applicability of computational thinking in rural settings with limited technological resources. Most studies have focused on urban or well-resourced environments, leaving their feasibility in structurally vulnerable contexts largely unaddressed. This study offers evidence that helps to fill that gap by implementing a pedagogical strategy in a rural Colombian school. The general objective was to evaluate the impact of a computational thinking-based strategy on the development of problem-solving skills among tenth-grade students. More specifically, the study examined whether the dimensions of computational concepts, required tasks, and evaluative capacities showed significant changes before and after the intervention.

Unlike exploratory or descriptive approaches, we adopted a cross-sectional quasi-experimental design, which allowed us to observe the effects of the intervention in real educational settings. Based on the theoretical review, previous work in the field, and the challenges observed at IEFPS, we formulated the following hypotheses (H):

H1: There are significant differences in the levels of understanding of computational concepts, required tasks, and evaluative capacities, as measured by the pre- and post-tests.

H0: There are no significant differences in these dimensions before and after the intervention.

This study contributes to ongoing debates on how to translate conceptual frameworks of computational thinking into formative strategies that are context-sensitive, measurable, and replicable in educational settings marked by structural vulnerability. Its empirical contribution aims to strengthen a nascent agenda of pedagogical innovation with territorial equity in Latin America.

This study followed a quantitative approach grounded in principles such as objective measurement, hypothesis formulation and testing, structured experimentation, and rigorous statistical analysis, as proposed by Jaime-Mirabal and Ladino-Luna [35] and Lim and Luo [36]. Its objective was to assess the impact of a pedagogical strategy based on computational thinking on three key dimensions: computational concepts, required tasks, and evaluative capacities, among upper secondary students in a rural Colombian setting.

A quasi-experimental design with a control group was employed. According to Albarracín-Villamizar et al. [37], this type of design is appropriate when random assignment of participants is not feasible, as it works with pre-existing groups in real educational contexts. Two parallel tenth-grade cohorts were selected, with comparable distributions in age, gender, and academic performance to ensure baseline equivalence. This methodological choice aimed to preserve equal access to the intervention while maintaining the natural learning environment and made it possible to evaluate the pedagogical strategy under authentic school conditions, as recommended by Restrepo-Tamayo et al. [38].

The sample consisted of 44 tenth-grade students aged between 14 and 17, evenly distributed into two groups: 22 in the experimental group and 22 in the control group. The sample corresponds to intact classroom groups in a rural public school and was deliberately defined to examine the pedagogical feasibility and internal coherence of a computational thinking intervention under authentic educational conditions, rather than to support population-level generalization. The experimental group participated in the pedagogical intervention, while the control group continued with their regular academic activities. Both groups were evaluated using a computational thinking test administered before and after the intervention. This instrument, developed by Román-González [10], has demonstrated high internal consistency across educational contexts, with reported Cronbach’s alpha values exceeding 0.80. The complete instrument, including the original Spanish version and an English translation, is provided in Appendix.

The test comprised multiple-choice items designed to evaluate computational concepts such as sequences, loops, conditionals, and functions; required tasks such as sequencing, completion, and debugging; and evaluative capacities such as abstraction, pattern recognition, decomposition, and algorithmic design. Statistical analysis was conducted using SPSS v.25. A repeated-measures analysis of variance (ANOVA) was performed to examine the effects of time (pre- and post-intervention), group (experimental versus control), and their interaction. This technique is well-suited for detecting differential changes attributable to the intervention in quasi-experimental designs with repeated measures.

In addition, the PROCESS macro developed by Hayes [39] was used to explore the potential influence of moderating and mediating variables such as gender, family structure, and academic average. Model 1 (interaction) and Model 4 (indirect effect) were applied. This analytical strategy served to rule out biases resulting from pre-existing sociodemographic or academic differences.

The methodological approach adopted in this study reflects the principles of applied research, aimed at improving educational conditions through interventions that are both context-sensitive and empirically testable, as emphasized by Hidalgo et al. [40] and Garibello et al. [41]. Designed to be replicable and adaptable to similar settings with limited technological and pedagogical resources, it reinforces the intervention’s relevance for rural Latin American contexts characterized by structural barriers to educational innovation.

The study complied with the fundamental principles of the Declaration of Helsinki. Participants’ rights, dignity, and well-being were protected through confidentiality protocols and the implementation of informed consent procedures involving both students and their parents. Ethical approval was granted by the institutional review board of the university where the authors are affiliated, under registration number FCSH1702T1001/COL0023742.

The implementation of the pedagogical strategy began on November 2, 2022, with an initial diagnostic test administered to both the quasi-experimental and control groups. The test used was the Computational Thinking assessment developed by Román-González, structured around the three dimensions outlined in the methodology, providing a clear baseline for designing the intervention.

Both groups consisted of students interested in the subject matter but with no prior experience in programming or algorithmic thinking. Based on the diagnostic results, a sequence of six sessions was developed and conducted between November 9 and 23, 2022. These sessions were grounded in a constructivist approach, emphasizing collaborative work and the use of the MakeCode editor as a pedagogical mediation tool. The intervention was organized around problem-based scenarios that encouraged the practical application of concepts, collective analysis, and the development of functional solutions.

Each session was designed to align with one or more pillars of computational thinking. In the first session, for example, students were introduced to the concepts of direction and basic sequencing through simple logic exercises using visual blocks. They were asked to create a sequence of instructions that triggered a response on a Micro:bit board when a button was pressed, reinforcing the connection between command, execution, and observable output.

Subsequent sessions focused on loops and conditionals. Students completed tasks that required designing routines based on specific criteria, such as displaying a repeating light pattern or performing a conditional count. These activities helped students recognize control structures and use loops strategically, while also encouraging peer discussions on execution logic.

The evaluative capacity dimension was addressed through exercises in decomposition and algorithmic representation. Students tackled open-ended problems that involved identifying patterns, abstracting relevant information, and designing solutions graphically represented using flowcharts. These tasks helped integrate logical reasoning with structured formalization.

To develop the required task dimension, students practiced sequencing and debugging in exercises that included deliberate coding errors. These dynamics encouraged critical thinking about computational logic and provided a way to formatively assess individual and collective progress. The strategy concluded on November 24 with the administration of a post-test using the same diagnostic instrument. This allowed for a comparison of baseline and final performance between the experimental and control groups, the latter having not participated in the intervention.

Across all three assessed dimensions, students in the experimental group showed sustained improvement. Statistical analysis, based on a repeated-measures mixed model, revealed not only significant effects over time but also clear differences between students who participated in the strategy and those who did not. These gains were consistently strong and educationally meaningful, with partial eta-squared values exceeding.81 across all dimensions, indicating a robust impact of the intervention. As shown in Table I, the most pronounced improvements emerged in the computational concepts dimension.

TABLE I Pre/Post-Test Outcomes by Dimension and Group

The experimental group improved from a mean of 0.447 to 0.785, a difference of 0.338 ($\Delta =$ .338, p <.001). In contrast, the control group declined from 0.475 to 0.413 ($\Delta =$ –.062, p <.001). This positive shift in the intervention group was particularly evident in their understanding of basic structures such as sequences, loops, and conditionals. The interaction between time and group was statistically significant and substantial in magnitude (F(1, 42) = 282.96, $\eta ^{2} =$ .871), confirming that the observed changes were attributable to the pedagogical experience rather than the mere passage of time.

In the required tasks dimension, which focused on the execution of logical instructions, sequencing, and debugging, the experimental group rose from 0.448 to 0.781 ($\Delta =$ .333, p <.001), while the control group declined from 0.481 to 0.416 ($\Delta =$ –.065, p <.001). The interaction effect was again significant (F(1, 42) = 295.27, p <.001, $\eta ^{2} =$ .875), underscoring the importance of collaborative work, teacher guidance, and the use of MakeCode as a technological mediator in developing practical computational skills.

Regarding the evaluative capacity dimension, which includes complex processes such as abstraction, decomposition, and algorithmic design, the experimental group improved from 0.442 to 0.766 ($\Delta =$ .324, p <.001), while the control group’s performance dropped from 0.470 to 0.399 ($\Delta =$ –.071, p <.001). The interaction between time and group (F(1, 42) = 186.40, p <.001, $\eta ^{2} =$ .816), along with the high variance explained in the experimental group ($\eta ^{2} =$ .856), points to a significant impact of the intervention beyond simple exposure to content or the passage of time.

Finally, the study explored whether the observed improvements were influenced by sociodemographic or academic characteristics using Hayes’s PROCESS macro (models 1 and 4). The results were conclusive: no significant moderation or mediation effects were found. In every case, R²-change values were marginal (between 0.0000 and 0.0038), F values ranged from 0.0036 to 0.1592, and all p-values exceeded.69. Confidence intervals included zero across all comparisons (e.g., gender in computational concepts: 95% CI = [–.1140,.1210]; academic average: 95% CI = [–.0156,.0252]). These findings reinforce the robustness and equity of the strategy’s educational impact, showing no bias based on gender, family structure, or prior academic performance.

The results confirm both hypotheses and demonstrate the formative impact of a context-sensitive intervention grounded in clear pedagogical principles. Observed effect magnitudes should therefore be interpreted in light of the study’s intensive and short-term design, the relative homogeneity of the participant groups, and the theoretical alignment between the pedagogical intervention and the assessment framework employed. Under these methodological conditions, large within-group effect sizes are plausible in educational research focused on instructional feasibility and internal coherence rather than population-level inference.

The observed progression across the three evaluated dimensions indicates that integrating computational thinking through accessible tools such as MakeCode, within collaborative dynamics, offers an effective path to enriching education in rural settings with structural limitations. This experience provides replicable evidence for future curricular initiatives aimed at incorporating computational thinking with attention to equity, pedagogical sustainability, and local relevance in educational environments marked by high structural vulnerability.

The findings support the alternative hypothesis that the pedagogical strategy based on computational thinking generated significant improvements in the evaluated dimensions. This not only validates the formative potential of the approach but also helps delineate more precisely the kind of impact it can achieve when implemented under genuine school conditions and with limited resources. The most notable progress occurred in the evaluative capacity dimension, which involved complex tasks such as abstracting information, identifying patterns, breaking down problems, and representing structured solutions. This improvement reinforces the view that computational thinking, conceived as a specific form of structured reasoning, can be translated into practical problem-solving skills, even in rural contexts with limited technological resources.

Unlike previous studies conducted in urban settings with stable access to advanced technologies [30], [31], this research offers evidence of the feasibility of implementing computational thinking in conditions of high structural vulnerability. Rather than aiming at population-level generalization, the present study was designed to examine the pedagogical feasibility and internal coherence of a computational thinking intervention under authentic rural school conditions. Whereas those studies focused on Scratch or robotics applications within consolidated infrastructure, this study employed a more accessible tool like MakeCode, integrated into a pedagogical sequence intentionally designed to be sustainable and responsive to the local context.

This distinction is not only methodological but also epistemological, shifting the focus from the tool itself to the cognitive practices that enable students to construct functional solutions. Such a shift foregrounds the role of instructional structure and task sequencing in shaping students’ problem-solving pathways and computational thinking practices, an interpretation that resonates with empirical accounts emphasizing pedagogy over technological sophistication [32], [33], [34].

In this same vein, the work of George-Reyes et al. [5], which stresses the need to adapt theoretical frameworks to real contexts, finds a concrete application here. This study advances in that direction by transforming the conceptual framework proposed by authors such as Wing [7], [8] and Seehorn et al. [6] into a measurable intervention that articulates computational concepts, problem-solving practices, and evaluative capacities in non-idealized scenarios. The gains observed across all three dimensions indicate that significant learning can occur without relying exclusively on sophisticated devices or optimal connectivity.

Another distinctive feature lies in the analytical structure employed. While studies such as those by Paucar-Curasma et al. [2] and Urquizo et al. [29] focused on general improvements in problem-solving, this research incorporated specific dimensions of computational thinking using a robust statistical model capable of isolating the effects of time and group. This analytical providen strengthens the reliability of the results and provides replicable evidence for the pedagogical efficacy of this approach. Furthermore, the absence of moderating or mediating effects from variables such as gender, family structure, or prior academic performance underscores the equity of the strategy, demonstrating that its impact is rooted in the learning experience rather than in students’ sociodemographic background.

Theoretically, the findings align with contributions from Román-González [10] and Barboza et al. [11], who argue for assessing computational thinking through a structure that accounts not only for the taught concepts but also for the cognitive tasks involved and the capacities being developed. This approach moves beyond instrumental or technical perspectives that reduce computational thinking to mere coding and instead positions it as a transversal competence applicable to open-ended problems, as articulated in Brennan and Resnick’s three-part model [9].

The central contribution of this study lies in showing that complex conceptual frameworks can be translated into actionable teaching strategies in challenging environments. This helps to bridge the gap between theory and implementation while opening new possibilities for incorporating computational thinking into the educational agenda of regions historically excluded from pedagogical innovation. The evidence affirms that the integration of instructional design, accessible technological mediation, and collaborative work can generate meaningful learning conditions without reproducing imported models that fail to reflect local realities.

This study confirmed that a pedagogical strategy based on computational thinking can significantly and equitably influence the development of problem-solving skills, even in rural settings with limited resources. The consistent improvement across the three evaluated dimensions, without mediation by sociodemographic or academic variables, demonstrates that the quality of learning does not rely exclusively on advanced technological infrastructure, but rather on intentional, structured, and contextual coherent pedagogical mediation.

Beyond the statistical effects, the intervention showed that computational thinking can be integrated as a cross-cutting competency without reducing it to coding or the use of sophisticated devices. The approach, grounded in problem-based scenarios, collaborative work, and accessible tools like MakeCode, enabled students to strengthen their abilities in analysis, abstraction, and solution design, leading to meaningful and applicable learning.

This finding is particularly relevant for educational institutions facing structural constraints, as it demonstrates that pedagogical innovation is possible without depending on costly technologies or ideal conditions. The experience developed in this study offers concrete insights for future curricular integrations that recognize computational thinking not as an isolated content area, but as a tool to enhance complex cognitive processes from the early stages of formal education.

Among the limitations of this research is the inability to establish strict causal relationships, due to the quasi-experimental design. The small sample size and absence of longitudinal follow-up also constrain the generalizability of the findings. Accordingly, the magnitude of the observed effects should be interpreted as evidence of short-term pedagogical impact under authentic school conditions rather than as proof of sustained, causal, or population-level effects. Nevertheless, the consistency of the results and the magnitude of the observed effects provide a strong foundation for future studies and practical applications.

It is worth exploring, in the medium term, the sustainability of these learning gains over time and their potential transferability to other subject areas. It will also be important to investigate how this strategy can be adapted to different educational levels and how teacher training can be strengthened to broaden its impact. These lines of inquiry can help consolidate the value of computational thinking as a formative resource, particularly in contexts where guaranteeing the right to quality education with territorial equity is still an urgent priority.

Appendix

Computational Thinking Assessment Instrument (Román-González, 2016)

SECTION A.

Original Spanish Version

Test de Pensamiento Computacional (Versión 2.0) de Román-González (2016) aplicado como Prestes y Postest en la Institución Educativa Francisco de Paula Santander.

Show All

Show All

Show All

Show All

Show All

Instrucciones para los estudiantes: Este cuestionario no es un examen escolar y no afectará sus calificaciones. Su propósito es comprender cómo razonan al resolver problemas lógicos y de programación. Lean cada pregunta con atención y seleccionen la opción que mejor represente su razonamiento. Trabajen individualmente y no consulten a sus compa ñeros.”

SECTION B.

English Translated Version

Computational Thinking Test (Version 2.0) by Román-González (2016), administered as a pretest and posttest at the Institución Educativa Francisco de Paula Santander.

Instructions for students:

This questionnaire is not a school exam and will not affect your grades. Its purpose is to understand how you reason when solving logical and programming problems. Read each question carefully and select the option that best represents your reasoning. Work individually and do not consult your classmates

Show All

Show All

Show All

Show All

Show All