Onlife Experiential Learning Model Centered on Artificial Intelligence: Association With Computational Thinking and Academic Satisfaction

Abstract

Spanish

This study examines the Onlife experiential learning model from a socioformative perspective, incorporating artificial intelligence (AI) to assess its impact on higher education teaching. The model is implemented as a hybrid experience that merges online and offline. The main objective was to analyze the practical implementation of the Onlife model at a Peruvian university by comparing empirical perceptions of students (n = 1,718) and faculty (n = 501), and to determine which aspects of the model are associated with both academic satisfaction and computational thinking, including possible gender differences. A survey design with specific questionnaires was used, and a four-dimensional structure was validated with strong reliability and validity. Results indicate that all four components of the model (Utility, Knowledge Construction, Innovation, and Accomplishment) contribute significantly to student academic satisfaction and to strengthening computational thinking. Faculty reported higher scores in Knowledge Construction and Utility, emphasizing the applicability of the model to their satisfaction, while students valued all components more evenly. Although female students showed slightly lower average satisfaction than male students, no substantial gender differences were found. These findings confirm that an Onlife experiential model supported by socioformation enhances the educational experience, increasing academic satisfaction and computational thinking competencies among students and faculty.

Plain Language Summary

Modelo de Aprendizaje Experiencial Onlife centrado en la inteligencia artificial: Asociación con el pensamiento computacional y la satisfacción académica

Este estudio aborda el modelo de aprendizaje experiencial Onlife desde la socioformación y el uso de inteligencia artificial, evaluando su impacto en la enseñanza universitaria, como experiencia híbrida que fusiona lo online-offline. El objetivo principal fue analizar la implementación práctica del modelo Onlife en una universidad peruana, comparando las valoraciones empíricas de estudiantes (n = 1718) y docentes (n = 501), y determinar qué factores del modelo se asocian con la satisfacción académica y el pensamiento computacional, incluyendo posibles diferencias de género. Mediante un diseño de encuesta con cuestionarios específicos, se validó una estructura de cuatro dimensiones con fiabilidad y validez sólidas. Los resultados indican que los cuatro componentes del modelo (utilidad, saberes, innovación y logro) contribuyen significativamente a la satisfacción académica del estudiante y al fortalecimiento del pensamiento computacional. Se halló que los docentes perciben niveles más altos en saberes y utilidad, enfocando su satisfacción en la aplicabilidad del modelo, mientras que los estudiantes valoran de forma más equilibrada todos los componentes. Asimismo, aunque las estudiantes mostraron ligeramente menor satisfacción promedio que los varones, no se registraron diferencias sustanciales por géneros. Se confirma que el modelo experiencial Onlife apoyado en la socioformación mejora la experiencia educativa, elevando la satisfacción académica y las competencias en pensamiento computacional tanto en estudiantes como en docents.

Keywords

onlife model experiential learning computational thinking academic satisfaction socioformation

Introduction

The Onlife reality represents a fundamental transformation in contemporary human experience, where digital and physical spheres converge in an indivisible continuum that redefines everyday existence (Floridi, 2015). This phenomenon transcends the mere digitization of processes and establishes a new ontology where the online and offline merge into a permanent hybrid experience. The relevance of this reality is evident in the accelerated adoption of immersive technologies, AI, and extended reality systems that permeate all aspects of social, work, and educational life (Latour, 2023). Recent projections suggest that by 2030, 90% of educational interactions will occur in hybrid spaces where the distinction between in-person and virtual will become obsolete, positioning the Onlife model as the dominant paradigm of higher education (Lee et al., 2024).

The Onlife model has been predominantly explained through connectivist principles, a theory that conceptualizes learning as a network formation process where knowledge resides in both human and non-human connections, significantly enriched by AI (Shang & Bai, 2025; Siemens, 2005). Connectivism provides the theoretical scaffolding for understanding learning in this hybrid reality, postulating that the ability to form connections among information sources is more critical than static, individually stored knowledge. However, connectivism presents significant limitations as an abstract theory. It lacks specificity on how learning should be structured between humans and machines in an AI context, focusing overly on the learning process without providing concrete guidance for pedagogical practice (Cabero-Almenara & Llorente-Cejudo, 2015). Additionally, connectivism does not articulate a shared vision or direction for education, emphasizing method while overlooking the ethical and social implications of networked learning (Holmes et al., 2019).

The socioformation pedagogical model emerges as a necessary metatheory to address experiential learning in the Onlife model, providing a solid shared vision that transcends academic matters and integrates all substantive university functions: research, community engagement, and institutional management (Tobon & Luna-Nemecio, 2021). It does not reject connectivism but rather articulates it alongside other theories into a higher-order metatheory such as socioformation, endowing universities with a clear ethical purpose to drive transformative actions in communities. Socioformation conceptualizes AI as a fundamental platform for high-impact social decision-making, promoting collective work and territorial linkage as the system base for generating meaningful transformations (Tobón & Lozano-Salmorán, 2024). This integration overcomes connectivism’s limitations by providing an ethical-pedagogical framework that guides the use of emerging technologies toward sustainable social development.

In particular, it is unknown which dimensions of the socioformation-based onlife teaching model (e.g., perceived utility, knowledge construction, applied innovation, and perception of accomplishment) most strongly explain academic satisfaction and how they relate to the development of computational thinking. Likewise, there is limited comparative evidence on the stability of these relationships between teachers and students, as well as on possible gender-related differences. In response to these gaps, this study analyzes the perception of the Onlife model in a university center, compares this perception between students and teachers, identifies the factors of the model associated with academic satisfaction in both groups, examines its relationship with the development of computational thinking, and evaluates differences according to gender.

This study contributes to strengthening the empirical basis of the Onlife–socioformative approach by offering useful analytical criteria to guide decisions on instructional design and institutional management of AI-mediated educational innovation in higher education.

Theoretical Framework

Socioformation presents itself as an alternative pedagogical metatheory that could guide universities through the transformations necessary to address the challenges posed by artificial intelligence (AI) and contribute to sustainable social development (Tobon & Luna-Nemecio, 2021). Unlike other pedagogical models such as socioconstructivism or situated learning, which focus predominantly on individual cognitive processes and meaning construction in context (Ausubel, 1976; Vygotsky, 1978), socioformation shifts the focus from learning to the holistic formation of individuals committed to territorial transformation. This perspective is particularly relevant in the current era, where traditionally valued technical competencies are increasingly automatable by AI systems (Larson et al., 2024), necessitating a reorientation toward uniquely human capabilities.

In this context, socioformation proposes to develop professionals capable of governing AI through metacompetencies that transcend technical skills replicable by algorithms. Among these metacompetencies are: (a) sociocritical thinking, understood as the capacity to dialectically analyze reality to reveal the structural causes of problems and orient action toward social transformation (Golden, 2025; Núñez-López et al., 2017); (b) interdisciplinary leadership for territorial transformation, which involves formulating and executing complex projects that integrate diverse bodies of knowledge toward sustainable social development (Mayboroda et al., 2025); (c) an ethical life project, conceived as a process of self-actualization achievable only through professional action oriented toward the common good (Varghese et al., 2025); and (d) metacognitive AI literacy, representing the capacity to exercise conscious, reflective, and ethical control over AI systems—extending thought through the machine without delegating moral direction (Rapanta et al., 2025).

Socioformation emerges as a developing alternative to the limitations of widely used pedagogical models. Socioconstructivism, derived from Vygotsky (1978), while incorporating the relevance of social interaction and cultural mediation for learning, remains centered on learning as an end in itself without necessarily demanding context transformation or ethically responsible action (Ainjärv & Laas, 2024). Similarly, situated learning uses context as a means to reinforce internal meanings but without prioritizing collective transformation or addressing the structural causes of problems (Golden, 2025). Connectivism, proposed by Siemens (2005), although explaining learning as the establishment of connections in digital networks, fails to clarify for what purpose these connections should be used or what impact they should generate beyond cognitive optimization. These perspectives prove insufficient for the AI era, which requires developing professionals capable of ethically directing technology and preventing risks such as purposeless automation, intellectual passivity, and uncritical submission to algorithmic systems (Larson et al., 2024; Mayboroda et al., 2025).

The Onlife experiential learning model analyzed in this study represents an operationalization of the theoretical principles of hybrid education, enriched by socioformation as an ethical-pedagogical framework. Its structure comprises four interdependent components: (a) Utility, which establishes the socioformative contextual problem that will guide the learning process, connecting challenges with real community needs; (b) Knowledge Construction, where knowledge is collaboratively built by integrating human and technological sources, ethically directed toward sustainable social development; (c) Innovation, which transcends mere technical application to focus on transforming specific territorial realities through original, contextualized solutions; and (d) Accomplishment, which reconceptualizes traditional evaluation toward the assessment of the social impact of learning (Floridi, 2015; Schlemmer et al., 2020). The integration of socioformation into this model provides a clear purpose oriented toward contributing to sustainable social development (Tobón & Luna-Nemecio, 2021) and developing metacompetencies that enable students and faculty to act as agents of territorial change.

Concept Clarification

Computational thinking. From a traditional perspective, computational thinking is defined as the ability to solve problems, design systems, and understand human behavior by applying fundamental concepts from computer science (Wing, 2006). This conception emphasizes cognitive skills such as problem decomposition, pattern recognition, and algorithmic design. However, from a socioformative standpoint, computational thinking is reconceptualized as part of a broader metacognitive literacy that integrates technical capacity with ethical judgment and orientation toward the common good (Rapanta et al., 2025). It is not merely about operating effectively in hybrid digital environments but doing so with critical awareness of the social implications of technology and with the capacity to direct AI systems toward purposes of territorial transformation (Varghese et al., 2025).

Innovation. Traditionally, innovation is associated with the incorporation of new technologies or the instrumental updating of processes, frequently reduced to cosmetic or operational changes (Pastrana et al., 2025). From a socioformative perspective, the metacompetency of continuous and deep innovation fundamentally differs from this superficial notion. Its essence lies in the rigorous pursuit of solutions that address territorial problems at their root, generating transformations that are sustainable over time. It is a recursive, dialogical, and ongoing process where each advance involves critical reflection, systemic feedback, and strategy reevaluation (Morin, 1999). Innovation is not validated by its technical spectacularity but by its demonstrated capacity to maximize impact on sustainable social development (Tobón & Luna-Nemecio, 2021).

Academic satisfaction. In conventional approaches, academic satisfaction is predominantly measured as the individual student’s perception of the quality of educational services received, focusing on aspects such as course organization and student-instructor relationships (Testa et al., 2023). From a socioformative perspective, satisfaction transcends this individualistic view to become linked with the perceived impact and relevance of the educational experience in relation to real territorial problems. Thus, satisfaction is associated with the educational model’s capacity to connect learning with authentic community needs, generate meaningful transformations, and develop metacompetencies that enable students to act as agents of social change (Pastrana et al., 2025).

Utility

In traditional pedagogical models, utility is understood as the perceived relevance of content for future professional or academic applications. Socioformation expands this notion by conceptualizing utility as the direct connection between learning challenges and real territorial problems requiring transdisciplinary solutions (Golden, 2025). From this perspective, classroom challenges are not abstract; rather, they seek to identify authentic territorial problems that demand the integration of diverse knowledge and the application of sociocritical thinking for effective resolution.

Knowledge Construction

While socioconstructivism conceives knowledge construction as a fundamentally social process mediated by interaction among peers and experts (Vygotsky, 1978), socioformation adds critical dimensions to this process. Knowledge construction integrates human and non-human sources (including AI systems), but always under ethical direction oriented toward sustainable social development. This component involves promoting collaborative work, using up-to-date scientific information, and creating educational products that contribute to territorial transformation—transcending mere individual knowledge acquisition.

Accomplishment

Conventionally, academic accomplishment is measured through individual performance indicators such as grades and submitted products. Socioformation reconceptualizes accomplishment toward the assessment of the social impact of learning. Not only are academic products evaluated, but also the real or potential contribution to improving community living conditions (Pastrana et al., 2025). AI facilitates multidimensional analysis of this impact, identifying systemic connections and unanticipated effects of proposed interventions (Tiukhova et al., 2024), but always under human ethical and metacognitive supervision.

Method

A survey study design was conducted using several scales organized into an ad-hoc questionnaire for this study. The sample consisted of 1,718 university students and 501 university professors, all from a private Peruvian university (Table 1). Participants were recruited using a non-probabilistic convenience sampling strategy from a private Peruvian university. Inclusion criteria for students were enrollment in undergraduate or graduate programs implementing the Onlife model, and voluntary participation with informed consent. Faculty participants were required to be actively teaching under the Onlife framework during the data collection period. Responses with missing data exceeding 10% were excluded. Given the single-institution context, findings should be interpreted as context-specific and not representative of higher education systems in Latin America as a whole. In any case, the university’s broad reach allowed for students from most regions of the Peru.

Table 1.

Description of Samples.

Characteristic	Students (n = 1,718)	Teachers (n = 501)
Gender:
Male	51.6%	58.5%
Female	48.4%	41.5%
Age:
Mean (SE)	26.4 (0.347)	50.4 (0.626)
Median	22	50
Variance	107	115
Minimum	17	26
Maximum	66	75
Academic degree:
Bachelor’s degree (Licenciatura)	1.0%	0.6%
Licentiate (Licenciado)	2.2%	1.8%
Master’s degree (Maestría)	40.5%	29.9%
Doctorate (Doctorado)	14.8%	9.2%
Program:
Undergraduate	39.9%	53.3%
Graduate (postgraduate)	11.8%	5.6%

Note. SE = Standard Error of the mean.

The questionnaire was distributed via the university’s institutional platform, ensuring that each participant (student or faculty) responded to the appropriate version. The study was approved by the ethics committee of the participating university. In all cases, digital informed consent was obtained from participants, ensuring voluntary participation, anonymity, and confidentiality. Personal data was processed in accordance with the Data Protection Act. Participants were informed about the research purpose, their right to withdraw without consequence, and that collected information would be used solely for academic purposes. All data were collected in the fourth quarter of 2024.

The following specific scales were used to measure the variables of interest and to collect general information:

Onlife Model Scale: This instrument consists of 17 items distributed across four dimensions. It is evaluated with a 4-point ordinal appreciation scale (similar to Likert) (1 = Rarely or almost never; 2 = Sometimes; 3 = Almost always; 4 = Always). The four dimensions correspond to the theoretical model components: (1) Utility: relevance of the challenges or problems posed in class, connection with prior knowledge and learning expectations, appropriateness of materials and class evidence; (2) Knowledge Construction: explanation of content, promotion of collaborative work, use of current scientific information (e.g., Scopus articles), use of multimedia presentations and creation of educational videos; (3) Innovation: incorporation of innovative evidence in class, application of topics to real situations, holding plenary sessions and integrative discussions; and (4) Accomplishment: assessment of learning evidence, verification of learning outcomes accomplishment, use of evaluation rubrics, and presentation of the next learning challenge. The Onlife Model Scale was developed based on a deductive process grounded in the theoretical principles of Onlife education and socioformative pedagogy. Item content was reviewed by three experts in educational technology and higher education pedagogy to ensure content relevance and clarity.

Academic Satisfaction Scale for Students: This instrument consists of three items aimed at evaluating the level of satisfaction with the implementation of the Onlife model at the university. The three items are: (1) the teaching by faculty; (2) the technology used in classes; and (3) the overall development of the classes. Responses used a 5-point ordinal scale (1 = “very dissatisfied” to 5 = “very satisfied”).

Academic Satisfaction Scale for Faculty: This scale also contains three items to gauge satisfaction with class development and digital technology use. Items are: (1) class development; (2) technology used in teaching; and (3) satisfaction with being a faculty member under the Onlife model. As with the previous scale, responses used a 5-point scale (1 = “very dissatisfied” to 5 = “very satisfied”).

Computational Thinking Self-Assessment Item: It was assessed using a single self-report item capturing participants’ perceived ability to operate effectively in hybrid digital learning environments. Although computational thinking is a multidimensional construct, single-item measures of perceived competence have been previously used in large-scale educational research. Nevertheless, this approach limits construct precision and should be interpreted as an indicator of perceived, rather than objective, computational thinking.. The question was: “I consider that I have a high level of computational thinking ability to function in hybrid educational environments.” Responses used a 5-point ordered scale (1 = Low, 5 = High).

Additionally, general questions on gender, age, degree being pursued or program taught were included depending on the sample.

Data Analysis

First, a confirmatory factor analysis (CFA) was conducted to evaluate the fit of the theoretical model composed of four latent dimensions (Utility, Knowledge Construction, Innovation, and Accomplishment). Given the ordinal nature of the variables, the diagonally weighted least squares (DWLS) estimator was used, which is appropriate for categorical data (Li, 2016; Rhemtulla et al., 2012).

Convergent validity was assessed using Average Variance Extracted (AVE) and Composite Reliability (CR). Following the criteria proposed by Fornell and Larcker (1981), AVE values equal to or greater than .50 indicate that a construct explains more than half of the variance of its indicators. Composite reliability values above .70 are considered acceptable, with values above .80 reflecting good internal consistency (Hair et al., 2019). Although these thresholds are commonly accepted guidelines rather than strict cutoffs, the obtained AVE and CR values provide robust evidence of convergent validity. In any case, all dimensions met these criteria in this study, supporting adequate convergent validity,

Measurement invariance was assessed by analyzing configural, metric, scalar, and strict invariance using the robust weighted least squares estimator (WLSMV). The model fit was assessed sequentially following Chen’s (2007) criteria, which suggest that changes in CFI ≤ .01 and RMSEA ≤ .015 between nested models indicate invariance.

To construct a composite measure of satisfaction, the three recorded satisfaction items (teaching, technology, class development) were combined. It was assumed that these items reflect a common latent construct of general satisfaction. Analysis proceeded in three steps. First, bivariate correlations were examined to assess the strength and direction of associations among the satisfaction indicators. Second, unidimensionality was evaluated using McDonald’s omega (ω), considering ω ≥ .70 acceptable (McDonald, 1999; Revelle, 2023; Revelle & Zinbarg, 2009). Finally, a principal component analysis (PCA) was performed to extract the first component capturing shared variance among the indicators.

Subsequently, a linear regression was conducted to analyze the contribution of the Utility, Knowledge Construction, Innovation, and Accomplishment dimensions to overall academic satisfaction. Gender was included as a dummy predictor (male = 0, female = 1).

For computational thinking, since it is ordinal, an ordinal logistic regression with a proportional odds model was used, with computational thinking levels as the dependent variable and the same predictors (the four dimensions and gender).

Although the Onlife dimensions were validated using confirmatory factor analysis as latent constructs, subsequent regression analyses were conducted using factor scores to enhance parsimony and interpretability. This analytic strategy allowed for separate examination of predictive relationships for students and faculty while maintaining manageable model complexity. Nevertheless, future research could benefit from integrated structural equation modeling approaches that simultaneously estimate measurement and structural components.

On the other hand, prior to conducting independent-samples t-tests, assumptions of normality and homogeneity of variances were examined. Given the large sample size, t-tests were considered robust to minor deviations from normality. Homogeneity of variances was assessed using Levene’s test, and Welch corrections were applied when necessary.

All analyses were performed in R (R Core Team, 2025). The lavaan package (v0.6-19) was used for regression analyses (Rosseel, 2012).

Results

Students

A confirmatory factor analysis (CFA) was performed to test a theoretical four-factor latent model (Utility, Knowledge Construction, Innovation, and Accomplishment) using the DWLS estimator. The model fit the data excellently, with indices within recommended ranges: CFI = .992, TLI = .990, SRMR = .017, and RMSEA = .072. The robust RMSEA was .079, very close to the acceptable threshold. The model significantly outperformed the null model (χ²(84) = 769.39, p < .001, scaled).

Standardized factor loadings of items on their respective factors were all high and significant (p < .001), ranging from .83 to .96, indicating a strong relationship between indicators and their latent constructs. The highest loadings were observed in items of the Accomplishment dimension (e.g., Item 16 = .96, Item 15 = .95) and Innovation (Item 13 = .94), suggesting a good representation of these factors. Additionally, high significant latent correlations were observed among factors (all > .81), indicating the four constructs are closely related yet conceptually distinguishable.

Given this, the plausibility of a general factor model was tested. A second-order model using DWLS was fit, yielding an excellent fit: CFI = .990, TLI = .988, SRMR = .022, and robust RMSEA = .082. Although RMSEA reached the upper limit of acceptability, other indices support adequacy (χ²(86) = 957.70, p < .001, scaled correction). Standardized loadings of first-order dimensions on the second-order factor were high and significant (Utility = .90, Knowledge Construction = .94, Innovation = .94, Accomplishment = .94), indicating that the general factor solidly explains the shared variance among dimensions. These results suggest that it is reasonable to consider a single overarching latent construct underlying the four student data dimensions.

Faculty

The same strategy was applied to evaluate the four-dimension model with the faculty dataset. The model showed a very good overall fit (CFI = .995, TLI = .994, SRMR = .056, RMSEA = .049). The robust RMSEA was higher (0.111), but the corrected CFI (0.907) and TLI (0.883) remained acceptable given ordinal estimation. Overall, results indicate the model adequately represents the four-factor structure.

Standardized loadings of items on their factors were high and significant for all cases (p < .001), ranging from .62 to .93. The highest loadings were found in Accomplishment and Innovation, while the lowest were in Knowledge Construction, though all surpassed the .60 threshold.

Latent correlations between factors were significant and relatively high (ranging from .77 to .88), suggesting that dimensions share common variance that could be represented by a general factor. A second-order CFA using DWLS yielded acceptable fit indices (CFI = .976, TLI = .970, robust RMSEA = .113, SRMR = .059). All standardized item loadings were significant (p < .001, range .619–.928), and first-order factors loaded strongly on the second-order factor (Utility = .892, Knowledge = .892, Innovation = .928, Accomplishment = .904).

These results support the validity of a hierarchical four-dimension structure centered on an overarching construct that may relate to faculty attitudes toward pedagogical improvement and innovation.

Measurement Invariance

A multigroup CFA for second-order models examined measurement equivalence of the instrument between faculty and students. The configural model showed good fit (CFI = .999, RMSEA = .017, SRMR = .042), indicating the underlying factor structure is similar across groups. When equality of factor loadings was imposed (metric model), fit remained adequate (CFI = .998, RMSEA = .022, SRMR = .054) with changes from the configural model (ΔCFI = −.001, ΔRMSEA = +.004) within acceptable limits, indicating metric invariance. Subsequently, when intercepts were constrained (scalar model), fit stayed acceptable (CFI = .999, RMSEA = .018, SRMR = .043), and changes (ΔCFI = −.001, ΔRMSEA = +.004) met criteria, providing evidence of scalar invariance.

Overall, scalar invariance holds, meaning the scales can be considered structurally equivalent for the student and faculty samples.

Perception of the Model: Faculty Versus Students

With scalar invariance established, latent means were compared between groups using students as the reference group (mean fixed at 0). Results (Table 2) show that faculty have significantly higher means in Knowledge Construction (β = 0.565, p = .005), Innovation (β = 1.204, p < .001), and Accomplishment (β = 1.505, p < .001) compared to students. No significant difference was found in the Utility dimension.

Table 2.

Comparison of Latent Means Between Faculty and Students (Students as Reference Group).

Dimension	Faculty mean (β)	SE	Z	p	95% CI	Significant difference
Utility	.327	0.264	1.239	.215	[–0.190, 0.845]	No
Knowledge	.565	0.199	2.838	.005**	[0.175, 0.955]	Yes
Innovation	1.204	0.249	4.841	<.001***	[0.717, 1.692]	Yes
Accomplishment	1.505	0.245	6.146	<.001***	[1.025, 1.984]	Yes

Note. Students’ latent means fixed at 0 (reference group). SE = standard error; CI = confidence interval.

p < .01; ***p < .001.

Gender Perspective: Comparison within Groups

Students

To construct a composite satisfaction index for students, the three ordinal satisfaction items (teacher effectiveness, classroom technology, class development) were analyzed. Spearman correlations among items ranged from ρ = .73 to .80, indicating strong associations. McDonald’s total omega (ω = .94) and hierarchical omega (ω_h = .94) demonstrated high internal consistency and unidimensionality. A PCA showed that one component explained 89% of the total variance, with loadings above .93 on all items. This justified synthesizing the original items into a single general satisfaction variable.

Using this composite satisfaction and the four model factors, descriptive and gender comparisons were conducted. Female students scored higher than male students on all factors of the model and on general satisfaction (Table 3).

Table 3.

Student Descriptives by Gender.

Variable	Females (M ± SD)	Males (M ± SD)	t	Cohen’s d
Utility	3.02 ± 0.79	2.90 ± 0.79	3.15**	0.15
Knowledge	3.11 ± 0.73	2.98 ± 0.75	3.62***	0.18
Innovation	3.03 ± 0.83	2.93 ± 0.80	2.64**	0.13
Accomplishment	3.11 ± 0.80	2.98 ± 0.79	3.45***	0.17
Global satisfaction	4.00 ± 0.80	3.96 ± 0.85	0.97	0.05

Note. M = mean; SD = standard deviation; t = t-test for independent sample.

p < .01; *** p < .001.

Independent t-tests showed significant differences between female and male students on Utility (t(1716) = 3.15, p = .002), Knowledge (t(1,696) = 3.62, p < .001), Innovation (t(1683) = 2.64, p = .008), and Accomplishment (t(1,687) = 3.45, p < .001). No significant difference was found in overall satisfaction (t(1,716) = 0.97, p = .334), as shown in Table 3. Although several gender differences reached statistical significance, effect sizes were small (Cohen’s d < .20), suggesting limited practical relevance.

Faculty

A similar composite satisfaction index was constructed for faculty using the three satisfaction items (being a faculty member, class development, technology satisfaction). Spearman correlations were moderate (.47 to .57), but McDonald’s total omega on the polychoric matrix was .83, indicating good consistency. PCA on the polychoric matrix showed one component explaining 75% of variance with high loadings (≥.85). Thus, a single latent satisfaction dimension was justified.

Given minimal differences in all analyzed variables and no significant t-test differences (all p > .14), we conclude that male and female faculty perceptions of the model are highly homogeneous (see Table 4 for descriptives).

Table 4.

Faculty Descriptives by Gender (no significant differences).

Variable	Females (M ± SD)	Males (M ± SD)
Utility	3.65 ± 0.48	3.58 ± 0.54
Knowledge	3.31 ± 0.62	3.31 ± 0.64
Innovation	3.57 ± 0.53	3.53 ± 0.59
Accomplishment	3.63 ± 0.49	3.60 ± 0.52
Satisfaction	3.17 ± 0.78	3.24 ± 0.78

Note. Means and standard deviations are shown. t-test results are omitted as they were all non-significant.

Student Satisfaction Model

A linear regression identified which model dimensions are significantly associated with students’ academic satisfaction. Gender (1 = female, 0 = male) was included given prior gender differences. Variance inflation factors (VIFs) indicated acceptable multicollinearity (all VIF < 3.7). The model was significant (F(5, 1,712) = 321.80, p < .001), explaining approximately 49% of the variance in overall satisfaction.

All model dimensions significantly predicted overall satisfaction: Innovation (standardized β = .23, p < .001), Accomplishment (β = .20, p < .001), Utility (β = .19, p < .001), and Knowledge (β = .15, p < .001). Gender had a small but significant effect (β = –.04, p = .021), indicating that, controlling other factors, female students reported slightly lower satisfaction than male students.

Faculty Satisfaction Model

A linear regression was also estimated for faculty academic satisfaction with the same predictors (including gender coded 1 = female, 0 = male), though gender differences in satisfaction were not previously found. Multicollinearity was not an issue (all VIF < 2.1). The model was significant (F(5, 495) = 10.21, p < .001), but explained only a modest 9% of satisfaction variance (R² = .09).

Among predictors, only Utility was significantly and positively related to faculty satisfaction (standardized β = .21, p < .001). Other variables did not reach significance: Knowledge showed a marginal trend (β = .10, p = .063); Innovation, Accomplishment, and gender had p-values above .17.

Student Computational Thinking Model

An ordinal logistic regression examined which model factors predicted students’ perceived computational thinking levels. The dependent variable was the self-rated digital competence integrated with computational thinking, categorized into four ordered levels (low, fair, medium-high, high). Predictors were the four model dimensions and gender (0 = male, 1 = female).

The model was significant (Deviance = 3,506.37, AIC = 3,522.37). Estimated coefficients indicated that all four factors were positively and significantly associated with a higher probability of students scoring higher on perceived computational thinking: Utility (β = 0.42, p < .001), Knowledge (β = 0.49, p < .001), Innovation (β = 0.41, p < .001), and Accomplishment (β = 0.31, p < .01).

Gender did not have a conventional significant effect (β = –0.17, p = .067), suggesting that although female students could have a slightly lower probability of being in the highest levels of perceived digital competence, this effect was not statistically significant (p < .05).

Faculty Computational Thinking Model

Similarly, an ordinal logistic regression was conducted for faculty, with perceived digital competence as the dependent variable and the same predictors. The model fit was reasonable (Deviance = 875.99, AIC = 891.99). Only two factors were significantly related to higher levels of perceived digital competence: Utility (β = 0.48, p = .036) and Knowledge (β = 0.82, p < .001).

Innovation, Accomplishment, and gender were not significant (all p > .05). In particular, gender (0 = male, 1 = female) did not predict significant differences in faculty self-perceived digital competence (p = .235). These results indicate that faculty who perceive higher Utility and Knowledge related to technology use are more likely to exhibit higher levels of digital competence integrated with computational thinking.

Discussion

This study addressed a critical gap in the literature regarding the empirical validation of experiential learning models in hybrid educational contexts enhanced by artificial intelligence. Previous research has predominantly focused on either purely online or face-to-face modalities, leaving underexplored the pedagogical dynamics of Onlife environments where digital and physical spheres converge into a continuous learning experience (Floridi, 2021; Schlemmer & Di Felice, 2023). The findings provide robust empirical evidence that the Onlife experiential learning model, grounded in socioformative principles, constitutes a valid and reliable framework for understanding and enhancing higher education in the AI era. The findings confirm the relevance of the AI-centered Onlife Experiential Learning Model: empirical validation shows that the four theoretical components of the model—utility, knowledge construction, innovation, and accomplishment —integrate holistically into a coherent hierarchical factorial structure

The theoretical integration of socioformation with the Onlife model offers a distinctive contribution that transcends the limitations identified in connectivism. While connectivism explains learning as network formation processes (Shang & Bai, 2025; Siemens, 2005), it lacks specificity regarding how learning should be structured between humans and machines, and fails to articulate a shared ethical vision for education (Holmes et al., 2019). Our findings demonstrate that socioformation effectively addresses these gaps by providing an ethical-pedagogical framework that orients technology use toward sustainable social development (Tobón & Luna-Nemecio, 2021). The strong factorial loadings observed across all four dimensions—Utility (.83–.96 for students; .62–.93 for faculty)—confirm that the model successfully operationalizes socioformative principles while maintaining psychometric rigor.

The differential patterns of satisfaction between students and faculty merit deeper theoretical interpretation. Faculty demonstrated significantly higher latent means in Knowledge Construction, Innovation, and Accomplishment, yet showed equivalent perceptions of Utility. This pattern suggests that faculty, with their greater pedagogical expertise and metacognitive awareness (Rapanta et al., 2025), may more readily recognize the methodological sophistication embedded in these dimensions. Conversely, the equivalent Utility scores indicate consensus regarding the model's relevance for connecting learning with authentic territorial problems (Núñez et al., 2023)—a core tenet of socioformation that emphasizes moving beyond abstract contextualization toward genuine territorialization (Golden, 2025). These findings contrast with Rao et al.’s (2024) observations of persistent faculty-student divergences in online education, suggesting that the Onlife model’s hybrid design may facilitate greater pedagogical alignment than purely virtual approaches.

The predictive role of Utility for faculty satisfaction, while all four dimensions predicted student satisfaction, reveals fundamentally different engagement patterns with AI-enhanced pedagogy. For faculty, perceiving AI as useful for achieving educational objectives appears to be the critical determinant of satisfaction—consistent with socioformation's emphasis on AI as an operational platform under human ethical direction rather than an autonomous agent (Mayboroda et al., 2025; Pastrana et al., 2025). Students, however, engage with the educational experience more holistically, valuing the multidimensional integration of challenge-based learning, collaborative knowledge construction, innovative methodologies, and outcome achievement. This multifactorial student perspective aligns with Testa et al.’s (2023) structural equation modeling findings that student satisfaction in technology-mediated environments depends on course organization, student-instructor relationships, and perceived accomplishment simultaneously.

The model’s contribution to computational thinking development extends existing literature predominantly focused on STEM contexts (Liao et al., 2022) to a broader undergraduate population. The finding that Knowledge Construction emerged as the strongest predictor of computational thinking (β = .49 for students; β = .82 for faculty) provides theoretical support for socioformation's reconceptualization of this construct. Rather than conceiving computational thinking merely as technical problem-solving capacity (Wing, 2006), socioformation integrates it within metacognitive AI literacy—the capacity to exercise conscious, reflective, and ethical control over AI systems (Rapanta et al., 2025; Varghese et al., 2025). The stronger faculty coefficients suggest that educators’ deeper engagement with collaborative knowledge co-construction, formative research integration, and methodological innovation translates more directly into perceived computational competence. This interpretation aligns with Massaty et al.’s (2024) argument that AI implemented within active learning environments enhances computational thinking through adaptive practice, immediate feedback, and complex problem-solving opportunities.

The absence of substantial gender differences, particularly among faculty, warrants consideration within broader debates regarding digital competence equity (Lin & Wong, 2024). While female students showed statistically significant but practically negligible differences in model dimension scores (Cohen’s d < .20), the gender coefficient in the satisfaction model (β = −.04) suggests underlying factors requiring further investigation. Potential explanations include differential self-efficacy beliefs, varying prior exposure to AI technologies, or subtle pedagogical dynamics not captured by the current measures. Importantly, the homogeneity observed among faculty indicates that the Onlife model does not introduce systematic gender-based disparities in professional engagement—a finding with significant implications for equitable AI integration in higher education.

Several limitations should temper interpretation of these findings. First, the single-institution sample, despite its geographic diversity within Peru, limits generalizability to broader Latin American or international contexts. Second, the reliance on self-report measures introduces potential social desirability and common method biases; future research should incorporate behavioral indicators of computational thinking and objective learning outcomes. Third, the newly constructed Onlife Model Scale, while demonstrating strong psychometric properties, requires additional validation across diverse educational settings. Fourth, the cross-sectional design precludes causal inferences; claims regarding transformative educational impact await longitudinal confirmation. Finally, computational thinking was assessed through a single self-report item, capturing perceived rather than objective competence—a limitation that constrains construct precision.

These findings carry significant implications for educational practice and institutional policy. For educators, the results suggest that intentional integration of all four model components—rather than selective emphasis—maximizes student satisfaction and computational thinking development. Professional development programs should emphasize not merely technical AI competencies but the metacognitive and ethical dimensions that distinguish human governance of technology from passive tool use (Larson et al., 2024; Mayboroda et al., 2025). For institutional leaders, the findings support investment in hybrid infrastructures that enable authentic territorialization of learning experiences, connecting academic challenges with community needs (Golden, 2025). Policy frameworks should promote clear guidelines for ethical AI integration while avoiding excessive prescription that might inhibit pedagogical innovation (Acevedo-Carrillo et al., 2025).

Future research should pursue several complementary directions. Multi-institutional and cross-national studies are essential for establishing the model’s broader applicability and identifying contextual moderators. Longitudinal designs tracking student trajectories would clarify causal relationships and assess long-term competency development. Investigation of specific socioformative variables—including ethical life project, sociocritical thinking, and transdisciplinary collaboration—would illuminate the mechanisms through which the Onlife model achieves its effects. Additionally, exploring how generative AI tools might be integrated within the socioformative framework represents a timely research frontier given their accelerating educational adoption (López-Regalado et al., 2024; Vieriu & Petrea, 2025).

Conclusions

Addressing the identified knowledge gap regarding empirically validated pedagogical models for AI-enhanced hybrid education, this study confirms that the Onlife experiential learning model, grounded in socioformative principles, provides a coherent theoretical and practical framework for contemporary higher education. The model's four-dimensional structure demonstrates robust psychometric validity across student and faculty populations, while simultaneously predicting both academic satisfaction and computational thinking development. The differential predictive patterns—with faculty satisfaction depending primarily on perceived utility and student satisfaction requiring multidimensional engagement—reveal the distinct ways these populations engage with AI-enhanced pedagogy. By situating artificial intelligence as an operational platform under human ethical direction rather than an autonomous educational agent, the socioformative approach offers a humanistic counterweight to purely technologically-deterministic visions of educational transformation. These findings provide empirical foundation for institutions seeking to navigate the complex intersection of hybrid learning, artificial intelligence, and sustainable social development.

The Knowledge Construction dimension—co-constructing student learning through collaborative work, formative research, and other active methods integrating human and non-human actions—is the main predictor of computational thinking development, with higher scores in faculty than in students. This finding suggests that computational thinking develops through the integrated practice of collaborative knowledge co-construction with digital tools, rather than through isolated technological instruction, highlighting the pedagogical and formative focus inherent in the model’s holistic concept of experiential learning.

Student academic satisfaction depends on a broader set of factors, suggesting a more integrated and multidimensional educational experience. In contrast, faculty satisfaction is primarily based on perceived utility. Computational thinking is strongly linked to perceptions of knowledge and utility in both groups, although innovation and accomplishment also play a role for students. Together, these results suggest that students value the technology-mediated educational experience from a multifactorial perspective, whereas faculty adopt a more focused approach centered on instrumental aspects of technology use.

Footnotes

Author Note

Any other identifying information related to the authors and/or their institutions, funders, approval committees, etc, that might compromise anonymity.

ORCID iD

Nemecio Núñez-Rojas

Ethical Considerations

The ethics committee of the participating university approved the study. The processing of personal data complied with the Personal Data Protection

Consent to Participate

Digital informed consent was obtained from all participants—both faculty and students of the university—ensuring that the process was voluntary, anonymous, and confidential.

Author Contributions

Nemecio Núñez Rojas: Conceptualization: Ideas; formulation or evaluation of the overarching research goals and aims. Investigation: Conducting the research and investigation process, specifically data/evidence collection. Project Administration: Management and coordination responsibility for the planning and execution of the research activity. Supervision: Oversight and leadership responsibility for the planning and execution of the research activity.

Writing – Original Draft: Preparation of the manuscript as the initial draft (including substantive translation). Writing – Review & Editing: Preparation, creation, and presentation of the manuscript for publication; critical review and commentary, including pre- and post-publication stages. Sergio Tobon: Data Curation: Management activities to annotate (produce metadata), clean, and maintain research data for initial use and later reuse. Validation: Verification, whether as part of the activity or separately, of the overall results, replication/reproducibility, experiments, or other research outputs. Writing – Original Draft: Preparation of the document, specifically the literature review section of the initial draft (including substantive translation). Erika Janet Valdivieso: Investigation: Conducting the research and inquiry process, specifically data/evidence collection. Resources: Provision of study materials, reagents, materials, instrumentation, computing resources, or other analysis tools. Writing – Original Draft: Preparation and drafting of the study’s conclusions as part of the initial manuscript (including substantive translation). Antonio Matas: Data Curation: Management activities to annotate (produce metadata), clean, and maintain research data (including software code when necessary to interpret the data) for initial use and later reuse. Formal Analysis: Application of statistical, mathematical, computational, or other formal techniques to analyze or synthesize study data. Methodology: Development or design of methodology; creation of models. Writing – Original Draft: Preparation and drafting of the Methods and Results sections as part of the initial manuscript (including substantive translation).

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

The database is not available for research purposes.

References

Acevedo-Carrillo

Cabezas-Torres

N. M.

La Serna

P. A.

Araujo

S. A.

(2025). Challenges and opportunities of artificial intelligence in Latin American higher education: A systematic literature review. Zenodo, 6(1), 1–10. https://doi.org/10.5281/zenodo.15508755

Ainjärv

Laas

(2024). Applying constructivism in driving teacher education: Analysis based on transcripts of videotaped lessons. Transportation Research Part F: Traffic Psychology and Behaviour, 100, 346–360. https://doi.org/10.1016/j.trf.2023.12.004

Ausubel

D. P.

(1976). Psicología educativa: Un punto de vista cognoscitivo. Trillas.

Cabero-Almenara

Llorente-Cejudo

M. D. C.

(2015). Information and communication technologies (ICT): Educational scenarios and learning theories. Revista Lasallista de Investigación, 12(2), 186–193.

Chen

F. F.

(2007). Sensitivity of goodness of fit indexes to lack of measurement invariance. Structural Equation Modeling, 14(3), 464–504. https://doi.org/10.1080/10705510701301834

Floridi

(Ed.). (2015). The Onlife Manifesto: Being Human in a Hyperconnected Era. Springer Open. https://doi.org/10.1007/978-3-319-04093-6

Floridi

(2021). Digital time: Latency, real-time, and the onlife experience of everyday time. Philosophy & Technology, 34(3), 407–412. https://doi.org/10.1007/s13347-021-00472-5

Fornell

Larcker

D. F.

(1981). Evaluating structural equation models with unobservable variables and measurement error. Journal of Marketing Research, 18(1), 39–50. https://doi.org/10.2307/3151312

Golden

(2025). Critical thinking for global citizenship education: A conceptual framework. Palgrave Macmillan. https://doi.org/10.1007/978-3-031-89642-2

10.

Hair

J. F.

Black

W. C.

Babin

B. J.

Anderson

R. E.

(2019). Multivariate data analysis (8th ed.). Cengage.

11.

Holmes

Bialik

Fadel

(2019). Artificial intelligence in education: Promises and implications for teaching and learning. OECD Publishing. https://unesdoc.unesco.org/ark:/48223/pf0000368021

12.

Larson

B. Z.

Moser

Caza

Muehlfeld

Colombo

L. A.

(2024). Critical thinking in the age of generative AI. Academy of Management Learning & Education, 23(3), 373–378. https://doi.org/10.5465/amle.2024.0338

13.

Latour

(2023). Reassembling the social: An introduction to actor-network-theory. Oxford University Press. https://doi.org/10.1093/oso/9780199256044.001.0001

14.

Lee

Matthew

Srivastava

Plastow

Strelan

Ploeckl

Lekkas

Palmer

(2024). The impact of generative AI on higher education learning and teaching: A study of educators’ perspectives. Computers and Education: Artificial Intelligence, 6, 100221. https://doi.org/10.1016/j.caeai.2024.100221

15.

C.-H.

(2016). Confirmatory factor analysis with ordinal data: Comparing robust maximum likelihood and diagonally weighted least squares. Behavior Research Methods, 48(3), 936–949. https://doi.org/10.3758/s13428-015-0619-7

16.

Liao

C. H.

Chiang

C. T.

Chen

I. C.

Parker

K. R.

(2022). Exploring the relationship between computational thinking and learning satisfaction for non-STEM college students. International Journal of Educational Technology in Higher Education, 19, 43. https://doi.org/10.1186/s41239-022-00347-5

17.

Lin

Wong

G. K. W.

(2024). Gender differences in computational thinking skills among primary and secondary school students: A systematic review. Education Sciences, 14(7), 790. https://doi.org/10.3390/educsci14070790

18.

López-Regalado

Ó.

Núñez-Rojas

López-Gil

Ó. R.

Sánchez-Rodríguez

(2024). Análisis del uso de la inteligencia artificial en la educación universitaria: Una revisión sistemática. Pixel-Bit. Revista de Medios y Educación, 70, 97–122. https://doi.org/10.12795/pixelbit.106336

19.

Massaty

M. H.

Fahrurozi

S. K.

Budiyanto

C. W.

(2024). The role of AI in fostering computational thinking and self-efficacy in educational settings: A systematic review. Indonesian Journal of Informatics Education, 8(1), 49–60. http://doi.org/10.20961/ijie.v8i1.89596

20.

Mayboroda

Guibin

Adamu

(2025). Artificial intelligence governance in higher education: The role of knowledge-based strategies in fostering legal awareness and ethical artificial intelligence literacy. Societies, 15(1), 144. https://doi.org/10.3390/soc15010144

21.

McDonald

R. P.

(1999). Test theory: A unified treatment. Lawrence Erlbaum Associates.

22.

Morin

(1999). Los siete saberes necesarios para la educación del futuro. UNESCO.

23.

Núñez-López

Ávila-Palet

J. E.

Olivares-Olivares

S. L.

(2017). El desarrollo del pensamiento crítico en estudiantes universitarios por medio del Aprendizaje Basado en Problemas. Revista Iberoamericana de Educación Superior, 8(23), 84–103.

24.

Núñez-Rojas

Meoño-Ballena

J. L.

Llatas-Altamirano

L. J.

Tejada-Pinto

P. E.

(2023). Modelo educativo Onlife: percepciones de los estudiantes y docentes sobre la metodología de enseñanza y la evaluación. Formación universitaria, 16(6), 33–46. https://dx.doi.org/10.4067/S0718-50062023000600033

25.

Pastrana

Chávez-Herting

Rodríguez

López

F. A.

(2025). Examining innovation practices in the IT sector: A socioformative and AI-driven analysis of talent development in Mexico and the United States. Social Sciences & Humanities Open, 11, 101512. https://doi.org/10.1016/j.ssaho.2025.101512

26.

R Core Team. (2025). R: A language and environment for statistical computing. R Foundation for Statistical Computing. https://www.R-project.org/

27.

Rao

L. N.

Shetty

Pai

Natarajan

Baliga

M. S.

Wahjuningrum

D. A.

Pawar

A. M.

(2024). Perceptions and challenges of online teaching and learning amidst the COVID-19 pandemic in India: A cross-sectional study with dental students and teachers. BMC Medical Education, 24, 637. https://doi.org/10.1186/s12909-024-05340-2

28.

Rapanta

Bhatt

Bozkurt

Chubb

L. A.

Erb

Forsler

Jandrić

(2025). Critical GenAI literacy: Postdigital configurations. Postdigital Science and Education, 7(4), 1296–1333. https://doi.org/10.1007/s42438-025-00573-w

29.

Revelle

(2023). psych: Procedures for psychological, psychometric, and personality research (R package version 2.3.8). Northwestern University. https://CRAN.R-project.org/package=psych

30.

Revelle

Zinbarg

R. E.

(2009). Coefficients alpha, beta, omega, and the greatest lower bound: Comments on Sijtsma. Psychometrika, 74(1), 145–154. https://doi.org/10.1007/s11336-008-9102-z

31.

Rhemtulla

Brosseau-Liard

P. É.

Savalei

(2012). When can categorical variables be treated as continuous? A comparison of robust continuous and categorical SEM estimation methods under suboptimal conditions. Psychological Methods, 17(3), 354–373. https://doi.org/10.1037/a0029315

32.

Rosseel

(2012). Lavaan: An R package for structural equation modeling. Journal of Statistical Software, 48(2), 1–36. https://doi.org/10.18637/jss.v048.i02

33.

Schlemmer

Di Felice

Ribeiro de Sousa

I. M.

(2020). Onlife education: The ecological dimension of digital learning architectures. Educar em Revista, 36, e76120. https://doi.org/10.1590/0104-4060.76120

34.

Schlemmer

Di Felice

(2023). The ecologies of presence in Onlife education. ETD – Educação Temática Digital, 25, 1–20. https://doi.org/10.20396/etd.v25i00.8667271

35.

Shang

L. E.

Bai

(2025). Generative AI and the future of connectivist learning in higher education. Journal of Asian Public Policy, 18(2), 329–351. https://doi.org/10.1080/17516234.2024.2386085

36.

Siemens

(2005). Connectivism: A learning theory for the digital age. International Journal of Instructional Technology and Distance Learning, 2(1), 3–10. http://www.itdl.org/Journal/Jan_05/Jan_05.pdf

37.

Testa

Macagno

Bertolino

Cacciamani

Grange

Perrucci

Cattelino

(2023). Academic satisfaction among university students during the COVID-19 pandemic: The influence of individual, social, and organizational variables. Frontiers in Education, 8, 1268585. https://doi.org/10.3389/feduc.2023.1268585

38.

Tiukhova

Pavani

López

Sigridur

Óskarsdóttir

Poelmans

Baesens

Snoeck

(2024). Explainable learning analytics: Assessing the stability of student success prediction models by means of explainable AI. Decision Support Systems, 182, 114229. https://doi.org/10.1016/j.dss.2024.114229

39.

Tobón

Luna-Nemecio

(2021). Proposal for a new talent concept based on socioformation. Educational Philosophy and Theory, 53(1), 21–33. https://doi.org/10.1080/00131857.2020.1725885

40.

Tobón

Lozano-Salmorán

E. F.

(2024). Socioformative pedagogical practices and academic performance in students: Mediation of socioemotional skills. Heliyon, 10, e34898. https://doi.org/10.1016/j.heliyon.2024.e34898

41.

Varghese

N. N.

Jose

Bindhumol

Cleetus

Nair

S. B.

(2025). The power duo: Unleashing cognitive potential through human-AI synergy in STEM and non-STEM education. Frontiers in Education, 10, 1534582. https://doi.org/10.3389/feduc.2025.1534582

42.

Vieriu

A. M.

Petrea

(2025). The impact of artificial intelligence (AI) on students’ academic development. Education Sciences, 15(3), 343. https://doi.org/10.3390/educsci15030343

43.

Vygotsky

L. S.

(1978). Mind in society: The development of higher psychological processes ( Cole

John-Steiner

Scribner

Souberman

, Eds.). Harvard University Press

44.

Wing

J. M.

(2006). Computational thinking. Communications of the ACM, 49(3), 33–35. https://doi.org/10.1145/1118178.1118215