Digital Teaching Practices and Student Academic Stress in the Era of Digitalization in Higher Education

Abstract

ICT and digitization are transforming higher education, but one aspect that has been little addressed is student academic stress in this context. The objective of this research was to explore the relationship between teaching practices and academic stress in virtual learning environments. A quantitative study based on structural equation modeling was conducted using 6605 students from a public university in Lima. Two scales were applied and validated to assess specific dimensions of teaching practices and levels of academic stress. The scales demonstrated robust psychometric properties. Exploratory and confirmatory factor analysis allowed the identification of key underlying factors and the validation of the proposed theoretical model. Findings reveal indirect negative influences between teaching practices and academic stress. External factors like overload, distractions, and technological issues disrupt teacher interaction, collaboration, and learning processes. Model fit indices confirm its robustness, emphasizing the need to optimize resources and improve virtual education environments. ICT in education enhances accessibility and personalization but demands addressing academic stress and optimizing virtual learning environments. Future research is called for to delve deeper into the integration of emerging technologies, particularly artificial intelligence and virtual reality, as well as mindfulness, digital skills, and mobile self-efficacy, to comprehensively address stress and student well-being.

1. Introduction

In an era defined by the proliferation of ICT, digital technologies, and e-learning, higher education institutions face a dual reality. While these advancements offer numerous benefits, such as flexibility, accessibility, and personalized learning, they also present significant challenges. One key issue is the impact of digitalization on student well-being, particularly academic stress. The integration of virtual learning environments requires students to navigate technological demands, reduced face-to-face interactions, and heightened self-regulation, all of which can exacerbate stress levels. This study examined how digital teaching practices influence student academic stress.

Student wellness and stress management are crucial aspects of ensuring meaningful learning experiences in higher education. University life, while enriching, is replete with academic, emotional, and social challenges that can generate high levels of stress in students. This stress, often influenced by factors such as academic load, performance expectations, and time constraints, can have significant implications for students’ mental health, academic performance, and quality of life. The transition to online learning environments has intensified these difficulties, introducing new sources of stress, such as lack of face-to-face interaction, difficulty managing time, and distractions at home. This phenomenon has made stress management even more critical in the online learning mode, where students face additional barriers that threaten their overall well-being and academic performance.

Multiple studies have addressed the association between stress and the well-being of university students, highlighting the significant impact this phenomenon has on their mental health, academic performance, and quality of life. Over time, the scientific literature has explored a wide variety of factors associated with academic stress, from transitions to online study modalities and emotional challenges to coping strategies and the development of personal competencies. In this context, flexibility in coping mechanisms, resilience, emotional intelligence, and social support emerge as central themes, all of which are critical to mitigating stress and promoting well-being. The following theoretical foundation analytically reviews these contributions, highlighting key findings from recent research and highlighting challenges and opportunities for the design of effective educational interventions.

Academic stress has been identified as a significant factor negatively influencing the psychological well-being of college students. This phenomenon intensified during the COVID-19 pandemic when social constraints and the transition to online modes of education increased stress levels and comprehensively affected students’ well-being [1,2]. Flexibility in coping mechanisms emerged as a crucial tool to mitigate the negative impact of stress on students’ mental health, highlighting the need for interventions that promote this adaptive capacity [1].

A cross-sectional study conducted among nursing students found a strong relationship between academic stress, goal orientation, and mental health. This finding highlights the importance of pedagogical strategies that foster stress regulation and clear academic goal setting to improve well-being [3]. Complementing this finding, five categories of variables associated with academic stress were identified: adaptation to change, study modality, learning resources, academic–life balance, and socioemotional variables, providing a comprehensive framework for understanding the challenges faced by students during critical situations such as the pandemic [2].

Emotional intelligence (EI) has also been considered a key resource for stress management. In a study with nursing students, it was shown that greater development of EI facilitates the use of more active and effective coping strategies, which increases subjective well-being [4]. This finding coincides with research that highlights resilience as a predictor of psychological well-being, suggesting that fostering emotional competencies and resilience in educational programs can contribute significantly to student well-being [5].

Likewise, studies such as [6] have explored the relationship between eudaimonic well-being, self-efficacy, and adaptive coping strategies, concluding that these capacities are essential to mitigate the effects of academic stress. In parallel, social support and compassionate self-acceptance emerge as critical variables that promote well-being despite academic stress [7].

Subgroup analysis also suggests that certain groups of students, such as females and non-binary individuals, experience higher levels of academic stress compared to others, calling for interventions targeted to these vulnerable groups [8]. The presence of personal psychological resources, such as self-efficacy and optimism, buffers the negative effects of academic stress on well-being, highlighting the value of integrating psychological interventions into the university curriculum [9].

Finally, although the impact of stress varies among different student profiles, the results are consistent in showing a significant inverse relationship between stress and emotional well-being [10]. This picture highlights the urgency of developing educational strategies that not only manage stress but also strengthen students’ coping skills and emotional resources, especially in a world increasingly inclined toward virtual education [11,12].

A very recent study enriches the analysis of the association between stress and student well-being by highlighting the crucial role of personality profiles in modulating these variables. Identifying three distinct profiles—Excessive, Average, and Resilient Controllers—provides insight into how individual differences influence stress coping strategies and psychological well-being. In particular, Resilient students demonstrated higher levels of well-being and a tendency to employ proactive problem-solving strategies, while Excessive Controllers used less adaptive strategies, such as self-criticism, which may intensify stress. This approach not only delves deeper into the individual dynamics that mediate the impact of stress but also underscores the importance of promoting resilience characteristics and effective coping strategies within educational interventions, especially in high-demand academic contexts [13].

Recent studies continue to highlight innovative strategies for developing student well-being. Authors such as Duchi and collaborators [14] introduced the concept of “study crafting”, which allows students to customize their educational experience based on their strengths, interests, and goals. This approach improves psychological well-being, independent learning skills, and academic satisfaction, as well as reducing burnout. For their part, other authors highlighted the inverse relationship between academic stress and emotional well-being, finding significant moderations by sociodemographic variables, which underscores the need to address stress with an approach tailored to the student’s profile. Finally, other authors examined the impact of a competitive climate in schools, concluding that student well-being does not depend directly on the perceived competitive environment, but on protective factors such as resilience, self-efficacy, and positivity of students, which mitigate the adverse effects of stress and enhance their integral development [15]. These findings highlight the importance of personalized approaches and the strengthening of emotional competencies as key pillars for well-being in contemporary educational contexts.

Recent studies [16,17] underscore the importance of educational approaches that promote student well-being. Laakso and collaborators [16] evaluated the impact of a school-based intervention, Flourishing Students, designed to improve psychological well-being, and hope and reduce depressive symptoms through 32 specific lessons implemented over an academic year. Their results showed that students in the intervention group presented lower levels of depressive symptoms and higher indicators of psychological well-being and hope compared to the control group, highlighting the effectiveness of structured and sustained interventions in educational contexts. Wahyuni and collaborators [17] investigated the role of the learning environment and task orientation as determinants of student well-being in an analysis of 1698 Indonesian students. The findings showed that a task-oriented learning environment significantly explained 33.4% of the variability in student well-being, establishing itself as a key factor. Both studies highlight that the careful design of interventions and learning environments can contribute substantially to improving student well-being, providing a framework for more effective pedagogical practices focused on the holistic needs of students.

Multiple factors can be found that are potentially stressful, and among them is the impact that the educational system can have on the well-being and academic performance of university students. This is referred to as academic stress [18]. Previous studies highlight the need to address academic stress comprehensively, not only as a threat to well-being but also as an opportunity to strengthen personal and academic competencies through the design of effective pedagogical strategies and institutional policies.

1.1. Literature Review

Student Academic Stress

Several studies address academic stress, including several highly cited studies that address this construct in the context of college students [8,11,19,20,21,22,23,24,25,26,27]. According to these studies, academic stress in college students is defined as the perception of an imbalance between academic demands and personal resources to cope with them, influencing their emotional well-being and academic performance.

This phenomenon includes dimensions such as pressure to achieve high standards, interpersonal conflicts in the educational environment, and the perceived lack of adequate resources, as well as factors associated with fear of failure and difficulties of integration into the university environment. The causes range from curricular and evaluative demands to social and academic transitions, exacerbated by events such as the COVID-19 pandemic, which increased uncertainty and emotional impact.

The consequences can manifest in anxiety, decreased performance, mental health problems such as emotional exhaustion and depersonalization, and the weakening of social support networks, particularly among more vulnerable populations such as women and international students. It is crucial to implement tailored strategies that strengthen coping and resilience, along with learning environments that consider cultural and gender differences.

For some students, accessing university becomes a stressful experience that manifests itself in signs of anxiety, so university institutions continue to design programs to manage stress and anxiety and improve mental well-being [28], because as one study ratifies, university can be a challenging time for students, associated with uncertainty and stress [29].

In particular, online education presents challenges such as time management, maintaining social interactions, and motivation, which can generate academic stress [30]. Living year-round with online learning can be overwhelming and uncomfortable for university students, with mental health consequences such as stress or fatigue, and this has an impact on their academic performance. Stress and fatigue levels in students can be high as they have to sit and face monitors every time they participate in a lecture with an online learning platform [31].

One study found that a significant number of participants noted an increase in their academic stress after switching to online academic programming (62%, n = 286) [32], while other research found that all academic stressors have a significant relationship with students’ level of online learning fatigue [33]; at the same time, another study shows that the risk of academic stress experienced by university students increased in online learning during the COVID-19 pandemic [34]. A further study shows that academic stress and its associated variables had unfavorable consequences on the psychological well-being of university students [2]. From there arises the need to study how to deal with the new normal: how to adapt to online and distance learning in the post-pandemic era [35].

Thus, it is demonstrated that academic stress is a risk factor for the health of students and their quality of life [36,37], associated with poor performance and mental health [38], while perceived social support and self-management are positive elements that may mitigate stress [39]. School climate may also moderate the relationships between students’ academic stress and their depressive symptoms [40], as well as academic motivation [41,42], authenticity [43], self-efficacy, management of physical distractions [36], and positive coping strategies [44,45].

Examinations are also a major source of academic stress. Assessment of learning outcomes is an essential element of curriculum design, but even well-designed assessments are a major source of stress and anxiety for university students [46]; this is directly related to course evaluations, submission of activities, assignments, course activities, and perception of limited time for the completion of assignments.

The five groups of variables associated with academic stress are (1) adaptation to change; (2) study modality; (3) learning resources; (4) academic–life balance; and (5) socioemotional variables [2]. In addition, specific scales, such as that developed by Busari, have allowed a more accurate assessment of academic stress in university students, providing a valuable background for the present research [47].

The perception of high academic demands, mentioned as a recurrent source of stress, is associated with student overload. This includes pressures from multiple deadlines, excessive workload, and expectations related to achieving high academic standards, which are frequent causes of academic stress.

Albeit indirectly, the learning environment during the transition to online modes due to the pandemic is a prominent example. Inadequacy of study spaces and external factors in the home environment (such as noise, interruptions, and family responsibilities) emerged as major obstacles contributing to academic stress, especially for students who did not have a suitable environment to study from home.

Regarding the teaching environment, students felt responsibility and academic overload, as though they had little time to finish homework, and that high levels of effort and dedication were required to obtain good grades, and were also affected by distractions generated by social networks, chats, and video games. The students in this research self-evaluated the consequences and solutions to emotional problems due to academic stress [48]. This is related to findings associated with academic stress in another study, where the majority of students reported poor time management skills, lack of confidence, and distractions [38].

From the studies analyzed, it is possible to identify dimensions of student academic stress in online environments:

• Student Overload: Associated with academic pressure, fear of failure, and overload of homework, exams, activities, etc.: Includes the level of demand related to assessments, deadlines, and students’ perceived overall performance. “I’m afraid of not meeting academic expectations.”
• Student Interaction: Associated with managing interpersonal relationships, including factors such as conflicts with teachers, problems with classmates, and difficulties in relational dynamics in the educational environment.
• Digital Resources: Adequacy of educational resources, including perceived inadequacies in study materials, infrastructure, or institutional support, especially in online education
• Environmental Distractions: Impact of external events. These include external circumstances such as the COVID-19 pandemic, which altered the normal dynamics of learning and added uncertainty to the academic process; external noises; and physical distractions.

A fundamental aspect is individual differences. Factors such as gender, culture, and personal skills (e.g., coping capacity and resilience) influence how students experience stress, hence the importance of taking these elements and sociodemographic variables into account in studies of academic stress.

1.2. Teaching Practices

Teaching practices are the result of the teacher’s conceptions and evaluations within a specific context and are composed of behavioral, cognitive, and attitudinal components [49,50,51]. To conceptualize teaching practices, it is essential to link them to the context (Becerril, 2005). Each teacher has unique characteristics that shape their interactions with students, the institution, and society, which are reflected in their beliefs and actions within the classroom. In this context, formative feedback from the teacher plays a crucial role, as it effectively guides the development of each student, bringing their formative potential to light and fostering meaningful and personalized learning [52,53].

At the university level, it is argued that teachers should not only be experts in their subjects but also strengthen their psycho-pedagogical training, to educate students in values, emotions, and social relations. Self-reflection is key to understanding and improving teaching practice, as it helps to analyze the elements that influence the pedagogical experience [54].

Teaching practices are “actions conditioned” to the educational context, which makes it difficult to define what constitutes a “good teaching practice” since it is a dynamic concept adaptable to pedagogical theories and applied approaches. These practices refer to pedagogical actions that solve specific problems, help schools become resources for communities, or articulate quality and access to learning in contexts of inequality. Good practices are not universal and need to be adapted to each educational context [51].

Four fundamental characteristics of good teaching practices are identified: first, they are conditioned by the context, since teaching actions depend on factors such as the students, the institution, and the curriculum; second, they have a two-dimensional nature, combining objective behaviors and subjective thoughts, according to the Theory of Reasoned Action; third, they follow the process–product paradigm, where teaching strategies facilitate learning; and finally, they must become praxis, where reflection and action are integrated to improve the educational reality [51].

It is difficult to define a practice or a teacher as “totally good”, so it is essential to identify specific actions that can be considered good teaching practices. These are important at all educational levels, as they allow for sharing experiences, improving the quality of teaching, and establishing medium- and long-term benchmarks. Good practices include a behavioral component, which encompasses actions both inside and outside the classroom, and a subjective component related to the beliefs and attitudes of the teacher. Promoting critical reflection on these practices is key to the continuous improvement of the educational process [51,55]

Perrenoud highlights several benefits that reflective practice can bring to teachers. Some of these benefits include: (1) complementing initial professional development; (2) facilitating the development of experiential knowledge; (3) fostering learning and collaboration among colleagues; (4) preparing teachers to assume political or ethical responsibilities; and (5) increasing the competencies needed for educational innovation. In short, reflection on practice enables teachers to identify problems in their own experiences and generate innovative solutions [55].

For their part, the importance of teachers considering their didactic knowledge from two perspectives is emphasized: public knowledge of teaching–learning theories and knowledge acquired through experience. Teachers must adapt this knowledge to their personal and contextual characteristics. At the university level, improvements in teaching practices are reflected in better student learning. Good teaching practices seek to foster deep learning and reduce superficial learning, which requires teachers to identify the cognitive, affective, and motivational characteristics that can influence their students’ motivation toward these learning approaches [50].

1.2.1. Teaching Practice in Virtual Environments

Teaching practice in virtual environments has evolved significantly thanks to technological advances and contemporary educational demands. Virtual education requires constant and strategic teacher training so that the teacher can perform optimally [56]. The results show that future teachers value digital learning content but are not sufficiently prepared for its creation; they develop tolerance in communication but do not have sufficient skills to show empathy; they appreciate independence in learning management and still need training in professional digital self-presentation [57], also considering the development of a pedagogical profile of experienced, enthusiastic, and cautious teachers [58], and holistically integrating three essential aspects for virtual environments: technological content, pedagogical content, and disciplinary content [59,60].

This educational model involves a comprehensive approach that combines teacher interaction, technology-mediated communication, student collaboration, and the implementation of innovative pedagogical processes. It also highlights the need for continuous support for teachers to ensure their adaptation and effectiveness in this environment. The main characteristics of this practice are presented below, organized into five key aspects that mark its impact on teaching and learning.

1.

Teacher Interaction

Teacher interaction in virtual environments is a central aspect that defines the quality of learning. The importance of frequent contact between teachers and students through digital platforms is highlighted, which contributes to a more effective and personalized educational experience [61]. In addition, teachers’ ability to generate professional discourse, through critical analysis of incidents in virtual contexts, fosters a reflective practice that enriches their interaction with students [62] and encourages collaborative online learning [63].

2.

Media Communication

The use of technological tools, both synchronous and asynchronous, transforms teacher–student communication. Tools such as forums, videoconferences, and virtual platforms allow not only the transmission of content but also the promotion of motivational and affective actions, essential for learning [64]. For example, the use of synchronous software enables real-time communication, which facilitates a more immersive and participatory learning experience [65].

3.

Student Collaboration

Collaboration among students in virtual environments is enriched by technologies that promote social interaction and community learning. Virtual worlds and serious games are effective tools for fostering collaboration and immersion, increasing intrinsic motivation and social connections among students [66]. This approach allows students to interact and solve problems together, which is essential for meaningful learning.

4.

Teaching Process

The teaching process in virtual environments requires rigorous planning that considers both content design and assessment methods. Teachers must adapt to new teaching models that integrate innovative technologies, such as virtual reality, which improves teaching quality and student satisfaction by offering more practical and realistic experiences [67]. In addition, content design must consider students’ prior knowledge to ensure relevance and accessibility [61].

5.

Teacher Support

Teacher support is crucial to the success of teaching in virtual environments. This support includes both training in digital competencies and access to pedagogical tools that facilitate adaptation to technological environments. For example, professional development programs focused on the use of synchronous software and the creation of an effective teaching presence are fundamental to improving pedagogy in virtual classrooms [65]. Likewise, the need for a dialogic approach and reflection on teaching practices as a means to generate professional discourses that strengthen their role is also highlighted [62].

1.2.2. Justification and Gaps in the Literature

Despite advancements in research on virtual education, significant gaps remain in the literature. First, there is a lack of in-depth studies exploring the relationship between teaching practices and academic stress among students in virtual learning environments. Furthermore, existing research lacks sufficiently comprehensive scales to assess academic stress resulting from virtualization. These scales often fail to integrate key dimensions that holistically address the multifaceted stressors of virtual learning environments, such as academic overload, technological distractions, and reduced interpersonal interaction. Similarly, there is a notable absence of robust and validated measures to evaluate teaching practices in virtual settings, hindering systematic analysis of this relationship.

1.3. Research Questions

What are the psychometric properties (construct validity and internal reliability) of the Teaching Practices Scale in Virtual Education (TPSVE), and how suitable is its structure for evaluating teaching practices in virtual education?

What are the psychometric properties (construct validity and internal reliability) of the Student Academic Stress Scale in Virtual Education (SASSEV), and how suitable is its structure for evaluating academic stress in virtual education?

What is the relationship between digital teaching practices and academic stress among university students in virtual learning environments, and what are the key factors influencing this dynamic according to a theoretical model validated through structural equation modeling?

1.4. Research Objectives

To evaluate the psychometric properties of construct validity and internal reliability of the Teaching Practices Scale in Virtual Education (TPSVE) to determine its applicability in studies on teaching practices in virtual educational settings.

To evaluate the psychometric properties of construct validity and internal reliability of the Student Academic Stress Scale in Virtual Education (SASSEV) supporting its capacity to comprehensively measure academic stress in virtual education.

To examine the relationship between digital teaching practices and academic stress among university students in virtual learning environments, identifying key factors influencing this dynamic and validating a theoretical model using structural equation modeling.

2. Materials and Methods

2.1. Methodological Approach

The present study adopted a quantitative approach, which allows for analyzing statistical relationships between variables using objective and replicable methods. This approach was appropriate to explore the relationship between teaching practice in virtual environments and academic stress in university students, using structural equation modeling as the main tool for analysis. This study was based on a non-experimental, cross-sectional, correlational research design.

2.2. Variables

Independent variable:

Teaching Practice in virtual environments is broken down into the following dimensions:

• Teacher Interaction;
• Media Communication;
• Student Collaboration;
• Teaching Process;
• Teacher Support.

Dependent variable:

Student Academic Stress is structured in the following dimensions:

• Student Overload;
• Student Interaction;
• Digital Resources;
• Environmental Distractions.

2.3. Sampling

The sample consisted of 6605 university students from the Universidad Nacional Mayor de San Marcos, selected using a simple random sampling method with allocation proportional to the number of students per faculty. The sample size was calculated using the formula for finite populations, considering a tolerance error of 0.01, a significance level of 0.05, and a success ratio of 0.5. This design ensures the representativeness of the data at the level of the entire institution.

2.4. Instruments for Obtaining Information

The Student Academic Stress Scale in Virtual Education (SASSEV) is designed to measure the dimensions of teaching practice. This scale has 21 items (Cronbach’s alpha: 0.889), distributed in five dimensions that evaluate key aspects of teaching practice in virtual environments. It uses a Likert-type scale from 1 to 5, where the values represent the degree of agreement or disagreement with the proposed statements:

• Strongly disagree;
• Disagree;
• Neither agree nor disagree;
• Agree;
• Strongly agree.

Example item: “The means of communication used by the teacher were useful”. A student answering “5” indicates that they perceive the teacher’s communication to be highly effective.

The Teaching Practices Scale in Virtual Education (TPSVE) is used to evaluate the identified dimensions of academic stress. This instrument consists of 15 items (Cronbach’s alpha: 0.879) grouped into four dimensions that reflect the main factors of academic stress in virtual contexts. It also uses a Likert-type scale from 1 to 5, in this case, measuring the frequency with which students experience stressful situations:

• Never;
• Rarely;
• Sometimes;
• Almost always;
• Always.

Sample item: “I feel overloaded with assignments and/or course activities”. A response of “4” indicates that the student frequently experiences this type of overload.

Both scales were subjected to rigorous validations, including reliability analysis (Cronbach’s alpha) and content validity (Lawshe’s index), ensuring their ability to measure the variables of interest accurately.

2.5. Fieldwork

The data collection was conducted in a virtual environment, considering the constraints inherent to digital platforms. Questionnaires were distributed electronically, ensuring both the anonymity and confidentiality of participants. The fieldwork spanned the second semester of 2023, with data systematically collected over three months. Measures were implemented to minimize bias and ensure consistency in responses, thereby enhancing the reliability of the findings.

2.6. Data Analysis

Data processing and analysis were performed using tools such as Excel 2024, SPSS version 26, and R-project version 4.4.2, AMOS version 24 following a detailed analytical sequence:

-

Calculation of reliability using Cronbach’s alpha and the theta and omega coefficients.

-

Obtaining the content validity index (CVI) to assess the quality of the instruments.

-

Exploratory factor analysis to identify the underlying structures of the variables.

-

Statistical description of the data collected.

-

Verification of assumptions of the parametric model.

-

Confirmatory factor analysis to validate the identified structures.

-

Application of structural equation models to analyze the relationships between variables and evaluate model fit. The structural equation modeling approach was chosen for its ability to simultaneously evaluate complex relationships between teaching practices and academic stress while accounting for the indirect effects and latent variables central to this study’s objectives.

3. Results

3.1. Results of Teaching Practice

3.1.1. Exploratory Analysis of the Data on Teaching Practice

The ICC value for single measures (0.271) indicates that the consistency of individual responses is low. In other words, students do not respond very consistently to the individual survey items, which could be due to variability in how they interpret the items or individual differences. Also, the ICC value for average measures (0.929) shows that when the items are considered as a whole (averaged), the scale has a very high reliability. This suggests that the 35 items are measuring the construct consistently when the responses are averaged.

In the initial reliability analysis of the 35-item Do-Cent practice scale, Cronbach’s alpha (0.935) and the theta coefficient (0.937) reflect excellent internal consistency, indicating that the items are highly correlated and accurately measure the construct of interest. Although the omega coefficient is slightly lower (0.784), it still indicates good reliability, being less sensitive to dimensionality. Taken together, the three coefficients suggest that the scale is highly reliable and suitable for assessing teaching practice.

From

Table 1. Statistics of the items deleted from the teaching practice.

Item code	Element Statistics		Element Deleted
	Media	Standard Deviation	Total Correlation of Corrected Elements	Squared Multiple Correlation	Cronbach’s Alpha If the Item Has Been Deleted
PD1	3.42	0.878	0.562	0.596	0.931
PD2	3.44	0.825	0.567	0.598	0.931
PD3	3.24	0.888	0.524	0.381	0.931
PD4	3.24	0.929	0.518	0.374	0.931
PD5	3.44	0.973	0.495	0.441	0.931
PD6	3.64	0.897	0.545	0.481	0.931
PD7	3.13	0.950	0.471	0.271	0.932
PD8	3.27	1.050	0.425	0.264	0.932
PD9	3.04	1.087	0.382	0.251	0.933
PD10	3.49	1.115	0.371	0.247	0.933
PD11	3.30	1.061	0.427	0.265	0.932
PD12	3.56	0.873	0.544	0.366	0.931
PD13	3.46	0.910	0.488	0.296	0.932
PD14	3.40	0.888	0.568	0.365	0.931
PD15	3.52	0.862	0.541	0.344	0.931
PD16	3.45	0.889	0.548	0.351	0.931
PD17	3.64	0.897	0.535	0.355	0.931
PD18	3.41	0.956	0.533	0.338	0.931
PD19	3.50	0.940	0.545	0.370	0.931
PD20	3.69	0.985	0.475	0.330	0.932
PD21	3.90	0.942	0.492	0.395	0.931
PD22	3.60	0.912	0.608	0.444	0.930
PD23	3.40	0.942	0.534	0.407	0.931
PD24	3.23	0.988	0.520	0.399	0.931
PD25	3.19	1.045	0.431	0.254	0.932
PD26	3.39	0.825	0.630	0.452	0.930
PD27	3.59	0.770	0.609	0.443	0.931
PD28	3.57	0.813	0.606	0.433	0.930
PD29	3.38	0.893	0.570	0.386	0.931
PD30	3.36	0.884	0.629	0.453	0.930
PD31	3.36	0.844	0.588	0.415	0.931
PD32	3.52	0.887	0.548	0.343	0.931
PD33	2.80	1.062	0.384	0.274	0.933
PD34	3.07	0.998	0.507	0.374	0.931
PD35	3.41	0.950	0.563	0.361	0.931

, it can be observed that the highest value in the mean corresponds to item PD21 (3.90), indicating that students tend to give it a higher score compared to other items, while item PD33 (2.80) presents the lowest score, suggesting that this item has a relatively lower evaluation. In terms of standard deviation, item PD10 (1.115) shows the highest variability in students’ responses, indicating a greater dispersion in scores, and item PD27 (0.770) presents the lowest variability, suggesting more consistent and uniform responses among students. The total corrected item correlation is highest for item PD27 (0.609), indicating that this item is highly correlated with the rest and contributes strongly to internal consistency, and the lowest value is for item PD9 (0.382), suggesting that this item has less of a relationship with the other items and could be affecting the consistency of the scale. In the squared multiple correlation, item PD22 (0.608) shows the highest proportion of its variance explained by the rest of the items, indicating a good alignment with the scale, and item PD10 (0.247) has the lowest value, suggesting that its variance is less explained by the other items, indicating a possible mismatch. Finally, Cronbach’s alpha if the item has been deleted is highest for item PD9 (0.933), indicating that removing it could improve the internal consistency of the scale, and the lowest value is for item PD22 (0.930), so no single item is noticeably decreasing the overall reliability of teaching practice.

The summary statistics are given herein of the 35 items that make up the teaching practice scale. The item means range from a minimum of 2.803 to a maximum of 3.901, with an overall mean of 3.401, suggesting that, in general, participants tend to score the items in the indifference (3) and agree (4) parts of the scale. As for the item variances, the values range from 0.593 to 1.244, with a mean variance of 0.875, indicating that the items present moderate variability in the responses. Finally, the correlations among the items range from a minimum of 0.043 to a maximum of 0.746, with a mean correlation of 0.293, which implies that, although some items are more correlated with each other, most have moderate correlations, suggesting some independence among them within the construct of teaching practice.

The Mardia (asymmetry = 16,905,812; kurtosis = 117,056), Royston (6,056,455), Henze–Zirkler (1040), and energy (13,349) tests are consistent and show very low p-values (<0.001), indicating that the data for the teaching practice items do not follow a multivariate normal distribution.

The variance inflation factor (VIF) values for the 35 items of the teaching practice scale range between 1.33 and 2.49, which indicates a low level of multicollinearity among the items, i.e., there are no strong linear relationships among them, which is a good sign for the consistency of the instrument. Furthermore, it is worth noting that a VIF of less than 5 is generally acceptable; in this case, all items are significantly below that threshold. This reinforces the conclusion that multicollinearity is not a problem and provides confidence that the estimates obtained from subsequent analyses will not be affected by the presence of multicollinearity.

3.1.2. Exploratory Factor Analysis of Teaching Practice

The KMO is 0.958, which indicates excellent sampling adequacy to perform a factor analysis since the items are highly correlated with each other. In addition, Bartlett’s test of sphericity is significant (p = 0.000), confirming that there are sufficient correlations among the variables to justify factor analysis. Taken together, these results suggest that further factor analysis is appropriate and is expected to provide a good representation of the underlying structure of the data.

In

Table 2. Initial communalities of the teaching practice items.

Item	Community	Item	Community	Item	Community	Item	Community
PD1	0.755	PD11	0.456	PD21	0.570	PD31	0.517
PD2	0.747	PD12	0.478	PD22	0.528	PD32	0.414
PD3	0.557	PD13	0.418	PD23	0.584	PD33	0.522
PD4	0.544	PD14	0.416	PD24	0.594	PD34	0.541
PD5	0.741	PD15	0.401	PD25	0.445	PD35	0.425
PD6	0.718	PD16	0.404	PD26	0.503
PD7	0.372	PD17	0.454	PD27	0.503
PD8	0.397	PD18	0.387	PD28	0.497
PD9	0.501	PD19	0.449	PD29	0.468
PD10	0.479	PD20	0.486	PD30	0.529

, it is shown that the highest commonality corresponds to item PD1, with a value of 0.755, indicating that 75.5% of its variance is explained by the extracted factors, meaning that item PD1 is well-represented by the common factors and contributes significantly to the model. On the other hand, the lowest communality is that of item PD7, with a value of 0.372, indicating that only 37.2% of its variance is explained by the factors, meaning that item PD7 has a low representation in the factorial model, which makes it a candidate for revision or possible elimination since its contribution to the model is limited.

In

Table 3. Initial total variance explained by the underlying factors of teaching practice.

Component	Initial Eigenvalues			Sums of Squared Extraction Charges			Sums of Loads Squared by Rotation
	Total	% Variance	% Accumulated	Total	% Variance	% Accumulated	Total
1	11.136	31.817	31.817	11.136	31.817	31.817	9.742
2	1.629	4.653	36.470	1.629	4.653	36.470	7.136
3	1.520	4.344	40.814	1.520	4.344	40.814	5.784
4	1.386	3.960	44.774	1.386	3.960	44.774	5.209
5	1.124	3.212	47.987	1.124	3.212	47.987	5.642
6	1.004	2.870	50.856	1.004	2.870	50.856	2.636
7	0.959	2.740	53.596
8	0.884	2.526	56.122
9	0.882	2.520	58.642
10	0.803	2.295	60.937
11	0.760	2.170	63.107
12	0.735	2.100	65.207
13	0.705	2.013	67.220
14	0.695	1.987	69.207
15	0.668	1.908	71.115
16	0.647	1.848	72.963
17	0.633	1.810	74.773
18	0.622	1.776	76.549
19	0.608	1.738	78.287
20	0.590	1.685	79.973
21	0.579	1.655	81.628
22	0.557	1.592	83.220
23	0.553	1.580	84.800
24	0.533	1.522	86.322
25	0.506	1.445	87.767
26	0.499	1.426	89.193
27	0.494	1.412	90.605
28	0.492	1.404	92.009
29	0.474	1.353	93.363
30	0.459	1.313	94.675
31	0.438	1.253	95.928
32	0.427	1.219	97.147
33	0.403	1.151	98.298
34	0.349	0.998	99.296
35	0.246	0.704	100.000

, the results of the analysis of total variance explained show that the first six components have eigenvalues greater than 1, indicating that these factors are significant and explain 50.86% of the total variance before rotation. The first component alone explains 31.82%, but after rotation, its contribution decreases to 9.74%, while subsequent components, such as the second and third, increase their explained variance to 7.14% and 5.78%, respectively. This reflects a better distribution of variance among the components after rotation, facilitating their interpretation. The remaining components, starting with the seventh, explain minimal proportions of the variance, suggesting that they are not relevant to the model. Together, these six factors account for a significant amount of the variance and allow for a more balanced interpretation after rotation.

Table 4 shows the factor loadings of the items on the underlying factors obtained using a principal components analysis with Promax rotation (an oblique rotation that allows for correlation between factors). Each item has a significant loading on one of the five extracted factors, suggesting the presence of five different latent constructs. All factor loadings are greater than 0.50, which gives it a better interpretation. The following factors are included:

-

Factor 1 “Teacher interaction” is composed of the items PD1 (0.909), PD2 (0.919), and PD3 (0.599). These three items have high factor loadings, especially PD1 and PD2, which exceed 0.90, indicating that they are very strongly related to the factor. This factor may represent a clear and definite construct to which the items are strongly associated. PD3, although it has a lower loading (0.599), is still well-aligned with this factor, contributing to the same latent construct.

-

Factor 2 “Media” is composed of items PD5 (0.901) and PD6 (0.838). This factor has two items with high factor loadings, indicating that both items are strongly aligned with the factor. Loadings above 0.80 suggest that these items measure the underlying construct represented by this factor very well. Given the high magnitude of the factor loadings, this factor is very well-defined by items PD5 and PD6.

-

Factor 3 “Student collaboration” is composed of the items PD8 (0.589), PD9 (0.679), PD10 (0.731), and PD11 (0.595). The items associated with this factor also present high factor loadings, particularly PD10 (0.731) and PD9 (0.679), indicating that these items are strongly related to the construct represented by this factor. All items in this factor exceed a loading of 0.50, suggesting that they are adequately represented by this factor.

-

Factor 4 “Teaching process” is composed of the items PD17 (0.589), PD19 (0.558), PD20 (0.722), PD21 (0.813), PD22 (0.640), PD26 (0.554), PD27 (0.625), PD28 (0.623), and PD32 (0.525). This factor has a greater number of associated items, all with factor loadings exceeding 0.50, indicating a good association with the factor. PD21 stands out with a loading of 0.813, suggesting a strong alignment with this construct. Other items such as PD20 and PD22 also have a strong relationship with the factor, indicating that this factor is well-defined and represents a clear construct.

-

Factor 5 “Teacher support” is composed of the items PD33 (0.810), PD34 (0.780), and PD35 (0.513). This factor also has items with high factor loadings, especially PD33 (0.810) and PD34 (0.780), indicating a strong association with the underlying construct. PD35, although it has a lower loading (0.513), is still adequately represented in this factor.

Table 4. Underlying factors in teaching practice.

Items	Underlying Factors
	1	2	3	4	5
PD1		0.909
PD2		0.919
PD3		0.599
PD5					0.901
PD6					0.838
PD8			0.589
PD9			0.679
PD10			0.731
PD11			0.595
PD17	0.589
PD19	0.558
PD20	0.722
PD21	0.813
PD22	0.640
PD26	0.554
PD27	0.625
PD28	0.623
PD32	0.525
PD33				0.810
PD34				0.780
PD35				0.513

Note: Extraction method: principal component analysis. Rotation method: Promax with Kaiser normalization.

Table 4. Underlying factors in teaching practice.

Items	Underlying Factors
	1	2	3	4	5
PD1		0.909
PD2		0.919
PD3		0.599
PD5					0.901
PD6					0.838
PD8			0.589
PD9			0.679
PD10			0.731
PD11			0.595
PD17	0.589
PD19	0.558
PD20	0.722
PD21	0.813
PD22	0.640
PD26	0.554
PD27	0.625
PD28	0.623
PD32	0.525
PD33				0.810
PD34				0.780
PD35				0.513

Note: Extraction method: principal component analysis. Rotation method: Promax with Kaiser normalization.

Beyond those, item PD18 was eliminated for not containing factor loadings greater than 0.30, item PD4 was eliminated for having a variance ratio of 1.07, item PD12 was eliminated for having a variance ratio of 1.08, item PD23 was eliminated for having a variance ratio of 1.43, item PD29 was eliminated for having a variance ratio of 1.23, item PD25 was eliminated for having a commonality of 0.352, item PD7 was eliminated for having a commonality of 0.363, item PD24 was eliminated for having a commonality of 0.376, item PD31 was eliminated for having a variance ratio of 1.39, item PD30 was eliminated for having a variance ratio of 1.05, item PD14 was eliminated for having a commonality of 0.381, item PD15 was eliminated for having a commonality of 0.379, item PD13 was eliminated for having a commonality of 0.378, and item PD16 was eliminated for having a commonality of 0.391. In total, 14 items were eliminated.

The final results of the exploratory factor analysis for teaching practice show that the Kaiser–Meyer–Olkin measure (KMO) of sampling adequacy is 0.920, which indicates that the data are adequate to perform a factor analysis since a value greater than 0.9 is excellent. Bartlett’s test of sphericity is highly significant (p < 0.001), with an approximate chi-square value of 12,841.328 and 210 degrees of freedom (gl). These results suggest that the correlations between the items are not identifiable, so it is appropriate to perform a factor analysis.

Table 5. Final commonalities of the teaching practice items.

Item	Communality	Item	Communality
PD1	0.806	PD20	0.472
PD2	0.808	PD21	0.571
PD3	0.516	PD22	0.505
PD5	0.802	PD26	0.519
PD6	0.782	PD27	0.509
PD8	0.452	PD28	0.500
PD9	0.541	PD32	0.436
PD10	0.571	PD33	0.599
PD11	0.450	PD34	0.630
PD17	0.411	PD35	0.498
PD19	0.436

shows the final commonalities of the items in the exploratory factor analysis of teaching practice. Higher values, as in items PD1 (0.806), PD2 (0.808), and PD5 (0.802), indicate that these items have a stronger relationship with the underlying factors and are better explained by them. On the other hand, items with lower commonalities, such as PD17 (0.411) and PD19 (0.436), suggest that these items are less related to the factors, although they are still acceptable. Overall, most of the commonalities are moderate, indicating that the extracted factors explain the variance of the items in the teaching practice scale reasonably well.

Table 6. Final total variance explained by the underlying factors of teaching practice.

Component	Initial Eigenvalues			Sums of Squared Extraction Charges		Sums of Loads Squared by Rotation *
	Total	% Variance	% Accumulated	Total	% Variance	% Accumulated	Total
1	6.850	32.620	32.620	6.850	32.620	32.620	5.819
2	1.423	6.778	39.398	1.423	6.778	39.398	4.215
3	1.273	6.064	45.462	1.273	6.064	45.462	3.126
4	1.189	5.662	51.123	1.189	5.662	51.123	3.561
5	1.077	5.128	56.251	1.077	5.128	56.251	3.329
6	0.841	4.004	60.255
7	0.776	3.694	63.950
8	0.709	3.377	67.326
9	0.670	3.189	70.515
10	0.668	3.182	73.697
11	0.641	3.054	76.751
12	0.627	2.988	79.739
13	0.609	2.901	82.640
14	0.581	2.768	85.408
15	0.540	2.573	87.982
16	0.522	2.483	90.465
17	0.508	2.421	92.886
18	0.463	2.203	95.088
19	0.423	2.015	97.104
20	0.361	1.719	98.822
21	0.247	1.178	100.000

shows the results of the final total variance explained by the underlying factors of teaching practice, extracted through principal component analysis. The first five components have eigenvalues greater than 1, indicating that they are the most significant, explaining 56.251% of the total accumulated variance. The first component explains 32.620%, while the remaining components explain between 5% and 7% each. After rotation, the values of variance explained are more balanced, with the first component explaining 5.819% and the following ones around 3–4%, suggesting that the rotation redistributed the factor loadings, facilitating the interpretation of the factors. In summary, these five components are key in explaining most of the variance in the teaching practice scale.

In

Table 7. Factor loadings by the method of extraction of the teaching practice items.

Item	Extraction Methods
	Principal Components Analysis	Unweighted Least-Squares	Principal Axis Factorization	Alpha Factorization	Image Factoring
PD1	0.909	0.861	0.869	0.864	0.737
PD2	0.919	0.904	0.897	0.891	0.744
PD3	0.599	0.366	0.366	0.375	0.351
PD5	0.901	0.797	0.793	0.777	0.587
PD6	0.838	0.772	0.776	0.787	0.563
PD8	0.589	0.445	0.445	0.431	0.321
PD9	0.679	0.543	0.543	0.548	0.367
PD10	0.731	0.601	0.601	0.607	0.394
PD11	0.595	0.445	0.445	0.430	0.325
PD17	0.589	0.541	0.541	0.525	0.491
PD19	0.558	0.468	0.468	0.495	0.423
PD20	0.722	0.594	0.595	0.636	0.537
PD21	0.813	0.802	0.802	0.795	0.698
PD22	0.640	0.625	0.625	0.604	0.558
PD26	0.554	0.544	0.544	0.528	0.483
PD27	0.625	0.610	0.610	0.585	0.555
PD28	0.623	0.601	0.600	0.585	0.539
PD32	0.525	0.458	0.458	0.457	0.403
PD33	0.810	0.626	0.626	0.630	0.500
PD34	0.780	0.711	0.710	0.719	0.527
PD35	0.513	0.428	0.428	0.425	0.360

, the results are robust, or the data show robustness to the method of analysis employed. This means that the underlying patterns in the data remain consistent regardless of the extraction method or technique used, which is a sign that the underlying model is robust and reliable.

The final reliability analysis of the teaching practice scale indicated high internal consistency with a Cronbach’s alpha of 0.889, suggesting that the 21 items consistently measure the underlying construct. The omega coefficient is 0.733, an acceptable value, but somewhat lower, which could indicate some variability in the factor loadings of the items, although still adequate. Finally, the theta coefficient of 0.897 confirms the high reliability of the scale, showing that the instrument is consistent and adequate for measuring teaching practice.

The confirmatory factor analysis of teaching practice is presented in Figure 1.

From Table 8, the following can be observed:

-

The F1 factor is well-represented by items PD1, PD2, and PD3, with PD1 being the strongest indicator (0.717) and PD3 the weakest (0.660). This means that all items contribute significantly to the construct, but PD1 is the item that most influences the definition of F1, while PD3 has a somewhat smaller contribution.

-

The F2 factor is mainly defined by PD6 (0.852), which has a very strong relationship with the factor, while PD5 also contributes significantly (0.741), although to a lesser degree. PD6 is the best indicator of F2, suggesting that the aspects captured by this item are crucial to the measurement of the construct.

-

Factor F3 is moderately well-represented by its four items, with PD8 being the strongest indicator (0.615) and PD9 the weakest (0.560). Although all items are statistically significant, PD8 has the strongest influence on the definition of F3, while PD9 contributes the least.

-

Factor F4 is well-represented by items PD21, PD22, PD26, PD17, PD19, PD20, PD32, PD28, and PD27, with PD26 being the strongest (0.691) and PD20 the weakest (0.505). Although all items contribute significantly to the construct, PD26 has the strongest influence on the definition of F4, while PD20 has a lower weight in comparison.

-

In the case of factor F5, PD35 is the strongest indicator (0.699) and PD33 the weakest (0.470), indicating that PD35 plays a crucial role in defining the factor. Although PD33 is significant, its lower weight suggests that it contributes less to the F5 construct.

Table 8. Weights of the model for measuring teaching practice.

Relation			Coefficient		S.E.	C.R.	p-Value
			Estimated	Standardized
PD1	←	F1	1.000	0.717
PD2	←	F1	0.945	0.721	0.022	42.057	***
PD3	←	F1	0.953	0.660	0.042	22.815	***
PD5	←	F2	1.000	0.741
PD6	←	F2	1.055	0.852	0.041	25.569	***
PD8	←	F3	1.000	0.615
PD9	←	F3	0.935	0.560	0.055	16.926	***
PD10	←	F3	0.986	0.575	0.057	17.237	***
PD11	←	F3	0.956	0.586	0.056	17.014	***
PD21	←	F4	0.949	0.528	0.054	17.622	***
PD22	←	F4	1.190	0.653	0.055	21.552	***
PD26	←	F4	1.110	0.691	0.052	21.494	***
PD33	←	F5	1.000	0.470
PD34	←	F5	1.262	0.625	0.069	18.181	***
PD35	←	F5	1.366	0.699	0.082	16.727	***
PD17	←	F4	1.000	0.571
PD19	←	F4	1.062	0.576	0.055	19.258	***
PD20	←	F4	0.961	0.505	0.054	17.897	***
PD32	←	F4	1.101	0.623	0.053	20.732	***
PD28	←	F4	1.067	0.653	0.050	21.498	***
PD27	←	F4	0.981	0.648	0.049	20.034	***

Note: S.E. = standard error, C.R. = critical ratio. *** Significance at 0.1%

Table 8. Weights of the model for measuring teaching practice.

Relation			Coefficient		S.E.	C.R.	p-Value
			Estimated	Standardized
PD1	←	F1	1.000	0.717
PD2	←	F1	0.945	0.721	0.022	42.057	***
PD3	←	F1	0.953	0.660	0.042	22.815	***
PD5	←	F2	1.000	0.741
PD6	←	F2	1.055	0.852	0.041	25.569	***
PD8	←	F3	1.000	0.615
PD9	←	F3	0.935	0.560	0.055	16.926	***
PD10	←	F3	0.986	0.575	0.057	17.237	***
PD11	←	F3	0.956	0.586	0.056	17.014	***
PD21	←	F4	0.949	0.528	0.054	17.622	***
PD22	←	F4	1.190	0.653	0.055	21.552	***
PD26	←	F4	1.110	0.691	0.052	21.494	***
PD33	←	F5	1.000	0.470
PD34	←	F5	1.262	0.625	0.069	18.181	***
PD35	←	F5	1.366	0.699	0.082	16.727	***
PD17	←	F4	1.000	0.571
PD19	←	F4	1.062	0.576	0.055	19.258	***
PD20	←	F4	0.961	0.505	0.054	17.897	***
PD32	←	F4	1.101	0.623	0.053	20.732	***
PD28	←	F4	1.067	0.653	0.050	21.498	***
PD27	←	F4	0.981	0.648	0.049	20.034	***

Note: S.E. = standard error, C.R. = critical ratio. *** Significance at 0.1%

From

Table 9. Covariance weights of the teaching practice measurement model.

Covariability			Covariance	Correlation	S.E.	C.R.	p-Value
F2	↔	F5	0.199	0.589	0.016	12.202	***
F3	↔	F4	0.198	0.645	0.015	13.051	***
F2	↔	F4	0.218	0.63	0.016	13.61	***
F4	↔	F5	0.208	0.887	0.015	13.65	***
F3	↔	F5	0.194	0.647	0.016	11.881	***
F1	↔	F5	0.213	0.731	0.016	13.344	***
F1	↔	F3	0.219	0.576	0.017	12.792	***
F1	↔	F4	0.23	0.771	0.016	14.033	***
F2	↔	F3	0.211	0.477	0.018	11.888	***
F1	↔	F2	0.267	0.624	0.019	14.24	***
e1	↔	e2	0.171	0.526	0.018	9.483	***
e11	↔	e12	0.119	0.201	0.017	7.071	***
e12	↔	e13	0.097	0.162	0.016	5.95	***
e13	↔	e14	0.131	0.259	0.015	8.728	***
e15	↔	e16	0.038	0.118	0.009	4.06	***
e16	↔	e17	0.055	0.161	0.01	5.431	***
e19	↔	e20	0.084	0.124	0.02	4.135	***
e11	↔	e13	0.072	0.131	0.014	5.172	***
e6	↔	e9	−0.124	−0.187	0.023	−5.47	***

Note: S.E. = standard error, C.R. = critical ratio. *** Significance at 0.1%

, it can be observed that factors F4 and F5 are the most strongly correlated factors (0.887), indicating that the items measuring these factors share a large part of variance and are highly interdependent, which suggests that the dimensions that capture these factors of teaching practice are practically complementary. On the other hand, F2 and F3 have the lowest correlation (0.477), implying that they measure more differentiated aspects of teaching practice. Likewise, the correlations between the error terms are low, meaning that the observed variables are well-represented by their respective latent factors, with minimal shared variance not explained by the model, i.e., the presence of low correlations between the errors indicates that these correlations are not unduly influencing each other, which helps to maintain the integrity and validity of the estimates of the model measuring teaching practice.

From

Table 10. Measures of adjustment of the teaching practice measurement model.

Measure	Value Obtained	Threshold		Interpretation
		Acceptable	Ideal
CMIN/DF	2.195	1–3	1–2	Acceptable
RMR	0.037	<0.08	<0.05	Ideal
GFI	0.955	>0.90	>0.95	Ideal
AGFI	0.939	>0.90	>0.95	Acceptable
PGFI	0.703	>0.50	>0.60	Ideal
NFI	0.794	>0.90	>0.95	Not Acceptable
RFI	0.746	>0.90	>0.95	Not Acceptable
IFI	0.876	>0.90	>0.95	Not Acceptable
TLI	0.843	>0.90	>0.95	Not Acceptable
CFI	0.873	>0.90	>0.95	Not Acceptable
PRATIO	0.810	0.50–1.00	0.80–1.00	Ideal
PNFI	0.643	>0.50	>0.60	Ideal
PCFI	0.707	>0.50	>0.70	Ideal
RMSEA	0.025	<0.08	<0.05	Ideal
Hoelter (0.05)	1032	>200	-	Ideal
Hoelter (0.01)	1106	>200	-	Ideal

, it can be seen that the estimated measurement model presents an acceptable overall fit, with several measures reaching the ideal thresholds, such as RMR, GFI, PGFI, PRATIO, PCFI, RMSEA, and Hoelter’s indices, indicating a good representation of the observed data. Other measures, such as CMIN/DF, AGFI, IFI, and CFI, are within acceptable ranges, suggesting a reasonable, though not ideal, fit. However, measures such as NFI, RFI, TLI, and CFI do not reach the minimum acceptable values, indicating that the model could benefit from additional adjustments. Overall, the model is valid, but with opportunities for improvement to optimize its fit.

3.2. Results of Student Academic Stress

3.2.1. Exploratory Analysis of Academic Stress Data

The ICC value for single measures (0.313) indicates that the consistency of individual responses is low. In other words, students do not respond very consistently to the individual survey items, which could be due to variability in how they interpret the items or individual differences. Also, the ICC value for average measures (0.905) shows that when the items are considered as a whole (averaged), the scale has very high reliability. This suggests that the 21 items are measuring the construct consistently when the responses are averaged.

In the initial reliability analysis of the academic stress scale with 21 items, Cronbach’s alpha (0.911) and the theta coefficient (0.913) reflect excellent internal consistency, indicating that the items are highly correlated and accurately measure the construct of academic stress. Although the omega coefficient is slightly lower (0.786), it still indicates good reliability, being less sensitive to dimensionality. Taken together, the three coefficients suggest that the scale is highly reliable and suitable for assessing academic stress.

From

Table 11. Academic stress item and suppressed item statistics.

Item Code	Element Statistics		Elements Deleted
	Media	Standard Deviation	Total Correlation of Corrected Elements	Squared Multiple Correlation	Cronbach’s Alpha If the Item Has Been Deleted
EA1	2.90	1.093	0.470	0.250	0.908
EA2	3.56	1.029	0.558	0.506	0.906
EA3	3.44	1.021	0.561	0.471	0.906
EA4	3.09	1.193	0.474	0.429	0.908
EA5	3.30	1.158	0.481	0.438	0.908
EA6	3.16	1.013	0.489	0.270	0.908
EA7	3.44	1.001	0.588	0.412	0.906
EA8	3.19	1.186	0.555	0.404	0.906
EA9	3.30	1.102	0.597	0.485	0.905
EA10	3.03	1.057	0.590	0.399	0.905
EA11	3.30	1.052	0.570	0.369	0.906
EA12	3.04	1.066	0.548	0.392	0.906
EA13	3.23	1.079	0.573	0.432	0.906
EA14	3.29	1.045	0.655	0.618	0.904
EA15	3.32	1.042	0.641	0.604	0.904
EA16	3.06	0.997	0.583	0.419	0.906
EA17	2.94	1.026	0.528	0.405	0.907
EA18	2.70	1.087	0.489	0.412	0.908
EA19	2.86	1.116	0.468	0.482	0.908
EA20	3.13	1.212	0.445	0.402	0.909
EA21	3.32	1.104	0.623	0.448	0.905

, it can be observed that the highest value in the mean corresponds to item EA2 (3.56), indicating that students tend to give it a higher score compared to other items, while item EA18 (2.70) presents the lowest mean, suggesting that this item has a relatively lower evaluation. In terms of standard deviation, item EA20 (1.212) shows the highest variability in students’ responses, indicating a greater dispersion in scores, and item EA16 (0.997) presents the lowest variability, suggesting more consistent and uniform responses among students. The corrected total item correlation is highest for item EA14 (0.655), indicating that this item is highly correlated with the rest and contributes strongly to internal consistency, and the lowest value is for item EA20 (0.445), suggesting that this item has less relationship with the other items and could be affecting the consistency of the scale. In the squared multiple correlation, item EA14 (0.618) shows the highest proportion of its variance explained by the rest of the items, indicating a good alignment with the scale, and item EA1 (0.250) has the lowest value, suggesting that its variance is less explained by the other items, indicating a possible mismatch. Finally, Cronbach’s alpha if the item has been deleted is highest for item EA20 (0.909), indicating that removing it could improve the internal consistency of the scale, and the lowest value is for item EA14 (0.904), so no single item is noticeably decreasing the overall reliability of academic stress.

The summary statistics are given herein of the 21 items that make up the academic stress scale. The item means a range of 2.701 to a maximum of 3.561, with an overall mean of 3.172, suggesting that, in general, participants tend to score the items in the sometimes (3) and almost always (4) parts of the scale. As for the item variances, the values range from 0.995 to 1.469, with a mean variance of 1.170, indicating that the items present moderate variability in responses. Finally, the correlations among the items range from a minimum of 0.157 to a maximum of 0.737, with a mean correlation of 0.330, which implies that, although some items are more correlated with each other, most have moderate correlations, suggesting some independence among them within the construct of academic stress.

The Mardia (asymmetry = 3,903,451; kurtosis = 60,398), Royston (3,178,622), Henze–Zirkler (1240), and energy (9727) tests are consistent and show very low p-values (<0.001), indicating that the data for the academic stress items do not follow a multivariate normal distribution.

The variance inflation factor (VIF) values 21 items of the academic stress scale range between 1.33 and 2.62, which indicates a low level of multicollinearity among the items, i.e., there are no strong linear relationships among them, which is a good sign for the consistency of the instrument. Furthermore, it is worth noting that a VIF of less than 5 is generally acceptable; in this case, all items are significantly below that threshold. This reinforces the conclusion that multicollinearity is not a problem and provides confidence that the estimates obtained from subsequent analyses will not be affected by the presence of multicollinearity.

3.2.2. Exploratory Factor Analysis of Academic Stress

The KMO is 0.926, which indicates excellent sampling adequacy to perform the factor analysis since the items are highly correlated with each other. In addition, Bartlett’s test of sphericity is significant (p = 0.000), confirming that there are sufficient correlations among the variables to justify the use of factor analysis. Taken together, these results suggest that it is appropriate to continue with factor analysis and it is expected to provide a good representation of the underlying structure of the data.

In

Table 12. Initial communalities of academic stress items.

Item	Communality	Item	Communality	Item	Communality
EA1	0.287	EA8	0.555	EA15	0.648
EA2	0.693	EA9	0.643	EA16	0.476
EA3	0.620	EA10	0.548	EA17	0.534
EA4	0.681	EA11	0.446	EA18	0.595
EA5	0.697	EA12	0.560	EA19	0.701
EA6	0.330	EA13	0.570	EA20	0.537
EA7	0.497	EA14	0.661	EA21	0.569

, it is shown that the highest commonality corresponds to item EA19, with a value of 0.701, indicating that 70.1% of its variance is explained by the extracted factors, meaning that item EA19 is well-represented by the common factors and contributes significantly to the factor model. On the other hand, the lowest communality is that of item EA1, with a value of 0.287, indicating that only 28.7% of its variance is explained by the factors, meaning that item EA1 has a low representation in the factorial model, which makes it a candidate for revision or possible elimination since its contribution to the model is limited.

In

Table 13. The initial total variance is explained by the underlying factors of academic stress.

Component	Initial Eigenvalues			Sums of Squared Extraction Charges			Sums of Charges Squared by Rotation *
	Total	% Variance	% Accumulated	Total	% Variance	% Accumulated	Total
1	7.673	36.539	36.539	7.673	36.539	36.539	6.113
2	1.624	7.732	44.271	1.624	7.732	44.271	5.956
3	1.402	6.678	50.949	1.402	6.678	50.949	4.597
4	1.148	5.467	56.417	1.148	5.467	56.417	3.154
5	0.903	4.300	60.717
6	0.875	4.169	64.886
7	0.770	3.668	68.553
8	0.678	3.228	71.781
9	0.656	3.125	74.906
10	0.620	2.952	77.858
11	0.549	2.617	80.474
12	0.513	2.443	82.917
13	0.497	2.368	85.285
14	0.494	2.353	87.638
15	0.452	2.153	89.791
16	0.423	2.013	91.803
17	0.397	1.891	93.695
18	0.380	1.808	95.503
19	0.362	1.723	97.226
20	0.330	1.572	98.797
21	0.253	1.203	100.000

Note: * Extraction method: principal component analysis. *When the components are correlated, the sums of squared loadings cannot be added to obtain a total variance.

, the results of the analysis of total variance explained show that the first four components have eigenvalues greater than 1, indicating that these factors are significant and explain 56.417% of the total variance before rotation. The first component alone explains 36.539%, but after rotation, its contribution decreases to 6.113%, while subsequent components, such as the second and third, increase their explained variance to 5.956% and 4.597%, respectively. This reflects a better distribution of variance among the components after rotation, facilitating their interpretation. The remaining components, starting with the fifth, explain minimal proportions of the variance, suggesting that they are not relevant to the model. Together, these four factors account for a significant amount of the variance and allow for a more balanced interpretation after rotation.

Table 14 shows the factor loadings of the items on the underlying factors obtained using a principal components analysis with Promax rotation (an oblique rotation that allows for correlation between factors). Each item has a significant loading on one of the four extracted factors, suggesting the presence of four different latent constructs. All factor loadings are greater than 0.60, which gives it a better interpretation. The following factors are included:

-

Factor 1 “Overload and time constraints” is composed of the items EA2 (0.885), EA3 (0.767), EA14 (0.687), EA15 (0.714), and EA21 (0.616). These five items have high factor loadings, especially EA2 and EA3, indicating that they are very strongly related to the factor. This factor may represent a clear and distinct construct to which the items are strongly associated. EA21, although it has a lower loading (0.616), is still well-aligned with this factor, contributing to the same latent construct.

-

Factor 2 “Interaction and Participation” is composed of the items EA8 (0.778), EA9 (0.804), EA12 (0.772), and EA13 (0.698). This factor has four items without very high factor loadings, indicating that the four items are strongly aligned with the factor. Item EA9 has the highest predominance, while item EA13 has the lowest predominance.

-

Factor 3 “Technological Resources” is composed of items EA17 (0.601), EA18 (0.757), EA19 (0.897), and EA20 (0.744). Item EA19 has the highest predominance in the factor, while item EA17 has the lowest predominance in the factor.

-

Factor 4 “Distractions in the Study Environment” is composed of the items EA4 (0.816) and EA5 (0.810). These two items indicate a strong association with the underlying construct.

Table 14. Underlying factors of academic stress.

Item	Underlying Factors
	1	2	3	4
EA2	0.885
EA3	0.767
EA4				0.816
EA5				0.810
EA8		0.778
EA9		0.804
EA12		0.772
EA13		0.698
EA14	0.687
EA15	0.714
EA17			0.601
EA18			0.757
EA19			0.897
EA20			0.744
EA21	0.616		0.300

Note: Extraction method: principal component analysis. Rotation method: Promax with Kaiser normalization.

Table 14. Underlying factors of academic stress.

Item	Underlying Factors
	1	2	3	4
EA2	0.885
EA3	0.767
EA4				0.816
EA5				0.810
EA8		0.778
EA9		0.804
EA12		0.772
EA13		0.698
EA14	0.687
EA15	0.714
EA17			0.601
EA18			0.757
EA19			0.897
EA20			0.744
EA21	0.616		0.300

Note: Extraction method: principal component analysis. Rotation method: Promax with Kaiser normalization.

Beyond those, item EA1 was eliminated for not containing factor loadings higher than 0.30, item EA10 was eliminated for having a variance ratio of 1.30, item EA6 was eliminated for not containing factor loadings higher than 0.30, item EA7 was eliminated for having a variance ratio of 1.44, and items EA16 and EA11 were eliminated for not being robust to estimation with non-parametric methods.

The final results of the exploratory factor analysis for academic stress show a Kaiser–Meyer–Olkin measure (KMO) of sampling adequacy at 0.876, which indicates that the data are adequate to perform a factor analysis. Bartlett’s test of sphericity is highly significant (p < 0.001), with an approximate chi-square value of 11,317.158 and 105 degrees of freedom (gl). These results suggest that the correlations between the items are not identifiable, so it is appropriate to perform the factor analysis.

Table 15. Final commonalities of academic stress items.

Item	Communality	Item	Communality	Item	Communality
EA2	0.685	EA9	0.650	EA17	0.509
EA3	0.635	EA12	0.578	EA18	0.591
EA4	0.754	EA13	0.606	EA19	0.728
EA5	0.758	EA14	0.691	EA20	0.571
EA8	0.595	EA15	0.675	EA21	0.591

shows the final commonalities of the items in the exploratory factor analysis of academic stress. Higher values, as in items EA5 (0.758), EA4 (0.754), and EA19 (0.728), indicate that these items have a stronger relationship with the underlying factors and are better explained by them. On the other hand, items with lower commonalities, such as Ea17 (0.509) and EA20 (0.571), suggest that these items are less related to the factors, although they are still acceptable. Overall, most of the commonalities are moderate, indicating that the extracted factors explain the variance of the items in the academic stress scale reasonably well.

Table 16. Final total variance explained by the factors underlying academic stress.

Component	Initial Eigenvalues			Sums of Loads Squared by Extraction		Sums of Loads Squared by Rotation *
	Total	% Variance	% Accumulated	Total	% Variance	% Accumulated	Total
1	5.664	37.757	37.757	5.664	37.757	37.757	4.440
2	1.539	10.260	48.016	1.539	10.260	48.016	4.136
3	1.329	8.863	56.879	1.329	8.863	56.879	3.642
4	1.084	7.226	64.105	1.084	7.226	64.105	2.166
5	0.804	5.362	69.467
6	0.746	4.973	74.440
7	0.645	4.302	78.742
8	0.508	3.389	82.131
9	0.492	3.279	85.411
10	0.432	2.883	88.294
11	0.410	2.734	91.028
12	0.384	2.561	93.589
13	0.370	2.470	96.058
14	0.336	2.241	98.300
15	0.255	1.700	100.000

Note: * Extraction method: principal component analysis. *When the components are correlated, the sums of squared loadings cannot be added to obtain a total variance.

shows the results of the final total variance explained by the underlying factors of academic stress, extracted through principal component analysis. The first four components have eigenvalues greater than 1, indicating that they are the most significant, explaining 64.105% of the total accumulated variance. The first component explains 37.757%, while the remaining components explain between 10.260%, 8.863%, and 7.226% each. After rotation, the explained variance values are more balanced, with the first component explaining 4.440%, the second component 4.136%, the third component 3.642%, and the fourth component 2.166%, suggesting that the rotation redistributed the factor loadings, facilitating the interpretation of the factors. In summary, these four components are key in explaining most of the variance in the academic stress scale.

From

Table 17. Factor loadings according to the method of extraction of academic stress items.

Item	Extraction Method
	Principal Components Analysis	Unweighted Least-Squares	Principal Axis Factorization	Alpha Factorization	Image Factoring
EA2	0.885	0.781	0.779	0.818	0.575
EA3	0.767	0.659	0.658	0.663	0.480
EA4	0.816	0.680	0.681	0.671	0.513
EA5	0.810	0.704	0.702	0.709	0.518
EA8	0.778	0.677	0.676	0.660	0.566
EA9	0.804	0.766	0.766	0.771	0.612
EA12	0.772	0.647	0.647	0.653	0.546
EA13	0.698	0.627	0.626	0.648	0.517
EA14	0.687	0.667	0.669	0.631	0.651
EA15	0.714	0.685	0.687	0.644	0.668
EA17	0.601	0.469	0.469	0.493	0.424
EA18	0.757	0.616	0.617	0.640	0.549
EA19	0.897	0.908	0.908	0.913	0.693
EA20	0.744	0.615	0.615	0.573	0.562
EA21	0.616	0.529	0.529	0.523	0.459

, it can be seen that the results are robust, or the data show robustness to the method of analysis employed. This means that the underlying patterns in the data remain consistent, regardless of the extraction method or technique used, which is a sign that the underlying model is robust and reliable.

The final reliability analysis of the academic stress scale indicates a high internal consistency with Cronbach’s alpha of 0.879, suggesting that the 15 items coherently measure the underlying construct. The omega coefficient is 0.752, an acceptable value, but somewhat lower, which could indicate some variability in the factor loadings of the items, although it is still adequate. Finally, the theta coefficient of 0.882 confirms the high reliability of the scale, showing that the instrument is consistent and adequate for measuring academic stress.

3.2.3. Confirmatory Factor Analysis of Academic Stress

Figure 2 shows the Academic Stress Measurement Model (standardized loads).

From Table 18, the following can be observed:

-

The F1 factor is well-represented by items EA3, EA14, EA15, EA21, and EA2, with EA21 being the strongest indicator (0.780) and EA3 the weakest (0.610). This means that all items contribute significantly to the construct, but EA21 is the item that most influences the definition of F1, while EA3 has a somewhat smaller contribution.

-

Factor F2 is represented by items EA9, EA12, EA8, and Ea13, where the factor is mainly defined by EA13 (0.733), which has a very strong relationship with the factor, while EA12 also contributes, although to a lesser degree (0.630).

-

Factor F3 is moderately well-represented by its four items, with EA17 being the strongest indicator (0.622) and EA18 the weakest (0.548).

-

Factor F4 is well-represented by items EA4 and EA5, with EA5 being the strongest (0.809) and EA4 the weakest (0.753).

Table 18. Weights of the academic stress measurement model.

Relation			Coefficient		S.E.	C.R.	p-Value
			Estimated	Standardized
EA3	←	F1	0.973	0.610	0.032	30.182	***
EA14	←	F1	1.267	0.780	0.050	25.447	***
EA15	←	F1	1.239	0.763	0.049	25.180	***
EA21	←	F1	1.346	0.780	0.052	26.061	***
EA2	←	F1	1.000	0.621
EA9	←	F2	0.970	0.679	0.038	25.512	***
EA12	←	F2	0.862	0.630	0.044	19.571	***
EA8	←	F2	1.000	0.653
EA13	←	F2	1.009	0.733	0.043	23.632	***
EA17	←	F3	1.000	0.622
EA18	←	F3	0.939	0.548	0.041	22.721	***
EA19	←	F3	0.957	0.550	0.053	18.145	***
EA20	←	F3	1.063	0.558	0.061	17.292	***
EA4	←	F4	1.000	0.753
EA5	←	F4	1.047	0.809	0.049	21.223	***

Note: S.E. = standard error, C.R. = critical ratio. *** Significance at 0.1%

Table 18. Weights of the academic stress measurement model.

Relation			Coefficient		S.E.	C.R.	p-Value
			Estimated	Standardized
EA3	←	F1	0.973	0.610	0.032	30.182	***
EA14	←	F1	1.267	0.780	0.050	25.447	***
EA15	←	F1	1.239	0.763	0.049	25.180	***
EA21	←	F1	1.346	0.780	0.052	26.061	***
EA2	←	F1	1.000	0.621
EA9	←	F2	0.970	0.679	0.038	25.512	***
EA12	←	F2	0.862	0.630	0.044	19.571	***
EA8	←	F2	1.000	0.653
EA13	←	F2	1.009	0.733	0.043	23.632	***
EA17	←	F3	1.000	0.622
EA18	←	F3	0.939	0.548	0.041	22.721	***
EA19	←	F3	0.957	0.550	0.053	18.145	***
EA20	←	F3	1.063	0.558	0.061	17.292	***
EA4	←	F4	1.000	0.753
EA5	←	F4	1.047	0.809	0.049	21.223	***

Note: S.E. = standard error, C.R. = critical ratio. *** Significance at 0.1%

Overall, each factor has at least one key indicator that strongly defines it, while other items contribute to a lesser extent, but are still significant in representing the construct of academic stress.

From

Table 19. Covariance weights of the academic stress measurement model.

Covariability			Covariance	Correlation	S.E.	C.R.	p-Value
F1	↔	F3	0.311	0.792	0.02	15.871	***
F3	↔	F4	0.309	0.568	0.024	12.922	***
F2	↔	F4	0.344	0.516	0.026	13.022	***
F1	↔	F2	0.377	0.785	0.023	16.24	***
F1	↔	F4	0.302	0.553	0.023	13.243	***
F2	↔	F3	0.361	0.755	0.024	15.177	***
e1	↔	e2	0.277	0.440	0.021	13.328	***
e3	↔	e4	0.162	0.384	0.019	8.613	***
e6	↔	e7	0.162	0.227	0.025	6.352	***
e6	↔	e9	−0.055	−0.087	0.023	−2.402	0.016
e7	↔	e8	0.058	0.089	0.021	2.736	0.006
e8	↔	e9	0.083	0.143	0.024	3.489	***
e10	↔	e11	0.191	0.271	0.028	6.715	***
e10	↔	e12	0.078	0.109	0.023	3.385	***
e11	↔	e12	0.243	0.300	0.031	7.914	***
e11	↔	e13	0.065	0.074	0.027	2.415	0.016
e12	↔	e13	0.379	0.423	0.033	11.414	***

Note: S.E. = standard error, C.R. = critical ratio. *** Significance at 0.1%

, it can be observed that factors F1 and F3 are the most strongly correlated factors (0.792), indicating that the items measuring these factors share a large part of the variance and are highly interdependent, suggesting that the dimensions capturing these factors of academic stress are practically complementary. On the other hand, F2 and F4 have the lowest correlation (0.516), implying that they measure more differentiated aspects of academic stress. Likewise, the correlations between the error terms are low, meaning that the observed variables are well-represented by their respective latent factors with minimal shared variance not explained by the model, i.e., the presence of low correlations between the errors indicates that these correlations are not unduly influencing one another, which helps to maintain the integrity and validity of the estimates of the model measuring academic stress.

In

Table 20. Measures of fit of the academic stress measurement model.

Measure	Value Obtained	Threshold		Interpretation
		Acceptable	Ideal
CMIN/DF	4.062	1–3	1–2	Not acceptable
RMR	0.051	<0.08	<0.05	Acceptable
GFI	0.962	>0.90	>0.95	Ideal
AGFI	0.938	>0.90	>0.95	Acceptable
PGFI	0.585	>0.50	>0.60	Acceptable
NFI	0.879	>0.90	>0.95	Not acceptable
RFI	0.826	>0.90	>0.95	Not acceptable
IFI	0.906	>0.90	>0.95	Acceptable
TLI	0.863	>0.90	>0.95	Not acceptable
CFI	0.905	>0.90	>0.95	Acceptable
PRATIO	0.695	0.50–1.00	0.80–1.00	Acceptable
PNFI	0.611	>0.50	>0.60	Ideal
PCFI	0.629	>0.50	>0.70	Acceptable
RMSEA	0.040	<0.08	<0.05	Ideal
Hoelter (0.05)	606	>200	-	Ideal
Hoelter (0.01)	671	>200	-	Ideal

, the estimated measurement model presents an acceptable overall fit, with several measures reaching the ideal and acceptable thresholds, such as RMR, GFI, AGFI, PGFI, IFI, CFI, PRATIO, PNFI, PCFI, RMSEA, and Hoelter’s indices, indicating a good representation of the observed data. Other measures, such as CMIN/DF, NFI, RFI, and TLI, do not reach the minimum acceptable values, indicating that the model could benefit from additional adjustments. Overall, the model is valid, but with opportunities for improvement to optimize its fit.

3.3. Structural Equation Model

The estimation of the structural model was performed using asymptotic free distribution estimation, due to the multivariate non-normality of the data.

From Table 21, the following can be observed:

-

There is a negative and significant relationship between academic stress (F2) and teaching support (F5pd), suggesting that higher levels of academic stress are associated with lower teaching support perceived by students.

-

Teaching support (F5pd) positively influences the use of technological resources (F3ea) dimension of students’ academic stress, meaning that the higher the teaching support, the higher the levels of academic stress in the use of technological resources.

-

Teaching support (F5pd) positively influences the interaction and participation dimension (F2ea) of students’ academic stress, meaning that the greater the teaching support, the higher the stress levels in the interaction and participation dimension.

-

Overload and time constraints (F1ea) as a dimension of students’ academic stress are positively related to teaching practice (F1); although the effect is small, it means that teaching practice in virtual environments increases academic stress in students in the dimension of overload and time constraints.

-

Interaction and participation (F2ea) as a dimension of students’ academic stress has a negative and significant relationship with student collaboration (F3pd), which could indicate that higher levels of academic stress in the interaction and participation dimension do not promote collaboration among students.

-

Distractions in the study environment (F4ea) as a dimension of students’ academic stress are negatively associated with media (F2pd), suggesting that more distractions decrease the positive perception of media.

-

The use of technological resources (F3ea) as a dimension of students’ academic stress is negatively associated with the teaching process (F4pd), which could suggest that excessive use of technology could interfere with the perception of a good teaching process.

-

Overload and time constraints (F1ea) as a dimension of students’ academic stress are negatively related to teacher interaction (F1pd), suggesting that more overload decreases teacher–student interaction.

-

Distractions in the study environment (F4ea) as a dimension of students’ academic stress have a negative relationship with teacher interaction (F1pd), indicating that distractions also negatively affect student–teacher interaction.

-

Both measurement models present acceptable and positive factor loadings, signifying the direct relationship between the dimensions of each construct and the construct itself.

Table 21. Structural equation modeling of university teaching practice and academic stress of university students in virtual education.

Relation			Coefficient		S.E.	C.R.	p-Value
			Estimated	Standardized
F5pd	←	F2	−0.184	−0.163	0.029	−6.429	***
F5pd	←	F1	0.918	0.584	0.048	19.025	***
F2ea	←	F2	0.926	0.726	0.039	23.892	***
F4ea	←	F2	0.767	0.495	0.044	17.513	***
F3ea	←	F5pd	0.147	0.133	0.025	5.767	***
F3ea	←	F2	0.733	0.587	0.036	20.126	***
F2ea	←	F5pd	0.132	0.117	0.025	5.361	***
F1ea	←	F1	0.097	0.059	0.038	2.572	0.01
F1ea	←	F2	1.000	0.839
F2pd	←	F1	1.025	0.595	0.047	21.699	***
F4pd	←	F1	1.012	0.838	0.038	26.304	***
F3pd	←	F2ea	−0.063	−0.073	0.019	−3.387	***
F3pd	←	F1	0.861	0.566	0.043	20.238	***
F2pd	←	F4ea	−0.071	−0.089	0.016	−4.403	***
F4pd	←	F3ea	−0.045	−0.064	0.013	−3.417	***
F1pd	←	F1ea	−0.065	−0.073	0.020	−3.341	***
F1pd	←	F4ea	−0.038	−0.056	0.014	−2.748	0.006
F1pd	←	F1	1.000	0.678

Note: F1 = teaching practice in virtual environments, F1pd = teaching interaction, F2pd = means of co-communication, F3pd = student collaboration, F4pd = teaching process, F5pd = Do-Cent support, F2 = student academic stress in virtual environments, F1ea = time overload and constraints, F2ea = interaction and participation, F3ea = technological resources, F4ea = distractions in the study environment.

Table 21. Structural equation modeling of university teaching practice and academic stress of university students in virtual education.

Relation			Coefficient		S.E.	C.R.	p-Value
			Estimated	Standardized
F5pd	←	F2	−0.184	−0.163	0.029	−6.429	***
F5pd	←	F1	0.918	0.584	0.048	19.025	***
F2ea	←	F2	0.926	0.726	0.039	23.892	***
F4ea	←	F2	0.767	0.495	0.044	17.513	***
F3ea	←	F5pd	0.147	0.133	0.025	5.767	***
F3ea	←	F2	0.733	0.587	0.036	20.126	***
F2ea	←	F5pd	0.132	0.117	0.025	5.361	***
F1ea	←	F1	0.097	0.059	0.038	2.572	0.01
F1ea	←	F2	1.000	0.839
F2pd	←	F1	1.025	0.595	0.047	21.699	***
F4pd	←	F1	1.012	0.838	0.038	26.304	***
F3pd	←	F2ea	−0.063	−0.073	0.019	−3.387	***
F3pd	←	F1	0.861	0.566	0.043	20.238	***
F2pd	←	F4ea	−0.071	−0.089	0.016	−4.403	***
F4pd	←	F3ea	−0.045	−0.064	0.013	−3.417	***
F1pd	←	F1ea	−0.065	−0.073	0.020	−3.341	***
F1pd	←	F4ea	−0.038	−0.056	0.014	−2.748	0.006
F1pd	←	F1	1.000	0.678

Note: F1 = teaching practice in virtual environments, F1pd = teaching interaction, F2pd = means of co-communication, F3pd = student collaboration, F4pd = teaching process, F5pd = Do-Cent support, F2 = student academic stress in virtual environments, F1ea = time overload and constraints, F2ea = interaction and participation, F3ea = technological resources, F4ea = distractions in the study environment.

From

Table 22. Covariance weights of the structural equation model of university teaching practice and academic stress of university students in virtual education.

Covariability			Covariance	Correlation	S.E.	C.R.	p-Value
e8	↔	e9	0.061	0.097	0.019	3.208	0.001
e3	↔	e5	0.028	0.075	0.011	2.541	0.011

Note: S.E. = standard error, C.R. = critical ratio.

, it can be observed that the correlations between the error terms are low but significant, meaning that the corresponding dimensions share something in common not measured, which is indirectly picked up by the errors.

In

Table 23. Goodness-of-fit indicators of the structural equation model of university teaching practice and academic stress of university students in virtual education.

Measure	Value Obtained	Threshold		Interpretation
		Acceptable	Ideal
CMIN/DF	1.271	1–3	1–2	Ideal
RMR	0.010	<0.08	<0.05	Ideal
GFI	0.996	>0.90	>0.95	Ideal
AGFI	0.989	>0.90	>0.95	Ideal
PGFI	0.354	>0.50	>0.60	Not Acceptable
NFI	0.979	>0.90	>0.95	Ideal
RFI	0.952	>0.90	>0.95	Ideal
IFI	0.995	>0.90	>0.95	Ideal
TLI	0.989	>0.90	>0.95	Ideal
CFI	0.995	>0.90	>0.95	Ideal
PRATIO	0.444	0.50–1.00	0.80–1.00	Not Acceptable
PNFI	0.435	>0.50	>0.60	Not Acceptable
PCFI	0.442	>0.50	>0.70	Not Acceptable
RMSEA	0.012	<0.08	<0.05	Ideal
Hoelter (0.05)	2472	>200	-	Ideal
Hoelter (0.01)	3008	>200	-	Ideal

, the structural equation model presents an ideal fit in most of the key indicators, such as CMIN/DF (1.271), RMR (0.010), GFI (0.996), AGFI (0.989), NFI (0.979), RFI (0.952), IFI (0.995), TLI (0.989), and CFI (0.995), indicating that the model fits the data very well. However, parsimony indicators, such as PGFI (0.354), PRATIO (0.444), PNFI (0.435), and PCFI (0.442), do not reach acceptable thresholds, suggesting that the model is somewhat complex. The Hoelter (0.05:2472) and Hoelter (0.01:3008) values indicate that the sample size is adequate for the model fit to be reliable. Finally, the model has an excellent overall fit but could be improved in terms of parsimony.

Figure 3 presents the structural equation model of university teaching practice and the academic stress of university students in virtual education.

Although no significant relationship is found between the latent variables V1 (teaching practice) and V2 (student academic stress), the model shows that some specific dimensions of student academic stress (V2) have a negative influence on some dimensions of virtual teaching practice (V1). This implies that there is no direct and global relationship between both variables, but indirect effects are observed between some of their dimensions.

Explanation of negative influences:

• Student overload (dimension of student academic stress) hurts “teacher interaction” (dimension of teaching practice). This suggests that when students experience high levels of overload and feel they do not have enough time to meet academic demands, interaction with teachers decreases.
• Environmental distractions negatively affect both teacher interaction and the means of communication used in virtual teaching practice. This could be explained by the fact that a study environment with many distractions has an impact on the student’s attention and participation, which limits their ability to interact effectively with the teacher and take advantage of the available means of communication.
• Student interaction (dimension of student academic stress) has a negative influence on student collaboration (dimension of teaching practice). This indicates that when students feel that they are not sufficiently engaged or do not participate adequately, this affects their ability to collaborate effectively with their peers.
• Digital resources (dimension of student academic stress) negatively influence the teaching process (dimension of teaching practice). This reflects that when students experience problems with technology, this is detrimental to teaching in a virtual environment since the effectiveness of the process depends largely on the availability and functionality of technological resources.

The model suggests that academic stress related to specific problems in the student’s environment (technological, time, distractions, etc.) has a detrimental effect on how students interact and collaborate in virtual teaching. Although virtual teaching practice does not directly generate academic stress, high levels of stress do appear to impair some aspects of the educational dynamic.

Therefore, institutions and teachers should consider these dimensions of academic stress to mitigate its effects on virtual teaching by adjusting the workload, improving technological resources, and promoting more suitable study environments for students.

In the model we analyzed, student academic stress arises mainly from contextual and personal factors that affect their ability to manage the demands of virtual education. These factors are not directly dependent on teaching practice but are more linked to the circumstances in which students find themselves during their remote learning process.

Although student academic stress affects certain aspects of their educational experience, the model indicates that there is no direct and significant relationship between virtual teaching practice and stress. This is because the source of stress is not in the quality of teaching or the pedagogical strategies implemented by the teacher, but in external factors that the student faces during the virtual learning process.

Teaching practice may be effectively designed and offer adequate support, interaction, and collaboration; however, student stress arises from conditions outside the direct intervention of the teacher. For example, a teacher may organize their virtual classes in a clear and accessible way, but if the student faces technological problems or a noisy environment, or feels they have too many responsibilities, these external factors will continue to affect their learning as the main source of their stress.

Overall, the model indicates that students’ academic stress in this context does not depend directly on how the teacher implements instruction, but on contextual and environmental elements that are beyond the teacher’s control. Teacher–student interaction may be affected by student stress, but this does not mean that the stress is caused by the teacher’s actions or by the educational practice itself.

This study validated two key scales for virtual education: the Teaching Practices Scale in Virtual Education (TPSVE), which measures teaching practices, and the Student Academic Stress Scale in Virtual Education (SASSEV), which assesses academic stress in virtual environments. Both tools are valuable for future research and the development of more effective pedagogical strategies and are presented below for use by the academic community (

Table 24. Final scales of this study.

Item Code
Teaching Practices Scale in Virtual Education (TPSVE)
TP1	The instructor communicated efficiently.
TP2	The instructor communicated effectively.
TP3	The instructor was easily accessible.
TP5	The communication channels used by the instructor were sufficient.
TP6	The communication channels used by the instructor were useful.
TP8	Communication among students was frequent.
TP9	The student introductions at the course start helped us connect.
TP10	I formed study groups with other students in the course.
TP11	Earning extra points for answering peers’ questions motivated me.
TP17	The instructor was well-informed about the course material.
TP19	Repeated evaluations on the same topic aided learning.
TP20	Problems with solutions helped me self-assess.
TP21	The instructor provided clear deadlines for assignments and activities.
TP22	The instructor provided clear instructions for assignments and activities.
TP26	The course information was clear and easy to understand.
TP27	The instructor linked the course activities to the topics covered.
TP28	The assignments and activities aligned with course topics.
TP32	The instructor provided supportive materials to assist in my learning.
TP33	The instructor offered personalized attention.
TP34	The course structure helped me avoid information overload.
TP35	Online video conferences held by the instructor were useful.
Student Academic Stress Scale in Virtual Education (SASSEV)
SAS2	Overload of assignments and activities in courses.
SAS3	Overload of course materials (slides, books, handouts, videos, etc.).
SAS4	The presence of third parties in my study environment.
SAS5	Noise in my study environment.
SAS8	Requirement to activate camera and microphone during online classes.
SAS9	Being selected to answer questions by instructors.
SAS12	Voluntary participation in online classes.
SAS13	Presentation of assignments and activities in courses.
SAS14	Limited time to complete individual assignments and activities.
SAS15	Limited time to complete group assignments and activities.
SAS17	Use of software and simulators in course topics.
SAS18	Effectiveness of digital tools (Kahoot, Prezi, Dropbox, etc.) for completing and presenting assignments.
SAS19	Performance and reliability of my technological devices.
SAS20	Stability and speed of my internet connection.
SAS21	Deadlines for assignments and activities in courses.

).

4. Discussion

Our study does not find a significant relationship between the latent variables virtual teaching practice and student academic stress. This finding is consistent with a previous study [68], which found that there is no direct relationship between the use of educational platforms and academic stress in university students. The coincidence between both studies suggests that the effects of academic stress in the virtual context may not depend exclusively on the use of technological tools, but on how pedagogical strategies are designed and implemented within these platforms.

However, one study found a direct relationship between virtual learning environments and students’ coping with stress, suggesting that the way virtual learning is structured may influence how students manage stress [69].

Therefore, educators and designers of virtual environments must consider not only the integration of technologies but also the proper management of academic demands and the creation of environments that minimize sources of stress. This approach would help reduce the negative effects of specific dimensions of academic stress on teaching practice, thus optimizing the quality of the educational process.

The results obtained in this study highlight several critical dimensions that contribute to students’ academic stress and their interaction with teaching practice in virtual learning environments. One of the main findings indicates that the academic overload experienced by students negatively affects their interaction with teachers, a key aspect of the educational process [70]. This overload suggests the need to balance and dimension the tasks assigned among the different school subjects, to mitigate its adverse effects on the teacher–student dynamic.

In addition, environmental distractions were identified as hurting both teacher interaction and media use. This finding supports the existing literature on the importance of reducing environmental distractions to facilitate cognitive performance [71] and suggests that strategies such as mindfulness can play a role in helping students manage these distractions. [72]. In this context, it would be useful to design interventions that promote more controlled and focused learning environments.

Interaction among students, another factor associated with academic stress, showed a negative influence on student collaboration. This result reinforces the need to foster connectivity and teamwork skills in remote learning contexts, as explored through digital resources [73] and even tools optimized by artificial intelligence [74].

Finally, problems related to digital resources, such as limited access or lack of functionality, adversely affect the effectiveness of the teaching process in virtual environments. This underscores the importance of ensuring the availability and proper design of these resources to facilitate a more effective and less stressful learning experience for students. These gaps in teaching practice and student interaction suggest the need to expand existing research [75] to develop interventions that are replicable at various grade levels and educational contexts.

Overall, the findings and complementary literature highlight the importance of comprehensive strategies that consider both student well-being and the optimization of teaching practice, particularly in the context of virtual education.

On the other hand, it is important to highlight that the five-factor structure of teaching practice (teacher interaction, media, student collaboration, teaching process, and teacher support) is congruent with that proposed by Zabalza, who emphasizes that effective teaching practices are adaptive and reflect cognitive and attitudinal components [51].

In addition, the finding on the importance of “Teacher Interaction” and “Teacher Support” aligns with Biggs, who suggests that the relationship between teacher and student is critical to promote deep learning and decrease stress [50]. However, Perrenoud emphasizes that reflective practice must be continuous to adapt pedagogical strategies to changing contexts, and this study does not explore in depth how teachers adapt their methods based on feedback [55].

The factors identified for academic stress (deadline management, workload, and perception of teacher support) reflect the dimensions of academic stress found in studies such as those of Naranjo, who noted that academic stress stems from various sources, such as workload and teacher expectations [76].

In a further previous study, 50% of participants were stressed by short turnaround times on academic assignments [77]. Another study showed that 41.5% of those surveyed presented moderate stress, the frequent causes being the short time to complete tasks and exams, and in the face of this, they presented permanent tiredness, drowsiness, and listlessness [78].

Another study found that more than 50% of students present a moderate level of academic stress, which is produced by fear of failure, family pressure, changes in eating habits, and new responsibilities in college, among others [79]. Meanwhile, the results of a survey indicated that 60% of respondents had a moderate level of academic stress, while 35.9% had a severe level, which were due to the short times given to complete work, the overload of tasks, the exams, the demanding teacher, the lack of clarity on the tasks, and the frequent use of technology, which affected the health of the respondents [80].

Other results indicated that 47.1% of participants had high levels of academic stress, while 37.8% had moderate levels of stress; it was also detected that women had higher levels of stress, and likewise, the participants who are between 16 and 25 years old had higher levels of stress [81]. In another study, it was found that the causes of academic stress were the teachers’ evaluations, the large amount of homework, the time given to complete it, the lack of understanding of the subjects in class, and the attitude and character of the teacher [82]. Similarly, another study—where the average age of the respondents was 21 years old—found that 96.32% of the respondents had academic stress, 66.94% of whom experienced this at the moderate level. The elements that made them stressed most frequently were the way teachers evaluated them and their formative practices [83].

The findings of this study align with and are further enriched by recent advancements in artificial intelligence (AI) applications in educational settings, particularly in addressing academic stress. The systematic review Integrating Artificial Intelligence to Assess Emotions in Learning Environments [84] highlights the potential of AI-driven tools to assess students’ emotional states in real-time, providing opportunities for adaptive and personalized support. Incorporating such technologies into virtual learning environments could complement the teaching practices identified in this study, offering proactive interventions to mitigate stress and improve engagement.

Similarly, a study emphasizes the effectiveness of integrating IoT-enabled AI systems for recognizing stress and emotional responses in students. These systems can enhance teacher–student interactions and collaboration by enabling more responsive and empathetic virtual environments, aligning with the dimensions of interaction and support validated in the structural equation model [85].

Additionally, another study demonstrates the transformative role of AI in promoting efficient workload management and reducing stress levels. AI-driven solutions, such as automated scheduling and intelligent tutoring systems, can directly address mediators of academic stress identified in this study, such as overload and resource adequacy, by optimizing task distribution and providing tailored academic resources [86].

These studies collectively underscore the potential of AI to bridge gaps in virtual education by enhancing emotional recognition, optimizing the workload, and providing personalized learning experiences. Future research should explore the integration of AI-driven tools within the framework proposed by this study to validate their effectiveness in diverse educational contexts. Moreover, interdisciplinary collaborations could refine these technologies to better align with pedagogical goals and the psychological well-being of students.

In the context of the increasing digitization of higher education, the mediating role of digital skills and mobile self-efficacy in the stress and academic engagement of university students is becoming more and more fundamental [87], aligned with several studies highlighting the power of virtual reality to help students cope with anxiety, depression, and stress [88,89,90], including scientific evidence combining virtual reality training with mindfulness-based training to reduce academic stress [91,92,93].

Looking ahead, the integration of virtual reality tools into digital teaching practices offers a promising horizon. These technologies integrated with mindfulness practices not only enhance interactive learning but also directly address the emotional and psychological needs of students. Thus, a holistic approach that combines effective pedagogical practices with innovative technological interventions, such as those based on VR, can transform higher education into a space that prioritizes both academic learning and students’ holistic well-being.

5. Conclusions

In summary, the study explored the relationship between virtual teaching practice and students’ academic stress, identifying that there is no significant relationship between these latent variables globally, but there are indirect effects in specific dimensions. Among the main findings, it is highlighted that factors such as academic overload, environmental distractions, and problems with digital resources negatively affect aspects of teaching practice such as interaction with students and the use of media. These results underscore the importance of designing adaptive pedagogical strategies that reduce sources of stress and optimize learning dynamics in virtual environments. Furthermore, this study confirms that key components of teaching practice, such as interaction and support, are fundamental to promoting deep learning and reducing academic stress.

Among the main limitations of the study is its cross-sectional design, which prevents the establishment of causal relationships between the variables analyzed. In addition, the context restricted to the virtual environment limits the generalization of the findings to other educational scenarios, such as face-to-face or hybrid modalities. Another relevant aspect is the lack of analysis on how teachers adjust their pedagogical strategies according to the feedback received, an element that has been pointed out as essential in the specialized literature. Likewise, although key dimensions of academic stress were identified, this study did not consider mediating or moderating variables, such as coping strategies or social support, which could offer a more complete view of the dynamics studied.

As for future perspectives, it is proposed that longitudinal research should be developed, permitting us to better understand how the relationships between teaching practice and academic stress evolve. It would also be valuable to replicate this study at different educational levels and in various cultural contexts to evaluate the applicability of the results in other scenarios. On the other hand, it is recommended that mediating and moderating variables should be incorporated, such as students’ level of digital competencies, coping strategies, and social support, to obtain a more holistic understanding. Finally, it is suggested that intervention programs should be designed and implemented with a focus on reducing environmental distractions, balancing academic demands, and optimizing the use of digital resources, to improve both student well-being and the effectiveness of teaching practice in virtual environments.

Reliability and validity analyses confirmed that the instruments are consistent and adequate to assess these practices and the constructs of this study, offering the scientific and educational community two robust scales for Teaching Practice in virtual environments, which are broken down into the following dimensions: Teacher Interaction, Media Communication, Student Collaboration, Teaching Process, Teacher Support, and for Student Academic Stress, structured in the following dimensions: Student Overload, Student Interaction, Digital Resources, and Environmental Distractions.

This study addressed three key research questions to advance the understanding of the relationship between digital teaching practices and academic stress in virtual environments.

(1)

What are the psychometric properties (construct validity and internal reliability) of the Teaching Practices Scale in Virtual Education (TPSVE), and how suitable is its structure for evaluating teaching practices in virtual education?

The TPSVE demonstrated strong psychometric properties, with a Cronbach’s alpha of 0.889, indicating high internal reliability. Construct validity was confirmed through exploratory and confirmatory factor analyses, identifying key dimensions such as teacher interaction, technology-mediated communication, and the teaching process. The scale’s structure is well-suited to evaluating teaching practices in virtual contexts, providing a robust framework for future research and assessments in this domain.

(2)

What are the psychometric properties (construct validity and internal reliability) of the Student Academic Stress Scale in Virtual Education (SASSEV), and how suitable is its structure for evaluating academic stress in virtual education?

The SASSEV also exhibited high internal reliability (α = 0.879) and significant construct validity. This instrument effectively captured essential dimensions of academic stress, such as overload, environmental distractions, and adequacy of digital resources. The scale is validated as a comprehensive tool for assessing the factors contributing to academic stress in virtual environments, facilitating the identification of critical areas for intervention.

(3)

What is the relationship between digital teaching practices and academic stress among university students in virtual learning environments, and what are the key factors influencing this dynamic according to a theoretical model validated through structural equation modeling?

The validated structural equation model (SEM) revealed a significant inverse relationship between digital teaching practices and academic stress. Factors such as high-quality teacher interaction, technology-mediated communication, and teacher support have a positively impact in reducing academic stress. However, this relationship is moderated by mediators such as academic overload, environmental distractions, and insufficient digital resources. The findings emphasize that well-designed teaching practices aimed at personalizing learning and minimizing technological barriers are essential for mitigating academic stress and improving student well-being.

The findings provide a solid theoretical framework to guide future research. Further studies should investigate contextual and sociodemographic differences that may influence the relationship between digital teaching practices and academic stress, as well as evaluate specific interventions, such as emotional support programs and personalized pedagogical strategies. From a practical perspective, educational institutions can design inclusive policies that not only optimize virtual environments but also strengthen the resilience, self-regulated learning, and the overall well-being of students.

Building on the established framework, future research could explore innovative directions to deepen the understanding of this relationship. One promising avenue is the integration of advanced learning analytics to identify patterns of academic stress in real-time and provide adaptive interventions. Additionally, longitudinal studies could investigate the long-term effects of digital teaching practices on academic stress and student well-being across diverse educational contexts. Exploring the role of emerging technologies, such as virtual reality (VR) and artificial intelligence (AI), in creating immersive and supportive learning environments is another potential area of study. Finally, interdisciplinary research that combines educational psychology, technology design, and policy development could yield holistic strategies to further enhance the effectiveness of virtual education while minimizing its adverse impacts on students.