Relevant factors in the decision to desert of engineering students in the context of emerging economies

Introduction: Early dropout from engineering programs is a major concern for higher education institutions (HEIs) in emerging economies. Understanding the factors that influence students’ decisions to abandon their studies is essential for designing effective retention strategies.

Methods: This study uses a quantitative approach through Partial Least Squares Structural Equation Modeling (PLS-SEM), based on the Spady model. A survey was conducted with 190 engineering students from a private university in Medellín, Colombia.

Results: The analysis revealed that academic and emotional support significantly influence students’ decisions to continue their studies. Institutional commitment and structured support mechanisms provided by HEIs emerged as critical determinants of student retention.

Discussion: The findings underscore the importance of implementing university welfare strategies aimed at strengthening students’ academic and emotional well-being. These strategies serve as key levers to reduce dropout rates and reinforce the role of HEIs in supporting engineering students throughout their educational trajectory.

1 Introduction

The retention of students is a crucial concern for universities, as it is evaluated by government agencies and international organizations to ensure the academic quality of these institutions. Dropout rates have become a pervasive issue affecting a majority of higher education institutions (HEIs) worldwide (Alban and Mauricio, 2018).

Students may drop out of education for various reasons, such as academic challenges, social pressures, economic constraints, and psychological factors. This phenomenon has far-reaching implications, not only for the students themselves but also for their families and the HEIs they attend. Student attrition is a significant issue for HEIs due to the drain on resources it represents. In regions like Latin America, where resources may be limited, student attrition is particularly problematic. This situation intensifies inequalities and exacerbates the barriers that vulnerable students face (Vila et al., 2019). As Cerpa (2015) elucidates, one of the primary obstacles confronting the Sistema de Educación Superior Colombiano is the high rate of undergraduate academic dropouts. This challenge is particularly noteworthy given the recent surge in higher education enrollment in the country.

Garzón and Pérez (2012) propose that student attrition can be attributed to various factors, including financial costs, program types, students’ biographical and social backgrounds, as well as the perceived value of education and educational qualifications. In response to these factors, Colombia has implemented the System for the Prevention of Student Attrition in Higher Education (SPADIES). SPADIES assesses and classifies the risk of student attrition at educational institutions (Suárez-Montes and Díaz-Subieta, 2015). This demonstrates the government’s commitment to reducing higher education dropout rates.

Engineering programs have the highest rates of student attrition. The need for increased attention and the development of targeted strategies to reduce dropout rates among early-level students is emphasized by the most pronounced attrition during the initial semesters (Cerpa, 2015). According to the Ministerio de Educación Nacional – MEN (2022), the annual student attrition rate for university programs was 8.79% in 2018. Technological programs had a rate of 10.75%, and professional technical programs exhibited a rate as high as 17.41%. These figures have increased due to the global state of emergency caused by COVID-19 (Rincon et al., 2020).

Additionally, recent research shows that dropout rates in engineering programs are influenced not only by academic and economic factors but also by psychological pressure and a specific culture of stress within these fields (Mirabelli et al., 2025). This stressful environment, characterized by high demands and expectations, can negatively impact students’ wellbeing and motivation, increasing their vulnerability to academic abandonment. Therefore, understanding the psychosocial dimension of the student experience in engineering is essential to developing comprehensive interventions that promote retention and academic success.

Furthermore, analyzing contextual and structural factors is crucial, especially in developing countries where resource limitations and unequal access to institutional support exacerbate the dropout problem (Kocsis and Molnár, 2025). In this regard, studies like this one are vital to provide local evidence that can inform educational policies tailored to the specific realities of each institution and region, thereby contributing to improved retention and graduation rates in high-demand and socially relevant programs such as engineering.

Therefore, it is crucial to identify the relevant factors that contribute to student attrition in engineering programs. This is essential for developing effective strategies to reduce the frequency of this issue in Colombia (Cerpa, 2015). With this objective in mind, this research aims to identify the most significant factors that influence the decision of engineering students to drop out at the Corporación Universitaria Americana - CUA on its Medellín campus in Colombia.

2 Theoretical framework

In higher education, attrition refers to a student’s decision to withdraw from their academic program (Maldonado et al., 2021). There are five main approaches to analyzing student attrition, each offering explanatory variables to understand the underlying causes. These approaches include the psychological, economic, sociological, organizational, and interactional perspectives, as explained by Himmel (2002). These categories can be grouped into three overarching clusters: personal, family, and institutional characteristics (Donoso and Schiefelbein, 2007). Models have been developed to comprehensively elucidate student attrition as a multidimensional phenomenon.

The organizational category is closely related to the college or university. According to Díaz (2009), pre-university characteristics intersect in this category and significantly affect students’ initial commitment to the institution and academic goals. Students’ perceptions depend on their level of academic and social integration with the institution. In this ongoing process, the institution’s commitment depends on a positive perception, which encourages the student to continue their educational journey within the program and institution (Donoso and Schiefelbein, 2007).

Interaction involves a student’s constructive relationship with both peers and faculty, fostering opportunities for engagement that provide support and motivation. Within this framework of social integration, students reconsider the possibility of abandoning their academic pursuits. This interaction is closely linked to the concept of academic integration and is complemented by informal interpersonal relationships that help to cultivate academic confidence. As a result, their commitment to the institution is strengthened (Pineda-Báez and Ortíz, 2009).

The sociological literature offers a comprehensive framework for comprehending the attrition process (Maldonado et al., 2021). Significant factors have been identified within this category, including social integration, family environment, university integration, normative consistency, institutional commitment, and peer support (Jorquera et al., 2018). Furthermore, in addition to psychological factors, other external elements have been identified, such as academic and social integration, gender, socioeconomic status, program quality, and average academic performance for each semester (Díaz, 2009; Spady, 1971).

The turnover phenomenon has been studied through the psychological approach, which examines the personality traits of individuals. Scholars, such as Fishbein and Ajzen (1975), have presented a perspective rooted in beliefs and attitudes. Accordingly, factors such as perceived usefulness, perceived ease of use, subjective norms, and attitude help explain their influence on the intention to withdraw from an academic program (Ramírez-Correa and Grandón, 2018). In addition, Jorquera et al. (2018) identified additional factors such as achievement behavior, academic self-concept, perceived difficulty of study, and level of aspiration.

Attinasi (1989) expanded the model to enhance our understanding of pre- and post-enrollment attitudinal patterns and behaviors, allowing for a more comprehensive analysis of attrition in terms of students’ evaluations of their post-enrollment college experience (Himmel, 2002). The study of attrition has continued to evolve. Ethington (1990) developed a more sophisticated framework that incorporated performance behaviors, such as persistence, choice, and achievement. Additionally, the framework included elements such as family support and encouragement, as well as values and expectations of success (Donoso and Schiefelbein, 2007).

The Spady model, developed by Spady (1971) (Figure 1), is one of the most widely accepted models for analyzing college student attrition. The researcher drew inspiration from Durkheim’s study of the principles of suicide and identified parallels that could be applied to the phenomenon of dropping out. Both studies emphasize that the decision to drop out is not only influenced by individual factors but also involves an individual’s disengagement from their social system. Therefore, this phenomenon can be better understood as a result of the difficulties encountered when integrating into society (Vásquez and Miranda, 2019).

Figure 1

Figure 1. Spady model (Spady, 1971).

Although several models have been proposed to analyze student attrition-such as the model proposed by Tinto (1975, 1994), the model proposed by Bean (1980), as well as the model proposed by Pascarella and Terenzini (1980), the integrated model of Cabrera et al. (1993), and the retention formula of Seidman (2005)—this study adopts Spady’s model (Spady, 1971) because of its emphasis on the interaction between academic, social, and personal factors. This model provides a fundamental structure particularly suited to the multidimensional context of attrition in engineering programs in emerging economies. However, future research could benefit from comparative analyses using more contemporary or integrative models.

2.1 Family background

According to Spady (1971), a student’s decision to drop out of an academic program can be influenced by their family environment. The family environment can affect a student’s social integration, relationships with the institution, peers, and teachers due to various influences, expectations, and demands. This cumulative effect ultimately leads to better integration into the university environment, which improves student retention (Himmel, 2002). The study identified predictors of student attrition mainly rooted in the early stages of students’ university life, typically within the first 2 years (Björklund and Salvanes, 2011). Therefore, the following research hypothesis is formulated:

H1: The academic potential of an individual is directly influenced by their family background.

H2: Normative consistency is also directly affected by family background.

2.2 Academic potential

Academic potential refers to a college student’s abilities and expected future performance that has not yet been demonstrated or anticipated. This concept illustrates the extent to which students believe they possess qualities that will enable them to develop skills and abilities in the academic field in the future (Akin and Akin, 2016). According to Mori Sánchez (2012), as students progress in their academic programs, the potential or shortcomings of their study plan become more apparent. As a result, students evaluate learning spaces more rigorously, especially due to the increased access they have in the labor market. This may have an impact on students’ academic performance. Therefore, the following hypothesis is proposed:

H3: Academic potential has a direct effect on academic performance.

2.3 Normative consistency

In this context, normative consistency refers to the motivation behind students’ choice of a topic (Georg, 2009). It has been noted that this regulation involves coherence and compliance with social norms. Students’ participation involves alignment with the social norm that is considered compatible with their motivations (Borden, 1988). Therefore, in the context of the Spady model, it refers to the level of compatibility between students’ dispositions, interests, attitudes, and expectations and the behaviors, expectations, and demands they may encounter through interactions with various individuals in the university environment (Borden, 1987). The model proposes that normative consistency has an impact on academic performance, social integration, intellectual development, institutional commitment, and the student’s integration with peers. As a result, the following research hypotheses are suggested:

H4: Normative consistency has a direct effect on academic achievement.

H5: Normative consistency has a direct effect on intellectual development.

H6: Normative consistency has a direct effect on social integration.

H7: Normative consistency has a direct effect on peer support.

H8: Normative consistency has a direct effect on institutional commitment.

2.4 Peer support

According to the study conducted by Valencia-Arias et al. (2023), peer support assesses students’ satisfaction with teachers’ teaching methods. As outlined by John et al. (2018), peer support involves providing understanding and care to someone in an empathetic manner through the exchange of emotional and psychological experiences. It serves as a system for respectfully giving and receiving help through mutual agreement. In this way, students can establish relationships, nodes, and interactions-whether cognitive, didactic, or social-especially when teachers are not always adequately prepared (Estrada-Molina and Fuentes-Cancell, 2022). According to the Spady model, peer support has a significant impact on academic achievement, intellectual development, and social integration (Hadjar et al., 2023). Therefore, the following hypotheses are proposed:

H9: Peer support has a direct effect on academic achievement.

H10: Peer support has a direct effect on intellectual development.

H11: Peer support has a direct impact on social integration

2.5 Intellectual development

Intellectual development is a by-product of educational outcomes (Carmo Nicoletti, 2019). Chalela-Naffah et al. (2020) suggest that the effectiveness of higher education institutions can be reflected in intellectual development. This is demonstrated through changes in institutional demands, curriculum flexibility, and academic program credit structure. Intellectual development involves problem-solving, informed decision-making, and understanding complex concepts (Martinez Ruiz et al., 2023). According to Spady’s abandonment process model, intellectual development is directly related to social integration (Black, 2023). Based on this correlation, we propose the following hypothesis:

H12: Intellectual development has a direct impact on social integration.

2.6 Academic performance

Academic performance refers to the evaluation of university students’ achievements throughout their academic program (Castrillón et al., 2020). Research indicates that academic performance is a crucial predictor of dropout rates (Martínez-Navarro et al., 2021). Therefore, institutions can proactively identify students at risk of dropping out by consistently monitoring their academic performance and implementing early interventions (Ortiz-Lozano et al., 2023). For engineering students, there is evidence that academic performance has a positive effect on their intent to persist, especially during their first year of study (Marschall et al., 2023). Therefore, we propose the following research hypothesis based on this body of evidence.

H13: Academic performance has a direct effect on the decision to desert.

2.7 Social integration

Beyond individual pre-enrollment and background characteristics, students’ experiences within higher education institutions are crucial in predicting their decision to drop out. Social integration encompasses the relationships between students and teachers, as well as interactions among students. High social integration is contingent on quality and, consequently, satisfaction (Piepenburg and Beckmann, 2022). Consequently, research emphasizes the significance of promoting academic and social integration as a preventive measure against dropout (Franz and Paetsch, 2023). Studies have consistently shown a significant relationship between social integration and university student satisfaction (Myrtveit et al., 2017). Therefore, we propose the following research hypothesis:

H14: Social integration has a direct effect on satisfaction.

2.8 Satisfaction

Student satisfaction is a crucial factor in determining the success and stability of academic institutions. According to Beck and Milligan (2014), students who are loyal and satisfied tend to achieve higher grades, better test scores, and have lower dropout rates than their less committed peers, which positively impacts institutional commitment. In this context, student institutional commitment is related to overall student satisfaction, sense of belonging, perceptions of educational quality, and willingness to return to the institution (Strauss and Volkwein, 2004). The research hypothesis proposed is as follows:

H15: Satisfaction has a direct effect on institutional commitment.

2.9 Institutional commitment

Based on the research conducted by Lizarte Simón and Gijón Puerta (2022), a student’s commitment to both the institution and their chosen academic path significantly influences their decision to persist to graduation. The theoretical framework suggests that students have varying expectations and commitments when it comes to the university environment. Institutional commitment, in particular, refers to the commitment to continue studying at a specific university (Schuster and King, 2022). Once enrolled, students’ academic commitment is likely to change as they gain experience in both the academic and social aspects of the institution. Research has shown that higher levels of institutional engagement are associated with lower attrition rates compared to less engaged students (Beck and Milligan, 2014). Therefore, we propose the following research hypothesis:

H16: Institutional commitment has a direct effect on the decision to leave.

2.10 Decision to desert

According to Spady (1971) the decision to desert is a social process, where the decision is the result of the interaction between academic/social integration and personal conditions. In addition, it is considered a dynamic process of institutional disengagement (Tinto, 1994). In this sense, the decision to desert is considered a predictor of the actual decision (Hadjar et al., 2023). According to Valencia-Arias et al. (2023) the decision to desert out is based on aspects such as family problems, lack of financial support, lack of vocational and professional guidance, differences in academic requirements between school and university, the number of credits needed to complete a degree, difficulty in paying tuition, difficulty in meeting additional costs, and psychological factors.

Alt text Figure 1. This figure depicts the model proposed by Spady in 1971. It provides a graphical visualization of the theoretical or conceptual framework that Spady developed in his research.

According to Spady, full integration necessitates meeting the demands of both the academic and social systems within higher education (Donoso and Schiefelbein, 2007). Studies have identified key factors that are now commonly used to describe and predict attrition in educational institutions (Maldonado et al., 2021).

3 Methodology

To achieve the research objective, we conducted a quantitative study with a correlational scope using a structural equation model. We administered a survey through Google Forms to students enrolled in the Systems Engineering and Industrial Engineering programs. A survey was conducted with 190 students, the majority of whom were enrolled in the Systems Engineering program (68%). The participants were predominantly male (76.5%), belonging to stratum 1 or 2 (53.5%), in their first five semesters (66.6%), and aged between 15 and 21 years (44.5%).

The proposed structural equation model is composed of 10 latent constructs derived from the model proposed by Spady (1971): Family Background, Normative Consistency, Social Integration, Satisfaction, Peer Support, Institutional Commitment, Intellectual Development, Academic Performance, Academic Potential, and Decision to Desert. Each of these constructs was operationalized by means of between 2 and 5 observable items, designed from the specialized literature and validated by experts. Table 1 presents the detailed correspondence between each construct and its observable indicators, together with the statements that were included in the data collection instrument.

Table 1

Table 1. Study factors and indicators.

The survey aimed to understand the reasons for dropping out by examining factors such as the secondary school attended, graduation age, marital status upon university entry, age of university entry, economic dependency, parental education level, and financial aid received. Table 2 provides additional socio-demographic information about the students.

Table 2

Table 2. Sample information.

Among the surveyed students, 83.2% reported receiving their secondary school diploma from a public school. Additionally, 61.3% of the students reported financial dependence on a family member or someone else. Upon entering university, the majority (80.6%) were single, and half of them (50.3%) had no financial dependents.

Regarding the age at which they received their secondary school diploma, 68.6% responded that they were between 17 and 19 years old. Similarly, when asked about their age upon entering university, the majority stated they were between 17 and 19 years old. This suggests that a significant portion of the surveyed students completed their secondary education and immediately began their professional education.

With regards to their parents’ education, 54% reported that their fathers had completed primary or secondary education, and 61.3% reported the same for their mothers. When asked about scholarships and loans for professional education, 88% of respondents reported not receiving a scholarship, and 74.3% reported not taking out a loan.

This socio-demographic information establishes a profile for the average engineering student at the Medellín campus of the CUA. The profile aligns with that of a recently graduated young individual who is financially dependent on a family member and whose parents do not hold professional degrees.

4 Results

The Partial Least Squares Structural Equation Modeling (PLS-SEM) method was used to estimate the parameters of the proposed structural model. PLS-SEM is a variance-based technique widely used in the social sciences when the research objective is predictive and the theoretical model is complex or exploratory (Hair et al., 2011). This method allows the estimation of causal relationships between latent variables by maximizing the explained variance of the dependent constructs. Unlike the covariance-based SEM, PLS-SEM is particularly useful when the sample size is small, the data distribution is non-normal, or when the model includes numerous indicators and constructs (Hair et al., 2016).

The PLS-SEM approach consists of two submodels: the measurement model (external model), which assesses the relationships between the observed variables (indicators) and their corresponding latent constructs, and the structural model (internal model), which estimates the relationships between the latent variables. The estimation procedure is based on the partial least squares method, conceptually linked to principal component analysis (PCA). Key PLS-SEM results include factor loadings, cross-loadings, composite reliability, Cronbach’s alpha, average variance extracted (AVE), R² and Q² coefficients, and path coefficients, which are evaluated through bootstrap procedures to assess their statistical significance (Hair et al., 2017).

A Structural Equation Model was used to predict dropout behavior among engineering students through variance analysis, employing the PLS-SEM approach. This technique reflects both theoretical and empirical conditions in the realms of social and behavioral sciences (Martínez Ávila and Fierro Moreno, 2018). Zeng et al. (2021) emphasize that PLS-SEM is a causal modeling method that aims to maximize the explained variance of dependent latent constructs rather than constructing a theoretical covariance matrix. Therefore, the study assesses the convergent and discriminant validity of the measurement model, conducts hypothesis testing, and performs prediction analysis of the structural model.

4.1 Convergent validity

Model estimation is a crucial step in PLS-SEM. It involves determining an initial solution and, more importantly, identifying the dimension of the data, which is the most appropriate number of factors (Ferrando and Anguiano-Carrasco, 2010). This process is carried out through principal component analysis, where components are seen as a linear combination of the observed indicators or variables (Alaminos et al., 2015). Factor loadings are essential in principal component analysis as they serve as correlation coefficients between variables and factors. The analysis clusters variables into a few latent constructs. Variables with loadings higher than 0.4 are grouped under the same factor, while those with lower loadings should be considered for elimination (Balasundaram, 2009). Table 3 presents the observed indicators selected for the 10 constructs of the student dropout model. The analysis excluded items AF1, AF5, PA1, AP2, and CI.

Table 3

Table 3. Convergent validity.

In this study, the observed variables—or indicators—correspond to specific survey items designed to measure the latent constructs derived from the Spady model, such as family background, academic potential, normative consistency, peer support, academic performance, intellectual development, social integration, satisfaction, institutional commitment, and dropout decision. Each construct was operationalized using between 3 and 5 items based on previous literature and expert validation. These indicators were grouped under their respective constructs through principal component analysis (PCA), which identifies uncorrelated factors (i.e., the latent constructs) based on their linear combinations of observed variables.

The factor loadings, in this context, represent the correlation coefficients between each indicator and its associated construct. A threshold of 0.4 was used to retain the indicators, following standard recommendations. The elimination of indicators with loadings below this value was considered to improve the convergent validity of the model. Specifically, items FB1, F5, PS2, and IC were excluded based on this criterion, as detailed in Table 3.

Although some variables presented cross-loadings higher than 0.4 on more than one construct, the final grouping was based on content validity, expert judgment, and theoretical consistency with Spady’s model. Thus, items AP1, AP2, and AP3, although with some loading on Family Background, were grouped under Academic Performance for conceptual consistency.

Convergent validity was assessed by calculating the AVE, a statistic that tests the true validity of the population model (Afthanorhan et al., 2020). The literature recommends that the AVE be above 0.5 (Amaro et al., 2015) In addition, we evaluated the variance inflation factor (VIF) to assess the formative collinearity of the indicators. Purwanto and Sudargini (2021) suggest that a VIF value of 5 or higher indicates a significant collinearity problem between formatively measured indicators.

They recommend a lower value, preferably below 3. To evaluate the internal consistency of the constructs, Cronbach’s Alpha (CA) and Composite Reliability (CR) were used. Although CA is widely used for measuring reliability, researchers have increasingly turned to other reliability coefficients, such as CR, due to their higher scores (Fauzi, 2022). Hair et al. (2014) argue that CR is a more appropriate measure of internal consistency reliability than CA for two reasons. Firstly, unlike CA, CR does not assume that all indicator loadings are equal in the population, which aligns with the working principle of the PLS-SEM algorithm that prioritizes indicators based on their individual reliabilities during model estimation. Secondly, it is important to note that CA is sensitive to the number of items in the scale and may underestimate internal consistency reliability. However, by using CR, PLS-SEM can accommodate different indicator reliabilities, thus avoiding the underestimation associated with CA. It is generally accepted that both statistics should have a value of 0.7 or higher (Gutiérrez Rodríguez et al., 2020).

All constructs were considered to be reflexive, given that the observable items reflect the same latent phenomenon and are interchangeable. This reflexive nature is evidenced by factor loadings above 0.7 for most of the indicators, as well as adequate CR and AVE values, presented in Table 3. In addition, discriminant validity was assessed using the Fornell-Larcker criterion (Table 4).

Table 4

Table 4. Fornell-Larcker criterion.

4.2 Discriminant validity

To increase confidence in the interpretation of findings, it is crucial to evaluate discriminant validity. This validity indicates the absence of statistical correlations between test scores, which means that the tests do not measure the same processes. In other words, the correlation between indicators of one construct must be higher than the correlation between those indicators and the proposed measures for another construct (Afthanorhan et al., 2021).

Various methods exist for assessing inter-construct validity after analyzing the internal consistency (convergent validity) of the model. Discriminant validity is established when all correlations between indicators of different constructs, such as variables X and Y, are significant, and each of these correlations is higher than all correlations between indicators of both variables (Rönkkö and Cho, 2022).

The Fornell-Larcker criterion is a commonly used statistic for assessing discriminant validity. According to this criterion, a latent variable must capture more variance in its associated indicator variables than it shares with other constructs in the same model to establish discriminant validity (Fornell and Larcker, 1981). This is achieved by comparing the AVE with the squared correlations with other constructs in the model. The criterion being met is indicated by higher values for each construct, as stated by Henseler et al. (2015) and demonstrated in Table 4.

4.3 Hypothesis testing

This section evaluates the structural model through hypothesis testing. Table 5 and Figure 2 illustrate the relationship between the p-values. The p-value test establishes that β > 0 with a significance level of 0.05. This test is complemented by the T-value test, which is calculated as β/σ, with a threshold >1.96 (Kock, 2016). The Bootstrapping method is utilized for calculating hypotheses, as it is the most commonly used method for estimating the standard error in PLS SEM (Kock, 2014).

Table 5

Table 5. Hypothesis testing.

Figure 2

Figure 2. Hypothesis testing - Spady model (1971).

Additionally, the model’s predictive capacity is evaluated using the coefficient of determination R². This assesses the model’s fit in the composite scores sample of endogenous constructs by predicting individual case values in the total sample using model estimates (Hair Jr, 2021). According to the literature, values of 0.75 are considered substantial, 0.5 moderate, and 0.25 weak (Hair et al., 2019). These results are also presented in Figure 2.

Alt text Figure 2. This figure illustrates the hypothesis testing process as applied specifically to the 1971 Spady model. It provides a visual representation of how the hypotheses associated with this model were evaluated and tested.

The study results allow for the identification of validated hypotheses and those that are not. The strongest relationship is observed between academic potential and academic performance, as evidenced by the indicators used in the questionnaire, specifically, items related to students’ perceptions of their self-discipline, perseverance, and ability to understand content when in optimal conditions. These results suggest that students who report stronger cognitive and motivational readiness are more likely to demonstrate consistent academic outcomes.

Similarly, the strong relationship identified between satisfaction and institutional commitment reflects questionnaire items addressing students’ access to financial support, perceived return on investment, and expectations of future employment. When students report satisfaction with institutional services and prospects, they tend to express higher levels of institutional commitment. These interpretations are grounded in the constructs and indicators operationalized through the validated survey instrument.

Additionally, there is a strong correlation between peer support and intellectual development. Peer interaction in a collaborative environment stimulates the exchange of ideas and perspectives, promoting deeper learning through discussion and joint analysis of concepts. Peer interaction in a collaborative environment stimulates the exchange of ideas and perspectives, promoting deeper learning through discussion and joint analysis of concepts. This contributes to the construction of knowledge more comprehensively. Furthermore, a strong relationship exists between normative consistency and peer support. When normative consistency exists, students share similar expectations and values, creating an environment conducive to empathy, support, and mutual understanding.

The influence of family background on academic potential is highlighted, as it provides a supportive environment and resources that facilitate educational development. A home with a positive atmosphere toward learning, where the importance of education is encouraged, establishes a solid foundation for academic growth.

In terms of the factors that influence engineering students dropping out, there is a significant relationship between academic performance and the decision to leave. Positive performance is often associated with greater student retention, while poor academic performance can lead to frustration, lack of confidence, and demotivation, increasing the likelihood of a student considering dropping out. Institutional commitment also affects students’ decisions to continue with higher education.

Regarding the model’s predictive capacity, the constructs Decision to Desert, Academic Performance, and Social Integration exhibit the highest levels of explained variance, as reflected in their R² values (all above 0.45). Additionally, the Q² values for these constructs exceed the 0.35 threshold, indicating high predictive relevance according to established benchmarks (Hair et al., 2021). It is important to clarify that these thresholds (0.02 = small, 0.15 = medium, 0.35 = large) apply to Q², not R². While R² evaluates the amount of variance explained, Q² assesses the model’s predictive accuracy through blindfolding. These metrics support the conclusion that the model has substantial predictive power for key outcomes, particularly in academic performance and Decision to Desert (see Supplementary Material).

Although R² and Q² are commonly used in PLS-SEM to assess model fit and predictive relevance, recent literature has questioned their sufficiency as sole indicators. Authors such as Hair et al. (2021, 2016) suggest complementing them with alternative measures such as PLSpredict or predictive-oriented segmentation, especially in complex models.

5 Discussion

The findings of this study confirm that academic performance is a key determinant in the decision to desert engineering programs. This relationship is supported by the significant path coefficient between Academic Performance and decision to desert (β = 0.386, p < 0.001), as reported in Figure 2 and Table 5. In our model, academic performance is measured through indicators related to psychological and emotional factors affecting student achievement (Lázaro Alvarez et al., 2020). These results suggest that students who exhibit difficulties in maintaining academic performance, often due to psychological or emotional stressors, are more likely to consider dropping out. This supports prior literature, which emphasizes the importance of academic behavior, progression rate, and prior preparation in mathematics and logical reasoning (Bedregal-Alpaca et al., 2020).

This study also identified Normative Consistency as a relevant construct influencing student retention. Normative consistency, defined as the alignment between students’ motivations and their expectations of the program, showed direct effects on multiple constructs, including Social Integration and Peer Support. This relationship is supported by the positive and statistically significant path between Normative Consistency and Peer Support in the SEM model (see Figure 2). In our model, this construct captures indicators such as program interest, enjoyment, and perceived fit, highlighting that students who perceive coherence between their values and their academic journey are more likely to persist. These findings align with earlier studies that emphasize the role of student-program alignment in preventing dropout (Chalela-Naffah et al., 2020).

Although constructs such as Normative Consistency and Peer Support did not show significant direct effects on the Decision to Desert, the structural model reveals their indirect influence through mediating variables, including Social Integration, Intellectual Development, and Institutional Commitment. For instance, Normative Consistency enhances Peer Support and Social Integration, which in turn increase Institutional Commitment, ultimately reducing the likelihood of dropout. These indirect pathways underscore the importance of incorporating mediation analysis into dropout models, as recommended in the literature.

Peer Support emerged as a significant factor, particularly influencing Intellectual Development (β = 0.419, p < 0.001) and Social Integration (β = 0.150, p = 0.059). In our model, this construct refers specifically to the role of fellow students in shaping academic and emotional experiences (Pascua-Cantarero, 2016). Although previous research highlights the importance of professors in retention (Riveros Sanabria et al., 2018; Caicedo Chacón et al., 2019), our study focuses on the impact of peer relationships, indicating that collaboration, shared study strategies, and interpersonal bonds contribute meaningfully to intellectual growth and social belonging.

The relationship between Satisfaction and Institutional Commitment was also statistically significant (β = 0.439, p < 0.001). Satisfaction, as operationalized in our study, encompasses perceptions related to tuition affordability, access to financial aid, and academic program prospects. These elements were found to strongly influence institutional commitment, a construct that reflects students’ loyalty and identification with the university. This connection supports findings in the Latin American context, where institutional commitment is frequently mediated by students’ satisfaction with their educational experience (Murillo et al., 2018; Apaza and Huamán, 2012).

The constructs decision to desert, Academic Performance, and Social Integration exhibited the highest predictive capacity in our model, with R² values above 0.45, as shown in Figure 2 and Supplementary Material. Additionally, Q² values surpassed the 0.35 threshold, indicating strong predictive relevance for these endogenous variables. It is important to note that Q² values above 0.35 suggest large predictive relevance, as established in the literature. These metrics reinforce the empirical robustness of the model.

Regarding the surveyed sample, the typical profile corresponds to a young male student from low-income households (stratum 1 or 2), recently graduated from a public high school, and dependent on family financial support. These characteristics reflect broader patterns in Colombian engineering education and contextualize the vulnerabilities identified in this research. The study contributes to the understanding of dropout risk among engineering students in private universities who often require additional academic support, especially in developing foundational competencies such as logical and mathematical reasoning (Hernández Betancur et al., 2016).

Although the model demonstrates strong predictive power, some constructs showed issues with convergent validity. In particular, Academic Potential exhibited an AVE below the 0.5 threshold, suggesting that some of its indicators may require refinement. Additionally, certain hypothesized paths, such as Normative Consistency → Institutional Commitment, while theoretically grounded, did not reach statistical significance (p = 0.077). These observations highlight areas for improvement in the model and indicate that future studies should consider updating the theoretical framework. While the Spady model provided a valuable structure for this analysis, its origin in the 1970s may limit its applicability to current educational contexts. More contemporary models could offer refined insights into the dropout phenomenon among engineering students.

It is important to note, however, that the model faces challenges related to convergent validity. Therefore, it is recommended to consider the application of more updated models. The Spady model, proposed in the 70s, may not fully capture the current situation. More recent models would likely be more aligned with the current dynamics and realities surrounding the dropout of engineering students. In addition, while the Spady model provided a valuable basis for this study, it did not include more recent or integrative models such as those developed by Tinto, Pascarella, Bean, Cabrera, or Seidman. These frameworks offer additional insights into student engagement, institutional adjustment, and persistence mechanisms. Future research could benefit from the application or comparison of these models to improve the theoretical soundness and relevance of dropout analyses in the current higher education landscape.

5.1 Limitations and future research agenda

This study provides valuable insights into the factors influencing engineering students’ Decision to Desert, particularly within the context of a private university in an emerging economy. However, several limitations should be acknowledged. First, the cross-sectional design restricts the ability to assess the evolution of students’ intentions or behaviors over time. Future studies would benefit from adopting longitudinal designs that can capture dynamic changes across the academic trajectory, particularly during critical transition points such as the first year or final semesters.

Second, the exclusive use of quantitative methods, though appropriate for testing causal relationships and validating the structural model, limits the interpretive depth regarding students’ lived experiences. Incorporating qualitative components, such as in-depth interviews or focus groups, would provide richer contextual data and help uncover underlying motivations, expectations, and emotional responses related to academic persistence or dropout.

Third, the sample was drawn from a single institution and was composed primarily of students in Systems and Industrial Engineering. This may limit the generalizability of the findings to other types of engineering programs or institutional contexts. Comparative studies across public and private institutions, urban and rural settings, or different engineering disciplines would offer a broader and more nuanced understanding of dropout determinants.

Therefore, it is essential to acknowledge the limitations related to the study’s sample size, as it consisted of only 190 participants, all of whom were from a private university in the city of Medellín. This restricts the possibility of generalizing the results to the population of engineering students in Colombia, particularly those enrolled in public universities, who may face different socioeconomic and institutional conditions. In this sense, it is recommended that future research include larger and more diverse samples, encompassing a broader population of universities and geographic regions, to enhance external validity.

This study focused on students currently enrolled in the Systems Engineering and Industrial Engineering programs. Students who had previously dropped out were not included, so the results focus on the identification of factors that predispose to dropout, rather than on their actual occurrence.

Finally, it should be recognized that while institutional commitment and academic performance are widely recognized as key determinants of student dropout, recent research suggests that psychological, socioeconomic, and family factors also play an important role (Gutiérrez-Monsalve et al., 2025). Beyond institutional and academic variables, the literature on student dropout increasingly points to the influence of mental health, economic hardship, job responsibilities, and family dynamics (Kocsis and Molnár, 2025). These factors interact with institutional elements and may intensify the risk of dropout, especially in students from vulnerable backgrounds. Therefore, while this study highlights institutional commitment and academic performance as central factors, the roles of unmeasured elements, such as financial pressure, emotional wellbeing, and family responsibilities, should not be ignored.

In addition, future research should consider the inclusion of socioeconomic and psychological variables not addressed in this model, such as financial stress, mental health conditions, external work commitments, family obligations, and academic motivation. These factors, often underexplored in structural models, may interact with institutional and academic variables in ways that either mitigate or exacerbate the risk of attrition. Their inclusion could enhance predictive power and provide a more holistic view of the dropout phenomenon, especially in settings marked by economic inequality and social vulnerability.

Finally, based on the study’s findings, future research could also investigate the effectiveness of specific institutional interventions aimed at strengthening academic performance, peer support, and intellectual development—three key variables in the model. Experimental or quasi-experimental designs could be employed to evaluate the impact of academic mentoring programs, financial aid schemes, psychological support services, or peer-led learning communities on student retention and engagement.

By addressing these methodological and contextual limitations and expanding the scope of analysis, future studies can contribute to the development of more comprehensive and inclusive strategies for preventing dropout in engineering education, with implications for both institutional policy and national higher education planning in emerging economies.

6 Conclusion

This study aimed to identify the key factors associated with the decision to desert out in engineering students, based on a structural equation model applied to undergraduate students of Systems and Industrial Engineering at a private Colombian university. The model revealed that Academic Performance, Social Integration, and Academic Potential were the constructs with the highest predictive capacity to explain dropout intentions. These results are supported by significant path coefficients and confirm the crucial role of individual and institutional dimensions in student retention.

Empirical findings suggest that students with higher academic potential, reflected in indicators related to health, discipline, and motivation, are better prepared for academic performance and perseverance in their studies. At the same time, social integration, measured through the student’s adaptation and sense of belonging, contributes positively to their continuity in the academic program.

Satisfaction with the academic experience was found to influence institutional commitment, particularly through concerns about tuition costs, financial aid, and future academic program expectations. Although other constructs, such as Peer Support and Normative Consistency, were not statistically significant in their direct trajectories, they could play an indirect role in shaping academic motivation and commitment, as evidenced by the observed indicators.

The model also highlighted the impact of psychological and emotional aspects, such as depression, family instability, and concentration difficulties, on academic performance, suggesting that school dropout depends not only on cognitive ability but also on the emotional and socioeconomic context of students.

Although this study does not propose formal intervention strategies, its results point to key areas where institutional efforts should focus: providing academic and emotional support, improving access to information on scholarships and financial aid, and adapting institutional policies to the realities of students from vulnerable backgrounds.

Finally, the model faced some challenges in terms of convergent validity on certain constructs, and its reliance on classical Spady theory may limit its ability to reflect the complexities of dropout in contemporary contexts. Future research should explore the application of more recent models and consider longitudinal approaches to better understand the dynamics of persistence and dropout.

Data availability statement

The original contributions presented in the study are included in the article/Supplementary material, further inquiries can be directed to the corresponding author.

Ethics statement

The studies involving humans were approved by Ethics Committee of the Escolme University Institution. The studies were conducted in accordance with the local legislation and institutional requirements. The ethics committee/institutional review board waived the requirement of written informed consent for participation from the participants or the participants’ legal guardians/next of kin because Consent was obtained from participants by agreeing to participate anonymously in the survey.

Author contributions

EM: Writing – original draft, Writing – review & editing. PR-C: Writing – original draft, Writing – review & editing. EA-C: Writing – original draft, Writing – review & editing. AV-A: Writing – original draft, Writing – review & editing. GM: Writing – original draft, Writing – review & editing. GS: Writing – original draft, Writing – review & editing. JJ: Writing – original draft, Writing – review & editing. MV: Writing – original draft, Writing – review & editing.

Funding

The author(s) declare that no financial support was received for the research and/or publication of this article.

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Generative AI statement

The authors declare that no Gen AI was used in the creation of this manuscript.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/feduc.2025.1582586/full#supplementary-material

References

Afthanorhan, A., Awang, Z., and Aimran, N. (2020). An extensive comparison of CB-SEM and PLS-SEM for reliability and validity. Int. J. Data Netw. Sci. 4, 357–364. doi: 10.5267/j.ijdns.2020.9.003

Crossref Full Text | Google Scholar

Afthanorhan, A., Ghazali, P. L., and Rashid, N. (2021). Discriminant validity: a comparison of CBSEM and consistent PLS using Fornell & Larcker and HTMT approaches. J. Phys. Conf. Ser. 1874:012085. doi: 10.1088/1742-6596/1874/1/012085

Crossref Full Text | Google Scholar

Akin, A., and Akin, U. (2016). Academic potential beliefs and feelings and life satisfaction: the mediator role of social self-efficacy. Int. J. Soc. Psychol. 31, 500–520. doi: 10.1080/02134748.2016.1190125

Crossref Full Text | Google Scholar

Alaminos, A., García, F. F., Perdú, C., and Fernández, Ó. S. (2015). “Análisis Multivariante para las Ciencias Sociales I” in Índices de Distancia, Conglomerados y Análisis Factorial (Cuenca: PYDLOS Ediciones).

Google Scholar

Alban, M., and Mauricio, D. (2018). Factors that influence undergraduate university desertion according to students perspective. Int. J. Eng. Technol. 10, 1585–1602. doi: 10.21817/ijet/2018/v10i6/181006017

Crossref Full Text | Google Scholar

Amaro, S., Seabra, C., and Abrantes, J. L. (2015). “Comparing CB-SEM and PLS-SEM Results: An empirical example,” in 2nd International Symposium on Partial Least Squares Path Modeling. Seville, Spain.

Google Scholar

Apaza, E., and Huamán, F. (2012). Factores determinantes que inciden en la deserción de los estudiantes universitarios. Apunt. Univ. Rev. Invest. 1, 77–86.

Google Scholar

Attinasi, L. C. (1989). Getting in: Mexican Americans’ perceptions of university attendance and the implications for freshman year persistence. J. Higher Educ. 60, 247–277. doi: 10.2307/1982250

Crossref Full Text | Google Scholar

Balasundaram, N. (2009). Factor analysis: nature, mechanism and uses in social and management science research. J. Cost Manage. 37, 15–25.

Google Scholar

Bean, J. P. (1980). Dropouts and turnover: the synthesis and test of a causal model of student attrition. Res. High. Educ. 12, 155–187. doi: 10.1007/BF00976194

Crossref Full Text | Google Scholar

Beck, H. P., and Milligan, M. (2014). Factors influencing the institutional commitment of online students. Internet High. Educ. 20, 51–56. doi: 10.1016/j.iheduc.2013.09.002

Crossref Full Text | Google Scholar

Bedregal-Alpaca, N., Tupacyupanqui-Jaén, D., and Cornejo-Aparicio, V. (2020). Análisis del rendimiento académico de los estudiantes de Ingeniería de Sistemas, posibilidades de deserción y propuestas para su retención. Ingeniare Revista chilena de ingeniería. 28, 668–683. doi: 10.4067/S0718-33052020000400668

Crossref Full Text | Google Scholar

Björklund, A., and Salvanes, K. G. (2011). “Chapter 3 - education and family background: mechanisms and policies” in Handbook of the economics of education, vol. 3, 201–247. doi: 10.1016/B978-0-444-53429-3.00003-X

Crossref Full Text | Google Scholar

Black, A. M. (2023). The role of bridging programmes in supporting student persistence and prevention of attrition: a UK case study. Stud. High. Educ. 49, 1519–1531. doi: 10.1080/03075079.2023.2269246

Crossref Full Text | Google Scholar

Borden, V. M. H. (1987). Student engagement in college: Concept and assessment. Amherst, MA: University of Massachusetts Amherst.

Google Scholar

Borden, V. M. H. (1988). “Student Engagement in College,” in 28th Annual AIR Forum. Phoenix Arizona.

Google Scholar

Cabrera, A. F., Nora, A., and Castaneda, M. B. (1993). College persistence: structural equations modeling test of an integrated model of student retention. J Higher Educ. 64:123.

Google Scholar

Caicedo Chacón, L. Y., Cárdenas Parra, C. N., Müller Rueda, J. S., and Ortiz Bernal, J. T. (2019). Aplicación para la gestión y el análisis de información relacionada con la deserción estudiantil universitaria. Rev. Colomb. Comput. 20, 6–19.

Google Scholar

Carmo Nicoletti, M. (2019). Revisiting the Tinto’s theoretical dropout model. High. Educ. Stud. 9, 52–64. doi: 10.5539/hes.v9n3p52

Crossref Full Text | Google Scholar

Castrillón, O. D., Sarache, W., and Ruiz-Herrera, S. (2020). Predicción del rendimiento académico por medio de técnicas de inteligencia artificial. Formación universitaria. 13, 93–102. doi: 10.4067/S0718-50062020000100093

Crossref Full Text | Google Scholar

Cerpa, W. F. (2015). Análisis multivariado para determinar los factores más relevantes de deserción estudiantil presentes en el programa de Ingeniería Industrial de una Universidad del Caribe colombiano. PRO 13, 86–98. doi: 10.15665/rp.v13i1.363

Crossref Full Text | Google Scholar

Chalela-Naffah, S., Valencia-Arias, A., Ruiz-Rojas, G. A., and Cadavid-Orrego, M. (2020). Factores psicosociales y familiares que influyen en la deserción en estudiantes universitarios en el contexto de los países en desarrollo. Rev Lasallista Investig. 17, 103–115. doi: 10.22507/rli.v17n1a9

Crossref Full Text | Google Scholar

Díaz, C. J. (2009). Factores de deserción estudiantil en ingeniería: Una aplicación de modelos de duración. Inf. Tecnol. 20, 129–145. doi: 10.4067/S0718-07642009000500016

Crossref Full Text | Google Scholar

Donoso, S., and Schiefelbein, E. (2007). Análisis de los modelos explicativos de retención de estudiantes en la universidad: Una visión desde la desigualdad social. Estud. Pedagóg. 33, 7–27. doi: 10.4067/S0718-07052007000100001

Crossref Full Text | Google Scholar

Estrada-Molina, O., and Fuentes-Cancell, D. R. (2022). El engagement y la deserción en los MOOCs: Revisión sistemática. Comunicar. 70, 111–124. doi: 10.3916/C70-2022-09

Crossref Full Text | Google Scholar

Ethington, C. A. (1990). A psychological model of student persistence. Res. High. Educ. 31, 279–293. doi: 10.1007/BF00992313

Crossref Full Text | Google Scholar

Fauzi, M. A. (2022). Partial least square structural equation modelling (PLS-SEM) in knowledge management studies: knowledge sharing in virtual communities. Knowledge Manage. E-Learning. 14, 103–124. doi: 10.34105/j.kmel.2022.14.007

Crossref Full Text | Google Scholar

Ferrando, P. J., and Anguiano-Carrasco, C. (2010). El Análisis Factorial como Técnica de Investigación en Psicología. Pap. Psicol. 31, 18–33.

Google Scholar

Fishbein, M., and Ajzen, I. (1975). Belief, attitude, intention and behaviour: An introduction to theory and research. Boston, MA: Addison-Wesley.

Google Scholar

Fornell, C. G., and Larcker, D. F. (1981). Evaluating structural equation models with unobservable variables and measurement error. J. Mark. Res. 18, 39–50. doi: 10.2307/3151312

Crossref Full Text | Google Scholar

Franz, S., and Paetsch, J. (2023). Academic and social integration and their relation to dropping out of teacher education: a comparison to other study programs. Front. Educ. 8:1179264. doi: 10.3389/feduc.2023.1179264

Crossref Full Text | Google Scholar

Garzón, L. P., and Pérez, A. M. C. (2012). Revisión de algunos estudios sobre la deserción estudiantil universitaria en Colombia y Latinoamérica. Theoria 21, 9–20.

Google Scholar

Georg, W. (2009). Individual and institutional factors in the tendency to drop out of higher education: a multilevel analysis using data from the Konstanz student survey. Stud. High. Educ. 34, 647–661. doi: 10.1080/03075070802592730

Crossref Full Text | Google Scholar

Gutiérrez Rodríguez, P., Villarreal, R., Cuesta Valiño, P., and Blozis, S. (2020). A PLS-SEM approach to understanding E-SQ, E-satisfaction and E-loyalty for fashion E-retailers in Spain. J. Retail. Consum. Serv. 57:102201. doi: 10.1016/j.jretconser.2020.102201

Crossref Full Text | Google Scholar

Gutiérrez-Monsalve, J. A., Garzón, J., Forero-Meza, M. F., Estrada-Jiménez, C., and Segura-Cardona, A. M. (2025). “Factors associated with dropout in engineering: a structural equation and logistic model approach” in Applied computer sciences in engineering WEA 2024 Communications in Computer and Information Science. eds. J. C. Figueroa-García, G. Hernández, D. F. Suero Pérez, and E. E. Gaona García (Cham: Springer), 225–236.

Google Scholar

Hadjar, A., Haas, C., and Gewinner, I. (2023). Refining the Spady–Tinto approach: the roles of individual characteristics and institutional support in students’ higher education dropout intentions in Luxembourg. Eur. J. High. Educ. 13, 409–428. doi: 10.1080/21568235.2022.2056494

Crossref Full Text | Google Scholar

Hair, J. F., Hult, G. T. M., Ringle, C. M., and Sarstedt, M. (2016). A primer on partial least squares structural equation modeling (PLS-SEM). Second Edn. Thousand oaks, CA: SAGE Publishing, 1–165.

Google Scholar

Hair, J. F. Jr. (2021). Next-generation prediction metrics for composite-based PLS-SEM. Ind. Manag. Data Syst. 121, 5–11. doi: 10.1108/IMDS-08-2020-0505

Crossref Full Text | Google Scholar

Hair, J. F. Jr., Matthews, L. M., Matthews, R. L., and Sarstedt, M. (2017). PLS-SEM or CB-SEM: updated guidelines on which method to use. Int. J. Multivariate Data Anal. 1, 107–123. doi: 10.1504/IJMDA.2017.087624

Crossref Full Text | Google Scholar

Hair, J. F., Ringle, C. M., and Sarstedt, M. (2011). PLS-SEM: indeed a silver bullet. J. Mark. Theory Pract. 19, 139–152. doi: 10.2753/MTP1069-6679190202

Crossref Full Text | Google Scholar

Hair, J. F., Risher, J. J., Sarstedt, M., and Ringle, C. M. (2019). When to use and how to report the results of PLS-SEM. Eur. Bus. Rev. 31, 2–24. doi: 10.1108/EBR-11-2018-0203

Crossref Full Text | Google Scholar

Hair, J. F. Jr., Sarstedt, M., Ringle, C. M., and Gudergan, S. P. (2021). Advanced issues in partial least squares structural equation modeling. 2nd Edn. London: SAGE Publications, Inc.

Google Scholar

Hair, J. F., Sarstedt, M., Hopkins, L., and Kuppelwieser, V. G. (2014). Partial least squares structural equation modeling (PLS-SEM): an emerging tool in business research. Eur. Bus. Rev. 26, 106–121. doi: 10.1108/EBR-10-2013-0128

Crossref Full Text | Google Scholar

Henseler, J., Ringle, C. M., and Sarstedt, M. (2015). A new criterion for assessing discriminant validity in variance-based structural equation modeling. J. Acad. Mark. Sci. 43, 115–135. doi: 10.1007/s11747-014-0403-8

Crossref Full Text | Google Scholar

Hernández Betancur, J. E., Rojas López, M. D., and López García, Y. (2016). Caracterización del desertor y diseño del sistema de alertas tempranas -SAT- para la Facultad de Minas de la Universidad Nacional de Colombia. Revista Logos Ciencia y Tecnología 8, 77–88. doi: 10.22335/rlct.v8i1.295

Crossref Full Text | Google Scholar

Himmel, E. (2002). Modelo de análisis de la deserción estudiantil en la educación superior. Calidad en la Educación. 17, 91–108. doi: 10.31619/caledu.n17.409

Crossref Full Text | Google Scholar

John, N. M., Page, O., Martin, S. C., and Whittaker, P. (2018). Impact of peer support on student mental wellbeing: a systematic review. MedEdPublish 7:170. doi: 10.15694/mep.2018.0000170.1

PubMed Abstract | Crossref Full Text | Google Scholar

Jorquera, Ó., Farías, J., and González, P. “¿Cuáles son los factores asociados al abandono de los estudiantes de programas de equidad en Chile? El caso de una universidad tradicional,” in Proceedings of the Congresos CLABES VIII; Universidad Technologica de Panama (2018).

Google Scholar

Kock, N. (2014). Stable P value calculation methods in PLS-SEM. Laredo, TX: ScriptWarp Systems.

Google Scholar

Kock, N. (2016). Hypothesis testing with confidence intervals and P values in PLS-SEM. Int. J. e-Collaboration. 12, 1–6. doi: 10.4018/IJeC.2016070101

Crossref Full Text | Google Scholar

Kocsis, Á., and Molnár, G. (2025). Factors influencing academic performance and dropout rates in higher education. Oxf. Rev. Educ. 51, 414–432. doi: 10.1080/03054985.2024.2316616

Crossref Full Text | Google Scholar

Lázaro Alvarez, N., Callejas, Z., and Griol, D. (2020). Factores que inciden en la deserción estudiantil en carreras de perfil Ingeniería Informática. Rev. Fuentes 22, 105–126. doi: 10.12795/revistafuentes.2020.v22.i1.09

Crossref Full Text | Google Scholar

Lizarte Simón, E. J., and Gijón Puerta, J. (2022). Prediction of early dropout in higher education using the SCPQ. Cogent Psychol. 9:2123588. doi: 10.1080/23311908.2022.2123588

Crossref Full Text | Google Scholar

Maldonado, S., Miranda, J., Olaya, D., Vásquez, J., and Verbeke, W. (2021). Redefining profit metrics for boosting student retention in higher education. Decis. Support. Syst. 143:113493. doi: 10.1016/j.dss.2021.113493

Crossref Full Text | Google Scholar

Marschall, M., Falk, S., and Klug, C. (2023). Social events, bridging courses, and academic skills trainings – participation in first-year courses and higher education dropout in Germany. Stud. High. Educ. 49, 1789–1804. doi: 10.1080/03075079.2023.2279246

Crossref Full Text | Google Scholar

Martínez Ávila, M., and Fierro Moreno, E. (2018). Aplicación de la técnica PLS-SEM en la gestión del conocimiento: un enfoque técnico práctico. RIDE 8, 130–164. doi: 10.23913/ride.v8i16.336

Crossref Full Text | Google Scholar

Martinez Ruiz, J. E., Cardenas Rodriguez, M. M., Junco Rosario, G. L., and Cabezas Cabezas, H. S. (2023). El desarrollo del pensamiento lógico a través del proceso de aprendizaje en los estudiantes universitarios. J. Sci. Res., 376–387. doi: 10.5281/zenodo.10420749

Crossref Full Text | Google Scholar

Martínez-Navarro, Á., Verdú, E., and Moreno-Ger, P. (2021). “Mining pre-grade academic and demographic data to predict university dropout” in Radical solutions for digital transformation in Latin American universities. eds. D. Burgos and J. W. Branch (Singapore: Springer), 197–215.

Google Scholar

Ministerio de Educación Nacional – MEN. (2022). Estadísticas De Deserción. Available at: https://www.mineducacion.gov.co/sistemasinfo/spadies/secciones/Estadisticas-de-desercion/

Google Scholar

Mirabelli, J. F., Johnson, E. M., Vohra, S. R., Sanders, J. L., and Jensen, K. J. (2025). Stressors and normalized stress in undergraduate engineering education culture: development of the engineering stress culture scale and undergraduate engineering stressors questionnaire. Int. J. STEM Educ. 12:19. doi: 10.1186/s40594-025-00540-8

Crossref Full Text | Google Scholar

Mori Sánchez, M. d. P. (2012). Deserción Universitaria en Estudiantes de una Universidad Privada de Iquitos. Rev. Digit. Investig. Docencia Univ. 6, 60–83.

Google Scholar

Murillo, M. G. E., Alfonso, A. D. M., Armando, R. P. L., and Chunata, N. M. I. (2018). Incidencia De Los Aspectos Psicosociales En La Deserción De Los Estudiantes Universitarios. Eur. Sci. J. 14:206. doi: 10.19044/esj.2018.v14n19p206

Crossref Full Text | Google Scholar

Myrtveit, S. M., Askeland, K. G., Knapstad, M., Knudsen, A. K., and Skogen, J. C. (2017). The Norwegian student introductory week: who takes part, and is participation associated with better social integration and satisfaction among students? Eur. J. Higher Edu. 7, 136–152. doi: 10.1080/21568235.2016.1252933

Crossref Full Text | Google Scholar

Ortiz-Lozano, J. M., Aparicio-Chueca, P., Triadó-Ivern, X. M., and Arroyo-Barrigüete, J. L. (2023). Early dropout predictors in social sciences and management degree students. Stud. High. Educ. 49, 1303–1316. doi: 10.1080/03075079.2023.2264343

Crossref Full Text | Google Scholar

Pascarella, E. T., and Terenzini, P. T. (1980). Predicting freshman persistence and voluntary dropout decisions from a theoretical model. J Higher Educ. 51, 60–75. doi: 10.1080/00221546.1980.11780030

Crossref Full Text | Google Scholar

Pascua-Cantarero, P. (2016). Factors related to dropping out in the freshman and sophomore years in the career of teaching mathematics at Universidad Nacional de Costa Rica. Rev. Electrón. Educare. 20, 1–23. doi: 10.15359/ree.20-1.5

Crossref Full Text | Google Scholar

Piepenburg, J. G., and Beckmann, J. (2022). The relevance of social and academic integration for students’ dropout decisions. Evidence from a factorial survey in Germany. Eur. J. Higher Educ. 12, 255–276. doi: 10.1080/21568235.2021.1930089

Crossref Full Text | Google Scholar

Pineda-Báez, C., and Ortíz, A. (2009). Programas exitosos de retención estudiantil: Las vivencias de los estudiantes. Revista Virtual Universidad Católica del Norte 28, 1–30.

Google Scholar

Purwanto, A., and Sudargini, Y. (2021). Partial least squares structural squation modeling (PLS-SEM) analysis for social and management research: a literature review. J. Ind. Eng. Manage. Res. 2, 114–123. doi: 10.7777/jiemar.v2i4.168

Crossref Full Text | Google Scholar

Ramírez-Correa, P., and Grandón, E. (2018). Predicción de la deserción académica en una universidad pública Chilena a través de la clasificación basada en árboles de decisión con parámetros optimizados. Formación universitaria. 11, 3–10. doi: 10.4067/S0718-50062018000300003

Crossref Full Text | Google Scholar

Rincon, I., Suárez-Castrillón, S., and Suarez, A. (2020). Impacto del programa Jóvenes en Acción en la deserción estudiantil en tiempos de Covid-19. Espacios. 41, 296–303. doi: 10.48082/espacios-a20v41n42p25

Crossref Full Text | Google Scholar

Riveros Sanabria, F., Agudelo Varela, O., and Castro, E. P. (2018). “Ova de Calculo Integral para disminuir la deserción en estudiantes de ingeniería,” in 16th LACCEI international multi-conference for engineering, education, and technology: “Innovation in education and inclusion”. Lima, Perú, p. 18–20.

Google Scholar

Rönkkö, M., and Cho, E. (2022). An updated guideline for assessing discriminant validity. Organ. Res. Methods 25, 6–14. doi: 10.1177/1094428120968614

Crossref Full Text | Google Scholar

Schuster, R., and King, B. R. (2022). First-year student retention, goal commitment, and institutional commitment: two meta-analyses. SN Soc. Sci. 2:145. doi: 10.1007/s43545-022-00446-0

Crossref Full Text | Google Scholar

Seidman, A. (2005). College student retention: Formula for student success. 1st Edn. Westport, CT: Praeger.

Google Scholar

Spady, W. G. (1971). Dropouts from higher education: toward an empirical model. Interchange 2, 38–62. doi: 10.1007/BF02282469

Crossref Full Text | Google Scholar

Strauss, L. C., and Volkwein, J. F. (2004). Predictors of student commitment at two-year and four-year institutions. J. High. Educ. 75, 203–227. doi: 10.1080/00221546.2004.11778903

Crossref Full Text | Google Scholar

Suárez-Montes, N., and Díaz-Subieta, L. B. (2015). Estrés académico, deserción y estrategias de retención de estudiantes en la educación superior. Revista de Salud Pública. 17, 300–313. doi: 10.15446/rsap.v17n2.52891

PubMed Abstract | Crossref Full Text | Google Scholar

Tinto, V. (1975). Dropout from higher education: a theoretical synthesis of recent research. Rev. Educ. Res. 45, 89–125. doi: 10.3102/00346543045001089

Crossref Full Text | Google Scholar

Tinto, V. (1994). Leaving college: Rethinking the causes and cures of student attrition. 2nd Edn. Chicago, IL: University of Chicago Press.

Google Scholar

Valencia-Arias, A., Chalela, S., Cadavid-Orrego, M., Gallegos, A., Benjumea-Arias, M., and Rodríguez-Salazar, D. Y. (2023). University dropout model for developing countries: a Colombian context approach. Behav. Sci. 13:382. doi: 10.3390/bs13050382

PubMed Abstract | Crossref Full Text | Google Scholar

Vásquez, J., and Miranda, J. (2019). “Student desertion: what is and how can it be detected on time?” in Data science and digital business. eds. G. Fp and B. Márquez (Cham: Springer International Publishing), 263–283.

Google Scholar

Vila, D., Cisneros, S., Granda, P., Ortega, C., Posso-Yépez, M., and García-Santillán, I. (2019). “Detection of desertion patterns in university students using data mining techniques: a case study” in Technology trends CITT 2018 Communications in Computer and Information Science. eds. M. Botto-Tobar, G. Pizarro, M. Zúñiga-Prieto, M. D’Armas, and M. Z. Sánchez (Cham: Springer), 420–429.

Google Scholar

Zeng, N., Liu, Y., Gong, P., Hertogh, M., and König, M. (2021). Do right PLS and do PLS right: a critical review of the application of PLS-SEM in construction management research. Front. Eng. Manag. 8, 356–369. doi: 10.1007/s42524-021-0153-5

Crossref Full Text | Google Scholar