One of the variables closely related to student achievement and the adaptive process in the university context is learner engagement or involvement in the task (van Rooij et al., 2018). This is expressed through the feeling of well-being to overcome obstacles (Salanova & Schaufeli, 2009; Schaufeli, 2017), beyond the conditioning factors and commitments acquired. It includes three dimensions: vigour or mental strength, which is manifested during the development of the task and allows to remain in the activity constantly; dedication or mental process, where the student performs the task with motivation, involvement and enthusiasm; finally, absorption or state of well-being where the student can abstract in the development of the activity, (Schaufeli, 2017). Learner engagement is an ability to engage emotionally, cognitively and behaviourally in a task, activity or situation in the university educational context. It relates to the active and positive involvement of students in their learning process and has been shown to be associated with higher academic performance and greater long-term knowledge retention. In the university context, learner engagement is key for the development of better learning strategies in students. There is a close relationship between learner engagement and motivation, being relevant its development in educational contexts, related to learning strategies (Sharp et al., 2020; Truta et al., 2018).

Another of the relevant variables for the development of learning strategies is motivation, which is defined as those perceived forces that induce a person to act and develop those strategies that are most suitable for achieving their academic goals (Ryan & Deci, 2019). In the university context, different studies have highlighted the incidence of greater or lesser motivation in the development of learning strategies. This construct is understood as a key factor in behaviour, depending on a set objective, focusing its content on the importance of internal resources for personal development, self-regulated behaviours and contextual aspects that favour or diminish motivation, according to the Theory of Self-Determination (Deci & Ryan, 1985). From this perspective, motivation is an internal factor that helps to develop self-motivation mechanisms for study, coexistence with peers and involvement in the task (Ben-Eliyahu et al., 2018). Research has shown that self-motivated students learn more, have a better understanding and retention of information, and experience less anxiety and distress in the academic context (Oriol-Granado et al., 2017). By achieving this, the desire to continue learning, which is one of the main goals of education, is fostered. Similarly, motivation is also influenced by external factors (Werner & Milyavskaya, 2018).

Motivation can be intrinsic, i.e., performing an action for satisfaction without expecting a tangible reward; and extrinsic, as a construct that is applied whenever an activity is performed to achieve some result (Zimmerman, 2008). Some research that has analysed motivation in university contexts from the perspective of achieving the goals set (Oriol-Granado et al., 2017; Werner & Milyavskaya, 2018), has noted the importance and prevalence of the contributions of goal orientation theory in the development of learning strategies. These theories (Ben-Eliyahu et al., 2018) explain learner motivation based on interests and commitments to the task.

Learning strategies consist of making decisions about the most appropriate means to achieve the objectives and goals established. It is essential to create appropriate situations for the development of learning strategies (Williams-Oyarce et al., 2022), which include control and socio-emotional elements, related to the student’s internal regulation. Research on learning strategies has corroborated their relevance, by differentiating people with different cognitive traits, which allows the transfer of knowledge according to complexity and adaptive processes, key to effective learning (Ergen & Kanadli, 2017). In the university context, a qualitative step forward takes place, where the student must establish different learning strategies that allow him/her to relate, apply and transfer knowledge to achieve the objectives set (De la Fuente & Justicia, 2003). Specifically, it has been corroborated that the development of learning strategies is related to cognitive and emotional factors, and that this relationship favours better academic performance (Ergen & Kanadli, 2017). Specifically, this research uses the 44-item Learning Strategies Scale (ACRA), which has shown good psychometric properties (De la Fuente & Justicia, 2003). Similarly, this scale has been positively related to socio-emotional support and academic engagement (Álvarez et al., 2015), in addition to increased positive emotions and greater motivation towards academic activity.

Personal resources are self-assessments related to one’s perceived ability to control and influence the surrounding context. These self-evaluations predict goal setting, goal development and increased task motivation (Ben-Eliyahu et al., 2018). Learner engagement is a persistent motivational state that students develop in relation to their academic activity, which manifests itself in the level of active participation in academic activities (Reeve, 2013). Therefore, learner engagement and academic motivation are factors that have shown a positive relationship, related to higher academic performance of university students (Oriol-Granado et al., 2017; Werner & Milyavskaya, 2018), being predictor variables of higher or lower performance adjustment and learning strategies.

Hypothesis 1 (H1): Academic motivation and learner engagement are related variables, which indicate the intensity and persistence of individuals’ effort to achieve their goals.

Learner engagement as a positive attitude of involvement towards the development of learning strategies and persistence towards the academic task, includes behavioural and affective elements, such as a positive predisposition towards the task; and cognitive elements, such as a preference for challenges, autonomy and involvement in the tasks (Sandoval-Muñoz et al., 2018). In other words, a high level of learner engagement is positively related to the predisposition towards homework and the development of study habits. High-achieving students are characterised by being more autonomous, having a positive self-perception of their own learning strategies and having extensive control over their study habits, managing adverse situations adaptively (Ferrer et al., 2020). To put this into practice, it is worth asking the question: does learner engagement influence the development of learning strategies in university students? Based on the assumption that students with high levels of learner engagement will be able to develop better study habits and task involvement, the following hypothesis is put forward:

Hypothesis 2 (H2): Higher levels of learner engagement will be related to the optimisation of better learning strategies and task control.

Learner engagement has been related to contextual factors that can either promote or reduce the motivation level towards the task (Rigo & Amaya, 2020; Perkmann et al., 2021). This is especially true in the university context, affecting affective, behavioral and cognitive components (Larson et al., 2019). Related to the motivational components (interest shown by students) or commitment to the task, the positive relationship with learning strategies has been demonstrated (Smith et al., 2020). Therefore, the improvement of academic strategies and skills will be determined by greater learner engagement (Agger & Koenka, 2019).

Hypothesis 3 (H3): Academic motivation and learner engagement will promote the development of better learning strategies.

Figure 1
Proposed Theoretical Model

The sample is made up of 648 university students from the Education Degrees, which belong to the Faculties of Humanities and Education Sciences in Andalusia (Spain). Of the distribution by sex in the education degrees, the majority are women, 417 (64.3%) and 231 (35.7%) are men, with an average age of 19.40 years (±3.67). 62.5% belong to the University Degree in Primary Education and 37.5% to the degree in Early Childhood Education. Specifically, the number of predictors of learning strategies in our model is 2. The results of the statistical power analysis (Cohen, 1988) show a power of .923 above 80% and at 5% significance level to observe R2 values of less than 10%. Therefore, no problems related to the adequacy of the sample size were found.

Échelle de Motivation en Éducation (EME) of Vallerand et al. (1989), which was adapted in Spanish by Núñez et al. (2005). It consists of 28 items, distributed in seven subscales that correspond to the three degrees of autonomy on which behaviours are based according to Deci and Ryan’s (1985) self-determination theory. Thus, motivation can be expressed from lack of control to self-determination, distributed in seven dimensions of four items each that assess the three types of MI (MI to knowledge, MI to achievement and MI to stimulating experiences), three types of ME (external regulation, introjected regulation and identified regulation) and amotivation. In our sample the reliability of the scale scores is Cronbach’s α = .944 and McDonald’s ω = .947.

Utrecht Work Engagement Scale (UWES) developed by Schaufeli and Bakker (2004). The Spanish version of the UWES-S scale for students (Belando et al., 2012) was used. It is a self-report questionnaire made up of 17 items that analyse the three dimensions that compose it: vigour, dedication and absorption. It has a Likert-type scale with seven response options. Schaufeli & Bakker (2004) reported that the UWES scale has an internal consistency reliability with Cronbach’s alpha values ranging from .80 to .90. In our sample the reliability of the scale scores is Cronbach’s α = .928 and McDonald’s ω = .932.

Learning strategies scale (ACRA) developed by De la Fuente and Justicia, (2003). It used the version for university students of the scale designed by Román and Gallego (1994) that measures the use of strategies during the learning process. It is an inventory of 44 Likert-type items with 7 response options that assesses three components of the strategies involved in learning according to the principles of information processing: cognitive and learning control strategies; learning support strategies; study habits. The internal consistency index for the total of 44 items is Cronbach’s α = .92 and McDonald’s ω = .91. In our sample the reliability of the scale scores is Cronbach’s α = .944 and McDonald’s ω = .947.

The ethical guidelines promoted and encouraged by national and international regulations for the conduct of research involving human subjects were followed. All data were treated in accordance with EU Regulation 2016/679 of the European Parliament and of the Council of 27 April 2016, both on Personal Data and Organic Law 3/2018 of 5 December on guaranteeing digital rights. Participants were assured that their responses would be kept anonymous and confidential, and that all information provided would be used for scientific purposes only. The instrument was administered individually through the platform Google (Google forms), subject to the informed consent of each participant. The researchers explained to the participants the purpose of the research, as well as the guidelines for proper compliance and confidentiality of the data, requesting the voluntary collaboration of the students. Data were collected and quality checked, always ensuring that the process conformed to the ethical principles for research as defined in the Declaration of Helsinki (World Medical Association, 2013).

Descriptive statistics (means and standard deviations) were obtained. Previously, the Hot-Deck multiple-entry method was applied to reduce bias while preserving the joint and marginal distributions (Lorenzo-Seva & Van-Ginkel, 2016), by analysing a priori the validity, reliability (Cronbach’s alpha and Omega coefficient) and internal consistency of each instrument, using Confirmatory Factor Analysis (CFA), to verify the psychometric properties of the questionnaire and obtain the factor loadings of each item. The normality analysis was carried out using multivariate hypothesis testing, resulting in a non-normal distribution. The analyses were carried out using the Amos programme (Version 25.0, IBM SPSS), the Jamovi software (The Jamovi Project, 2020) in its Version 1.2 and SmartPLS (version 3.3.6). The coefficients considered in this research were the χ2/df ratio, the root mean square error of approximation (RMSEA), the comparative fit index (CFI) and the Tucker-Lewis index (TLI). The goodness of fit of the model was considered satisfactory when the TLI and CFI ≥ .95, and the RMSEA was close to .07 (Kline, 2016). We used the Partial Least Squares (PLS) technique for explanatory and predictive purposes for the dependent variables and types of relationships, direct and indirect. Statistical significance required a 95% confidence level (significance p<.05).

To assess the robustness of the factor loadings and the significance between variables, the Bootstrapping procedure was used with 2000 subsamples (Hair et al., 2021). This resulted in the structural model (Figure 2), where the variables considered in this study are reported. The predictive relevance and standardised regression coefficient or path coefficient of learning strategies [(Q2 = .295); (R² = .456)]; learner engagement [(Q2 = .314); (R² = .364)], in the estimation of the measurement model, indicated a moderate model fit. In this regard, R² values above .67 indicate a substantial model fit and above .33 a moderate fit.

Table 4 presents Cronbach’s alpha, Omega coefficient, external loadings and composite reliability index (CFI) scores. In relation to the convergent validity or degree of certainty that the proposed indicators measure the same latent variable or factor, through the estimation of the average variance extracted (AVE), the values must be greater than .5, according to the criteria of Becker et al. (2018). That is, a high value of (AVE) will have a better representation of the loading of the observable variable.

Table 4
Correlation weights, reliability estimates and convergent validity statistics

Discriminant validity (Table 5) shows the difference between the latent variable, to determine the statistical differentiation of each factor with respect to the others, indicating in bold the square root of the mean variance extracted (Martínez-Ávila & Fierro-Moreno, 2018).

Table 5
Measurement model. Discriminant validity

Discriminant validity (Table 6) was analysed through the analysis of the cross-loadings of each of the latent variables and their respective observed variables, where the loadings were higher than the rest of the variables (Ramírez-Asís et al., 2020).

Table 6
Cross-loadings (latent and observable variables).

Table 7 shows the results of the hypothesis testing, following the criteria of Hair et al. (2021), where the causal relationship with the latent variables can be observed. The t-test was obtained (values higher than 1.96 indicate the coherence of the reflective model. In this research, the results that showed a higher value were: learner engagement → learning strategies (β = .344, t = 3.937, p<.001); motivation → learner engagement (β = .603, t = 8.311, p<.001) and motivation → learning strategies (β = .409, t = 4.388, p<.001).

Table 7
Path coefficient (standardised regression coefficient)