English self-efficacy is understood as the beliefs an individual holds about the effectiveness of their abilities to successfully perform a task in English (Wang et al., 2014). Similarly, academic self-efficacy is defined as the belief in one's capabilities to attain academic success (Bandura, 1977; Palenzuela, 1983). Previous research suggests that elevated levels of English self-efficacy correlate with a more frequent and diverse use of learning strategies (Yang and Wang, 2015). English self-efficacy is intrinsically linked to academic self-efficacy (Wang et al., 2014, 2017; Mendoza-Torres et al., 2023). Similarly, a study demonstrated a positive and significant relationship between self-efficacy beliefs and English self-efficacy (Liu, 2023). Other study demonstrated positive correlation between self-efficacy for English learning and the strategies employed for language acquisition (Nasimi et al., 2024). The literature indicates that English self-efficacy is a direct and significant predictor of the use of cognitive strategies, such as critical thinking, metacognitive self-regulation, and time management (Wang et al., 2013; Lee et al., 2020). Additionally, there exists a positive association between English self-efficacy and the employment of metacognitive strategies (Li, 2023).

Recent studies indicate that academic self-efficacy may serve as a mediating factor in the relationship between English self-efficacy and language learning strategies. For instance, it has been demonstrated that academic self-efficacy mediates the interaction of psychological and behavioral variables in university students' learning (Zheng et al., 2024; Chavez-Yacolca et al., 2025). Furthermore, self-efficacy mediates the relationship between motivation, as evidenced by a positive teacher-student relationship, and learning outcomes learning (Ma et al., 2018). This implies that the influence of English-specific self-efficacy (a psychological-motivational variable) on learning strategies (a behavioral variable) may be mediated through academic self-efficacy. In this context, academic self-efficacy—by embodying the student's confidence in their ability to successfully engage with academic tasks—functions as a conduit that either amplifies or diminishes the impact of English-specific self-efficacy on the employment of strategies.

This dynamic can be thoroughly examined through the lens of social cognitive theory (SCT), which posits that human behavior emerges from a triadic interaction among personal, behavioral, and environmental factors, with self-confidence playing a pivotal role (Bandura, 1986). In a subsequent study, Wang et al. (2014) conceptualized self-efficacy for English within the framework of Social Cognitive Theory (SCT). They posited that a student's confidence in their ability to learn the language is cultivated through mastery experiences, observational learning, verbal persuasion, and the individual's emotional or physiological states. The perception of academic competence serves as a conduit that integrates the perception of specific competence in learning English into a broader belief system, thereby facilitating competence and the utilization of learning resources within this academic context (Korpipää et al., 2020). Furthermore, English learning strategies, which include metacognitive, cognitive, and social resources, are actions undertaken by students during the learning process. These strategies are intrinsically connected to students' beliefs regarding their ability to perform the tasks (Cancino et al., 2022).

Academic self-efficacy emerges as the most proximate variable mediating the relationship between English self-efficacy and English learning strategies, as elucidated by the theoretical considerations outlined below. Firstly, this is attributed to the hierarchical and general nature of the self-efficacy concept as proposed in Social Cognitive Theory (SCT). The differentiation between specific beliefs regarding language proficiency and those pertaining to the general capacity to organize and execute actions necessary for academic success enables the latter to mediate the transfer of specific beliefs to a broader context, such as the application of learning strategies (Zimmerman, 2000; Huang, 2024). Usher and Pajares (2008) support this hierarchical model by asserting that domain-specific beliefs, such as English self-efficacy, are incorporated into broader levels of self-efficacy. Furthermore, academic self-efficacy functions as a regulatory variable influencing the selection, persistence, and adaptation of learning strategies (Pintrich and De Groot, 1990), thereby serving as a motivational filter that promotes the employment of effective strategies for learning English (Schunk, 1995).

Empirical research on the link between English self-efficacy and language learning strategies is limited, particularly in Latin American settings. However, given the documented connections between academic self-efficacy and English learning strategies, as well as between English self-efficacy and academic self-efficacy, it is reasonable to propose that English self-efficacy may influence language learning strategies.

The existing literature has largely overlooked the interaction between academic self-efficacy beliefs and English-specific self-efficacy, as well as their combined impact on the adoption of language learning strategies. This study seeks to address this gap by proposing a manifest variables mediation model to explore the mediating role of academic self-efficacy in the relationship between English self-efficacy and language learning strategies. Subsequently, the study evaluates the fit of a structural model that examines the interrelationships among academic self-efficacy, English self-efficacy, and language learning strategies in university students including an indirect effect analysis (Figure 1). Specifically, the following hypotheses are proposed:

This study employs a quantitative approach to empirical research, utilizing an associative strategy and a cross-sectional design, as data were collected from participants on a single occasion. The research methodology involved two sequential designs: initially, an explanatory design with observable variables, followed by an explanatory design with latent variables (Ato et al., 2013).

A non-probabilistic or convenience sampling technique was utilized, wherein participants were selected based on their accessibility and willingness to participate (Creswell and Creswell, 2023). An online form was created using Google Forms, and the link was disseminated at the specified date and time. For in-person classes, the link was shared via WhatsApp, whereas for virtual classes, it was distributed through the Zoom chat feature. The dataset employed in this research is the same as that used in a previous validation study of the Strategy Inventory for Language Learning (Saez-Zevallos et al., 2025). However, the purpose and framework of the current study differ, as it emphasizes mediation analysis through the use of two complementary approaches: a mediation model and a structural equation modeling (SEM) framework that identifies indirect effects. To ascertain the minimum required sample size, a sample size calculator for structural equation models was utilized (Soper, 2024), with parameters set at an anticipated effect size of 0.3, a desired statistical power of 0.8, two latent variables, ten observed variables, and a significance level of 0.05. This calculation indicated a recommended minimum sample size of 100. Nonetheless, the data collected comprised 610 university students aged between 18 and 50 years (M = 22.8; SD = 5.90), of whom 399 (31.8%) were in early adulthood, 355 (58.2%) were women, 560 (91.8%) were single, 261 (42.8%) were from the Faculty of Health Sciences, and, in terms of geographical distribution, 342 (56.1%) were from the coastal region (Table 1).

The survey methodology was employed, utilizing the following measurement scales:

The Strategy Inventory for Language Learning (SILL) was employed to assess language learning strategies. This instrument, developed by Oxford (1986), was translated into Spanish (García-Herrero and Jiménez-Vivas, 2014) and validated in Peru among university students (Saez-Zevallos et al., 2025), to evaluate the strategies employed in learning the English language. The SILL comprises 18 items rated on a 5-point Likert scale: “never or almost never” = 1, “generally do not do this” (less than half the time) = 2, “sometimes” (about half the time) = 3, “often” (more than half the time) = 4, and “always or almost always” = 5. These items are distributed across six dimensions: memory strategies (Items 1 to 3), cognitive strategies (Items 4 to 6), compensation strategies (Items 7 to 9), metacognitive strategies (Items 10 to 12), affective strategies (Items 13 to 15), and social strategies (Items 16 to 18). The complete scale exhibits evidence of reliability (α = 0.916, ω = 0.917) and validity for its internal structure (CFI = 0.971, TLI = 0.966, RMSEA = 0.034, SRMR = 0.037).

The English Self-Efficacy Scale, as developed and validated in Peru with Peruvian university students (Mendoza-Torres et al., 2023), was employed to assess English self-efficacy. This scale comprises 36 items formatted in a Likert scale with five response options: “I cannot do it at all” = 0, “I cannot do it” = 1, “Relatively sure I can do it” = 2, “I can do it” = 3, “Totally sure I can do it” = 4. The scale comprises three dimensions: Reading (items 1–13), Oral Communication (items 14–22), and Writing (items 23–36). It demonstrates reliability (α = 0.98, ω = 0.98) and validity in its internal structure (CFI = 0.92, TLI = 0.92, RMSEA = 0.06, and SRMR = 0.04).

To measure academic self-efficacy, the Perceived Self-Efficacy Scale for Academic Situations (EAPESA) (Palenzuela, 1983) was used, validated in the Peruvian university context (Dominguez-Lara et al., 2012; Dominguez-Lara, 2014). This instrument assesses perceived self-efficacy in academic situations through 9 items with a unidimensional structure (Navarro-Loli and Dominguez-Lara, 2019), measured on a four-point Likert scale: “never” = 1; “sometimes” = 2; “quite often” = 3; and “always” = 4. It demonstrated evidence of reliability (ω = 0.88) and validity of its internal structure (CFI = 0.978, GFI = 0.969, AGFI = 0.949, RMSEA = 0.056, SRMR = 0.029).

The data collection process adhered to the guidelines set forth by the Declaration of Helsinki and received approval from the research ethics committee of the university affiliated with the authors (Reference 2023-CEEPG-00010), thereby ensuring compliance with ethical and data protection standards. An online form was developed using Google Forms, which initially included a request for informed consent. This informed participants about the altruistic nature of their involvement, the study's objectives, the procedures, and the questionnaires to be completed. Emphasis was placed on the ethical use of data, the voluntary nature of participation, confidentiality, anonymity, the exclusive use of data for research purposes, the right to withdraw at any time, and the protection of data and access to results, in accordance with the ethical standards required for conducting research. Following the informed consent, the form included sections pertaining to sociodemographic data and the data collection instruments.

The data analysis was performed utilizing Jamovi software, version 2.4.14 (The Jamovi Project, 2023), which serves as an interface for the “R” software (R Core Team, 2023). Initially, the reliability of the instruments was assessed using the internal consistency method, employing both alpha (α) and omega (ω) coefficients. Following the reliability assessment, a descriptive statistical analysis was conducted, encompassing mean, standard deviation, skewness, and kurtosis. Skewness and kurtosis values were deemed to indicate an approximately normal distribution of the variables if they fell within the range of ±1 (George and Mallery, 2024). The relationships between variables were examined through Pearson's correlation analysis.

Two complementary approaches to mediation analysis were employed: one utilizing observable variables and the other employing latent variables through structural equation modeling (SEM). The mediation analysis with observable variables was conducted using the jAMM GLM Mediation Model module (Gallucci, 2020), which implemented Bias-corrected Bootstrap (BC) analysis with 5,000 samples. For the evaluation of the structural model with latent variables, the “lavaan” library (Rosseel, 2012) was utilized via the SEMlj module (Gallucci and Jentschke, 2021). The theoretical model of the study was analyzed using the maximum likelihood (ML) estimator in conjunction with bias-corrected bootstrapping (BC) with 5,000 repetitions, a suitable procedure for numerical variables with an approximately normal distribution (Kline, 2023). This method is recommended for assessing indirect effects in mediation analysis through SEM, as it does not require the assumption of normality for the distribution of the products of coefficients that determine the values of indirect effects (Preacher and Hayes, 2008). The structural equation modeling (SEM) approach was implemented as a content-based parcel model, parceling items into their respective dimensions (Matsunaga, 2008), reducing bias based on the empirical evidence and theoretical support of the dimensionality of the constructs (Landis et al., 2000; Little et al., 2013). This decision was based on the understanding that parcel models, as opposed to item-level models, effectively minimize sampling and parsimony errors and decrease the probability of correlated residuals (Rioux et al., 2020; Little et al., 2022). The model fit was evaluated using the comparative fit index (CFI), the root mean square error of approximation (RMSEA), and the standardized root mean square residual (SRMR). CFI and TLI values greater than 0.90, SRMR less than 0.080, and RMSEA 90% CI less than 0.10 were employed, given the complexity of the structural model (Mueller and Hancock, 2019; Kline, 2023), with 95% confidence intervals for standardized coefficients. This SEM analysis facilitates the control of measurement error and rigorously assesses the validity of the structural model (Kline, 2023). The reliability of the constructs was calculated using α (Cronbach, 1951) and ω (McDonald, 2009) coefficients from the psych statistical package (Revelle, 2022), for which magnitudes > 0.70 are considered adequate (Hair et al., 2019; Stensen and Lydersen, 2022).

English self-efficacy demonstrated a positive association with the employment of strategies for learning English. This finding aligns with the results reported by Montaño-González and Cancino (2020) in a study involving Chilean university students, although that particular study did not specifically address English self-efficacy nor did it incorporate regression analysis. Conversely, research conducted with Iranian university students indicated that general self-efficacy positively predicted the use of English learning strategies (Shirzad et al., 2022). This research emphasizes the link between English-specific self-efficacy and the utilization of strategies for mastering the language.

The relationship between English self-efficacy and academic self-efficacy was found to be positive. This finding is consistent with the results reported by Wang et al. (2017) and Mendoza-Torres et al. (2023), who illustrated those specific forms of self-efficacy, such as English self-efficacy, are positively associated with more general forms of self-efficacy, including academic self-efficacy. Similarly, Chen and Usher (2013) identified a relationship between self-efficacy resources and enhanced scientific self-efficacy. Furthermore, other studies have indicated that these self-efficacy resources are positively linked to reading self-efficacy (Shehzad et al., 2019) and writing self-efficacy (Sun et al., 2021; Barreda-Parra et al., 2023).

Academic self-efficacy demonstrated a positive association with English learning strategies. This finding aligns with the results reported by Martins and Santos (2019) among Brazilian university students. In the context of China, elevated levels of self-efficacy among university students were associated with an increased utilization of cognitive, affective, and compensation strategies for English learning (Shi, 2018). Furthermore, this variable has been linked to avoidance learning strategies and metacognitive self-regulation, which, although not encompassed within Oxford's theoretical classification, are related to metacognitive and affective strategies (Bai et al., 2022).

The primary contribution of this study is the identification of academic self-efficacy as a mediating factor in the relationship between English self-efficacy and language learning strategies. English self-efficacy is connected to the use of cognitive strategies, such as critical thinking, metacognitive self-regulation, time management and metacognitive strategies (Wang et al., 2013; Lee et al., 2020; Li, 2023). Furthermore, it is associated with academic self-efficacy (Wang et al., 2014, 2017; Mendoza-Torres et al., 2023), highlighting the intermediary function of academic self-efficacy in this dynamic. This concept is consistent with the notion of transfer within Social Cognitive Theory, wherein beliefs in one's specific abilities in English can be transferred to other academic areas requiring similar cognitive processes, such as general academic self-efficacy, thereby facilitating the development of skills that enhance performance across various academic domains (Bandura, 1986; Franco and Rodrigues, 2018).

The distinction between specific beliefs about language proficiency and those related to the general ability to organize and execute actions necessary for academic success allows the latter to facilitate the transfer of specific beliefs to a broader context, such as the application of learning strategies (Zimmerman, 2000; Huang, 2024). In alignment with this, the present research adds to the existing literature by revealing that English self-efficacy, while encompassing more specific domain beliefs than academic self-efficacy, benefits from the latter's role in enhancing strategic learning behavior. This finding aligns with previous research outcomes (Zimmerman, 2000; Schunk and Pajares, 2002). Furthermore, the study empirically validates the presence of a bidirectional relationship between specific self-efficacy and general academic self-efficacy. Specifically, self-efficacy in particular domains predominantly enhances academic self-efficacy, thereby facilitating the employment of cognitive, metacognitive, and behavioral strategies for learning (Zimmerman and Schunk, 2011). Additionally, creates opportunities for future research to investigate how varying degrees of self-efficacy, particularly general academic self-efficacy, affect the learning processes and outcomes of university students.

Beyond the social cognitive account of self-efficacy, research has highlighted affect as a proximal mechanism through which efficacy beliefs influence learning strategies. According to Bandura (1997), self-efficacy is shaped by mastery experiences, vicarious learning, verbal persuasion, and physiological and affective states. However, affect should not be treated merely as a background context but as a core process that enables or constrains self-regulation. Control–value theory (Pekrun, 2006) explains how perceived control over learning tasks and their value generate achievement emotions, which facilitate or hinder strategy deployment. Lo (2023) posits that emotional factors, such as self-control, significantly influence self-efficacy, which in turn affects the degree of commitment exhibited toward learning. Similarly, another study suggests that students with high levels of self-efficacy experience reduced anxiety and increased enjoyment in learning Lo, (2022) thereby facilitating the adoption of effective learning strategies. Future studies could conceptualize English self-efficacy as indirectly influencing strategy use through academic self-efficacy and affective mechanisms, particularly enjoyment, emotional regulation, and perceived control.

The present study elucidates significant implications derived from the findings reported. It underscores the necessity of employing a diverse array of strategies to enhance academic performance in language acquisition (Agustin et al., 2021). The study demonstrates that the direct effect of English self-efficacy on language learning strategies is substantial, there exists a positive correlation between English self-efficacy and academic self-efficacy, academic self-efficacy positively forecasts language learning strategies, and there is a notable indirect effect of English self-efficacy on language learning strategies mediated by academic self-efficacy. These findings suggest that higher education institutions should (a) optimize the development of self-efficacy skills to enhance the utilization of language learning strategies, ensuring that students independently monitor their use, (b) implement training programs to fortify academic self-efficacy and promote the use of language learning strategies to improve performance and facilitate the learning of additional languages such as English, and (c) design instructional activities that integrate linguistic skills, thereby facilitating the use of multiple strategies with the aim of fostering effective and meaningful learning (García-Herrero and Jiménez-Vivas, 2014).

Educational institutions can enhance students' academic and English self-efficacy through the implementation of peer mentoring (Huang, 2023), structured linguistic tasks (Nguyê~n et al., 2022), and digital tools (Arbulú Pérez Vargas et al., 2024). Digital technologies provide personalized practice opportunities, personalized feedback, and interactive AI Tools (Yaseen et al., 2025). When effectively applied, these strategies contribute to the improvement of students' academic and language performance (Casa-Coila, 2025).

This study may signify a substantial advancement in instructional design within English for Academic Purposes (EAP) and English for Specific Purposes (ESP) contexts by incorporating pedagogical principles that prioritize affect as a crucial connection between course structure and students' emotional wellbeing. In this context, it aims to foster the development of linguistic competencies and enhance academic self-efficacy. This proposal acknowledges that language learning in higher education is not merely cognitive but also deeply affective, and that addressing factors such as enjoyment, anxiety, and perceived control can significantly influence the sustained adoption of learning strategies. Thus, aligning course design with an affective rationale enhances its practical feasibility and provides stakeholders with a means to improve student performance and experiences.