Exploring the role of social media in mathematics learning: effects on self-efficacy, interest, and self-regulation

Abstract

Background

Social media’s integration into education prompts exploration of its effects on mathematics learning outcomes. While Mathematics Learning with Social Media (MLSM) can create engaging environments, its impact on key factors such as mathematics self-efficacy (MathSE), mathematics interest (MathI), and self-regulation in mathematics learning (SR) remains underexplored. This study investigates the direct and indirect effects of MLSM on these variables to enhance mathematics education.

Methods

A quantitative research design was employed, involving 398 university students who engaged in Mathematics Learning with Social Media (MLSM) activities. Participants were randomly selected from two universities in China, with a gender distribution of 43% male and 57% female. The sample included students from diverse academic programs, including science, technology, engineering and mathematics (STEM) (37.9%), economics and management (43.7%), and humanities and social sciences (17.3%). Data were collected through validated survey instruments that measured mathematics self-efficacy (MathSE), mathematics interest (MathI), and self-regulation in mathematics learning (SR). A two-step structural equation modeling (SEM) approach was used to analyze the relationships between MLSM and the three variables, examining both direct effects and mediation pathways.

Results

The findings revealed that MLSM had significant direct effects on both MathI (β = 0.343, p < 0.001) and SR (β = 0.578, p < 0.001) indicating that the use of social media in mathematics learning fosters students’ interest in mathematics and enhances their ability to self-regulate their mathematics learning processes. Furthermore, while the direct influence of MLSM on MathSE was not substantial, its indirect effects through MathI (MLSM→MathI→MathSE, β = 0.092, p < 0.001), SR (MLSM→SR→MathSE, β = 0.418, p < 0.001) and both SR and MathI (MLSM→SR→MathI→MathSE, β = 0.115, p < 0.001) were found to be significant. These results highlight the mediating roles of interest and self-regulation in strengthening students’ confidence in their mathematics abilities.

Conclusions

This study underscores the transformative potential of social media as a valuable tool for mathematics education. By effectively integrating MLSM into instructional practices, educators can foster students’ interest, enhance their self-regulation skills, and indirectly build their self-efficacy in mathematics. The findings provide actionable insights for designing innovative educational strategies and policies that leverage social media to create dynamic and learner-centered environments, particularly in mathematics. This research contributes to the broader understanding of social media’s role in the digital age, emphasizing its capacity to support self-directed and effective learning in academic disciplines.

Introduction

The ways individuals acquire knowledge and develop skills are undergoing rapid transformation in the digital age [1, 2]. Social media has emerged as a key player in this evolution, revolutionizing educational experiences by transcending traditional classroom boundaries and providing universally accessible platforms for learning [3, 4]. Unlike conventional educational settings, social media fosters casual, spontaneous, and interactive learning experiences, enabling incidental, informal, and open-ended interactions that have become increasingly prevalent in modern education [5–14].

Mathematics education has long been perceived as rigid and challenging [15–17]. The traditional teacher-centered instructional approach may limit students’ ability to connect with the abstract mathematics in an engaging and personal way [18–21]. However, the integration of social media into education presents an opportunity to reimagine mathematics learning by fostering informal learning environments that prioritize exploration, collaboration, and self-directed learning. Unlike traditional teacher-centered approaches, social media enables a more interactive learning experience, empowering students to take greater control of their mathematics learning processes [18].

This shift highlights the need to re-evaluate the mathematics learning outcomes in the context of transitioning from traditional instructor-led methods to more autonomous, student-centered approaches [22–25]. Despite the growing recognition of social media’s potential in education, empirical investigations into its effectiveness within mathematics education are still limited [26–28]. Furthermore, most of the prior studies have focused on K-12 students, with limited attention to the unique experiences and needs of college students [26–28]. College students operate in distinct educational contexts where self-regulation and independent learning are critical, yet the impact of social media on their mathematics learning remains underexplored [29]. Additionally, existing studies have often emphasized formal education contexts, leaving a gap in understanding how informal, technology-mediated learning can shape mathematical attitudes and beliefs at the tertiary level [30–34].

To address these gaps, this study introduces a theoretical model designed to explore the impact of social media on mathematics education, focusing on its potential to facilitate flexible, student-driven learning in technology-rich environments. Specifically, this research investigates the role of social media in mathematics learning, examining its effects on three critical variables: students’ mathematics self-efficacy (MathSE), mathematics interest (MathI), and self-regulation in mathematics learning (SR). Particularly, it is important to highlight that MathI plays a crucial role in students’ engagement and success in mathematics education. High levels of interest in mathematics have been shown to enhance motivation, persistence, and overall academic performance [35, 36]. By fostering a genuine interest in the subject, educators can help students develop a deeper understanding of mathematical concepts and improve their problem-solving skills [37]. In the context of social media integration, MathI is particularly important because it can drive students to actively participate in online learning activities, seek out additional resources, and engage with peers and educators in meaningful discussions about mathematics.

By leveraging data from 398 participating students randomly selected from two Chinese universities and employing a two-step structural equation modeling approach, this study aims to uncover both the direct and indirect effects of mathematics learning with social media (MLSM). The findings shed light on the capacity of social media to enhance learning outcomes, foster self-directed learning, and create more dynamic and effective learning environments in mathematics education.

Literature review and hypothesis development

Our framework is grounded in insights from the control-value theory [38, 39], which provides a robust foundation for understanding the dynamic interplay between students’ learning environments, motivation, beliefs, emotions, and behaviors. This theory emphasizes that students’ appraisals of control and value are central to shaping their achievement emotions and subsequent learning outcomes. On the one hand, control appraisals refer to students’ perceptions of their ability to influence their learning outcomes, such as their confidence in their skills and capacity to succeed (e.g., self-efficacy). These appraisals are informed by both internal factors, like prior knowledge and skills, and external factors, such as the supportiveness and flexibility of the learning environment. On the other hand, value appraisals reflect the degree of importance, interest, or utility students attach to a particular learning task [38, 40–44]. According to the control-value theory, the learning environment plays a pivotal role in shaping these appraisals. A supportive, engaging, and autonomy-promoting environment can enhance students’ control beliefs by providing opportunities to develop competence and experience success. Simultaneously, it can increase the perceived value of learning tasks by making them meaningful, enjoyable, and aligned with students’ intrinsic interests and long-term aspirations. These positive appraisals contribute to the development of adaptive achievement emotions—such as excitement, curiosity, and pride—which are critical for sustaining motivation and persistence in learning [38, 39].

Furthermore, the learning environment does not only affect students’ appraisals but also directly influences their behaviors, particularly their adoption of self-regulated learning strategies. Self-regulated learning encompasses critical skills like goal-setting, self-monitoring, strategic planning, and reflection, which enable students to take control of their learning processes and adapt to challenges effectively. A supportive learning environment can encourage students to engage more actively in self-regulated learning behaviors, promoting greater autonomy and effectiveness in learning [42]. For instance, in a technology-rich or informal learning environment, such as social media-supported mathematics learning, the availability of diverse resources—such as interactive tutorials, video demonstrations, and real-time problem-solving forums—can provide students with a wealth of materials tailored to different learning needs and preferences. The opportunities for exploration offered by social media, such as access to various learning communities, gamified activities, and personalized content recommendations, empower students to take ownership of their learning journey. This autonomy allows students to set goals, monitor their progress, and apply effective strategies to overcome obstacles, thereby enhancing their self-regulation skills. Hence, an engaging and supportive learning environment can empower students to actively manage their learning processes and achieve better academic outcomes [43, 44].

In the context of mathematics education, control-value theory has been employed to examine the environmental factors related to mathematics learning [45, 46]. Aligned with control-value theory, our framework posits that MLSM serves as an informal learning environment that influences learners’ motivation, beliefs, and behaviors. Specifically, MLSM can enhance students’ interest in mathematics, boost their mathematics self-efficacy, and support their self-regulated learning by providing opportunities for exploration, collaboration, and autonomy [38, 47–49]. Building on this theoretical perspective, our model examines the effects of MLSM on three key constructs: mathematics self-efficacy (MathSE), mathematics interest (MathI), and self-regulation (SR).

Mathematics learning with social media (MLSM)

Social media

Social media broadly encompasses web-based platforms that facilitate the interactive exchange of information through communication networks and communities [50, 51]. Platforms such as Facebook, Twitter, YouTube, Instagram, and LinkedIn transformed how individuals communicated, exchanged information, and connected with one another [14]. Social media platforms provide features such as instant messaging, video sharing, blogging, and social networking, allowing users to participate in real-time conversations, share multimedia content, and create virtual communities on the basis of common interests [52]. The interactive nature of social media facilitates a two-way flow of information, making it a powerful tool for collaboration and community building. With billions of active users globally, social media has become an integral part of daily life, impacting various aspects of society, including education.

The impact of social media on informal learning in mathematics is significant, enhancing accessibility, engagement, and the immediacy of information exchange [53–55]. These platforms enable learners to access mathematical problems and solutions, participate in peer-to-peer discussions, and receive updates from mathematical communities, fostering a culture of continuous learning and curiosity about mathematics beyond the classroom [3, 31, 56–58]. Specifically, social media facilitates spontaneous mathematics learning activities [58–60]. Individuals explore mathematical content at their own pace, seek out resources, engage with content creators, and participate in discussions that are not bound by formal educational structures [6, 7, 12, 26–28, 34, 61]– [62]. During periods such as the COVID-19 pandemic, social media has proven crucial in maintaining continuity in learning by providing platforms for virtual engagement and resource sharing [63].

Moreover, social media allows for the extension of mathematical learning beyond traditional limits, offering continuous access to a diverse range of resources and expert insights [50, 64]. This approach is particularly valuable for learners who find mathematics challenging or inaccessible within formal educational contexts [3, 31, 56, 57]. Platforms also serve as informal tools for gauging understanding and progress, utilizing interactive posts and quizzes to provide immediate feedback and adapt learning experiences to diverse needs [65, 66].

Learning with social media

Learning with social media refers to the use of these platforms to facilitate informal learning experiences. Unlike traditional educational settings that rely on structured curricula and formal assessments, learning with social media is often learner-led, flexible, and driven by the individual’s interests and needs. Social media platforms provide access to a vast array of resources, including educational videos, articles, online courses, discussion forums, and expert insights, which learners can explore at their own pace [52]. The communal and interactive features of social media fostered collaborative learning environments where users could engage in discussions, ask questions, and share knowledge with peers and experts from around the world. This informal learning environment supports self-regulated learning strategies, allowing learners to take control of their educational journeys, set their own goals, and seek out resources and support as needed [67].

The integration of social media into education has shown numerous benefits, including increased accessibility to information, enhanced engagement, and the ability to connect with a global learning community. Integrating social media into mathematics learning shifted the approach to more inclusive, interactive, and self-directed environments. It underscores the need for ongoing research to understand its impacts fully and to develop best practices for its use in promoting mathematical understanding and engagement [6, 7, 68, 69]. For example, platforms such as YouTube offered educational channels that provided tutorials and lectures on a wide range of subjects, whereas Twitter and LinkedIn were used to follow and join professional groups for networking and knowledge sharing [8]. Social media also supports experiential and collaborative learning by enabling users to participate in virtual labs, simulations, and group projects. The real-time feedback and peer support available on these platforms significantly enhanced the learning experience, making it more interactive and dynamic [65].

Mathematics learning with social media (MLSM)

Mathematics learning with social media (MLSM) specifically refers to the engagement and utilization of social media platforms by individuals to enhance and expand their mathematical knowledge and skills [3]. This involved browsing content about mathematics or mathematicians, seeking online resources for problem solving, engaging in discussions about mathematical topics, and following updates from mathematical organizations, all of which were facilitated through social media channels [31]. By leveraging the capabilities of social media, the MLSM provides learners with a dynamic approach to enhance their mathematical knowledge and skills, engage with a global community, and foster a life-long love for mathematics. MLSM utilizes the interactive and communal features of social media to create an accessible learning environment that supports ongoing mathematical education and engagement [31].

Digital platforms, including social media, are used to perform and demonstrate mathematical concepts interactively and innovatively [70]. Platforms such as YouTube offer numerous channels dedicated to mathematics education, providing tutorials, problem-solving techniques, and explanations of complex mathematical concepts, thereby enhancing learning [71]. Twitter and Facebook groups enable users to engage in discussions, inquire about mathematical problems, and share solutions with a community of learners and experts [8]. These platforms also allowed learners to stay informed about the latest developments in the field of mathematics by following academic institutions, professional organizations, and renowned mathematicians, reshaping how students thought about and understood mathematics [72]. For example, LinkedIn was used to connect with professionals in the field, access educational content, and join groups focused on mathematics education.

The impact of MLSM on learners was significant, enhancing accessibility to high-quality educational resources, promoting continuous engagement with mathematical content, and supporting collaborative learning [47]. The communal aspect of social media fostered a supportive learning environment where users sought help, shared knowledge, and collaborated on mathematical problems [69]. This interactive approach to learning demystified complex mathematical concepts, making them more accessible and understandable [70]. Furthermore, the informal nature of MLSM allows learners to explore mathematical topics at their own pace, catering to individual learning styles and needs. Research has shown that MLSM positively influences learners’ attitudes towards mathematics, increases their self-efficacy, and improves their problem-solving skills [74]. The ability to access diverse resources and interact with a global community of learners and experts enhances motivation and fosters a deeper understanding of mathematical concepts.

While MLSM has demonstrated positive effects on mathematics learning, the increasing integration of AI-driven tools, such as ChatGPT and YouTube tutorials, introduces new complexities that warrant careful consideration. Specifically, it is crucial to examine how these technologies might facilitate or hinder cognitive offloading [75, 76]. Cognitive offloading occurs when individuals delegate cognitive tasks to external aids, thereby reducing their engagement in deep, reflective thinking [77]. Although AI tools can enhance learning outcomes by providing personalized instruction and immediate feedback, over-reliance on these tools may lead to a decline in mathematical skills [78]. For example, studies have shown that the availability of information through social media can negatively impact memory retention and the inclination to process information deeply [79]. Similarly, reliance on AI-driven solutions for mathematical problems may undermine students’ ability to develop foundational mathematical skills and conceptual understanding [80].

The conflicting findings on social media’s role in learning further complicate this landscape. While some studies suggest that social media enhances learning outcomes in STEM fields by fostering engagement and providing access to diverse resources [73]– [74], others caution that it may cause distraction and cognitive overload, particularly when students are exposed to excessive or irrelevant information [79, 80]. This duality underscores the importance of understanding how social media and AI tools affect students’ learning in mathematics. On one hand, these technologies offer unparalleled opportunities for personalized and interactive learning experiences [73]– [74]. On the other hand, they may inadvertently hinder the development of essential cognitive skills if not used judiciously [78, 79]. To navigate this complex terrain, our research aims to position itself within this broader debate by exploring the effects of social media in mathematics learning.

Mathematics interest (MathI)

Mathematics interest (MathI) refers to an individual’s emotional and cognitive engagement with mathematics, characterized by a positive disposition, enjoyment, and intrinsic motivation toward learning and engaging in mathematical activities [81, 82]. It is the degree to which an individual finds mathematics appealing, valuable, or relevant, and this interest influences their willingness to persist in learning and applying mathematical concepts [83, 84]. Actually, MathI is both an affective and cognitive construct. It encompasses emotional responses (e.g., enjoyment of solving problems) and cognitive evaluations (e.g., perceiving math as valuable or useful) [81, 82]. MathI acts as a driver for intrinsic motivation and deep learning in mathematics. When students are genuinely interested, they are more likely to engage actively in mathematics learning activities, persist through challenging mathematical problems, and develop high levels of mathematical competence [81, 82].

Previous research presented mixed findings on how STEM informal learning opportunities affected interest in STEM fields. Specifically, some studies reported that STEM informal learning experiences positively fostered students’ interest in STEM because of the enjoyable nature of these activities [28]. Conversely, some other studies reported that the effect of STEM informal learning opportunities on STEM interest was not statistically significant [85]. Specifically, some scholars have proposed that social media can lead to distraction and cognitive overload [76, 86, 87], which may ultimately diminish students’ interest in the subject matter. The highly interactive and information-dense nature of social media platforms can overwhelm students, making it difficult for them to focus on specific learning tasks [88]. This constant stream of information, combined with the potential for interruptions and multitasking, can fragment students’ attention and reduce their ability to engage deeply with the content [89]. As a result, the initial enthusiasm for learning through social media may wane, leading to a decline in motivation and interest over time. This highlights the importance of carefully managing the use of social media in educational contexts to ensure that its benefits outweigh its potential drawbacks [76]. Hence, the effect of social media-supported mathematics informal learning on MathI needs further investigations.

Control-value theory suggests that a student’s interest and motivation are shaped by their educational surroundings [38] (e.g., MLSM). For example, an environment that stimulates curiosity, such as the MLSM supported by ICTs, could make it easier and more enjoyable for students to access appealing mathematical content [38, 39, 90]. Therefore, it is highly probable that students will cultivate MathI through the MLSM, although empirical verification of this hypothesis remains limited. We propose:

Hypothesis 1

(H1). MLSM has a direct effect on MathI.

Mathematics self-efficacy (MathSE)

Mathematics self-efficacy (MathSE) denotes an individual’s confidence in his or her capacity to successfully accomplish mathematics-related tasks, including assignments, courses, and exams [91–93]. This belief encompasses confidence in mastering the necessary knowledge and skills required for success in mathematics [91, 94]. Individuals with strong MathSE can effectively understand and solve mathematical problems, perform well on mathematics exams, and meet the challenges presented in mathematics courses. Drawing on the social cognitive framework, self-efficacy emerged from four principal elements: mastery encounters, vicarious learning, societal encouragement, and physical conditions [95, 96]. In particular, experiencing positive emotions such as joy, excitement, and satisfaction during an activity tends to increase self-efficacy levels, whereas feelings of anxiety, sadness, and discontent can diminish them [96]. Logically, learners with a profound enthusiasm for mathematics tended to derive pleasure from mathematics learning activities, which in turn enhanced their self-efficacy. The association between enthusiasm for mathematics and self-efficacy received empirical validation from prior research [96, 97]. On this basis, we hypothesize the following:

Hypothesis 2

(H2). MathI has a direct effect on MathSE.

On the other hand, the impact of informal learning environments on MathSE was profound. Research suggested that children’s informal science and mathematics experiences significantly contributed to their interest and self-efficacy in these disciplines [24, 27]. These experiences, often characterized by practical, hands-on activities and demystified mathematics and science, render these subjects more approachable for students, thereby increasing their confidence and readiness to engage with these subjects in formal educational settings. On the other hand, prior studies highlighted the role of technology in enhancing self-efficacy within informal educational contexts [4]. Specifically, technology can provide personalized learning experiences that enhanced self-efficacy by enabling learners to monitor their progress and better understand their learning habits [2]. Additionally, findings have indicated that involvement in tech-based learning activities has been linked to improvements in academic achievement [58–60]. Engagement with digital platforms was demonstrated to enhance not only interaction but also mathematical skills among students. Studies conducted previously indicated that the incorporation of digital tools in mathematics education led to enhanced problem-solving capabilities and superior academic outcomes [10, 16]. Moreover, technology-based environments offer distinct opportunities for students to engage with mathematical content in manners that are both more engaging and pertinent to their daily experiences, thereby potentially diminishing anxiety and augmenting self-efficacy [97].

However, some scholars have proposed that reliance on AI-driven solutions for mathematical problems may undermine students’ ability to develop foundational mathematical skills and conceptual understanding [80]. For example, studies have shown that the availability of instant solutions through AI tools, such as ChatGPT, can lead to a decline in students’ problem-solving abilities [80]. This over-reliance on external aids may result in students becoming less proficient in fundamental mathematical operations and less capable of independently tackling complex problems [80]. Consequently, students’ MathSE may decrease, as their confidence in their own mathematical abilities is eroded.

Control-value theory suggests that a student’s beliefs about their own abilities and the value they ascribed to a task (e.g., self-efficacy) are significantly influenced by their educational environment [38]. According to this theory, these beliefs, including self-efficacy, are shaped through interactions within educational settings. For example, when students perceive learning materials as accessible and comprehensible, they are likely to feel more competent, thereby enhancing their self-efficacy [38]. Therefore, engaging in a supportive educational setting such as the MLSM could promote the cultivation of self-efficacy in learners.

Thus, within the context of mathematics, it is plausible that students might feel more assured about their mathematical skills or performance following their participation in MLSM. We hypothesize:

Hypothesis 3

(H3). The MLSM has a direct effect on MathSE.

Self-regulation in mathematics learning (SR)

SR was characterized through the mechanism by which students established objectives in math education and then endeavoured to oversee, manage, and direct their cognitive processes, motivation, and actions, all within the framework of their goals and the characteristics of their math learning context [98]. According to control-value theory, students employ diverse strategies across different educational scenarios [38]. Furthermore, educational environments offering autonomy and support are seen as conducive to fostering self-regulated learning [38]. Substantial evidence has supported the notion that the integration of ICTs into educational activities enhances students’ ability to self-regulate [99–101].

Some prior studies explored the impact of MLSM and identified a significant link between these activities and SR [49]. These studies utilized structural equation modelling to demonstrate how digital platforms could facilitate self-regulatory behaviors by providing unique opportunities for students to engage with mathematical content outside traditional classroom settings, thus promoting a more autonomous and personalized learning experience [49]. In the context of MLSM, students often lack real-time assistance or direction from educators and instead have to utilize various cognitive and metacognitive tactics (e.g., establishing goals and planning, tracking progress, self-evaluation) to advance their independent learning [102]. However, the link between social media-supported learning and self-regulation, particularly in the realm of mathematics education, remains largely unconfirmed. We hypothesize:

Hypothesis 4

(H4). The MLSM directly influences the SR.

Recent findings have indicated that self-regulation significantly influences interest levels [103]. A comprehensive systematic review revealed that learners’ inherent drive and enthusiasm markedly improved following training in self-regulated learning strategies [104]. Specifically, a recent study also suggested that self-regulation serves a vital function in mediating the influence of technolgy-based learning environments on learners’ interest [105]. Learner interest can be stimulated through active engagement and appealing educational settings [106]. Actually, as some research pointed out, in learner-centric learning spaces, learners must employ self-regulatory abilities to effectively interact with these environments [105].

Given the central role of self-regulation in enhancing engagement and interest within educational settings, further exploration of its specific applications in mathematics education is imperative. Within the context of technology-enhanced learning environments, the necessity for robust self-regulation has been accentuated. For instance, Carneiro et al. underscored that self-regulatory practices were crucial for effectively navigating and benefiting from these technologically enriched educational spaces, ultimately increasing learners’ interest and engagement [105]. This highlighted the potential of self-regulation as a critical mediator that could significantly influence learners’ interactions with and outcomes from technology-based educational resources.

Moreover, the link between self-regulation and interest in STEM fields was supported by empirical studies that focused on various educational technologies [4]. For example, one prior research demonstrated that the deployment of a ubiquitous-physics application significantly enhanced students’ self-efficacy and achievements in physics. This evidence supported the hypothesized direct impact of self-regulation on mathematics interest, suggesting that enhanced self-regulatory capacities could facilitate deeper engagement and sustained interest in mathematics learning environments. Meanwhile, if learning settings are engaging and stimulating but learners lack the self-regulatory skills needed for participation (for example, if learners struggle to maintain focus owing to insufficient self-discipline), their interest in the subject may be adversely affected [105, 106]. Nonetheless, the specific link between self-regulation and interest amid the realm of mathematics education persists as definitively established. We hypothesize:

Hypothesis 5

(H5). The SR directly impacts MathI.

Self-regulation has been identified as having a significant relationship with academic self-efficacy [107]. Prior research suggested that self-efficacy is a result of self-regulatory practices, as self-regulatory behavior itself represents a form of enactive mastery experience, recognized as a critical source of self-efficacy [95]. Some scholars explained that when students executed their tasks and achieved objectives within the self-regulation process, they evaluated the outcomes and formed beliefs about their ability to manage future similar educational activities [105, 107]. This connection has been supported by empirical evidence. For example, some research reported that learners who effectively employed self-regulated learning tactics often exhibited increased confidence in their academic capabilities [38, 108, 109]. Notably, research has demonstrated that self-regulation skills directly influence students’ beliefs in their mathematical capabilities [108]. This finding was pivotal, as it underscored the role of self-regulated learning strategies in fostering a robust sense of MathSE.

Furthermore, the integration of technology in learning environments has introduced new opportunities and challenges for the development of self-regulation within educational settings [4]. Technological interventions could be a viable method to enhance self-regulated learning practices, subsequently increasing students’ self-efficacy in subjects such as mathematics. However, to our knowledge, this association has not yet been specifically explored within the realm of mathematics education. The following is postulated:

Hypothesis 6

(H6). SR directly impacts MathSE.

Figure 1 presents our conceptual research model, which provides a comprehensive framework for understanding the potential impacts of mathematics learning with social media (MLSM) on three critical outcomes: mathematics interest (MathI), mathematics self-efficacy (MathSE), and self-regulation in mathematics learning (SR). Grounded in the control-value theory, the model hypothesizes that MLSM may exert both direct and indirect effects on these variables, highlighting its multifaceted role in shaping students’ cognitive, emotional, and behavioral engagement with mathematics.

Fig. 1.

The conceptual model

According to the control-value theory, the learning environment—characterized by the accessibility, interactivity, and collaborative nature of MLSM—can influence students’ control beliefs (e.g., self-efficacy) and value appraisals (e.g., interest in mathematics). MLSM may directly enhance MathI by fostering curiosity, enjoyment, and the perceived relevance of mathematics through engaging content and social interactions. It may also directly strengthen MathSE by providing a supportive environment where students can gain confidence through peer discussions and immediate feedback.

Furthermore, the interactive and resource-rich nature of MLSM could directly affect SR by encouraging students to take charge of their learning through goal-setting, time management, and self-monitoring. Actually, social media is inherently designed to capture attention, often pulling users toward non-academic content such as entertainment, social interactions, or irrelevant materials. Students who cannot self-regulate may spend excessive time on unrelated activities, reducing the time and energy they devote to meaningful mathematics learning. Hence, in a social media-supported mathematics learning environment, students are required to actively engage in self-regulation to effectively navigate and utilize the diverse resources and interactive features provided by social media. This means effective MLSM may foster SR.

The model also emphasizes the intricate and dynamic interrelationships between MathI, MathSE, and SR, suggesting that changes in one variable can mediate the effects of MLSM on the others. For instance, effective MLSM may foster SR, and further leads to increased MathI and enhanced MathSE. This indicates that MLSM may not only directly impact MathI and MathSE, but also indirectly impact them via SR. By exploring these intricate relationships, our research aims to provide a nuanced understanding of how MLSM can serve as a transformative tool in fostering a holistic and effective mathematics learning experience.

Method

Participants

Our study was designed as a cross-sectional investigation, a methodology commonly employed in the fields of science and mathematics education research [57, 110–112]. We recruited 459 students from two universities in China using a random sampling approach. The use of random sampling helps mitigate potential selection biases. Participants were randomly selected from a comprehensive list of students enrolled in the respective universities, ensuring a diverse representation across different academic programs. It is important to note that randomly selected students had the right to refuse participation in the survey. Students who voluntarily participated in our study were invited to complete an online survey via a link shared with them. No incentives were provided for participation. After applying the listwise deletion technique to remove 61 incomplete surveys [113], we analysed data from 398 respondents. The gender distribution among these participants was 43% male and 57% female. With respect to academic majors, 37.9% of the participants were enrolled in STEM programs, 43.7% in the fields of economics and management, and 18.4% in the disciplines of humanities and social sciences. This diverse makeup suggests that our sample is relatively representative, given the slight gender disparity and broad coverage of academic disciplines. Among our participants, 19.1% were 18 years old and below, 25.4% were 19 years old, 23.6% were 20 years old, 16.8% were 21 years old, and 15.1% were 22 years old and above. In terms of time spent on mathematics learning with social media (MLSM), 13.4% of the students reported engaging in MLSM for less than one hour per week, 46.9% for 1–2 h, 35.8% for 2–3 h, and 3.9% for more than three hours. While our sample included a diverse range of academic disciplines, the gender disparity may limit the generalizability of our findings to broader populations. For a detailed breakdown of the demographic information, please refer to Table 1.

Table 1.

Demographic information of the participants

Demographic variable		Number (Percentage)
Gender	Male	171 (43.0%)
	Female	227 (57.0%)
Major	Science, Technology, engineering and mathematics (STEM)	151 (37.9%)
	Economics and management	174 (43.7%)
	Humanities and social sciences	73 (18.4%)
Age	18 years old and below	76 (19.1%)
	19 years old	101 (25.4%)
	20 years old	94 (23.6%)
	21 years old	67 (16.8%)
	22 years old and above	60 (15.1%)
Time spent on mathematics learning with social media	Less than one hour per week	53 (13.3%)
	1–2 h per week	187 (47.0%)
	2–3 h per week	142 (35.7%)
	More than three hours per week	16 (4.0%)
Total		398 (100%)

Instrumentation development and data acquisition

Following the approach suggested by Jiang et al. [109], the development of our investigative apparatus unfolded across four phases. Initially, drawing from the pertinent literature, we crafted an English-language questionnaire featuring four key variables (refer to Table 2). A Chinese variant was subsequently created, adhering to established translation and retranslation protocols [114]. This version was then scrutinized by academic and linguistic specialists, leading to adjustments on the basis of their feedback.

Table 2.

Items for the preliminary instrument utilized during the pilot study

Concepts	MLSM	MathI	MathSE	SR
Number of Questionnaire Items	6	5	5	5
Abbreviated Title of Questionnaire Items	MLSM1, MLSM2, MLSM3, MLSM4, MLSM5, MLSM6	MathaI1, MathI2, MathI3, MathI4, MathI5	MathSE1, MathSE2, MathaSE3, MathSE4, MathSE5	SR1, SR2, SR3, SR4, SR5
References for item development	[47, 50, 85, 120, 121]	[57, 111, 114, 122]	[109, 111, 122, 123, 124]	[125–127]

A preliminary trial was subsequently administered to 139 university students, guided by [115], resulting in modifications or exclusions of certain items on the basis of clarity and comprehension issues highlighted by the participants. Specifically, items with factor loadings below 0.40 were excluded, as they failed to meet the statistical criteria for item retention [116]. Additionally, elements exhibiting low statistical significance were excised following an initial analytical evaluation [116]. After the trial, seven specific items were discarded (MLSM2, MLSM6, MathI4, MathI5, MathSE1, and MathSE5). The pilot study revealed initial reliability metrics (Cronbach’s α = 0.986), indicating strong internal consistency across the items retained.

Informed by the pilot study outcomes, an enhanced seven-point Likert scale was developed for the main survey. Participants were randomly recruited in the two universities, with all respondents voluntarily participating after being briefed on privacy safeguards and their participation rights. Following data gathering, we applied the compiled dataset of 398 responses to validate the refined research tool. It should be noted that the reliability and validity of our research instrument was ensured by the confirmatory factor analysis (for more details, please see the Results section) [117–119].

Data analysis

Using AMOS 23 graphics, we implemented a two-phase structural equation modelling (SEM) strategy to rigorously examine the complex interactions among the four variables [117, 118]. Prior to initiating the first phase, we conducted tests for univariate normality to verify the normality assumption, which is essential for the accurate application of structural equation modelling [117]. During the initial phase, confirmatory factor analysis (CFA) was applied to ascertain the reliability and validity of our research instrument [119], ensuring that our measurement models were both valid and reliable. In the subsequent phase, we tested our hypotheses and determined the direct, indirect, and total impacts of MLSM on MathI, MathSE, and SR, employing the approach to interpret the complex SEM paths. This two-phase approach enabled us to disentangle the nuanced effects that the MLSM exerts on learner outcomes, providing a robust framework to explore the hypothesized relationships.

In our data analysis process, we meticulously utilized various fit indices to evaluate the adequacy of our model fit, as initially recommended by seminal studies [128]. These indices comprise a thorough set of metrics, such as the chi-square to degrees of freedom ratio (χ2/df), comparative fit index (CFI), Tucker‒Lewis index (TLI), standardized root mean square residual (SRMR), and root mean square error of approximation (RMSEA). These metrics were applied in accordance with established guidelines found in the literature. It is suggested slightly more stringent criteria for CFI and TLI, proposing that these indices ideally exceed 0.90 for a better representation of an excellent model fit [128], and advised that for a model to be considered a good fit, the χ2/df ratio should be less than 5. It is also argued for the RMSEA to be as low as 0.06 to denote a close fit, emphasizing the need for careful interpretation of model approximation errors [129, 130]. If the criteria are more lenient, SRMR and RMSEA values below 0.08 are also acceptable [128].

Results

The results of the measurement model

Before initiating the initial phase of the analysis, assessments for univariate normality were conducted. It is highlighted that significance tests focusing on kurtosis can be overly sensitive to the size of the sample, potentially rendering them less effective in the context of structural equation modelling [131]. Therefore, it was advised to closely examine the kurtosis measurements for each variable individually [131, 132]. To maintain the integrity of the model’s fit, ensuring that none of the kurtosis values exceeded a threshold of 3.0 was critical [132]. The analysis results from AMOS indicated that all absolute values of kurtosis fell below 3.0, signifying an absence of significant departures from normality.

At the beginning of our structural equation modelling process, the confirmatory factor analysis (CFA) method was utilized to evaluate the measurement model and affirm the validity of our instrument (see Table 3). The CFA results demonstrated that our four-factor models, MLSM, MathI, MathSE, and SR, achieved satisfactory fits with the collected data, as evidenced by the following fit indices: χ² = 262.181, df = 84, χ²/df = 3.121, RMSEA = 0.073, CFI = 0.980, TLI = 0.975, and SRMR = 0.030 (as for the threshold values for the fit indices, see the “Data Analysis” section). The analysis of the composite reliability (CR) and average variance extracted (AVE) measures for each construct clearly surpassed the stipulated benchmarks, with CR values above 0.70 and AVE values exceeding 0.50, respectively, thereby providing robust evidence of convergent validity [116]. Furthermore, the coefficients of interconstruct correlation were systematically found to be less than the square roots of the AVE values, as detailed in Table 4, effectively meeting the Fornell–Larcker criterion and thereby confirming discriminant validity [133]. Additionally, the Cronbach’s alpha (α) metrics for each construct notably exceeded the minimum recommended threshold of 0.70, which underlines the consistent reliability of the constructs within the scope of this study. This comprehensive assessment solidifies the structural integrity and reliability of the measurement model employed in the research.

Table 3.

The results of the confirmatory factor analysis

Concepts	Items	Mean	SD	Standardized factor loadings	Cronbach’s alpha (α)	CR	AVE
MLSM	MLSM1	5.582	1.493	0.865	0.963	0.964	0.870
	MLSM3	5.648	1.443	0.944
	MLSM4	5.508	1.492	0.966
	MLSM5	5.477	1.520	0.953
MathI	MathI1	5.254	1.634	0.955	0.984	0.985	0.955
	MathI2	5.309	1.549	0.994
	MathI3	5.284	1.557	0.979
MathSE	MathSE2	5.681	1.431	0.953	0.967	0.968	0.953
	MathSE3	5.648	1.443	0.971
	MathSE4	5.508	1.492	0.936
SR	SR1	5.794	1.265	0.939	0.950	0.950	0.795
	SR2	5.701	1.314	0.951
	SR3	5.497	1.412	0.912
	SR4	5.709	1.286	0.791
	SR5	5.698	1.372	0.855

Table 4.

Fornell–Larcker evaluation of discriminant validity

Concepts	MLSM	MathI	MathSE	SR
MLSM	0.933
MathI	0.678	0.977
MathSE	0.619	0.784	0.953
SR	0.667	0.767	0.860	0.892

Note. The bolded figures along the diagonal represent the square roots of the average variance extracted (AVE)

The results of the structural model

During the second phase of our structural equation modelling process, we focused on evaluating the structural model and scrutinizing the proposed hypotheses. As delineated in Table 5, the structural model demonstrated an appropriate fit to the data, as evidenced by the following goodness-of-fit metrics: χ2 = 262.181, df = 84, χ2/df = 3.121, RMSEA = 0.073, CFI = 0.980, TLI = 0.975, and SRMR = 0.030. The analysis revealed that mathematics learning with social media (MLSM) had a significant effect on mathematical interest (MathI) (β = 0.343, p < 0.001) and self-regulation in mathematics learning (SR) (β = 0.578, p < 0.001). Conversely, the influence of the MLSM on mathematics self-efficacy (MathSE) (β=-0.010, p = 0.794) did not reach statistical significance. Additionally, SR significantly affected MathI (β = 0.343, p < 0.001) and MathSE (β = 0.724, p < 0.001) scores.

Table 5.

Overview of hypothesis testing outcomes

Concepts	Pathway	Standardized coefficients (β)	C.R.	Findings
H1	MLSM→MathI	0.343***	6.739	Supported
H2	MathI→MathSE	0.267***	6.817	Supported
H3	MLSM→MathSE	-0.010	-0.261	Not Supported
H4	MLSM→SR	0.578***	14.758	Supported
H5	SR→MathI	0.748***	12.496	Supported
H6	SR→MathSE	0.724***	13.348	Supported

Note. *p < 0.05; **p < 0.01; ***p < 0.001

One notable finding from our study is the lack of a direct effect from MLSM to MathSE. This result suggests that while MLSM significantly influences MathI and SR, its direct impact on MathSE is not as pronounced. On the one hand, this discrepancy may be attributed to the complex nature of self-efficacy, which is often influenced by multiple factors beyond interest and engagement. According to Bandura’s social cognitive theory, self-efficacy is shaped by mastery experiences, vicarious experiences, verbal persuasion, and physiological states [95, 96]. In the context of mathematics learning, students’ self-efficacy may be more strongly influenced by their direct experiences with mathematical tasks, feedback from teachers, and their overall confidence in their abilities. For instance, mastery experiences, such as successfully solving complex problems, and verbal persuasion, such as encouragement from educators, play crucial roles in shaping students’ beliefs about their mathematical capabilities. These factors may overshadow the impact of social media-based learning on self-efficacy, suggesting that traditional classroom interactions and individual achievements remain pivotal in building students’ confidence in mathematics. On the other hand, the lack of a direct effect from MLSM to MathSE suggests that the relationship may be fully mediated by SR and MathI.

Investigating the impacts of mathematics learning with social media, we found that self-regulation (SR) significantly influenced mathematics interest (MathI) (β = 0.748, p < 0.001), and MathI notably affected MathSE (β = 0.267, p < 0.001). Hence, hypotheses H1, H2, H4, H5, and H6 received empirical support, whereas H3 was not supported.

Given the lack of support for H3, we adjusted our initial research model by excluding the pathway from MLSM to MathSE. The model was refined to reflect standardized coefficients, as depicted in Fig. 2. This alteration prompted a reevaluation examining the direct, indirect, and cumulative impacts of the MLSM on MathI, MathSE, and SR, as detailed in Table 4. While the direct influence of MLSM on MathSE was not substantial, its indirect effects through MathI (MLSM→MathI→MathSE, β = 0.092, p < 0.001), SR (MLSM→SR→MathSE, β = 0.418, p < 0.001) and both SR and MathI (MLSM→SR→MathI→MathSE, β = 0.115, p < 0.001) were found to be significant. Additionally, the indirect influence of MLSM on MathI through SR was also significant (MLSM→SR→MathI, β = 0.432, p < 0.001).

Fig. 2.

The results of the structural model

Table 6 shows the normalized Immediate, Indirect, and Cumulative Impacts of the MLSM on MathI, MathSE, and SR.

Table 6.

Normalized immediate, indirect, and cumulative impacts of the MLSM on mathi, mathse, and SR

Concepts	MLSM on MathI	MLSM on MathSE	MLSM on SR
Direct effect	0.343***	-0.010	0.578***
Indirect effect	0.432***	0.625***	–
Total effect	0.775***	0.615***	0.578***

Note. *p < 0.05; **p < 0.01; ***p < 0.001

Discussion and conclusions

Theoretical contributions

This study makes a significant theoretical contribution by extending the control-value theory [38, 39] into the context of technology-enhanced learning, particularly mathematics learning with social media (MLSM). By demonstrating how MLSM influences students’ control appraisals (e.g., self-efficacy) and value appraisals (e.g., mathematics interest), this research validates and refines the control-value theory in a modern, technology-driven educational environment. Our findings provide empirical evidence that informal, technology-rich learning environments play a critical role in shaping students’ control and value appraisals and self-regulatory behaviors, thus expanding the applicability of the control-value theory beyond traditional classroom settings.

Our study also fills a notable gap in the literature by applying the control-value theory to the underexplored domain of informal mathematics learning. While most previous research has focused on formal educational contexts [38–44], our work highlights the dynamics of students’ self-efficacy, interest, and self-regulation in informal learning environments supported by social media. This theoretical extension underscores the unique affordances of MLSM, such as its capacity to foster exploration, collaboration, and autonomy, which align with the principles of the control-value theory and enrich its framework with new insights specific to digital learning ecosystems.

We developed and justified a conceptual model that delineates the direct and indirect effects of MLSM on MathI, SR, and MathSE. This model contributes to the theoretical understanding of how learning environments mediated by technology interact with core psychological constructs, offering a comprehensive framework for analyzing the complex relationships among these variables. By incorporating both cognitive and emotional (e.g., MathI) and behavioral (e.g., SR) mediators, the model provides a deeper theoretical lens for examining the interplay between students’ learning environments and their motivational and behavioral outcomes.

This research contributes to the ongoing discourse on the efficacy of informal learning environments, which are frequently enabled by social media platforms [12, 33]. Actually, the impact of social media on STEM education remains a topic of debate. Some research highlights its potential to improve learning outcomes by increasing student engagement and expanding access to educational resources [73,74]. However, other studies warn that excessive or off-topic content can lead to distractions and cognitive overload, negatively affecting learning [76, 79, 80]. This dual nature highlights the need to carefully examine how social media influences mathematics education. While these platforms can facilitate personalized and interactive learning experiences [73,74], their misuse may also impair the cultivation of critical cognitive skills [76, 79, 80]. For example, students who frequently seek help in social media when solving mathematical problems may become less proficient in basic arithmetic operations and problem-solving strategies [80].

By demonstrating that the MLSM significantly enhances MathI (β = 0.343, p < 0.001) and SR (β = 0.578, p < 0.001) and indirectly links to MathSE (MLSM→SR→MathI→MathSE, β = 0.115, p < 0.001; MLSM→SR→MathSE, β = 0.418, p < 0.001; MLSM→MathI→MathSE, β = 0.092, p < 0.001), our study provides empirical support for the argument that informal learning settings can be as effective as traditional educational environments in some aspects, especially in STEM education [6, 24, 63].

Inconsistencies persist among previous findings regarding the impact of informal STEM learning on students’ interest. While some studies reported positive effects [28], others did not detect significant impacts [85]. For the first time, our findings establish that MLSM can influence MathI both directly and indirectly. This highlights that the effects of MLSM on MathI extend beyond formal educational settings to include informal learning environments, thereby providing further support for the potential of technology-enhanced informal learning. Furthermore, our findings reinforce the control-value theory, which posits that students’ interest is shaped by their learning environments [38, 39]. This theoretical confirmation underscores the importance of leveraging MLSM to create engaging and interest-driven learning opportunities in mathematics across diverse contexts. Our study found that MLSM positively impacts MathI. This highlights social media’s potential to engage students in challenging subjects like mathematics [19, 21]. By leveraging the unique capabilities of social media to facilitate engaging, personalized, and contextually rich learning experiences, educators were able to effectively increase students’ MathI. Integrating informal learning through technology can enrich the experience and potentially boost interest in mathematics [5]. This blend of social interactions and academic content created a dynamic learning environment that was conducive to developing a sustained interest in mathematics [5]. Furthermore, the role of informal learning support, which is often mediated through digital social platforms, can develop a positive attitude and increasing interest in mathematics among learners [6]. For instance, as Morris et al.’s research demonstrated, informal learning environments contributed significantly to enhancing MathI by providing relatable, real-world contexts that engaged learners more deeply than traditional classroom settings did [6].

The direct effects of technology-related classroom learning activities on MathSE have been highlighted in prior studies [10, 16, 24, 27, 58–60]. Additionally, research has shown that informal STEM learning activities can positively influence students’ self-efficacy, including MathSE [97]. Previous findings also suggest that students’ academic performance tends to improve after engaging in technology-supported informal learning, as these activities provide increased access to diverse subject knowledge [58–60]. Building on these conclusions, it was hypothesized that students might feel more confident in their mathematics performance and abilities after engaging in MLSM. However, contrary to our initial expectations, the direct effects of MLSM on MathSE were not statistically significant. This finding challenges the assumption that technology-supported learning environments directly enhance students’ self-efficacy. Instead, our study identified significant indirect effects of MLSM on MathSE, mediated by mathematics interest (MathI) and self-regulation (SR). These results provide indirect evidence supporting the efficacy of technology-enhanced informal learning, suggesting that its influence on MathSE is more nuanced and operates through interconnected pathways rather than direct causation. Furthermore, our findings underscore an important distinction between the impacts of technology-supported learning in formal and informal settings. While formal classroom environments may facilitate direct improvements in self-efficacy, informal learning environments like MLSM appear to foster self-efficacy through the enhancement of interest-related and behavioral factors. This highlights the complexity of technology-enhanced learning and points to the need for future research to explore the unique mechanisms at play in informal educational contexts.

Our findings also underscore the critical roles of the SR and MathI as mediators, illustrating that the pathways through which the MLSM influences MathSE are complex and multidimensional. This observation corroborates recent studies that underscore the importance of intermediary processes in shaping learning outcomes in the context of informal learning [100, 101, 134]. Notably, the significant mediating effect of SR within our model confirmed the vital importance of self-regulation skills in maximizing the educational advantages of social media [99, 100], a finding of particular relevance amidst the growing prevalence of digital learning environments where self-regulation is both essential and a targeted outcome [99, 102]. While MLSM may not directly influence MathSE, its indirect effects through key mediators demonstrate its potential to cultivate students’ confidence and capabilities in mathematics. These findings offer new insights into the differential roles of formal and informal technology-supported learning activities, paving the way for a deeper understanding of how educational technologies can be leveraged to enhance self-efficacy in mathematics and beyond.

Our findings align with previous research showing that incorporating ICTs into learning activities can significantly foster students’ self-regulation [6, 8]. Self-regulation, which involves cognitive processes such as planning, monitoring, and reflecting [6, 8], plays a critical role in understanding how technology-enhanced informal learning environments operate. Our study specifically confirms this association within the context of MLSM at the university level, an area that has been relatively overlooked in existing research [9]. By focusing on informal mathematics learning, this research extends the understanding of self-regulation beyond traditional classroom settings, demonstrating that MLSM provides students with opportunities to practice and enhance their self-regulatory skills. These opportunities are critical for fostering autonomous learning behaviors, particularly in technology-enhanced environments [24, 27]. More importantly, our study goes beyond prior research by demonstrating that self-regulation is not only a key outcome of MLSM but also an indispensable mediator in its broader impact. This finding underscores the central role of self-regulation in bridging the gap between informal technology-supported learning environments and students’ overall learning outcomes. Actually, students without self-regulation in the context of MLSM cannot achieve the desired learning goals because self-regulation is fundamental to navigating the vast and dynamic nature of social media platforms effectively. Without self-regulation, students may struggle to focus on relevant mathematical content, becoming distracted by non-educational elements commonly present in social media environments. By positioning self-regulation as a pivotal mechanism, this research contributes to a deeper understanding of how MLSM can be strategically designed to enhance students’ learning experiences and outcomes in mathematics.

In summary, this study offers a comprehensive model illustrating how mathematics learning with social media (MLSM) influences critical educational outcomes, including mathematics interest (MathI), self-regulation (SR), and mathematics self-efficacy (MathSE). By delineating both the direct and indirect effects of MLSM, this research provides a nuanced understanding of its impact, addressing gaps in the literature on informal and technology-enhanced mathematics learning. This study makes a unique contribution by validating the control-value theory in the context of MLSM. Furthermore, the integration of empirical evidence highlights the transformative potential of MLSM, extending its theoretical implications beyond mathematics education to broader applications in other subject areas and educational settings. By advancing the discourse on MLSM, this study lays a strong foundation for future research to explore how social media-supported learning can be leveraged to foster engagement, autonomy, and academic success across diverse disciplines.

While our study provides valuable insights into the relationships between MLSM, MathI, SR, and MathSE, it is essential to acknowledge the limitations imposed by the cross-sectional design. Specifically, because exposure and outcome were measured simultaneously, our findings cannot establish temporal precedence or causality. This limitation is particularly relevant when interpreting the mediation analysis results, as the observed associations may not reflect true causal pathways. Future research employing longitudinal designs or quasi-experimental methods would be necessary to more robustly test the causal mechanisms underlying these relationships. By doing so, we can enhance the theoretical contributions of our work and provide more definitive insights into the role of informal learning settings in mathematics education.

Practical implications

The findings of this study offer several practical implications for educators, policymakers, and technology developers aiming to leverage social media as an effective tool for mathematics learning.

The significant impact of MLSM on students’ MathI suggests that social media platforms can be effectively integrated into educational practices to make mathematics more engaging, interactive and interesting. For example, educators can use platforms like YouTube to create and share math tutorial videos, interactive quizzes, and problem-solving challenges, thereby fostering sustained interest and active participation. Additionally, teachers can curate content from social media platforms such as Instagram and Pinterest, which offer visual representations of mathematical concepts and activities, making learning more interactive.

This study also highlights the effects of MLSM on students’ SR, a critical skill for independent and lifelong learning. Educators can provide training on self-regulated learning strategies by incorporating reflective journaling, progress tracking, and peer evaluation into MLSM activities. For instance, teachers can use social media groups to facilitate peer feedback and collaborative problem-solving, helping students develop self-regulatory habits.

The indirect relationship between MathSE and MLSM highlights the importance of fostering interest and self-regulation as mediators. Educational programs and instructional designs should focus on creating supportive environments that build students’ confidence in mathematics. For example, gamified experiences, collaborative challenges, or mentorship opportunities through social media can strengthen students’ belief in their abilities while maintaining their interest. Educators can also host live Q&A sessions with mathematics experts on social media platforms or use interactive mathematics games and puzzles to enhance engagement.

The transformative potential of social media in mathematics learning underscores the need for educationally oriented features on social media platforms. Developers can create tools such as progress-tracking dashboards, interactive whiteboards, or AI-driven personalized content recommendations to support self-directed learning and improve mathematics education outcomes. Social media can facilitate peer collaboration and knowledge sharing in mathematics. Teachers and administrators can encourage students to form online study groups or participate in mathematics-focused social media communities, where they can collaborate on problem-solving, share resources, and seek support from peers and educators. Additionally, educators can leverage AI-based tutoring systems to provide personalized learning experiences tailored to individual student needs. These systems can adapt to student performance in real-time, offering targeted feedback and support through social media platforms. This approach allows for remote and flexible learning, enhancing the accessibility and effectiveness of mathematics education. As informal learning environments become increasingly prominent, schools and universities can provide students with workshops or courses that teach them how to leverage social media for academic purposes. This includes helping students identify credible resources, engage in meaningful discussions, and manage their time effectively while learning mathematics in informal contexts.

However, it is important to acknowledge potential challenges, such as digital literacy gaps, which may hinder students’ ability to fully benefit from these resources. Addressing these gaps through targeted interventions and training programs is essential to ensure equitable access to digital learning opportunities.

Future directions and limitations

While this study contributes valuable insights, it acknowledges certain limitations that could be addressed in future research. The cross-sectional design, while effective for this study, limits the ability to infer causality. The simultaneous measurement of exposure and outcome in a cross-sectional study makes it difficult to establish temporal precedence, which is essential for demonstrating directional effects between MLSM, MathI, SR, and MathSE. Redesigning the study using a longitudinal approach could properly establish temporal precedence in the mediation model and provide deeper insights into how the effects of MLSM on educational outcomes evolve over time [113]. Furthermore, the focus on Chinese university students may limit the applicability of these results to younger learners or different cultural settings. Extending this research to other educational levels and countries could provide a broader understanding of the impact of MLSM across different age groups and educational contexts. Additionally, qualitative studies could complement this research by exploring students’ personal experiences and perceptions of MLSM in their learning processes. Future research could also consider employing mixed methods approaches to combine quantitative and qualitative insights, offering a more comprehensive analysis of the phenomenon. This methodological improvement would allow researchers to triangulate findings and address the limitations of either approach alone [115, 134, 135].

Haozhe Jiang

is a ZJU100 young professor at Zhejiang University’s College of Education. His research has appeared in many prestigious Education journals. Notably, his paper “Online learning satisfaction in higher education during the COVID-19 pandemic: A regional comparison between Eastern and Western Chinese Universities” was recognized as an ESI Highly Cited Paper by Web of Science in 2022, 2023 and 2024. He’s also an Editorial Member of the International Journal of Smart Technology and Learning. Jiang’s expertise spans educational technology, educational psychology, and higher education, and he excels in technology-enhanced learning and educational psychology. Google Scholar: https://scholar.google.hk/citations?user=fBHt7XkAAAAJ%26;hl=zh-CN%26;oi=ao.

Author contributions

Conceptualization, Haozhe Jiang; methodology, Haozhe Jiang; software, Ling Dai; validation, Ling Dai; formal analysis, Ling Dai, Biao Zhu; investigation, Haozhe Jiang; resources, Biao Zhu, Jia Guan; data curation, Ling Dai, Jia Guan; writing—original draft, Ling Dai, Haozhe Jiang; writing—review and editing, Ling Dai, Haozhe Jiang; supervision, Rundong Liao, Guoxing Xu, Haozhe Jiang; project administration, Ling Dai, Wu Jin, Biao Zhu, Rundong Liao, Guoxing Xu, Haozhe Jiang, Jia Guan; funding acquisition, Ling Dai, Wu Jin, Biao Zhu, Rundong Liao, Guoxing Xu, Haozhe Jiang. All the authors have read and agreed to the published version of the manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (Grant number: 72374072).

Data availability

Data from this study are available upon request for researchers who meet the eligibility criteria. Please contact the corresponding authors directly via email.

Declarations

Ethics approval and consent to participate

This study was conducted in accordance with the Declaration of Helsinki and received approval from the Ethics Committee of East China Normal University (protocol code: HR 347–2022, approval date: June 8, 2022). Informed consent was obtained from all the subjects involved in the study.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.