Social cognitive theory

SCT, first proposed by Bandura [16], posits that human activities are shaped by a triadic reciprocal interaction among personal, behavioral, and environmental factors, emphasizing the critical role of learning experiences in developing individual behavior [17]. The theory argues that individuals acquire new cognitions and perceptions of their environment through observational learning during specific actions, influencing their future behaviors [18]. The SCT encompasses three key dimensions: personal, behavioral, and environmental factors [16,19]. Rather than operating in isolation, personal cognition, behavior, and environmental conditions dynamically interact and mutually influence one another, forming a reciprocal framework for understanding behavior formation.

SCT has been widely applied across diverse fields and has yielded significant insights into various forms of human behavior. In academic entrepreneurship, Liao, Nguyen [20] found that scholars strengthen their ESE through continually transforming research outputs, while observational learning—such as examining peers’ entrepreneurial cases and engaging in technology transfer activities—stimulates EI. In artificial intelligence entrepreneurship, Sutrisno, Siminto [21] reported that ChatGPT directly enhanced students’ EI and indirectly reinforced them by stimulating innovative behavior. This study focuses on the relationship between SMU and EI among culinary arts students, a context in which SCT is particularly well-suited. On one hand, students’ self-efficacy and entrepreneurial motivation are often strengthened through the display of their work on social media and the feedback they receive from others, exemplifying the role of personal factors within SCT. On the other hand, by observing others’ entrepreneurial experiences and participating in online interactions, students engage in learning and imitation processes that reflect behavioral factors. Furthermore, social media functions as an essential environmental resource platform for entrepreneurship, illustrating the role of environmental factors. Taken together, SCT’s triadic reciprocal model provides a systematic framework for explaining the formation of EI among culinary arts students in SMU.

Moreover, as academic research has advanced, SCT has increasingly been integrated with other theories to enhance its explanatory and predictive power. For example, in entrepreneurship education research, Liu [22] combined Goal Orientation Theory, Social Capital Theory, and SCT to construct an integrated model for explaining the formation of students’ EI. The findings confirmed that the integrated theoretical model exhibited greater explanatory capacity than single-theory approaches. Similarly, Nguyen and Nguyen [23] combined SCT with the TPB to uncover students’ entrepreneurial passion drivers, significantly improving the explanatory power for emotion-driven entrepreneurial behaviors. These findings provide an important theoretical basis for adopting an integrated SCT–TPB framework in the present study.

Despite the expanding application of SCT and its significant contributions across various fields, research on entrepreneurial behavior still exhibits notable gaps. First, most studies have concentrated on business or engineering students. At the same time, insufficient theoretical attention has been given to students’ entrepreneurial mechanisms and cognitive processes in non-traditional disciplines such as culinary arts [15]. Second, existing literature has primarily focused on offline environmental factors—such as family background and school education—while relatively few studies have examined the relationship between digital media environments and entrepreneurial behavior [24]. Given the ubiquity of SMU among students and the distinctive cognitive characteristics of culinary arts students, investigating the link between SMU and EI in this group offers an important opportunity to enrich SCT’s research perspective. This study seeks to fill this theoretical gap by analyzing the specific mechanisms through which SMU shapes EI, thereby providing both new theoretical contributions and empirical evidence for the advancement and practical application of SCT.

Theory of planned behavior

The TPB, first proposed by Ajzen [25], is primarily employed to explain why individuals form specific behavioral intentions and how these intentions translate into actual behaviors. The theory posits that behavioral intention is jointly determined by an individual’s attitude toward the behavior, SN, and PBC, which in turn influence the occurrence of the behavior [26]. As a result, TPB provides a parsimonious and widely validated framework for explaining the motivational structure underlying entrepreneurial intention [27].

In recent years, the TPB has been widely applied across multiple research domains and has achieved notable success in predicting and explaining human behavior. In sustainable entrepreneurship, Yasir, Babar [28] found that a positive attitude toward sustainable entrepreneurship directly strengthened EI, while supportive SN indirectly reinforced intention by enhancing perceived social support. In the entrepreneurship education, Slomski, Tavares de Souza Junior [29] demonstrated that PBC not only directly enhanced EI but also indirectly strengthened it by fostering a positive attitude. In the digital entrepreneurship, Garcez, Franco [30] revealed that favorable EA and supportive SN both directly promoted EI and indirectly enhanced it through improved PBC. These findings suggest that TPB provides a robust framework for examining how cognitive evaluations and perceived control shape entrepreneurial intention. In the present study, TPB is particularly suitable for analyzing the entrepreneurial intention of culinary arts students in a social media context, where attitudes, perceived social expectations, and control perceptions are likely to be shaped by digitally mediated experiences.

In addition, as academic research has progressed, the TPB has increasingly been integrated with other theories to enhance its explanatory and predictive power. For example, Ferri, Spanò [31] expanded the explanatory scope of the TPB in general higher education by incorporating the skill dimension, thereby deepening the understanding of students’ EI formation. However, the limited consideration of subjective traits and contextual environmental factors restricts the generalizability of this model and its explanatory strength across diverse entrepreneurial domains. In entrepreneurship education research, Maheshwari and Kha [32] combined the TPB with self-efficacy theory to examine the psychological and behavioral drivers of university students’ EI. This theoretical integration significantly improved the accuracy of predicting EI. Nonetheless, the model remains underdeveloped in incorporating broader macro-environmental contexts, such as policy frameworks and cultural settings, which must be strengthened to enhance its explanatory power across different contexts of entrepreneurship.

Despite the expanding applications and significant advancements of the TPB, existing research still reveals notable limitations. Most studies have focused on the influence of entrepreneurship education in higher education on students’ EI. At the same time, relatively little theoretical attention has been paid to new contextual factors such as SMU [32]. Given the pervasiveness of social media, examining the relationship between SMU and EI among culinary arts students offers an important opportunity to enrich the TPB research perspective. This study seeks to fill this theoretical gap by analyzing how culinary students’ attitudes, SN, and PBC evolve within SMU, thereby uncovering how these factors influence EI. In doing so, it provides new theoretical contributions and empirical evidence for the further development and practical application of the TPB.

Theoretical integration of social cognitive theory and the theory of planned behavior

Although social cognitive theory (SCT) and the theory of planned behavior (TPB) have each been extensively applied to explain entrepreneurial intention, their explanatory focus differs in important ways. TPB primarily addresses the motivational and intentional structure of entrepreneurship by clarifying how attitudes, subjective norms, and perceived behavioral control jointly shape individuals’ entrepreneurial intention. However, TPB provides limited insight into how these cognitive evaluations are formed through learning experiences and capability development, particularly in educational contexts.

In contrast, SCT emphasizes how entrepreneurial cognition is constructed, highlighting the roles of observational learning, experiential engagement, and self-efficacy in shaping individuals’ beliefs about their entrepreneurial competence. SCT is therefore well suited to explain how students’ entrepreneurial abilities and confidence are cultivated through social media–supported interaction and feedback. Nevertheless, SCT alone does not explicitly model how such capability-related cognitions are translated into deliberate entrepreneurial intention.

By integrating SCT and TPB, the present study captures both capability formation and intention formation within a unified framework. SCT explains how social media use contributes to the development of entrepreneurial self-efficacy and abilities, while TPB clarifies how these capability-based cognitions, together with attitudes, subjective norms, and perceived behavioral control, shape entrepreneurial intention. This integrative perspective provides superior explanatory power by linking learning-driven cognitive development with intentional decision-making, thereby offering a more comprehensive explanation of entrepreneurial intention among culinary arts students.

Entrepreneurial intention

EI refers to an individual’s psychological tendency and determination to start a business or engage in entrepreneurial activities, and it is widely recognized as a critical antecedent of entrepreneurial behavior [33]. Research on EI has drawn upon multiple theoretical frameworks, including the TPB [1,13,33], SCT [11,34], Social Capital Theory [22], and the Conservation of Resources Theory [5]. The TPB and SCT are the most widely applied and relevant to the present study. The TPB, as a foundational framework for explaining the formation of EI, posits that EI is determined by an individual’s attitude toward entrepreneurial behavior, SN, and PBC. For example, drawing on the TPB, Boucif, Wan Nawang [1] found that attitude directly predicts EI and indirectly influences it by shaping PBC. However, the TPB primarily emphasizes psychological cognition while paying insufficient attention to external environmental dynamics—such as policy, social resources, and cultural context—thus limiting its capacity to explain the mechanisms underlying EI fully. By contrast, SCT suggests that EI emerges from the reciprocal interaction of personal factors, environmental influences, and behavioral outcomes. For instance, Nwosu, Obidike [34], building on SCT, investigated how environmental factors (e.g., internships in commercial firms) and personal factors (e.g., entrepreneurial self-efficacy, creativity, and passion) shape students’ EI. Their findings showed that internships indirectly fostered EI by enhancing students’ ESE and creativity. Nevertheless, SCT emphasizes internal cognitive processes and direct interactions with the environment, while giving comparatively less attention to social influences such as group pressure and SN. As a result, it struggles to explain how these social factors drive cognitive convergence among individuals.

Scholars have increasingly employed quantitative methods to examine the factors influencing students’ EI in recent years. Common quantitative approaches include correlation analysis [35], regression analysis [11,36], mediation analysis [37], and SEM [38,39]. For example, Ghouse, Barber Iii [11] used regression analysis to investigate the formation mechanisms of rural students’ EI. They found that self-efficacy directly enhanced individuals’ confidence in their entrepreneurial abilities, strengthening EI. Through large-scale data collection and statistical analysis, quantitative research allows for precise measurement and examination of the strength of EI, its influencing factors, and the interrelationships among them. However, such research often places heavy emphasis on numerical analysis while overlooking the complex social, psychological, and cultural factors underlying EI, which may limit its ability to capture the multidimensional nature of the phenomenon entirely.

Studies have shown that personal, environmental, and sociocultural factors are important antecedent predictors of EI. First, personal factors include personality traits [40], self-efficacy [41], motivation [42], gender [43], and attitudes [44]. For instance, Schlaegel, Engle [40] examined the influence of personal factors—including personality traits and abilities—on EI. They found that innovative personality traits indirectly enhanced entrepreneurial intention by increasing the likelihood of identifying entrepreneurial opportunities. However, personal factors alone cannot account for the broader macro-environmental and sociocultural contexts in which EI are formed. Second, environmental factors include government policies [45], the economic environment [46,47], entrepreneurship education [29], and institutional settings [48]. For example, based on the theory of entrepreneurial intention, Požega and Ribić [46] investigated the impact of the economic environment on EI. They found that regional economic development levels influenced students’ perceptions of entrepreneurial opportunities and risks, shaping their EI. Nevertheless, environmental factors cannot fully explain internal motivations and individual choices; even in regions with favorable economic conditions and strong policy support, not all individuals pursue entrepreneurship. Third, sociocultural factors include social support [49], SN [50], social values [51], and social networks [52]. For instance, Neneh [49] examined the influence of entrepreneurial passion, ESE, and social support on students’ EI and found that social support directly enhanced confidence and motivation toward entrepreneurship, thereby strengthening EI. At the same time, social support indirectly promoted EI by reinforcing the effect of entrepreneurial passion on self-efficacy. However, while sociocultural factors shape individuals’ perceptions, attitudes, and values toward entrepreneurship, they cannot fully predict entrepreneurial behavior. Even in supportive sociocultural environments, not all individuals are willing to bear the risks and pressures associated with entrepreneurship.

Social media use, subjective norms, and entrepreneurial intention

Social media is an effective tool for connecting individuals, gathering additional information, and analyzing knowledge [53]. While prior studies have operationalized social media use in terms of frequency, duration, and intensity [54], the present study conceptualizes social media use (SMU) as entrepreneurial-oriented engagement with social media platforms, focusing on acquiring restaurant entrepreneurship knowledge, observing entrepreneurial practices, and interacting with relevant professional communities, rather than general recreational use.

SMU refers to students’ engagement with social media platforms, specifically in frequency, duration, and intensity [54]. Several studies have demonstrated a significant association between SMU and EI [55,56]. For example, drawing on the TPB, Loan, Libo-on [55] found that students who used social media to access entrepreneurial information and knowledge while building broader networks for support and resources exhibited stronger EI. Similarly, Al Halbusi, Soto-Acosta [56], examining the determinants of e-entrepreneurial intention among Iraqi university students, revealed that exposure to peers’ or entrepreneurs’ social media content fostered more positive affective evaluations of e-entrepreneurship, thereby enhancing EI. In line with these findings, this study proposes the following hypothesis:

H1: SMU has a significant positive effect on the EI of culinary arts students.

SN refers to the expectations and social pressure perceived from significant others or social groups when an individual makes decisions [57]. Prior studies have demonstrated that SMU significantly influences students’ SN [55,58]. For example, drawing on the TPB, Loan, Libo-on [55] found that SMU enhanced students’ SN by expanding their social networks and increasing interactions on digital platforms. Similarly, Barrera-Verdugo and Villarroel-Villarroel [58] examined the relationship between SMU frequency and the EA of Chilean business and engineering students, revealing that social media, as an external environmental factor, amplified students’ perceptions of social pressure through information dissemination and social interaction, thereby strengthening their SN. In line with these findings, this study proposes the following hypothesis:

H2: SMU has a significant positive effect on the SN of culinary arts students.

Numerous studies have demonstrated that SN significantly positively influence EI [32,59]. For instance, drawing on the TPB, Maheshwari and Kha [32] examined how entrepreneurship education support shaped the EI of Vietnamese university students and found that SN fostered EI by enhancing students’ entrepreneurial confidence. Similarly, Duan [59] investigated the factors influencing Chinese university students’ EI and reported that SN strengthened EI by increasing students’ confidence and motivation while reducing concerns about entrepreneurial risks. Accordingly, this study proposes the following hypothesis:

H3: SN significantly positively affect the EI of culinary arts students.

H4: SN mediate the relationship between SMU and EI among culinary arts students.

Social media use, entrepreneurial self-efficacy, entrepreneurial attitude, and entrepreneurial intention

ESE refers to an individual’s belief in their capability to perform tasks and achieve outcomes associated with entrepreneurship [60]. Prior studies have demonstrated that SMU significantly affects SES [61,62]. For instance, Kurnia Khafidhatur, Irsyad [61] showed that positive interactions on social media can enhance students’ ESE. Similarly, drawing on SCT and the TPB, Hu, Jabor [62] found that SMU, by providing access to skill-learning resources and peer interaction support, strengthened vocational students’ confidence in their entrepreneurial abilities, thereby improving their ESE. Based on these findings, this study proposes the following hypothesis:

H5: SMU has a significant positive effect on the ESE of culinary arts students.

EA refers to an individual’s overall positive or negative evaluation of specific entrepreneurial behaviors or actions [63]. It is not merely an expression of intention but also reflects an inherent disposition toward entrepreneurship as a particular goal [64]. Prior research has established a significant relationship between ESE and EA [65–67]. For example, within an integrated framework of the TPB and SCT, Cao [65] found that higher ESE strengthens individuals’ confidence in their entrepreneurial capabilities, fostering more positive EA. Similarly, drawing on the Cognitive-Affective Processing Systems Theory, Song and Lu [67] demonstrated that ESE shapes EA by boosting confidence, reducing fear of failure, enhancing expectations of success, and improving adaptability. Accordingly, this study proposes the following hypothesis:

H6: ESE has a significant positive effect on the EA of culinary arts students.

Multiple studies have shown that students’ EA significantly and positively influence their EI [68,69]. For example, drawing on the TPB, Lim, Kim [68] found that positive EA strengthened nursing students’ confidence in their abilities and willingness to take risks, enhancing EI. Similarly, Qudsia Yousaf, Munawar [69], examining the effects of entrepreneurship education, EA, and culture on Pakistani students’ EI, reported that students with more positive attitudes toward entrepreneurship were more inclined to identify and evaluate entrepreneurial opportunities. Moreover, positive affect fostered by EA stimulated intrinsic motivation, reinforcing students’ determination to overcome obstacles and embrace challenges, strengthening EI. Accordingly, this study proposes the following hypothesis:

H7: EA has a significant positive effect on the EI of culinary arts students.

H8: ESE and EA mediate the relationship between SMU and EI among culinary arts students.

Social media use, perceived behavioral control, and entrepreneurial intention

PBC refers to the perceived ease or difficulty of performing a specific behavior [25], reflecting an individual’s evaluation of their ability to carry out entrepreneurial activities [70]. Numerous studies have shown that SMU significantly influences PBC [55,71,72]. For example, Huang and Zhang [71] demonstrated that SMU provides students with entrepreneurial knowledge and information, directly enhancing their ESE. In turn, students with higher self-efficacy feel more confident in overcoming challenges and difficulties, improving their PBC [72]. Similarly, Loan, Libo-on [55] found that SMU enhanced Vietnamese students’ PBC through multiple mechanisms, such as providing information, building relationships, sharing experiences, and offering convenience, collectively boosting their confidence in entrepreneurial capabilities. Based on these findings, this study proposes the following hypothesis:

H9: SMU has a significant positive effect on the PBC of culinary arts students.

Prior studies have confirmed a significant relationship between PBC and students’ EI [73,74]. For example, drawing on the TPB, Lim, Kim [73] examined the factors influencing nursing students’ EI. They found that students with higher PBC were more likely to enhance their skills and knowledge through entrepreneurship education, strengthening their ESE and ultimately their EI. Similarly, Sargani, Jiang [74] investigated gender differences in the EI of Pakistani agricultural students. They revealed that students with higher PBC were more inclined to seek and accept support from family, friends, and society. This reinforced their positive attitudes toward entrepreneurship and, in turn, increased their EI. Based on these findings, this study proposes the following hypothesis:

H10: PBC has a significant positive effect on the EI of culinary arts students.

H11: PBC mediates the relationship between SMU and EI among culinary arts students.

Based on the above literature review and hypotheses, the hypothetical model of this study is shown in Fig 1.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Research model.

https://doi.org/10.1371/journal.pone.0342055.g001

Participants

Before completing the questionnaire, participants were provided with detailed information about the study, including its purpose and the protection of participant privacy. Only those who gave written informed consent were allowed to proceed, and participants retained the right to withdraw at any time. Data was collected from March 10 to April 25, 2025, via the Wenjuanxing platform (https://www.wjx.cn/). This platform was chosen because of its professional data collection functions, strong adaptability for Chinese users, and robust privacy safeguards, making it particularly suitable for large-scale and remote data collection while ensuring research rigor and data quality. A snowball sampling method was employed, whereby initial participants referred new participants who met the eligibility criteria, creating a chain-referral network that gradually expanded the sample size while maintaining sample relevance. Participants were also encouraged to recommend other suitable respondents to join the survey. As a non-probability sampling technique, snowball sampling may introduce selection bias by over-representing individuals with stronger social connections or greater interest in entrepreneurial activities; therefore, the findings should be interpreted with appropriate caution regarding their generalizability. In addition, Wenjuanxing’s response-time recording function was used to monitor and assess data quality. Respondents who completed the survey and provided valid answers were given a random cash reward.

This study collected data from 558 undergraduate students at three universities in Sichuan Province, China. To ensure sample quality, we followed the procedure of Wu, Li [75] and applied three screening criteria. First, based on preliminary testing, the average completion time for the questionnaire ranged from two to five minutes. Participants who completed the survey in less than 90 seconds were considered to have responded carelessly, and their data were excluded. Second, the questionnaire contained one reverse-coded item; participants who failed to provide the correct reverse-coded response were deemed inattentive, and their data were removed. Third, questionnaires with identical responses across all items were also excluded. After this rigorous screening process, 59 invalid responses were discarded, leaving 499 valid questionnaires for further analysis. In the hypothesized model of this study, the maximum number of arrows pointing to an endogenous latent construct was four. According to Hair [76], a minimum sample size of 191 is required to achieve an R² value of 0.10 at the 1% significance level. The final sample size of 499 participants far exceeded this threshold, ensuring the robustness and reliability of the study’s findings.

Among the 499 valid questionnaires, 170 were male (34.1%) and 329 were female (65.9%). In terms of grade distribution, 227 were freshmen (45.5%), 134 were sophomores (26.9%), 122 were juniors (24.4%), and 16 were seniors (3.2%) (Table 1). This representative sample provides a solid foundation for analyzing SMU’s impact on culinary arts students’ EI. The sample’s diversity and size help ensure that this study’s findings broadly apply to the broader population of university students within this research context.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Demographic information of participants (N = 499).

https://doi.org/10.1371/journal.pone.0342055.t001

Measurement instruments

The measurement instruments consisted of two parts. The first part collected participants’ demographic information, while the second part gathered self-reported data on each construct. Established and validated scales were adopted to assess these constructs, with appropriate modifications to fit the specific research context and objectives. This approach ensured that the measures accurately reflected participants’ conditions within the study setting.

In addition to basic demographic information, the measurement instruments included six key constructs: SMU, SN, ESE, EA, PBC, and EI. Compared with the seven-point Likert scale, the five-point Likert scale offers notable advantages in improving reliability and validity [77], reducing bias, and enhancing the sensitivity of statistical analyses [78]. Accordingly, all constructs in this study were measured using a five-point Likert scale, ranging from (1) “strongly disagree” to (5) “strongly agree.”

The SMU scale was adapted from Zhang, Abbas [79]. It was designed to measure the extent to which culinary arts students use social media in restaurant entrepreneurship. The scale consists of five items (e.g., “Social media provides me with opportunities to learn about restaurant entrepreneurship knowledge”). The Cronbach’s alpha coefficient for this scale was 0.869, indicating good reliability. Notably, the items emphasize learning-oriented and entrepreneurship-related uses of social media, rather than general recreational engagement, ensuring consistency between the conceptual definition and the operationalization of SMU in this study.

The SN scale was adapted from Nguyen and Nguyen [23]. It was employed to assess the social pressure or support perceived by culinary arts students from family, friends, and peers when making decisions about restaurant entrepreneurship. The scale comprises three items (e.g., “If I choose to pursue restaurant entrepreneurship, my family will support me”). The Cronbach’s alpha was 0.861, indicating good reliability.

The PBC scale was adapted from Tchokoté, Bawack [80]. It was used to measure the extent to which culinary arts students perceive themselves as capable of managing and controlling the process of restaurant entrepreneurship. The scale consists of six items (e.g., “It would be easy for me to start and run a restaurant business”). The Cronbach’s alpha was 0.938, indicating excellent reliability.

The ESE scale was adapted from Duong [81] and was employed to measure the level of self-efficacy perceived by culinary arts students in restaurant entrepreneurship. The scale comprises five items (e.g., “I have strong creativity and innovative ability”). The Cronbach’s alpha was 0.903, indicating high reliability.

The EI scale was adapted from Duong [81] and was designed to measure the extent of culinary arts students’ willingness to engage in entrepreneurial activities in the restaurant sector. The scale consists of six items (e.g., “I am willing to do whatever it takes to become a restaurant entrepreneur”). The Cronbach’s alpha was 0.906, indicating high reliability.

The EA scale was adapted from Tchokoté, Bawack [80] and was used to assess culinary arts students’ attitudes toward entrepreneurship in the restaurant sector. The scale comprises five items (e.g., “For me, the benefits of becoming a restaurant entrepreneur outweigh the drawbacks”). The Cronbach’s alpha was 0.867, indicating good reliability.

Data analysis

This study employed a two-stage analytical approach to test the hypotheses and construct the predictive model. First, Partial Least Squares–Structural Equation Modeling was used to identify the linear relationships between exogenous and endogenous constructs. Partial Least Squares-Structural Equation Modeling offers substantial advantages in small sample sizes, non-normal data distributions, exploratory research, and complex models [76]. With a sample size 499 and a model comprising six constructs and 30 items, this study qualifies as an exploratory analysis of a complex model [82]. Therefore, Partial Least Squares-Structural Equation Modeling was appropriate for the data analysis. However, structural equation modeling has inherent limitations in capturing nonlinear and non-compensatory relationships, critical for understanding social media’s influence on culinary arts students’ EI.

To address these limitations, the second stage of this study incorporated ANN. ANN enable non-compensatory models to capture linear and nonlinear relationships, enhancing predictive accuracy [83]. Moreover, prior studies have highlighted the robustness of ANN in handling complex data patterns and prediction tasks [84]. By integrating SEM with ANN, this study increased the depth and precision of the data analysis, allowing for a more comprehensive understanding of how social media influences the EI of culinary arts students and the relative importance of the contributing factors.

Ethics statement

This study was reviewed and approved by the Human Research Ethics Committee of Sichuan Tourism University (Approval No. SCTU-2025-01-0001). The study complies with the National Statement on Ethical Conduct in Human Research (2007). Written informed consent was obtained from all participants prior to their involvement in the research.

Measurement model

The evaluation of the measurement model followed established guidelines.

Step 1: Indicator reliability. The outer loadings of all indicators should be ≥ 0.70 [85].

Step 2: Internal consistency reliability. Two tests were used: Cronbach’s alpha (α) and composite reliability (CR). For both measures, the threshold value should be ≥ 0.70 [87].

Step 3: Validity.

1. (1) Convergent validity: The Average Variance Extracted (AVE) should be ≥ 0.50 [88].
2. (2) Discriminant validity: Two methods were applied to assess discriminant validity:

The Fornell–Larcker criterion [88].

The Heterotrait–Monotrait (HTMT) ratio of correlations [89].

First, following Hair [85] recommendations, the outer loadings of the items should exceed 0.70. In this study, all 30 indicators had outer loadings greater than 0.70.

Second, internal consistency reliability and convergent validity were assessed using Cronbach’s alpha, CR, and AVE. As shown in Table 2, all values satisfied the minimum criteria for indicator reliability and internal consistency reliability. Moreover, the AVE for all constructs was ≥ 0.50, confirming adequate convergent validity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 2. Reliability and average variance extracted.

https://doi.org/10.1371/journal.pone.0342055.t002

Third, Table 3 presents the correlation matrix used to assess discriminant validity based on the Fornell–Larcker criterion. According to Hair [76], the square root of the AVE for each construct should be greater than its highest correlation with any other construct in the model. The results clearly met this standard, confirming adequate discriminant validity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 3. Discriminant validity (Fornell-Larcker criteria).

https://doi.org/10.1371/journal.pone.0342055.t003

The HTMT, proposed by Henseler, Ringle [89], is a criterion for assessing discriminant validity. The HTMT is defined as the ratio of the average heterotrait–heteromethod correlations to the average monotrait–heteromethod correlations. Heterotrait–heteromethod correlations refer to correlations between indicators across different constructs, whereas monotrait–heteromethod correlations refer to correlations between indicators within the same construct. This study’s HTMT values were calculated using Smart Partial Least Squares (see Table 4). All the HTMT values fell within the acceptable threshold of ≤ 0.85 [89], confirming discriminant validity.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 4. Discriminant validity (The Heterotrait–Monotrait criteria).

https://doi.org/10.1371/journal.pone.0342055.t004

Structural model

The structural model was assessed following the recommendations of Hair [76] using the following steps:

1. (1) Collinearity assessment: Variance Inflation Factor (VIF) values should be less than 3.3 to indicate the absence of multicollinearity issues.
2. (2) Significance and relevance of structural relationships: Path coefficients were evaluated based on their significance levels, with p < 0.05 considered statistically significant.
3. (3) Coefficient of determination (R²): The model’s explanatory power was assessed using R² values, with threshold levels of 0.190 (weak), 0.333 (moderate), and 0.670 (substantial).

First, collinearity was assessed using the VIF. A VIF value ≥ 3.3 indicates potential collinearity issues. As shown in Table 5, all VIF values were below the acceptable threshold of 3.3 [76]. Therefore, no collinearity problems were detected.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 5. VIF.

https://doi.org/10.1371/journal.pone.0342055.t005

Second, Table 6 presents the path coefficients (β values) for the relationships among the constructs in the model. The significance of the path coefficients was assessed using the bootstrapping procedure in Partial Least Squares. The t-values and p-values were used to test whether the path coefficients (β) were statistically significant at the 5% level. A 5% significance level requires p < 0.05 and t > 1.96. Among the direct predictors of EI, EA exerted the strongest effect (β = 0.667, t = 20.964, p = 0.000), followed by PBC (β = 0.244, t = 7.412, p = 0.000). In addition, SMU significantly predicted ESE (β = 0.344, t = 7.458, p = 0.000), PBC (β = 0.303, t = 7.364, p = 0.000), and SN (β = 0.587, t = 14.641, p = 0.000). However, SMU (β = –0.038, t = 0.847, p = 0.397) and SN (β = 0.055, t = 1.419, p = 0.156) did not significantly predict EI.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 6. Hypothesis testing.

https://doi.org/10.1371/journal.pone.0342055.t006

Third, the coefficient of determination (R²) represents the proportion of variance in an endogenous construct explained by its associated exogenous constructs [76]. Values of approximately 0.67 are considered substantial, around 0.33 moderate, and around 0.19 weak. As shown in Table 7, SMU, SN, PBC, EA, and ESE explained 70.7% of the variance in EI, indicating substantial explanatory power.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 7. R².

https://doi.org/10.1371/journal.pone.0342055.t007

Mediation analysis

To assess the mediating roles of SN, PBC, ESE, and EA in the relationship between SMU and EI among culinary arts students, this study employed the bootstrapping procedure in Partial Least Squares-Structural Equation Modeling [90]. Both direct and indirect effects were examined to determine the type and magnitude of mediation. As shown in Table 8, PBC (β = 0.074, t = 5.100, p = 0.000) and the combined mediation of ESE and EA (β = 0.135, t = 5.542, p = 0.000) exerted significant indirect effects in the relationship between SMU and EI. However, the direct effect of SMU on EI (β = –0.038, t = 0.847, p = 0.397) was insignificant. These findings indicate that PBC, ESE, and EA fully mediate the relationship between SMU and EI among culinary arts students. Finally, SN (β = 0.032, t = 1.385, p = 0.166) showed no significant direct or indirect effects, suggesting that SN do not mediate this relationship.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 8. Mediation analysis.

https://doi.org/10.1371/journal.pone.0342055.t008

Artificial neural network analysis

Given the potential nonlinear relationships between exogenous and endogenous variables, this study used the significant factors from the SEM path analysis as input neurons in the ANN model. Specifically, a feed-forward multilayer perceptron (MLP) architecture with one hidden layer was adopted, where SEM predictors served as input neurons and EI was specified as the output neuron. The use of SEM predictors as ANN inputs was intended to ensure theoretical consistency while allowing the ANN to explore potential nonlinear and non-compensatory relationships beyond linear path estimation. ANN are robust to noise, outliers, and relatively small sample sizes, and they can be applied to non-compensatory models in which a decrease in one factor does not need to be offset by an increase in another. The ANN analysis was implemented using IBM SPSS Neural Network Module. ANN algorithms can capture linear and nonlinear relationships and do not require data to follow a normal distribution [91]. Through iterative training, the model learns using the feed-forward back-propagation algorithm to predict outcomes [92]. Multilayer perceptrons with sigmoid activation functions were employed in both the input and hidden layers [93]. Multiple learning iterations were conducted to minimize errors and improve predictive accuracy [94]. This study used 70% of the sample for training, while the remaining data were reserved for testing. A ten-fold cross-validation procedure was performed to avoid potential overfitting, and the root mean square error was obtained [95]. As shown in Table 9, the average root mean square error values for the training and testing processes were 0.1357 and 0.1351, respectively, indicating stable and comparable predictive performance across training and testing samples.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 9. Root mean square of error values.

https://doi.org/10.1371/journal.pone.0342055.t009

A sensitivity analysis was conducted to evaluate the predictive power of each input neuron (Table 10). The normalized importance of each neuron was obtained by dividing its relative importance by the highest observed importance and expressing the result as a percentage [96]. The results revealed that EA was the most important predictor, with a normalized importance of 100%. This was followed by PBC (25.4%), ESE (20.7%), SMU (17.2%), and SN (13.1%).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 10. Sensitivity analysis.

https://doi.org/10.1371/journal.pone.0342055.t010

Theoretical implications

First, this study is the first to examine the EI of culinary arts students from the perspective of SMU. Previous research on EI among university students has primarily focused on factors such as individual traits, educational and economic environments, and sociocultural influences. In contrast, SMU—as a behavioral variable that reflects patterns of information acquisition and social interaction—has received limited attention in this domain. By systematically analyzing its impact on the EI of culinary arts students, this study fills a gap in the literature on specialized disciplines. It provides a novel theoretical lens for understanding the motivational drivers of entrepreneurship.

Second, this study employed a hybrid SEM–ANN approach to capture the linear and nonlinear relationships and the non-compensatory effects between SMU and EI. This methodological integration provides a novel perspective for research on the EI of culinary arts students. Beyond enhancing the accuracy of predictions, the combined method offers precise empirical evidence to inform entrepreneurship education in universities and policymaking by government institutions. Moreover, it serves as a methodological reference for future studies examining SMU’s psychological and behavioral impacts, offering broad academic value.

Third, this study integrates SCT and the TPB to explore the influence of SMU on students’ EI. Specifically, by combining the core constructs of the two theories—SN, PBC, ESE, and EA—this study developed a comprehensive model that examined the direct effects of SMU and revealed its indirect influence through multiple mediating pathways. This theoretical integration uncovers the psychological and behavioral drivers of EI among culinary arts students. It provides a new perspective for understanding the relationship between SMU and EI.

Fourth, the study’s inconsistent findings—such as the non-significant direct effects of SMU and SN on EI—open new avenues for future research. These divergences enrich the theoretical understanding of the relationship between SMU and the EI of culinary arts students, highlighting boundary conditions and contextual factors that may shape entrepreneurial cognition. In this study, boundary conditions refer to the contextual and disciplinary circumstances under which established theoretical relationships may weaken or operate through indirect rather than direct pathways. Specifically, the skill-intensive and practice-oriented nature of culinary arts education, together with culturally implicit forms of social influence, shapes how entrepreneurial intention is formed and limits the direct applicability of normative pressures. The study identifies these nuances and provides important academic insights and practical directions for subsequent research.

Practical implications

This study investigated the impact of SMU on the EI of culinary arts students and found that, although SMU did not directly influence EI, it indirectly affected intention through mediating variables such as PBC, ESE, and EA. These findings not only enrich the application of social media research in student entrepreneurship but also carry important practical value for policymakers, educational institutions, educators, and culinary arts students with entrepreneurial aspirations.

Given that SN do not directly influence the EI of culinary arts students, while PBC, ESE, and EA play decisive roles, policymakers should focus on empowering individuals rather than merely fostering a general entrepreneurial climate. Specifically, policies should strengthen students’ practical abilities and entrepreneurial confidence. For example, a dedicated “Culinary Entrepreneurship Training Subsidy” program could provide financial support for students who participate in restaurant kitchen rotations or complete standardized dish development projects, enabling them to gain hands-on experience and enhance their control over essential entrepreneurial skills. In addition, initiatives such as a “Youth Culinary Entrepreneurship Competition” could be organized, offering media exposure and access to entrepreneurial resources for outstanding participants, thereby stimulating students’ ESE and motivation.

Given that PBC was found to significantly influence the EI of culinary arts students, educational institutions and educators should prioritize strengthening students’ PBC. Specifically, they should focus on cultivating students’ entrepreneurial competencies and enhancing their ability to manage entrepreneurial processes. For instance, institutions could introduce a “Culinary Entrepreneurship Skills Development” course, requiring students to acquire restaurant entrepreneurship skills and industry insights through case studies and hands-on practice. Moreover, institutions could establish an “Entrepreneurial Competency Growth Portfolio,” which tracks students’ progress using both online engagement data from social media (e.g., likes and comments on culinary creations) and offline performance assessments (e.g., time-limited menu design and cost estimation tasks). This would reinforce students’ confidence in their entrepreneurial capabilities. Additionally, educators could provide personalized mentoring tailored to students’ specific weaknesses. For example, students lacking knowledge of supply chain management could be guided with supplier-matching case studies and tutorials on inventory optimization.

As an EA was identified as the strongest predictor of EI, culinary arts students should maintain optimism to strengthen their intentions and increase the likelihood of entrepreneurial success. Specifically, students can reinforce their EA through coursework and participation in restaurant entrepreneurship projects. For example, they may proactively enroll in restaurant entrepreneurship management courses or attend industry expert seminars to foster cognitions such as “entrepreneurship brings more benefits than drawbacks” and “restaurant entrepreneurship is highly attractive.” In addition, culinary arts students can engage in campus-based restaurant entrepreneurship initiatives or culinary competitions, using these practical experiences to strengthen their belief that “entrepreneurship will lead to success” and to cultivate an entrepreneurial spirit characterized by optimism and perseverance.

Limitations and future research

While this study provides valuable insights into the impact of SMU on the EI of culinary arts students, several limitations should be acknowledged, offering opportunities for future research. First, the data were collected using a cross-sectional design, which does not capture how the influence of social media on EI may evolve. Future studies could adopt longitudinal designs to examine SMU’s temporal dynamics and long-term effects on EI. Second, this study primarily employed a quantitative research design. Although quantitative methods allow for hypothesis testing and generalization, they may not fully capture the nuanced mechanisms through which different types of SMU affect EI. Future research could adopt mixed-method approaches, incorporating qualitative techniques such as interviews or focus groups, to provide richer contextual insights and explanatory depth. Third, the sample in this study was gender-imbalanced and drawn exclusively from Sichuan Province. In addition, the use of a snowball sampling approach may further constrain the representativeness of the sample. Given that gender and regional differences may influence students’ psychological, physiological, and social behaviors, the findings should be interpreted as context-dependent and may not fully represent the broader population of culinary arts students. Replicating this study with more diverse, gender-balanced, and geographically representative samples would help examine the robustness and boundary conditions of the proposed relationships and strengthen the validity and generalizability of the results.