The more self-control, the more anxious?— A network analysis study of the relationship between self-control and psychological anxiety among Chinese university students (2024)

Introduction

With the growing prevalence of anxiety symptoms among university students, self-control is an important potential influence. This study aims to understand the network structure of self-control and anxiety, and to identify the core symptoms within this network. It will provide a theoretical basis for the prevention and intervention of anxiety symptoms in university students.

Method

We used network analysis to study anxiety and self-control in 3,792 university students from six schools in Heilongjiang Province, China. We checked for linear connections in the networks using a restricted cubic spline. We conducted the analyses and made graphs using R software.

Results

(i) The total sample network showed that higher levels of self-control in university students were associated with lower anxiety levels and were validated by the restrictive cubic spline. (ii) There was the strongest negative correlation (edge weight = -0.42) between Without thinking (SC7) and Panic (A5), and the edge weight coefficients of this self-control component and anxiety symptoms were greater than those of other self-control components. (iii) Physical exertion (A6) and Scared (A7) were identified as the core symptoms of the overall network, with expected influence of 1.08 and 1.08 (Z-score). (iv) A significant difference was observed between the anxiety positive network structure and the total sample network structure, with the strongest positive correlation between Iron self-control (SC3) and Breathing difficulty (A2) (edge weight = 0.22), with the strongest negative correlation between Certain things (SC2) and Situations (A4) (edge weight = -0.35). (v) The self-control component Iron self-control (SC3) had only one positive edge in the rural network, and only two positive edges in not one child network.

Conclusion

The present study offered a new perspective on the relationship between self-control and anxiety using network analysis for the first time. The control component Without thinking (SC7) was an important concept influencing the negative correlation of anxiety, and Physical exertion (A6) and Scared (A7) were core symptoms in the total network. Heterogeneity analyses showed a tendency for the more self-controlled to be more anxious in the anxiety positive sample. These results may be a potential target for preventing and intervening anxiety in university students.

As the prevalence of Common mental disorders (CMDs) among university students rises significantly [1, 2], an increasing number of studies focus on the mental health of this demographic. Among all CMDs, the incidence of anxiety symptoms is particularly prominent. A review found that university students globally had an average anxiety rate of 32%, which could peak at 55% in certain regions [3]. This is notably higher than the general population and continues to increase. The university period represents a transitional period of academic and social identity transformation, a time that is prone to the onset of mental health issues such as anxiety and depression. These conditions not only diminish personal well-being but also exert negative impacts on society. For example, anxiety in university has been associated with an increased risk of suicidal ideation and suicide attempts among students [4, 5]. A meta-analysis showed that many Chinese university students were experiencing moderate levels of anxiety, which was becoming more severe as academic and employment pressures increased [6]. Moreover, family financial status, access to healthcare, and personal lifestyle habits are also recognized as potential contributing factors to anxiety [7, 8], indicating that the etiology of anxiety among university students is complex and multifaceted. Given the complexity of anxiety symptoms among university students, the undergraduate education phase is a crucial period for mental health prevention and intervention, as it helps students complete their education, secure a competitive edge in the job market, and actively adapt to societal changes. Hence, an in-depth examination of the factors contributing to university students’ anxiety to further clarify and develop effective intervention strategies.

University students’ self-control is significantly negatively correlated with anxiety, a finding of great importance for studying the mental health issues of this population. University students with higher self-control abilities tend to exhibit lower levels of anxiety [9,10,11], emphasizing the key role of self-control in promoting emotional health, learning ability, and social adaptability. It is worth noting that among the sample experiencing anxiety, there is a complex correlation between self-control and anxiety. On one hand, high anxiety itself weakens an individual’s attention control level, leading to impaired inhibitory control functions. This means that under high anxiety states, enhancing self-control might be more difficult and might even exacerbate anxiety due to overexertion [12]. On the other hand, self-control has been proven to help alleviate university students’ exam anxiety [13, 14]. This indicates that within the anxious cohort, the dynamic between self-control and anxiety is not uniform; in certain scenarios, an increase in self-control might paradoxically intensify anxiety symptoms, while in others, it could be instrumental in their reduction. Therefore, it is necessary for us to construct a network model of self-control and anxiety, providing more targeted mental health support and intervention measures for university students by clarifying the relationship between self-control and anxiety.

As an advanced research method, network analysis transcends the traditional paradigm of psychological research, offering a new perspective and standards of measurement [15, 16], and it can intuitively reveal the connections between individual elements and their underlying causes [17]. In exploring the comorbidity of anxiety and depression among university students, as well as the core symptoms that exist between the two, network analysis has become an indispensable tool [18, 19]. However, sociodemographic factors, such as family economic status, educational background, and cultural differences, might lead to heterogeneity in network analysis results [20, 21]. The variability of these factors across different groups may affect the interplay between anxiety and self-control and their network structure. Although existing research has made some progress in identifying the core characteristics of anxiety, considering the heterogeneity of sociodemographic factors, further in-depth research is still needed on how anxiety and self-control interact. Building a network analysis model that considers sociodemographic factors in the relationship between self-control and anxiety would help us to more comprehensively understand this relationship and fill the gaps in the research field. We aim to reveal the important pathways connecting self-control with anxiety, identify the core symptoms of anxiety, and provide a theoretical basis for the prevention and intervention of anxiety symptoms among university students. Based on these aims, the following hypotheses were proposed in this study: (i) Self-control is negatively correlated with anxiety symptoms in university students. (ii) Within the sample exhibiting anxiety, self-control is linked to anxiety symptoms in both negative and positive ways. (iii) Sociodemographic factors contribute to heterogeneity differences in the network structure.

Participants and procedure

In December 2021, we randomly sampled and surveyed university students from six schools in Heilongjiang Province, China, using an online questionnaire. The survey was conducted with strict adherence to ethical guidelines to ensure participants’ consent and privacy, which was crucial for the authenticity of their responses. Before the survey, an online system was established to facilitate a secure and efficient data collection process. To maintain the integrity of the data and prevent multiple submissions, the survey system was designed to allow each internet address to submit only one questionnaire. This measure ensured that each response was unique and from a different participant.

The online questionnaire was structured to include demographic variables, self-control, and depression-anxiety-stress levels among university students. The order of questions within the survey was determined based on a pre-test that aimed to minimize potential order effects. The pre-test involved a smaller group of participants and was used to assess whether the sequence of questions influenced the responses. Based on the findings, which indicated no significant order effects, we maintained a logical and consistent sequence throughout the survey.

This survey was supported by six schools in Heilongjiang Province, China, and included 4,431 university students from those schools, all aged between 17 and 23. Informed consent was obtained from all participating students before the survey commenced. To ensure data quality, we meticulously reviewed each questionnaire after collection, excluding those that were incomplete, irregular, or contained illogical responses. This rigorous process resulted in a total of 3,792 valid questionnaires being collected and completed anonymously, yielding a validity rate of 85.58%.”

Measurement

Brief self-control scale (BSCS)

The Brief Self-Control Scale (BSCS), developed by Tang et al., was a widely used measure of self-control, a construct linked with positive psychological outcomes [22, 23]. The BSCS scale was divided into two dimensions: Self-control and Impulse Control. The four items for Impulse Control were reverse-scored questions, and we reverse-adjusted these items. Higher scores indicated greater self-control. There were seven items in total in the Brief Self Control Scale, which we represented in our study using SC1-SC7 (Table1). Domestic and foreign scholars demonstrated that the Simplified Self Control Scale has good reliability and validity and can be used as a tool to study self-control levels [24,25,26,27]. This study used it to measure the self-control ability of university students, and the internal consistency of the BSCS data was good (Cronbach’s alpha = 0.667).

Depression, anxiety, and stress scale (DASS-21)

Anxiety levels of university students were measured using the DASS-21 scale, which many studies have demonstrated to be a recognized tool for measuring depression, anxiety, and stress symptoms in the sample population [28,29,30,31,32,33]. This study mainly utilized the anxiety items within the DASS-21 scale, which we represent as questions A1-A7 (Table1). Each question was rated on a scale from 0 (never) to 3 (always), with higher scores indicating higher levels of anxiety. The internal consistency of the Anxiety data was good, with a Cronbach’s alpha = 0.935).

Network analyses

All analyses were performed using R software (version 4.3.2). Firstly, we conducted demographic and scale analysis on the sample, and used Pearson correlation coefficient calculation to create a correlation heatmap. We set a score of 10 as the cutoff point for our analysis. Participants with DASS-21 scores at or above this threshold were classified as having significantly abnormal stress levels, forming the “Anxiety group”. Those with scores below 10 were considered to have basically normal stress levels, comprising the “Non-Anxiety group”. To outline the demographic profile of our sample, we used statistics such as totals, composition ratios, and p-values. The correlation heatmap displays numbers, asterisks, and colors, which respectively report the correlation coefficient, significance, and correlation strength.

Secondly, we utilized the qgraph package [34] in R 4.3.2 software to create the relevant network (“spring” layout from package qgraph). In the network structure diagram, nodes A1-A7 represented anxiety entries on the DASS-21 scale, nodes SC1-SC7 represented items on the BSCS scale, and edges connecting those nodes illustrate the relationships between the two items: red edges line indicated a negative correlation between nodes, while green edges line indicated a positive correlation. Additionally, the thickness of the edge line reflected the strength of the association between nodes. We used the ‘centrality auto’ and ‘centrality Plot’ functions in the qgraph package to calculate and visualize the centrality of nodes in the network, thereby understanding their importance within the network structure. Centrality indicators such as strength, closeness, betweenness, and expected influence (EI) are used to assess node importance within a network, and EI has proven particularly effective for analyzing both positively and negatively correlated edges within a network [35,36,37,38]. Therefore, we computed these four centrality metrics and used EI as an important metric for identifying influential nodes with negative edge lines. To facilitate comparison of different centrality indices on a uniform scale, we converted the results into z-scores.

Thirdly, we analyzed the network structure using the bootnet package [23] for accuracy and robustness tests. We applied the bootstrapping method for 1,000 autonomous samples (8 cores) for the difference test of centrality indicators. We also calculated the correlation stability coefficients (CS) to determine the stability of the centrality index. Studies have shown that the CS coefficient should be higher than 0.25, preferably above 0.5 [39]. Restricted cubic spline is a tool that can effectively respond to linear relationships and correlations between dependent and independent variables [40,41,42,43]. We used the plotRCS package to produce a restricted cubic spline to explore the linear relationship between anxiety and self-control and validate the results of the network analysis.

Descriptive statistical analysis

The descriptive summary of demographic information was shown in Table2. In this study, we analyzed data from 3,792 participants. Among them, 1,674 (44.1%) were male and 2,118 (55.9%) were female. In terms of age, 2,886 (76.1%) of the total sample were aged 17–20, while 906 (23.9%) were aged 21–23. The sample included 1,711 (45.1%) from rural areas and 2,081 (54.9%) from urban areas. In terms of the only child in a family, 2,094 (55.2%) were one child, while 1,698 (44.8%) were not one child. Of these, 3110 (82.0%) had a subjective socioeconomic status of medium and above. Referring to the grading criteria of DASS-21 scale, 563 (14.9%) of the study participants felt anxious. The sex difference was smaller in the Non-Anxiety group than in the Anxiety group (P < 0.05). Additionally, the mean age was lower, and the subjective socioeconomic status was higher in the Non-Anxiety group compared to the positive group. However, the differences in registered residence and being an only child in a family between the two groups were not statistically significant (P > 0.05).

Table1 showed the mean and standard deviation for each entry in the network analysis. Figure1 showed the strength of the associations between anxiety symptoms. The strongest associations within the anxiety entries were Scared (A7) - Physical exertion (A6), Physical exertion (A6) - Panic (A5), and Scared (A7) - Panic (A5), with their regularized partial correlation coefficients were 0.83, 0.80 and 0.78, respectively. The strongest associations within self-control were Iron self-control (SC3) - Resisting temptation (SC1), Work effectively (SC5) - Resisting temptation (SC1), and Work effectively (SC5) - Iron self-control (SC3), with their regularized partial correlation coefficients were 0.66, 0.65 and 0.64 respectively. The strongest associations between self-control and anxiety were Dryness (A1) - Without thinking (SC7), Breathing difficulty (A2) - Without thinking (SC7), and Panic (A5) - Without thinking (SC7), and their regularized partial correlation coefficients were − 0.34, -0.32 and − 0.32, respectively.

Network structure

Based on the total sample, we constructed a visual network model of the relationship between self-control components and anxiety symptoms in university students (Fig.2). In the anxiety community, there was one closely linked cluster and one relatively isolated node A4, and all non-zero correlations between the nodes were positive. In the self-control community, there were two clusters with relatively tightly connected internal nodes (Cluster 1 included SC2, SC4, SC6, and SC7, Cluster 2 included SC1, SC3, and SC5), and there was a negative non-zero correlation between the two clusters. Notably, there were extensive associations between anxiety symptoms and self-control components, and all non-zero correlations between the nodes were negative. The three strongest edges were Panic (A5) - Without thinking (SC7) (edge weight = -0.42), Dryness (A1) - Without thinking (SC7) (edge weight = -0.42), and Physical exertion (A6) - Without thinking (SC7) (edge weight = -0.42). Our network analysis revealed that SC7’s seven correlation edges with anxiety symptoms had higher edge weights than the other edges (Fig.2, Supplementary Table S1).

We conducted an in-depth study to understand the differences between the subsamples (Fig.3). Our study found significant differences in network structure between the anxiety positive sample and the total sample. Specifically, the anxiety positive sample showed both negative and positive correlation edges between self-control and anxiety clusters, which was not the case for the total sample. Within the self-control community, two distinct clusters were identified, each with tightly interconnected nodes: Cluster 1 comprised Certain things (SC2), Pleasure (SC4), Wrong (SC6), and Without thinking (SC7), while Cluster 2 included Resisting temptation (SC1), Iron self-control (SC3), and Work effectively (SC5). Notably, the sentiment framing of the questions in these clusters aligns with their respective correlations to anxiety. Specifically, Cluster 2, which contains questions with positive sentiment framing, exhibited positive correlations with anxiety. Conversely, Cluster 1, consisting of questions with negative sentiment framing, showed negative correlations with anxiety. In the anxiety positive network, we identified the three strongest positive correlations between self-control components and anxiety symptoms: Iron self-control (SC3) with Breathing difficulty (A2) (edge weight = 0.22), Iron self-control (SC3) with Trembling (A3) (edge weight = 0.22), and Work effectively (SC5) with Breathing difficulty (A2) (edge weight = 0.21). On the other hand, the three strongest negative correlations were Certain things (SC2) with Situations (A4) (edge weight = -0.35), Certain things (SC2) with Trembling (A3) (edge weight = -0.29), and Pleasure (SC4) with Dryness (A1) (edge weight = -0.28). Significantly, in the anxiety positive sample, Without thinking (SC7) showed a lower edge weight and no longer had the strongest negative edge, indicating a clear contrast to the network structure of the total sample. Unusually, the structure of the anxiety negative network was similar to that of the total sample network, but it had relatively small edge weights. The three strongest negative edges in the anxiety negative network were Without thinking (SC7) - Dryness (A1) (edge weight = -0.35), Without thinking (SC7) - Panic (A5) (edge weight = -0.34), and Without thinking (SC7) - Breathing difficulty (A2) (edge weight = -0.33) (see Supplementary Materials). We noticed that the network structures for sex, registered residence, and only child in a family were much like the total sample. However, unlike the urban sample, the rural sample only showed a positive correlation, which was between Iron self-control (SC3) and Trembling (A3) (edge weight = 0.03). The not one child sample had two positive correlations: Iron self-control (SC3) - Trembling (A3) (edge weight = 0.04) and Iron self-control (SC3) - Breathing difficulty (A2) (edge weight = 0.02). The female network was different from the total sample and the other subsamples, all the connections between self-control cluster 1 (SC1, SC3, SC5) and Cluster 2 (SC2, SC4, SC6, SC7) were positive (Fig.3, Supplementary Table S2-S9).

Network centrality

We calculated the z-score EI value for each entry in the total sample network to evaluate their relative importance (Fig.4). The two items that had the highest EI value were Physical exertion (A6 = 1.08) and Scared (A7 = 1.08). The EI value of Without thinking (SC7 = -1.41) was the lowest. This result indicates that Physical exertion (A6) and Scared (A7) were the core symptoms in the total sample network structure, while Without thinking (SC7) was statistically the least important node in the current network. In addition, the study conducted bootstrap difference testing on edge weights and node EIs (1000 bootstrap samples, α = 0.05). The test results showed that EI’s CS coefficient was 0.75, which indicated that the estimated value for node EI was quite stable (Fig.5).

We calculated the Z-score EI value for each subsample entry. Bootstrap difference test results confirmed that the EI value for a single subsample node was sufficiently stable (Supplementary Fig.1, Table S10). In anxiety positive network, Breathing difficulty (A2 = 1.00) and Physical exertion (A6 = 0.85) had the highest EI value, while Certain things (SC2 = -1.72) had the lowest. In anxiety negative network, Trembling (A3 = 1.16) and Physical exertion (A6 = 1.16) had the highest EI value, with Work effectively (SC5 = -1.28) being the least. The EI value of urban network was close to the total sample, with the highest value being for Physical exertion (A6 = 1.12) and Scared (A7 = 1.07), and the lowest value being for Without thinking (SC7 = -1.40). In rural network, Trembling (A3 = 1.14) and Scared (A7 = 1.09) had the highest EI value, while Without thinking (SC7 = -1.41) had the lowest EI value. The one child network had similar expected effects as the total sample network, with Physical exertion (A6 = 1.09) and Scared (A7 = 1.05) having the highest EI value and Without thinking (SC7 = -1.42) having the lowest EI value. In the not one child network, Trembling (A3 = 1.13) and Scared (A7 = 1.11) had the highest EI value, while Without thinking (SC7 = -1.39) had the lowest EI value. In male network, Trembling (A3 = 1.05) and Physical exertion (A6 = 1.03) had the highest EI value, while Without thinking (SC7 = -1.34) had the lowest EI value. The female network had similar EI values to the total sample network, with Physical exertion (A6 = 1.16) and Scared (A7 = 1.13) having the highest EI value and Without thinking (SC7 = -1.53) having the lowest EI value.

Restricted cubic spline model

This study employed a restricted cubic spline to visualize the linear relationship between self-control and anxiety (Fig.6). We observed that as the total score of self-control increases the risk of anxiety among university students diminished (P for overall < 0.001). Furthermore, there was a negative nonlinear trend between self-control and anxiety levels (P for nonlinear < 0.001). This finding strongly supported the primary hypothesis of the study: the greater the self-control of university students, the less likely they were to experience anxiety.

Based on the above cues, we conducted further research on each subsample (see Supplementary Fig.2). Firstly, in the anxiety positive sample, the estimation curve gradually decreased as the total self-control score increased. When the total self-control score was greater than “20”, the estimated curve of the anxiety positive sample reached a turning point, after which the likelihood of anxiety symptoms increased with the total score. Secondly, in the anxiety negative sample, the estimation curve rose gradually as the total self-control score increased. When the total self-control score went above “15”, the anxiety negative sample’s curve hit a turning point. From there, the higher the total score, the lower the chance of having anxiety symptoms was. Finally, for the urban, one child, and female samples, the curves did not show a specific turning point and the chance of anxiety went down as the total self-control score got higher. The estimated curves for the rural, not one child, and male samples were similar to the anxiety negative sample.

In this study, we aimed to explore the significant links between university students’ self-control abilities and their levels of anxiety, as well as to identify the core symptoms of anxiety among university students. By integrating network analysis with the restricted cubic spline, we were able to identify the interactions between these factors and accurately pinpoint the core symptoms within the network. This approach can provide valuable references for the intervention and prevention of anxiety symptoms in this demographic.

In the total sample network structure, all edges between self-control and anxiety were negatively correlated, and the reliability of this negative correlation was confirmed by restricted cubic splines. This result supported the initial research hypothesis that anxiety symptoms in university students were negatively correlated with self-control. Previous studies have also confirmed this negative correlation [44, 45], which suggests that enhancing the self-control ability of university students will be beneficial for intervening in the anxiety symptoms of this group. Undergraduate education offers courses on enhancing self-control to provide theoretical and practical guidance for university students to improve their behavioral and cognitive skills. Self-management (SM) training is one of the most effective strategies for improving self-control, and existing research has fully confirmed its effectiveness in improving lifestyle and social adaptability [46, 47]. Especially when used as a university course, it can effectively enhance the self-efficacy of university students, driving their personal and professional development [48]. For instance, courses can guide university students in goal setting and planning, emotional and stress management, and improving communication skills, among others. In addition to SM training, physical exercise can also strengthen the self-control ability of adults [49]. Domestic and international research has found that physical training is positively correlated with the life satisfaction of university students, and the self-control ability of university students is a mediating factor in this relationship [50,51,52]. This means that cultivating the exercise habits of university students can not only enhance their self-control ability but also help them self-regulate their emotions. Therefore, the combination of physical education and self-management training may be an effective way for universities to intervene in the anxiety levels of university students.

Centrality analysis is an effective method for revealing the core symptoms of anxiety. In the total sample network, Physical exertion (A6) and Scared (A7) exhibit significant EI, indicating that interventions targeting Physical exertion (A6) and Scared (A7) could have broad positive effects on other related factors within the network. Without thinking (SC7) was significantly lower than other nodes in the network, indicating that targeting Without thinking (SC7) may have little benefit for other variables considered in the network [37]. It was worth noting that compared to other self-control components, Without thinking (SC7) had higher strength at all edges of the network structure. On the one hand, this result indicates that Without thinking (SC7) has a lower ability to conceptualize anxiety than other components in this study, which may be related to the relatively small proportion of anxious individuals in the sample selected for the study. On the other hand, this may be due to some mediating effect between self-control and anxiety. For instance, anxiety serves as a predictor for blood pressure can induce blood pressure issues among university students through self-control [53]. It has also been found that self-control mediates the association between trait anxiety and procrastination [54]. Moreover, self-control also plays a mediating role in anxiety and mobile phone addiction [55, 56]. Therefore, improving the self-control level of university students can be an effective way to directly or indirectly reduce their negative behaviors such as procrastination, sleep, internet addiction, and declining grades. For individual university students, the mediating factors that affect their self-control and anxiety related relationships vary. Analyzing these mediating factors can help develop personalized anxiety intervention strategies for university students, thereby achieving the goal of targeted intervention for their anxiety symptoms.

In the anxiety positive sample, there was a strong positive correlation between the self-control component Iron self-control (SC3) and anxiety symptoms Breathing difficulty (A2) and Trembling (A3). This suggests that university students with greater self-control are more susceptible to feelings of anxiety and may experience more severe anxiety symptoms, such as a predisposition to shortness of breath and sensations of bodily tremors. This substantiated the second hypothesis of the study, which posited that university students with anxiety disorders exhibited both negative and positive associations between self-control and anxiety symptoms. According to related studies, trait self-control is a longitudinal predictor of anxiety, and this relationship may be mediated by affective regulation [9]. This reminds us of that students with strong self-control are less likely to experience anxiety due to their adeptness at managing emotions and dealing with challenges such as competition and stress. However, those who find it difficult to control their emotions are at a higher risk of experiencing intense anxiety symptoms. It is particularly important to identify individuals with high levels of self-control within the anxiety group, as they may experience more severe anxiety symptoms due to poor emotional regulation when facing difficulties. Considering the challenges in promptly evaluating the individual differences in anxiety symptoms, we suggest that universities and relevant organizations should foster the development and application of digital interventions. These include online therapies, guided therapy apps, and chatbots [57]. Such interventions can help mitigate the aversion that university students may have towards psychological monitoring and treatment, thereby enhancing their accessibility to psychological care. Additionally, they enable administrators, healthcare providers, and researchers to closely monitor psychological changes, allowing for the timely adjustment of therapeutic strategies.

In analyzing the impact of sociodemographic factors on the structure of networks, we found that self-control often exhibited heterogeneity. Particularly in the samples from rural and not one child, we observed some positive edges that did not appear in the total sample or other samples, which were associated with Iron self-control (SC3). Although these correlations are relatively weak (edge weight < 0.05), this subtle difference should not be overlooked. It not only confirms the existence of heterogeneity in self-control but also suggests that we should pay more attention to differences in scales. This subtle heterogeneity indicates that sociodemographic factors play an important role in shaping network structures, as predicted by our Hypothesis 3. Research had found that even after controlling for self-control and social control, gender, age, and ethnicity still maintained significant relationships with crime [58]. This further confirms the aforementioned viewpoint. The study of Kaygusuz, Canani et al. [59] investigated the relationship between university students’ self-control and their socio-demographic and academic characteristics. The results showed that negative life experiences and gender affected self-control. In contrast, other variables such as age, place of birth, socio-economic status (SES), family type, and longest residence did not affect self-control. In addition, a study of 542 Chinese undergraduate students and 446 American undergraduate students revealed cultural differences in self-control [60]. Chinese participants demonstrated lower attitudinal self-control but higher behavioral self-control than American participants. In conclusion, self-control is a complex multidimensional concept influenced by various factors, and its diversity and complexity should be taken into account when analyzing its heterogeneity.

The network analysis structure also reflected significant gender differences in our study. These differences were manifested in the network structure of the female sample, where a positive correlation was observed between self-control cluster 1 (including SC1, SC3, and SC5) and cluster 2 (including SC2, SC4, SC6, and SC7). Based on the scale items, we discovered that Cluster 1 comprised questions with negative sentiment framing, while Cluster 2 included questions with positive sentiment framing. In the network structure of the female sample, we observed a unique phenomenon not seen in other samples: the positive and negative sentiment framings of self-control exhibited a positive correlation. In several studies, women had high emotional intelligence and good behavioral cognitive skills compared to men [61, 62], but they tend to be at a disadvantage when it comes to mental health [63]. Notably, women were diagnosed with anxiety and related subtypes more frequently than men [64]. A growing body of evidence suggests that social support is an effective factor in alleviating anxiety [65, 66]. Moreover, social support from friends, teachers, and family can enhance the life satisfaction and sense of well-being among university students [67]. It is important to recognize that traditional interventions may not alleviate anxiety symptoms in women with strong self-control abilities. Therefore, psychological monitoring during university years must pay attention to this group and provide them with more social support. By focusing on social participation as an intervention measure, universities can create a favorable competitive and cooperative environment for them, guiding their recovery of mental health in practical activities and social interactions.

Author notes

1. Meilin Zhang and Jienite Pan contributed equally.

Authors and Affiliations

1. College of Humanities and Management, Guilin Medical University, Guilin, 541199, China

   Meilin Zhang,Jienite Pan,Wuxiang Shi&Yinghua Qin

2. Department of General Practice, The Affiliated Luohu Hospital of Shenzhen University Medical School, Shenzhen, Guangdong Province, 518000, China

   Botang Guo

3. Psychological Science and Health Management Center, Harbin Medical University, Harbin, 050017, China

   Botang Guo

Contributions

All authors contributed to this manuscript. BTG and YHQ were responsible for the overall design of the study. MLZ analyzed the results and drafted the manuscript. BTG made significant contributions to data collection, WXS and JNTP assisted with the literature review. MLZ and JNTP contributed to the interpretation of the results and the writing of the manuscript. YHQ and BTG revised the paper. All authors agreed to publish the current version of this manuscript.

Corresponding authors

Correspondence to Yinghua Qin or Botang Guo.

Ethics approval and consent to participate

All participants gave written informed consent. The studies involving human participants were reviewed and approved by the Ethics Committee of Harbin Medical University (HMUIRB20200002).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions