According to the definition proposed by Kaplan and Haenlein (2010), social media can refer to any social networking sites or various of Internet-based applications, which was considered to developed based on the Web 2.0 platforms and the ideology of “User Generated Content” (Kaplan and Haenlein, 2010; Tartari, 2015). A typical characteristic of social media is that it allows all users to participate in the creation of the presented content and communicate with each other among social media users. Now, social media has become a part of people’s lives in modern society, through which people can maintain interpersonal relationships, document those meaningful moments in life, gain knowledge and information from the external world, and share whatever you are willing to share with others (Boyd and Ellison, 2007). Besides, it is readily for social media users to perceive social support from others on the platform, which is considered to be a protective factor for our self-esteem and subjective well-being (Rozzell et al., 2014; Carr et al., 2016; Tao and Cheng, 2018).

Despite the fact that people benefit a lot from social media, researchers also reveal some detrimental influences on individuals’ mental health and social adaptation resulting from excessive social media (O'Keeffe et al., 2011; Amedie, 2015; Tartari, 2015). For example, previous research consistently shows that individuals who spend too much time on Facebook are inclined to exhibit classic symptoms of depression, called “Facebook depression” (Jelenchick et al., 2013; Steers et al., 2014; Alfasi, 2019). Additionally, due to the positive self-presentation tendency on social media, individuals with excessive social media use often suffer from upward social comparison and mistakenly perceive that they are inferior than others round them, which further harms their self-esteem (Jan et al., 2017). In some cases, social media provides an ideal place for some criminals to hide their identity and spread criminal ideas, such as cyber bullying, cyber terrorism, and drug dealing (Amedie, 2015).

The term social anxiety refers to a phenomenon existing in social situations that one person has a strong desire to make a good impression on others while the person is quite not sure whether he/she has the ability to this end (Yen et al., 2012; Morrison and Heimberg, 2013). The cognitive-behavioral model of social anxiety posits that individuals with social anxiety tend to exhibit three tendencies: (1) applying an excessively high standard for their performances in social situations, (2) presupposing that the encountered others will give negative evaluations for their social performances, and (3) being inclined to believe that others’ evaluations about themselves are true (Clark and Wells, 1995; Rapee and Heimberg, 1997). As a result, when interacting with others in social situations, socially anxious individuals often experience fear or anxiety that they will be negatively evaluated or judged by others (Clark and Wells, 1995; Alden and Taylor, 2004; Hofmann, 2007; O’Day and Heimberg, 2021). And for these individuals, this excessive fear for poor social performances will further reduce the possibility to develop new interpersonal-relationships and also have a detrimental influence on the existing interpersonal-relationships (O’Day and Heimberg, 2021). It should be pointed out, despite the fact that social anxiety is detrimental to developing normal social interaction, anxiety symptoms have been widely observed in nonclinical samples and most people actually have experienced more or less social anxiety in daily life (Purdon et al., 2001).

Prior literature has documented that there is a significant correlation between social media use and social anxiety (Davidson and Farquhar, 2014; Dobrean and Pasarelu, 2016; Jiang and Ngien, 2020; Sternberg et al., 2020; O’Day and Heimberg, 2021). For instance, in a questionnaire survey conducted in Singapore, researchers found that those participants with higher Instagram use intensity were inclined to report higher social anxiety, suggesting excessive social media use may elevate users’ social anxiety (Jiang and Ngien, 2020). Similarly, in another study by Davidson and Farquhar (2014), researchers examined the relationship between Facebook use intensity and social anxiety and revealed a significant correlation between them. Even for those individuals having suffered from social anxiety, using social media will further enhance their social anxiety although they initially intend to compensate for their “deficits” in social skills via the online interactions on social media (Carruthers et al., 2019). It should be pointed out, several existing studies examining the effect of social media use on social anxiety adopted a single questionnaire approach, so such studies in nature are a correlational design (Davidson and Farquhar, 2014; Qiu et al., 2017; Jiang and Ngien, 2020). And a single study conducted in the lab consisted of clinical samples (individuals suffering from social anxiety) but not community samples (Carruthers et al., 2019). As a consequence, we so far cannot draw a causal inference about whether social media use will increase social anxiety of users in daily life. To fill this gap, the current research sought to provide first evidence for the causality between social media use and social anxiety. Correspondingly, our first hypothesis was that,

According to the propositions of social comparison theory, we often intentionally or unconsciously compare ourselves with others around us in order to assess our self-worth or gain our self-enhancement (Festinger, 1954). When comparing ourselves with someone who is better than us on specific domains, we can call this comparison as upward social comparison; in contrast, when comparing ourselves with someone who is worse than us on specific domains, we can call this comparison as downward social comparison (Wills, 1981; Wood, 1989). Past research suggests that compared to downward social comparison, upward social comparison is more likely to be associated with a series of negative health outcomes, such as envy, lower well-being, depressive symptoms, and lower self-esteem (Feinstein et al., 2013; Charoensukmongkol, 2018; Tao and Cheng, 2018; Alfasi, 2019; Schmuck et al., 2019). In contrast, comparing oneself with others inferior than the self has been found to be helpful for elevating self-esteem, receiving mental gratifications, and maintaining good mood states (Gibbons and Gerrard, 1989; Lew et al., 2007; Huang and Zhou, 2018). Notably, although a large body of research demonstrates the negative effect of upward social comparison and the positive effect of downward social comparison, the opposite patterns are also possible in given circumstances. As an example, for an individual with a self-improvement motive, comparing the self with a better one will provide hope and inspiration for the individual (Wood, 1989).

On social media, users can autonomously determine which aspects of them will be presented on the platform, and if necessary, they also can employ the beauty filter to construct perfect self-images following their aesthetic preference (Arroyo and Brunner, 2016; Yao et al., 2020). As a result of this characteristic of social media, it is a common phenomenon that social media users tend to present positive self-presentations on various of social media platforms (Haferkamp and Krämer, 2011). In this situation, social media users often mistakenly perceive that the lives of others always seem to be exciting and colorful, but their own lives seem to be mediocre and boring, which will result in the generation of upward social comparison (Chou and Edge, 2012; Steers et al., 2014).

Upward social comparison, a phenomenon commonly observed on social media platforms, has been found to be a risk factor inducing social anxiety (for a review, see McCarthy and Morina, 2020). The Cognitive-behavioral model of social anxiety has defined the fear of evaluation as a core feature of the disorder, and upward social comparison seems to readily induce this fear, especially the fear of negative evaluations from others (Clark and Wells, 1995; Rapee and Heimberg, 1997; Weeks et al., 2009; McCarthy and Morina, 2020). For example, Antony et al. (2005) used a questionnaire survey method to examine the relationship between social comparison and social anxiety among individuals with high and low social anxiety, and found that for individuals in both groups, upward social comparison was significantly and positively correlated with social anxiety. In another lab experiment by Mitchell and Schmidt (2014), researchers asked participants to view a profile of a fellow student with excellent performance (the upward social comparison condition) or with normal performance, and then measured participants’ social anxiety via the Social Interaction Anxiety Scale (Mattick and Clarke, 1998). The results showed that male participants with social anxiety tended to report higher social anxiety after experiencing upward social comparison in the lab. Following the above reasoning, we proposed our second hypothesis that,

In addition to directly inducing social anxiety, upward social comparison may also lead to lower self-esteem of social media users, and lower self-esteem may further lead to higher social anxiety. Krause et al. (2021) have proposed that social comparison can produce the opposite influences on self-esteem depending on the comparison direction—upward social comparison will harm self-esteem and downward social comparison will benefit self-esteem. Consistent with this proposition, empirical research revealed that, due to the positive self-presentation on social media, social media users in general experienced upward social comparison rather than downward social comparison when they use social media (Jan et al., 2017). And upward social comparison has been widely demonstrated to be correlated with lower self-esteem (Mitchell and Schmidt, 2014; Alfasi, 2019; Midgley et al., 2021). As an example, using an experimental approach, researchers found that participants reported lower self-esteem when they were assigned to the upward comparison condition (reading an introduction about an excellent college student) than the downward comparison condition (reading an introduction about an average college student; Mitchell and Schmidt, 2014).

Sociometer theory proposes that self-esteem represents “an internal, subjective gauge of interpersonal acceptance and Rejection”—high self-esteem means an individual perceives good interpersonal relationships and the individual can receive social acceptance from others, while low self-esteem means an individual perceives poor relationships and the individual may suffer from social exclusion (Leary and Baumeister, 2000; Leary, 2005). So, according to this theory, due to the threat of social exclusion, low self-esteem should have a close connection with social anxiety. In line with the proposition of the sociometer theory, empirical research also suggests that individuals with lower self-esteem tend to report higher social anxiety (van Tuijl et al., 2014; Fatima et al., 2017). Recently, with 388 Singaporeans as participants, Jiang and Ngien (2020) used the questionnaire survey to clarify the mechanisms underlying the relationship between Instagram use intensity and social anxiety, and found that upward social comparison on social media had a significantly negative prediction on self-esteem and self-esteem further had a significantly negative prediction on social anxiety. On the basis of previous research, we hypothesized that,

Participants of Study 1 were 470 college students (182 males, 282 females, 6 participants did not provide gender information) from a university in Shandong Province of China-Mainland, whose average age was 20.99 years (SD = 1.33), ranging from 18.58 to 30.08 years. All participants took part in this survey on the voluntary basis and they signed the informed consent prior to the formal survey. Participants could receive 5 RMB as a return (approximately 0.8 USD) after completing the survey. Eight participants (3 males, 5 females) were regarded as invalid participants resulting from their excessive answer omissions, so there were 462 participants involved in final data analysis.

We carried out our questionnaire survey in a group of 40~60 participants. Before the formal survey, we told participants that we were conducting a survey about social media use and social anxiety. We also explicitly alleged that the collected data was used only for academic purpose, and if participants were willing to continue the survey, they needed to assign the informed consent. After that, we instructed participants to fill out several scales mentioned above and also report their demographic information. When they completed all measurements, they would receive their rewards (5 RMB). The whole survey lasted for approximate 10–15 min.

In Study 1, we used a questionnaire survey to examine the relationship between social media use and social anxiety, and the underlying mediating-mechanisms. Consistent with Hypothesis 1, we found that there was a significantly positive correlation between social media use and social anxiety, which implies that individuals with high use intensity of social media were more likely to display social anxiety than individuals with low use intensity of social media.

Besides the direct effect between social media use intensity and social anxiety, we also found that higher social media use intensity was significantly correlated with higher upward social comparison, showing the same results-pattern with previous research (Lee, 2014; Jan et al., 2017). The higher upward social comparison, on one hand, directly induced social media users’ social anxiety, resulting in the mediating role of upward social comparison between social media use and social anxiety (Hypothesis 2); on the other hand, the higher upward social comparison was further correlated with lower self-esteem and lower self-esteem finally induced social anxiety, thus supporting the chain-mediating role of upward social comparison and self-esteem between social media use and social anxiety (Hypothesis 3).

It should be pointed out, Study 1 employed the questionnaire survey to examine the relationship and the underlying mechanisms between social media use and social anxiety, which should belong to the correlational research-design. As we have discussed in the introduction section, the correlational research-design allowed us to test whether there was a significant correlation between social media use and social anxiety, but did not enable us to confirm the causality between them. So, Study 1 indeed provided preliminary supports for our hypotheses. Given that, we aimed to conduct a lab experiment to further demonstrate the causality between social media use and social anxiety.

We determined the sample size of Study 2 based on the calculation of G*power 3.1 (Faul et al., 2009), which suggested that the presupposed medium effect size d = 0.8 and the significance at 0.05 level required at least 88 participants in each condition to conduct an independent-samples t-test. Given that, we finally recruited 180 college students (79 males, 101 females, M_age = 19.74, SD = 0.93, ranging from 18.33 to 22.42) to participate in Study 2. As a return for their participation, they would receive 10 RMB (approximate 1.4 USD) when they complete all tasks. Due to excessive answer omission, 5 participants were regarded as invalid participants and excluded from the following data analysis. As a result, a total of 175 participants (79 males, 96 females) were included in final data analysis.

Study 2 was conducted in the lab. At the appointed time, participants arrived at the lab in a group of 6~8 people. Upon arrival, we introduced the academic purpose of the study for participants, and they needed to assign the informed consent prior to the formal task. After that, we sent the video links to participants and instructed them to view these short videos on their own cellphone. Then, participants successively reported their upward social comparison, perceived self-esteem, perceived anxiety, and demographic information. When they completed all tasks, we gave each participant 10 RMB as a return for their participation. We explained any questions raised by participants in a detailed way. The whole experiment lasted for about 25~30 min.

In Study 2, we conducted a lab experiment in which we assigned participants to the experimental and control conditions, and then compared participants’ social anxiety between the two conditions. In the experimental condition, participants were exposed to several wonderful and excited experiences shared by others, which actually also belonged to the content that they often view on social media platforms in daily life. In the control condition, participants were exposed to several short videos related to natural scenery. The results showed that participants viewing wonderful and excited experiences of others (the experimental condition) reported higher social anxiety than participants viewing natural scenery documentaries (the control condition), which suggested that social media use in daily life may increase users’ social anxiety. By demonstrating the causality between social media exposure and social anxiety, Study 2 thus provided compelling evidence for Hypothesis 1.

When social media exposure condition was coded as a dummy variable, we investigated the mediating roles of upward social comparison and self-esteem between social media exposure condition and social anxiety. The results revealed the single-mediating role of upward social comparison and the chain-mediating role of upward social comparison and self-esteem between social media exposure condition and social anxiety, thus providing collaborating evidence for our Hypothesis 2 and 3. Additionally, Study 2 also revealed an unexpected finding that self-esteem played a mediating role between social media exposure and social anxiety. We attempted to explain this unexpected finding in terms of the self-evaluation maintenance model (SEM) proposed by Tesser and Campbell (1980) and Tesser et al. (1988), which contends that an individual’s self-esteem will be threatened when confronting other in-group members who are superior than the individual on ability domains. For example, previous research has found that college students may perceive lower self-esteem when they were exposed to other excellent college students (Blanton et al., 2000). In the experimental condition of Study 2, one of the presented short-videos documented a college graduate sharing the experiences that how she successfully passed the post-graduate entrance examination and entranced Tsinghua University, a top university that many Chinese college students have been longing for. Obviously, the protagonist is an excellent college student and this may yield a threaten for participants’ self-esteem. As a consequence, participants in the experimental condition perceived lower self-esteem, and lower self-esteem further contributed to higher social anxiety. Considering this unexpected mediating role of self-esteem was not our major concern, we thus did not make further investigation on this issue. We will keep an eye on this issue in future research.

There was another finding that may need our explanation. That is, considering that participants in the control condition just viewed natural scenery documentaries, we asked participants in both conditions to report the tendency of upward social comparison when using social media rather than viewing short videos. Even so, participants in the experimental condition still report significantly stronger upward social comparison than participants in the control condition. This difference, in nature, may reflected a priming effect, a widespread phenomenon in daily life (for a detailed discussion, see Higgins, 2012). Past research has suggested that when individuals are exposed a specific kind of stimuli, the knowledge accessibility relevant to the stimuli will be elevated and such knowledge constructs correspondingly get priority in the following information processing (Higgins et al., 1982; Higgins and Brendl, 1995; Bargh et al., 1996). For instance, the research regarding stereotypes demonstrates that individuals may form impressions or make decisions following the guidance of stereotypes when stereotypic knowledge constructs are accessible, but if counter-stereotypic knowledge constructs are accessible, individuals may also rely on counter-stereotypic information to process information (Power et al., 1996; Finnegan et al., 2015). With respect to the current research, participants in the experimental condition were presented with the content that college students often browse on social media platforms, and such videos were more likely to activate knowledge constructs relevant to social media use than natural scenery documentaries. As a result, those experiences suffering from upward social comparison may also be activated (Steers et al., 2014; Liu et al., 2017). From this perspective, it seemed to be reasonable and acceptable for the finding that participants in the experimental condition reported more upward social comparisons than participants in the control condition.

As we have discussed in the introduction section, although past research has consistently demonstrated the significant correlation between social media use and social anxiety, the causal evidence that higher social media user causes higher social anxiety is still scarce (Davidson and Farquhar, 2014; Xie and Karan, 2019; Jiang and Ngien, 2020; Sternberg et al., 2020). The current research extended prior research in the field of social media use and social anxiety by providing the first evidence that excessive social media use could increase social anxiety. Moreover, in Study 2, participants were temporarily exposed to the content that they often view on social media platforms or the natural scenery documentaries that they are less likely to view on social media platforms. The results showed that, even this temporary exposure in the artificial laboratory-setting, still induced higher social anxiety of participants, which significantly strengthened our confidence to conclude that social media use will increase social anxiety. Interestingly, it has been widely documented that individuals with social anxiety are inclined to avoid the face-to-face interaction and construct their social networks on social media platforms, so that they can alleviate their social anxiety (Shepherd and Edelmann, 2005; Caplan, 2007; Carruthers et al., 2019). However, according to our results, social anxious individuals may experience less social anxiety in a specific online interaction, but their social anxiety may be further exacerbated in the long run. In other words, the online interaction indeed cannot replace the face-to-face interaction.

We noticed that when the indirect effects were taken into account, the direct effect of social media use on social anxiety was still significant across two studies, which was inconsistent with previous research by Jiang and Ngien (2020). We speculated that this inconsistency may be because the participants included in the research were all college students. Specifically, compared to other age groups, college students often have more free time and also spend more time on various of social media platforms (She et al., 2023). And correspondingly, prior research suggests that social anxiety actually is quite common among college students and the majority of them have experienced social anxiety in social situations from time to time (Purdon et al., 2001). For individuals with social anxiety, when encountering some cues relevant to social interactions, they may spontaneously imagine an anxiety-provoking social situation and look at themselves from an observer’s perspective, which will further increase their social anxiety (Hackmann et al., 2000; Hirsch et al., 2004; Morrison and Heimberg, 2013). In the current research, maybe our explicit declaration that participants would complete the survey relevant to social media use and social anxiety immediately induced college students’ social anxiety in both studies, resulting in relatively strong direct effect of social media use on social anxiety. Of course, we must realize that this explanation is tentative and premature, and needs further investigation in future work.

Consistent with previous research, the present research found that individuals with higher social media use intensity experienced more upward social comparisons, and upward social comparisons further increased social anxiety (Jiang and Ngien, 2020). In simple words, social media use would increase users’ social anxiety via the mediating role of upward social comparison. It should be pointed out, although prior research has revealed a series of negative effects on mental health, several studies have revealed some factors that may alleviate the negative effects of upward social comparison on mental health (Cramer et al., 2016; Kong et al., 2021; Latif et al., 2021). For example, prior research indicates that if individuals realize that they are making social comparison with superior others, they may hold positive evaluations for themselves and tend to believe they can do as well as the reference target (Collins, 2000; Suls et al., 2002). Following this logic, in daily life, we may can relieve social anxiety of social media users by explicitly reminding them that they are suffering from upward social comparison when using social media. Besides, we also can alleviate social anxiety of social media users by encouraging them to use social media with a self-improvement motive, because past research has suggested that, for individuals with a self-improvement motive, even exposed to superior others, they are still inclined to believe that they can improve the ability of themselves (Lockwood and Kunda, 1997; Cramer et al., 2016).

Beside the single-mediating role of upward social comparison, we also found that social media use could increase social anxiety through the chain-mediating role of upward social comparison and self-esteem. This mediation path inspired us that we may can alleviate social anxiety of social media users by elevating their self-esteem. For instance, past research found that positive feedback received from social media platforms enhanced users’ self-esteem while negative feedback received from social media platforms decreased users’ self-esteem (Valkenburg et al., 2006). And compared to casual acquaintances, individuals are more likely to receive feedback from close friends (Carr et al., 2016). Thus, for individuals with social anxiety, it is better for them to construct a “small but intimate” online social network than a “large but loose” online social network. Because the former enables users to receive more positive feedback from close friends, which will be helpful for alleviating social anxiety. Moreover, close friends are considered to be a primary source of social support and social support is an important protective-factor for self-esteem (Goodwin et al., 2004; Rozzell et al., 2014). On a broader level, considering that social media use may increase users’ social anxiety through decreasing self-esteem, those measures that are helpful for elevating self-esteem may deserve an attempt to examine whether they also can relieve social anxiety of social media users.

There were several limitations existing in the current research. Firstly, the indirect effect was relatively small in comparison with the direct effect across two studies. As we have explained in the previous section, this may be because some cues relevant to social interactions immediately induced social anxiety of college students (e.g., the instruction that we explicitly told participants that they would take part in a survey about social media use and social anxiety), resulting in the relatively small indirect effect across two studies. Despite this, considering that previous research has revealed similar results-pattern with the current research (Jiang and Ngien, 2020), we still have confidence for our findings and their implications for deepening our understanding about how social media use will increase social anxiety of social media users.

Secondly, in Study 2, we revealed an unexpected mediating-effect that social media use produced an effect on social anxiety via the mediating role of self-esteem. According to our explanation, this unexpected effect may result from presenting an introduction about an excellent undergraduate who had successfully entered Tsinghua University. That means, viewing different kinds of content on social media might have different psychological consequences for social media users. Additionally, the event that an individual successfully entered Tsinghua University may occasionally raise social status concerns of participants, which in turn increased social anxiety of participants. Given that, in future work, we can attempt to explore whether the effect of social media use on social anxiety will vary with the presented content on social media.

Thirdly, excessive social media use has been found to have close connections with a series of negative mental consequences. However, to date, there is little causal evidence that social media use will increase social anxiety. So, the primary goal of the present research was to provide causal evidence for whether and how social media use will increase social anxiety. Nevertheless, we must realize that prior research has shown that individuals with social anxiety are more likely to use social media to compensate for their social deficit (Lee-Won et al., 2015; Stănculescu and Griffiths, 2022), displaying the reversed pattern with the present research. For this phenomenon, we speculate that the relationship between social media use and social anxiety may not be a simple causality in a single direction. Rather, there may be a mutually reinforcing relationship between social media use and social anxiety—individuals with social anxiety tend to excessively use social media, which will further exacerbate their social anxiety. Considering this possibly mutually reinforcing relationship between social media use and social anxiety, it may be necessary to develop a unified theoretical framework to interpret this relationship.

Fourthly, in Study 2, the baseline social anxiety of participants was not taken into accounted, which may constitute a confounding factor for our findings. Because prior literature has indicated that individuals with high social anxiety are more vulnerable to risk factors in external environment than those with low social anxiety (e.g., Mitchell and Schmidt, 2014). On a broad level, to thoroughly answer the question whether social media use will elevate social anxiety, we need to take possible individual differences into account in future research (e.g., lower trait self-esteem individuals are more likely to suffer from social anxiety than higher trait self-esteem individuals; Fatima et al., 2017). Given such potential limitations, in future research, we need to replicate our findings in both community and clinical samples, and compare potential results-pattern differences between the two samples.

Finally, due to the regulation of Chinese government, many social media platforms popular in Western society fail to provide service in China-Mainland. Correspondingly, those social media platforms popular in China-Mainland actually have a relatively small user base. Given that, although the current research showed the similar results-pattern with previous research in Western society (Davidson and Farquhar, 2014; Xie and Karan, 2019; Sternberg et al., 2020), it may be still necessary for us to examine to what extent our findings can be generalized to Western society.