Introduction

The rapid development of 5G mobile internet and personal intelligent terminal equipment significantly supports the adoption of live streaming platforms. Live streaming has become an important component of China’s internet economy. According to CNNIC’s 50th “Statistical Report” (2022), the number of live streaming users in China reached 716 million by June 2022, accounting for 68.1% of the total number of internet users. Lu et al., (2018) demonstrate that live streaming in China has significant differences in content, style, and format when compared to live streaming in North America and Europe. In contrast, live streaming in China is more widely used and integrates entertainment short video, goods-selling, online social networking, and knowledge dissemination, which goes deeply into all aspects of social life and serves as the main channel for obtaining and transmitting information. As digital platforms for real-time recording and uploading audio and video, live streaming platforms connect and construct viewers’ virtual presence experiences with the distant world, based on the characteristics of hyper temporal and spatial attributes, para-authenticity, real-time interactivity, and connectivity (Cunningham et al., 2019; Lim et al., 2020), which significantly affects social media users’ interactions and willingness to convey information (Lin et al., 2014). Live streaming transmits images and sounds in real time through a variety of communication technologies, enabling the viewer to interact in real time on the platform. In the live streaming system, the live host and the viewer can obtain a sense of participation through real-time interaction, providing a unique immersive interactive experience that can trigger viewers’ behavioral intention. For example, the viewer can support their favorite host through monthly subscriptions or gift-giving (Wongkitrungrueng et al., 2020). This kind of physical and situational experience can help to bridge the psychological gap between the live host and viewer, so as to promote the establishment of a closer relationship between the viewer, the live host, and the platform (Liu et al., 2020), and enhance viewers’ social support willingness. Therefore, as a new media for real-time broadcast and interaction, the impact of the unique immersive reality and real-time interactive social experience of webcasts on viewer behavior requires further study.

A wealth of existing work has explored the characteristics of live streaming from the perspective of regional, cultural, professional, and gender performance aspects, among others (Wohn, Freeman (2020); Hsu et al., 2020). Studies also address live viewers, largely discussing the factors influencing viewer participation, based on technology adoption, user attitudes (Xu and Ye, 2020), and user motivations (Hilvert-Bruce et al., 2018; Chen and Lin, 2018). However, current studies ignore the new phenomenon of live streaming, as characterized by “immersive” experiences, and the real psychological state of the viewer in online interaction (Wongkitrungrueng et al., 2020). In fact, the interaction between the live host and viewer is a critical element of the live platform. In this process, the viewers are regarded as “real people” who can perceive the existence of others and thus experience individual psychological feelings, such as intimacy and psychological participation (Short et al., 1976), resulting in pseudo intimacy (Horton and Whol, 1956), which affects the viewers’ cognition and behavior, thus increasing the willingness of social support for the live platform (Hassanein and Head, 2007). Therefore, network social presence is undoubtedly an appropriate perspective from which to study the viewer’s social support willingness.

We have discussed that, on live platforms, the viewer’s willingness to support the live host is affected by network social presence. Furthermore, in the live broadcasting field, we need to reveal the practices of and mechanisms behind all parties’ actions, which work together to build emotional connections. Parasocial interaction is influenced by interaction experience and network social presence; additionally, the viewer’s network social presence is an important prerequisite for parasocial interaction (Xiong, 2016). The sense of belonging, immersion, and other aspects of network social presence generated by online interaction when watching live streaming reflects whether the viewer can have a sense of intimacy or direct feeling in interpersonal interaction. Therefore, network social presence, as an individual’s intention to maintain relationships, can cultivate a large number of positive and loyal users, and serves as an important factor for the construction of parasocial interaction. Above all, this study explores the impact of viewers’ network social presence on parasocial interaction and social support willingness based on the live broadcast environment in China. The structure of the paper is as follows: the following section presents a literature review and our research hypotheses. In “Research Design and Methods”, we propose a research model and further detail the research variables. In the “Data Analysis and Results” section, we report the sample and verify the research hypothesis. The “Discussion and Conclusion” section explains the contribution, inspiration, and limitations of this work.

Literature review and research hypothesis

Social support willingness of live viewer

The existing literature has defined and presented social support in numerous ways and from different angles. Early studies interpreted social support from a functional perspective and claimed that social support is related to material, psychological, and spiritual support (Hoffman et al., 1988), which can convey care and love to the recipient (Shumaker and Brownell, 1984), and make the recipient realize that they are part of an interpersonal network. Broadly, social support includes tangible support and intangible support. Tangible support, also known as physical support, refers to a type of resource to enhance self-esteem and provide ways to meet material needs, such as instrumental assistance, goods, and property. Intangible support refers to emotional care and the belief that support is available (Barrera, 1986). Furthermore, social support can be divided into offline social support and online social support. Online social support mainly focuses on the potential and willingness to obtain information or emotional support through interpersonal relationships (Williams et al., 2006), which is also known as social capital (Ellison et al., 2014). This study explores the social support willingness provided by live viewers to live hosts in online communities. Therefore, the social support willingness in this study refers to willingness to provide rather than accept behavior, that is, one’s willingness to provide support at the information, assistance, and emotional levels (Introne et al., 2016; Barak et al., 2008).

Existing studies mostly focus on the willingness to provide informational support (Introne et al., 2016), emotional support (Barak et al., 2008), and tangible social support (Lu et al., 2018). During live streaming, the viewer can interact with the live host or other viewers through messages, or respond to the host’s questions and requests (Haimson and Tang, 2017).In addition to watching, the viewer on the live streaming platform can also express their support and appreciation for the live host through likes, comments, and gifts (Haimson and Tang, 2017; Yu et al., 2018), and also provide immediate help at the live host’s request. Based on the above research, the viewer’s willingness to provide social support is reflected in three aspects: instrumental, emotional and economic support willingness (Wohn et al., 2018). Instrumental support willingness refers to the willingness of the viewer to provide direct assistance or practical action to the live host and help others by solving problems; emotional support willingness refers to the viewer’s emotional willingness to comfort, encourage, or care for the live host; economic support willingness refers to the willingness of the viewer to provide rewards and other financial support for the live host. Therefore, this study will conduct a more detailed investigation of the viewer’s willingness to provide social support from these three aspects: instrumental, emotional and economic social support willingness.

Network social presence and social support willingness

Short et al., 1976 first proposed the term “social presence” and defined it as the saliency of objects in media communication and the subsequent saliency of interpersonal relationships. However, there are now different perspectives and dimensions for the definition of social presence. Initially, scholars explored social presence within the characteristics of media and revealed the communication effect of different media through comparing the differences between remote communication media and face-to-face communication (Short et al., 1976). Some scholars began with a more psychological perspective and claimed that users’ perception of media is more critical than the attributes of the media itself; these writers further defined social presence as the feeling of being with others in the media environment, including the degree of trust in the process of interaction (Yeboah & Afrifa-Yamoah, 2023).

The development of networks and the ontological subversion of virtual reality has promoted more in-depth and specific discussions on the impact of social networks on individuals (Zhou et al., 2019). These studies focus on the extent to which individuals can perceive the existence of others in the process of using social networks; the individuals’ psychological feelings, such as intimacy; and the individuals’ psychological involvement, forming the so-called “network social presence”. In other words, network social presence is a sense of authenticity that individuals achieve through the social network, which makes individuals feel immersed in the the digital setting (Pettey et al., 2010) and even enhances individuals’ willingness to take action on social network (Cheung et al., 2015). Some scholars have applied network social presence to online interactions and marketing research. E-commerce studies have found that online shopping behaviors within social networks are highly similar to those within real-world settings. Some shopping websites and brands stimulate consumer behavior by creating a network social presence and maintaining relationships with consumers (Algharabat, 2018). Jiang et al. (2022) affirm that social presence affects the continued use and purchase intentions of Chinese consumers. Therefore, network social presence is an important factor driving individual consumption behavior intention.

As previously acknowledged, network social presence triggers a higher willingness to consume and promote the provision of more social support. Previous studies have shown that, in virtual networks, network social presence can increase the stability and satisfaction of the relationship between the two parties, and then increase the willingness of users to use, consume, and recommend products and services in the future (Choi et al., 2016; Chen et al., 2023), and improve the potential willingness of social support. Therefore, if the viewer is closely related to the platform and has a high sense of network social presence, the viewer is more willing to provide more social support to the platform and hosts. Based on the above analysis, network social presence plays an important role in the willingness of social support. Consequently, the hypotheses are developed as follows.

H1. Network social presence has a positive impact on the viewer’s willingness to support the host (a: emotional support willingness; b: instrumental support willingness; c: economic support willingness).

Mediation: parasocial interaction

Parasocial interaction was first proposed by Horton and Whol, 1956 who defined parasocial interaction as a “simulacrum of conversational give and take.” Horton and Strauss (1957) further indicated that parasocial interaction is a solely one-sided experience of the audience; most examples of this type of experience are based on the audience’s own illusion. Rubin and McHugh (1987) echoed this finding, describing parasocial interaction as a one-way interpersonal relationship between media performers and their TV audiences. Currently, parasocial interaction has been introduced into live streaming contexts, emphasizing the “illusion” and unilateral intimacy between the viewer and live host (Chen et al., 2021; Sheng et al., 2022). Furthermore, parasocial interactions can influence viewers’ emotional responses, attitudes, and behaviors (Chang and Kim, 2022; Sheng et al., 2022), the center of attention in this study.

The development of the internet and social network has promoted increasing numbers of scholars to explore the impact of new media use on individual users from the perspective of network social presence (Gao et al., 2017). In addition, investigations of network social presence highlight that the virtual space built through technology, similarly to real communication, can make media users perceive strong sociality, authenticity, and intimacy, and produce a strong sense of belonging, reflecting the degree to which individuals use media to build interpersonal relationships. Therefore, network social presence, as an important social psychological factor in the use of individual new media, has a far-reaching impact on interpersonal communication, which deserves more attention. The network social presence can enhance the immersion and authenticity of group communication, improve the interactive experience between live hosts and strengthen the network density, which can build a close connection between members, maintain the rapid flow of information and resources, and then generate a sense of identity with the group. In the live streaming, the viewer and the host in the live room perceive each other’s existence, cause emotional reactions, and gradually build a parasocial relationship by continuously participating in online discussions. Therefore, there is a positive relationship between network social presence and parasocial interaction. In line with the above, the following hypothesis was created.

H2. Network social presence has a positive effect on the parasocial interaction between the viewer and live host.

Cohen (2004) believes that parasocial interaction is most suitable for analyzing media figures who directly talk to the audience, such as newscasters and hosts. Rubin and Step (2000) found the parasocial interaction of radio hosts leads to changes in audience attitudes and behaviors. In recent years, the vigorous development of social networks, especially the widespread use of live platforms, has narrowed the distance between media figures and audiences, prompting new research on parasocial interaction. Lee and Watkins (2016) highlights the potential of social networks in establishing two-way communication and balancing the relationship between media users and media properties. Stever and Lawson (2013) claim that YouTube, TikTok, and other social networking sites allow the audience to approach the personal life of the media personality within the scope of the media personality, so as to obtain more social support. Furthermore, social media has expanded the phenomenon of parasocial interaction from solely concerning the world of TV characters to becoming a real tool for marketing brands to consumers (Lee and Watkins, 2016). For example, Hsu et al. (2020) found that vloggers can deepen viewers’ identity and sense of belonging and cultivate viewers’ fluid experience by establishing parasocial interaction, thereby urging viewers to purchase and enabling addiction. Especially in the live streaming environment, parasocial interaction affects the social interaction between the live host and the viewer (Hu et al., 2017; Lim et al., 2020), and can promote the impulsive consumption of the audience (Xiang et al., 2016).

In view of the above discussion, we believe that parasocial interaction has a mediating effect between network social presence and social support willingness. The existing studies provide a logical basis for investigating this mechanism. It is generally believed that the higher the viewer’s perception of network social presence, the higher the frequency of interaction on the network, so the viewer is more likely to feel that they are in a “real” network society, which is also known as network society presence. As for the impact of parasocial interaction on network social presence and social support willingness, studies have found that the degree of interactivity and vividness of online advertising are regulated by the audience’s social presence, and influences their attitudes and behavioral intentions (Lu et al., 2016). In addition, according to the existing research, network social presence can influence users’ satisfaction and sense of belonging, increase the possibility of contact, and strengthen the parasocial interaction between the audience and the live host, so as to trigger more social support (Lin et al., 2014). Therefore, the interpersonal interaction in the live streaming environment should be included in the influence of network social presence. According to the different degrees of network social presence, viewers with high network social presence are more suitable to perform tasks related to interpersonal interaction. Above all, network social presence may not only directly affect the social support willingness of viewers but also indirectly affect the social support willingness by enhancing the parasocial interaction. Therefore, we propose the following hypothesis:

H3. Parasocial interaction plays a mediating role between network social presence and social support willingness (a: emotional support willingness; b: instrumental support willingness; c: economic support willingness).

Moderation: emotional response

There are different opinions regarding whether different emotional experiences produce different physiological reactions. Every emotion is multi-dimensional; Mehrabian (1995) believed that an emotional response has three aspects: pleasure, arousal, and dominance, namely, the PAD emotional-state model. Pleasure refers to the positive or negative performance of emotions such as happy, satisfied, and satisfied; arousal refers to the level of individual physiological activation and alertness, on a scale of drowsiness to excitement; dominance is the state of individual control over situations or others. This model is widely used in environmental psychology; although it is intended to represent the dimensions of emotional response rather than a complete typology of emotional responses (Eroglu et al., 2003), its simple structure and widespread use make it an appropriate choice in this context. Russell (1979) believed that pleasure and arousal adequately capture the range of appropriate emotional responses, and Eroglu et al., (2003) demonstrated that, when studying emotions in the network, pleasure and arousal are commonly used to present individual emotional responses. Therefore, when exploring the emotional response of Chinese live cast viewers, this study only draws on the pleasure and arousal emotions in the PAD emotional state model, and the dominance dimension is not included. As the core construct of emotional response, the degree of pleasure and arousal is generated on the basis of cognition, which in turn affects cognition. They are the internal factors that regulate and control cognition, so as to achieve different psychological dynamic response (Pan and Huang, 2017). Multiple studies have explored how emotional constructs interact with and influence user attitudes and behaviors, confirming that individual behavior is regulated by positive emotions (Gavriel-Fried and Ronen, 2016).

In the context of webcasts, the viewer’s social and psychological state is an important determinant of how viewers choose a live host to meet their needs; that is, the viewer can be aware of their needs, consider various channels and content, evaluate the choice of functionality, and choose the media that they believe can provide the satisfaction they seek. In this framework, parasocial interaction is considered to meet the emotional needs of the viewer and can reduce anxiety (Suggs & Guthrie, 2017). If the live host can provide positive emotions for the viewer and activate individual energy, viewers will have high satisfaction with the live host, which is the main reason for viewers to form parasocial interaction and provide social support. Specifically, a parasocial interaction relationship is formed between the live viewer and the live host, which conveys valence and arousal to the viewer, thus making the viewer more inclined to provide social support. At the same time, the financial media environment enabled by new technology promotes highly autonomous participation mechanisms, and the emotional perception of the webcast platform also promotes the continuous use and recommendation willingness of viewers (Han et al., 2015). Therefore, under the influence of different degrees of valence and arousal, the influence of the parasocial interaction perception on viewer’s social support willingness is also accordingly different. Specifically, in the case of high positive emotional response, the impact of parasocial interaction on social support intention increases. However, in the situation of low positive emotional response, the incremental impact of parasocial interaction on social support decreases. Based on the above analysis, we hypothesize the following:

H4. Emotional response amplifies parasocial interaction’s effect on social support willingness (a: emotional support willingness; b: instrumental support willingness; c: economic support willingness).

Due to the experiential communicability and flexibility (Dale and Pymm, 2009) of live streaming, viewers feel a sense of belonging and pleasure when watching the live content. In this process, the relationship between viewer and the media has become closer and has broken through the constraints of time and space, which can guide the viewer’s online social presence and trigger an obvious positive emotional response. Furthermore, emotion, as one of the factors affecting user behavior (Gavriel-Fried and Ronen, 2016), is guided by a positive relationship between viewer and host that urges both parties to work together to better meet each other’s needs. Previous studies have shown that emotional response can not only affect audience satisfaction but also adjust audience’s willingness to support (Cheikh-Ammar and Barki, 2016). As a high level of network social presence can arouse the viewer’s positive emotional response, the viewer may therefore maintain a long-term good relationship with the host and provide social support. Hence, we hypothesize the following:

H5. Emotional response has a moderation role in the process of the effect of network social presence on social support willingness (a: emotional support willingness; b: instrumental support willingness; c: economic support willingness).

According to the above theoretical basis and research assumptions, this study proposes the following research model, as shown in Fig. 1.

Fig. 1: Research model.
figure 1

The impact of network social presence on live streaming viewers’ social support willingness: a moderated mediation model.

Full size image

Methodology

Sample and data collection

This study takes users who are over 18 years old and have watched the live as the research sample. The data for the study was collected through a snowball sampling procedure. Because of the unavailability of a valid sample frame and the difficulty of conducting random sampling for all live streaming viewers, this study used a non-probability sampling approach. The snowballing sampling is simple, inexpensive, and usable, and it is also helpful in determining the relationships between various events and situations (Sahu et al., 2021). Further, due to the snowball effect of participant referrals, a higher number of responses were obtained. Previous studies confirm that the snowballing sampling method is effective and appropriate for multivariate data processing and estimating the results (Almaqtari et al., 2023; Sahu et al., 2021; Chan, 2020; Noy, 2008; Wright and Stein, 2004).

We designed the questionnaire on QuestionnaireStar (https://www.wjx.cn/; a professional data collection website in China). Respondents can access our questionnaire homepage through an online link. The primary researcher used personal communication with “seeds” to assist the data collection process. The questionnaire survey started on June 9, 2022 and ended on June 24, 2022. They sent the questionnaire to respondents via social medias and asked respondents to send it to another potential participant after completing the survey. Respondents’ participation was completely consensual, anonymous, and voluntary. Moreover, the questionnaire does not cover the highly sensitive personal identity information such as the name, home address, telephone number, ID number of the respondents, ensuring the confidentiality and anonymity. The data obtained in this survey is only for academic research purposes. In addition, all procedures performed in studies involving human participants followed the ethical standards of the institutional and/or national research committee and with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.

To obtain valid samples, we created two screening questions: “Are you over 18 years old” and “Have you watched live streaming before?” If one of the respondent’s answers is “no,” the participant is directed to the end of the survey. After excluding the samples who are under the age of 18, had not watched the live streaming, and expressed an abnormal response time, 515 valid samples remained, and the sample pass rate was roughly 83.5%. Hair et al., (2006) mentioned that the factor analysis requires a minimum sample size of at least five times the number of measurement items in the study. This present study has 25 measurement items adapted from previous literature, thus, the appropriate number of sample size would be at least 125 respondents. Thus, the current sample size of 515 participants was suitable for conducting the research. Therefore, 515 surveys were considered the final sample for the present study. As the Kaiser-Meyer-Olkin measure of sampling adequacy value is 0.954, it is greater than 0.7, this sample is considered statistically adequate for estimating the results. Furthermore, this test shows high significance at the 1% level (p-value = 0.000, <0.01), indicating the suitability and adequacy of the sample.

Demographic information of the sample

To increase the generalizability of the findings, respondents with diverse backgrounds (age, gender, education, residence, etc.) were selected (as per Mladenović et al., 2020). As shown in Table 1, 227 respondents (44.08%) were male. Most respondents were between 18 and 39 years old (78.84%). In terms of education level, the number of respondents with a bachelor’s degree was the largest, 256 (49.71%); followed by below a bachelor’s degree, 186 (36.12%); and then a master’s degree or above, 73 (14.17%). Most of the respondents watched live streaming platforms for an average of 2.6 days a week: 196 samples (38.06%) watched live streaming platforms for less than one day, 199 samples (38.64%) for 1–3 days, and 120 samples (23.3%) for more than three days. 390 samples (75.7%) spent an average time per viewing of less than one hour, 94 samples (18.25%) less than one hour, and 31 samples (6.02%) more than two hours. According to the 2021 Research Report on the Development of China’s Online Live Broadcasting Industry, in 2021, 74.4% of China’s online live broadcasting users were 39 years old or younger: 47.1% of users were male, 52.9% were female users, 78.1% watched each live broadcast for less than one hour, and 17.2% watched for one to two hours (iiMediaResearch (2022)). Therefore, the basic characteristics of the sample in this study are broadly consistent with the current composition of China’s network live streaming users, indicating that the sample is representative.

Table 1 Characteristics of respondents (n = 515).
Full size table

Variable measurement

To increase the validity and reliability of the results, each construct in the model has multiple items adapted from previous studies with minor changes to fit the research context. All the questionnaire information was translated from English to Chinese with manual testing to optimize wording and grammar to ensure the linguistic accuracy and comprehensibility of the questionnaire. This study used a seven-point Likert scale to measure survey items, where one indicates strongly disagree/inconsistent and seven indicates strongly agree/consistent.

Network social presence

The measurements of network social presence were adopted from studies by Hassanein and Head (2007), Lu et al., (2016), and Gao et al., (2017). Some scale items were deleted per the CFA, leaving eight items. Respondents were asked to rate how they felt about the following statements; a seven-point Likert-type scale (one denoting “strongly disagree” and seven denoting “strongly agree”) was utilized for the following eight statements: (1) “in the live studio, I will pay close attention to the existence of others”; (2) “I felt someone approaching me in the live studio”; (3) “in the live studio, I have a sense of reality of face-to-face communication with others”; (4) “in the live studio, I have a feeling of social interaction”; (5) “in the live studio, I have a warm feeling”; (6) “in the live studio, I feel close to others”; (7) “in the live studio, the knowledge shared by the live host can benefit me”; (8) “in the live studio, I have a high degree of recognition for the behavior and view of the live host.” An overall network social presence composite measure was created by averaging the seven items together, Cronbach’s α = 0.888, M = 3.434, SD = 1.035.

Parasocial interaction

Because the initial parasocial interaction scale was created considering television news announcers (Rubin et al., 1985), some measurement items are not suitable for online live streaming (for example, “I miss seeing my favorite newscaster when he or she is on vacation”) or present a specific program format or content (for example, “When the newscasters joke around with one another it makes the news easier to watch”); these items were excluded. To measure the quasi-social interaction of research objects, previous research adapted the initial 20 measurement items from Rubin, Perse, and Powell, (1985) and reduced them to several measurement items, as found in studies by Choi et al., (2019), Kim and Song, (2016). Reducing the number of measurement items is suggested for addressing response behavior and data quality problems (Cheah et al., 2018; Drolet and Morrison, 2001), especially for participants from the general public (Messer et al., 2012). This study selected five current measurement items based on the characteristics of online live streaming, and these five measurement items showed strong reliability in this study (Cronbach’s α = 0.913), indicating that the internal consistency of these five measurement items is highly suitable for measuring quasi-social interaction. The five items are as follows: (1) “I look forward to watching the live broadcast on her/his live channel”; (2) “Watching the live stream makes me feel that the live host is accompanying me”; (3) “I think the live host is like my friend”; (4) “I will pay attention to the news of my favorite live host”; (5) “When I watch the live stream, I feel like I am a member of their live team.” A seven-point Likert-type scale (1 denoting “strongly disagree” and 7 denoting “strongly agree”) was used for all items. To create a composite measure of parasocial interaction, the five items were averaged together, M = 3.529, SD = 1.224.

Emotional response

The measurements of emotional response were derived from studies by Mehrabian (1995) and Jin et al., (2020). The PAD emotional-state model is widely applied in environmental psychology, although it is intended to represent the dimensions of emotional response rather than a complete typology of emotional responses (Eroglu et al., 2003). However, its simple structure and widespread use make it a suitable choice in this context. Initially, we designed three items to measure pleasure and arousal, but some scale items were deleted per the confirmatory factor analysis (for example, “Live streaming content makes me interested”, “Watching live streaming makes me happy”, and “Watching live streaming makes me feel stimulated”). A total of three measurement items remained, and these three measurement items showed strong reliability in this study (Cronbach’s α = 0.813), indicating that the internal consistency of these three measurement items is highly suitable for measuring emotional response. The five items are as follows: (1) “I enjoy myself in the live studio”; (2) “The live streaming content makes me feel novel and fresh”; (3) “Watching the live stream can make my life and work full of power.” A seven-point Likert-type scale (one denoting “strongly inconsistent” and seven denoting “strongly consistent”) was used for all items. In this study, three items were summed and averaged to establish a comprehensive index of emotional response, M = 3.797, SD = 1 .104.

Social support willingness

The measurements of emotional support willingness, instrumental support willingness, and economic support willingness were adopted from Wohn et al., 2018 study. A seven-point Likert-type scale (one denoting “strongly disagree” and seven denoting “strongly agree”) was used for all items. Among them, emotional support willingness includes three items: (1) “I am willing to send some encouraging words on the bullet screen to show my support for the live host”; (2) “I am willing to try to interact with the live host to make them feel concerned”; (3) “I am willing to express my support for the live host in some way.” This study sums the three items and averages them to construct the index of emotional support willingness (Cronbach’s α = 0.898, M = 3.790, SD = 1.360). Instrumental support willingness includes three items: (1) “If the live host really needs it sometimes, I am willing to try to help him/her”; (2) “If the live host needs to complete a time-limited task, I am willing to help him/her”; (3) “If there is a problem in the live studio, I am willing to help the live host solve it.” This study sums the three items and averages them to construct an indicator of instrumental support willingness (Cronbach’s α = 0.917, M = 3.548, SD = 1.328). Economic support willingness includes three items: (1) “I am willing to reward the live host by giving virtual currency and help him/her make a living”; (2) “I would like to reward the live host to express my gratitude by giving virtual currency”; (3) “I am willing to reward the live host and support his/her efforts by giving virtual currency.” This study sums the three items and averages them to construct the index of economic support willingness (Cronbach’s α = 0.931, M = 2.862, SD = 1.423).

Results

Analysis of variable correlation

Table 2 shows the correlation between the six variables. There is a positive correlation between network social presence and emotional support willingness (r = 0.597, p < 0.01), instrumental support willingness (r = 0.602, p < 0.01), economic support willingness (r = 0.518, p < 0.01), and parasocial interaction (r = 0.703, p < 0.01). Parasocial interaction is also positively correlated with emotional support willingness (r = 0.703, p < 0.01), instrumental support willingness (r = 0.700, p < 0.01), and economic support willingness (r = 0.577, p < 0.01). These findings provide preliminary data support for the subsequent hypothesis verification.

Table 2 Correlation coefficient matrix and square root of each variable AVE.
Full size table

Reliability, validity, and common method bias

In this study, the software Mplus 8.3 was used for confirmatory factor analysis to test the reliability and validity. According to Fornell, Larcker, (1981) and Hair et al., (2019), the AVE value of all variables being greater than 0.5 (Table 3) indicates that the six variables have good aggregate validity. The factor load of the item corresponding to each variable is greater than 0.5, and the square root of AVE is greater than the correlation coefficient between all variables (Table 2), indicating that the discriminant validity of each variable is good. In addition, the combined reliability (CR) of all variables and Cronbach’s α were greater than 0.7, indicating that the internal consistency of the questionnaire is high and each variable has good reliability. Therefore, the variables measured in this study are valid and credible.

Table 3 Reliability and validity test.
Full size table

This study used a single data source for six variables, so it had the possibility of common method bias (CMB). To ensure the accuracy of the research conclusion, we have taken several steps to solve the potential CMB. First, we used anonymous questionnaires to improve the objectivity and freedom of respondents to answer questions. Second, we used different types of response scales (highly agree–highly disagree; very consistent–very inconsistent). Third, we used the test of multicollinearity through the variance inflation factor (VIF) to check if CMB may be a threat (Kock, 2015). The VIFs were lower than 3.3, indicating that common method variance in the data was not detected as CBM (Kock, 2015).

Hypothesis testing

Direct effect test

H1 and H2 were verified by hierarchical regression analysis in SPSS 26.0 software because linear models usually require normal distribution of dependent variables. However, the distribution of parasocial interaction (Kolmogorov-Smirnov z = 0.180, p < 0.001), emotional support willingness (Kolmogorov-Smirnov z = 0.185, p < 0.001), instrumental support willingness (Kolmogorov-Smirnov z = 0.198, p < 0.001), and economic support willingness (Kolmogorov-Smirnov z = 0.169, p < 0.001) significantly deviated from the normal distribution. Therefore, this study adopts the bootstrapping method to perform regression analysis of 1000 samples under a 95% confidence interval. Bootstrapping is a nonparametric statistical method. Its basic principle is that when the assumption of normal distribution is not tenable, a certain number of samples are resampled within the scope of the original sample data, and the parameters obtained by averaging each sampling are taken as the final estimation results.

The results of regression analysis (Table 4) show that network social presence has a significant effect on emotional support willingness (β = 0.787, p < 0.001), instrumental support willingness (β = 0.769, p < 0.001), and economic support willingness (β = 0.708, p < 0.001). That is to say, the stronger the network social presence, the more likely the viewer will have social support for the live host. Therefore, H1a, H1b and H1c are supported.

Table 4 Hierarchical regression analysis.
Full size table

The results of the regression analysis also show that the network social presence can positively influence the parasocial interaction (β = 0.831, p < 0.001). This influence demonstrates that the enhancement of network social presence can strength parasocial interaction between the live viewer and the live host. Therefore, H2 is supported.

Mediating effect of parasocial interaction

SPSS PROCESS macro-Model 4 was applied to analyze the mediating effect of parasocial interaction on the relationship between network social presence and emotional support willingness, instrumental support willingness, and economic support willingness. The mediating effect model results (Table 5) show that the parasocial interaction of live viewers regarding the network social presence and emotional support willingness (β = 0.510, 95% CI [0.405, 0.612]), instrumental support willingness (β = 0.487, 95% CI [0.388, 0.594]), and economic support willingness (β = 0.408, 95% CI [0.288, 0.531]) played a significant positive mediating role. Therefore, H3a, H3b, H3c are supported.

Table 5 Results of mediating effect.
Full size table

Moderating effect of emotional response

Through the model 15 in PROCESS V4.0 of SPSS 26.0, this study continues to test whether the moderating effect of emotional response is tenable under the mediation effect of parasocial interaction. We performed a bootstrap test by taking the score of the average emotional response with plus or minus one unit of standard deviation. The results show that (Table 6) the parasocial interaction and emotional response has a significant predictive effect on the instrumental support willingness (β = 0.096, t = 2.142, p < 0.05). Therefore, emotional response plays a significant positive moderating role in “network social presence-parasocial interaction-instrumental support willingness”.

Table 6 Results of moderated mediation effect.
Full size table

Table 7 shows that the mediating effect of parasocial interaction is significant at three levels of emotional response. Specifically, when the level of emotional response is low (less than 1 standard deviation), the mediating effect of parasocial interaction is supported (β = 0.346, 95% CI [0.195, 0.503], excluding 0). The mediating effect of parasocial interaction is supported when the emotional response is averaged (β = 0.434, 95% CI [0.315, 0.553], excluding 0) and higher than 1 standard deviation (β = 0.522, 95% CI [0.376, 0.653], excluding 0). Moreover, the mediating effect of parasocial interaction increases significantly with the improvement of emotional response. That is to say, the mediating effect of parasocial interaction is the strongest when emotional response is high (β = 0.522). In summary, the network social presence indirectly affects the instrumental support willingness of the live viewers by influencing the parasocial interaction. The strength of this indirect effect depends on the level of emotional response. The higher the emotional response, the stronger the mediation effect of parasocial interaction. However, other moderating effects of emotional response were not significant. Therefore, H5a is verified and H5b and H5c are not supported.

Table 7 Mediation effect of parasocial interaction at different levels of emotional response.
Full size table

Discussion

This paper examines the impact of network social presence, parasocial interaction, and emotional response on livestream viewers’ social support willingness, as well as the interaction mechanisms between these influencing factors. Through 515 valid samples, we have supported some previous research conclusions but also creatively proposed some research arguments, inspected and tested them, and finally constructed a possible pathway that affects the willingness of live viewers to provide social support to the live host. The specific research conclusions are as follows.

This study found that network social presence can positively promote the willingness of live viewers to support the host at three levels of social support. These findings suggest that network social presence is an important factor in explaining social support willingness in live streaming environments. The findings show that the network social presence can give the viewer a real experience in the media intermediary environment, not only allowing viewers to have a positive attitude towards the media, improving the pleasure of media viewing, but also enhancing the persuasive effect of media information (Westerman et al., 2015). Previous research has also confirmed the importance of network social presence in intermediary environments (Cummings and Wertz, 2022). The findings support the proposition that network social presence positively impacts the economic support willingness of live viewers, and this is consistent with previous studies, which suggests that shaping viewers’ network social presence is an effective strategy for live streaming platforms to maintain their cooperation with the viewers and subsequently trigger more purchasing addiction (Huang et al., 2022; Algharabat, 2018). However, apart from the economic support willingness, limited research has explored the effect of network social presence and other forms of support willingness. Therefore, this study further investigated the relationship between network social presence and emotional and instrumental support willingness, and found that network social presence also had a positive predictive effect on emotional and instrumental support willingness in the live streaming environment. Compared to traditional media, live streaming platforms have the advantage of being three-dimensional, interactive, and real-time, allowing viewers to observe the host’s facial expressions, body gestures, and their offices (or homes) while watching live streaming, and to hear their voices in real time. These rich sensory stimuli make online conversations similar to face-to-face interactions (Zhang et al., (2022)), generating a sense of identity and companionship with the host, as well as the sense of coexistence and connection, immersing viewers in a virtual interaction. This sense of social presence in the virtual space as a factor helping the development of close social bonds (Maloney and Freeman, 2020) has a positive impact on the viewer’s social support willingness. This research conclusion expands upon and enriches the research and application aspects of this relationship.

Moreover, network social presence was found to have a significant and direct impact on parasocial interaction. This is consistent with previous research results (Kim and Song, 2016; Lee, 2013). These findings resonate well with the notion that social presence can affect the formation of PSI (Lombard and Ditton, 1997). As the media form evolves in the direction of “humanization” proposed by Levinson, the necessity of “personal participation” in social activities decreases with the evolution of media (Meyrowitz, 1986). Live streaming not only enables the live host to release information in the form of text and pictures but also through voice and video that can convey richer social clues over a variety of communication technologies. High network social presence communication that uses humor, emojis, and phatic communication to express interconnectedness with viewers can foster a sense of intimacy, which is part of the parasocial interaction experience (Rubin, 2002). This humorous or warm communication style encourages the viewer to believe that the live host is friendly and warm, which helps to narrow the psychological distance between them (Lu et al., 2016), thereby providing the viewers with an imaginary sense of intimacy and social bond with the live host, even considered the live host to be a friend and companion.

On the other hand, parasocial interaction plays a significant positive mediating role in the relationship between network social presence and emotional, instrumental, and economic support willingness. That is to say, the network social presence can indirectly affect the viewer’s social support willingness by influencing their parasocial interaction. Specifically, during a live stream, the live host not only provides the viewer with functional benefits but also creates interactive relationships with viewers which can help create emotional experiences such as setting off the atmosphere, arising resonance, enhancing the authenticity and intimacy of the communication, and improving the interactive experience between the viewer and host. The stable parasocial interaction as a spiritual relationship model can be maintained to enhance the viewer’s willingness to provide support for the live host (Horton and Whol, 1956).

Significantly, this study also examines the emotional response of the live viewer and investigates whether the mediating effect of parasocial interaction on the relationship between network social presence and social support willingness is moderated by emotional response. Previous studies largely focus on the direct effect of emotional response on individual behavior (Zhang et al., 2012; Klein et al., 2009). There is a gap in the literature regarding whether emotional response indirectly affects social support willingness through parasocial interactions. This study offers new insights that emotional response only moderates the relationship between parasocial interaction and instrumental support willingness. Specifically, emotional response strengthens the relationship between parasocial interaction and instrumental support willingness. In the live streaming environment, the interaction between the live host and the viewer makes the viewer feel as if they are close friends in real life (Horton and Wohl, 1956), arousing the viewer’s emotional pleasure, and thus enhancing the willingness and motivation to provide instrumental support. Nevertheless, emotional response has no moderating effect on the relationship between parasocial interaction and emotional and economic support willingness. The first reason for this lack of a moderating effect may be that, for the sake of performance, the live host is often oriented by task interaction; that is, the host needs to interact with multiple viewers synchronously, utilizing the limited live time to complete the explanation and promotion of goods, ignoring the establishment of emotional social interaction with the viewer. This one-to-many asymmetric communication interaction mode weakens the real-time interaction experience of some viewers and reduces the trust of the viewer (Yu et al., 2018; Chen et al., 2017). Second, live streaming is unlike other online community activities with reciprocal support (Introne et al., 2016). The live host does not provide economic or instrumental feedback to the viewer. Third, Chinese internet users have been accustomed to a “free” online consumption mode for a long time, and many viewers instinctively have a resistance to the reward and gift-giving mechanism in the live platform. The interaction loss, one-way payment and free inertia make it difficult for emotional response to play a moderating role in the process of parasocial interaction affecting emotional and economic support willingness.

Research implications

Theoretical implications

This study has three theoretical contributions. Firstly, social support willingness is a multidimensional structure. Currently, academia has not reached consensus on these dimensions, but research has consistently shown that there is a difference between tangible support (such as instrumental assistance, goods, services, money) and intangible support (such as emotional care, information; Barrera, 1986; Weiss, 1974). Around webcasting, our study has supported social support willingness as a three-dimensional structure, including emotional support willingness, instrumental support willingness, and economic support willingness. One advantage of a multidimensional conceptualization of social support intention is the ability to better capture the different behavioral intentions of live viewers, which overcomes the limitations of viewing social support intention as a one-dimensional structure. Moreover, previous research on the social support willingness of live viewers has mainly focused on economic support, such as consumer purchase willingness (Huang et al., 2022; Chen et al., 2018) and gift-giving behavior (Zhou et al., 2019) in online live streaming, but little progress has been made in studying the factors that affect the instrumental and emotional support willingness of live viewers. The findings of this study provide a deeper understanding of what factors affect the three dimensions of social support willingness of live viewers. Therefore, our research complements existing research on the viewer’s social support willingness in the online live streaming environment.

Secondly, this study provides a new path for the study of network social presence. As an emerging type of social television, live streaming is more attractive than other media such as video games, online shopping, and online broadcasting because it provides both entertainment and immersive experiences (Haimson and Tang, 2017). Past research has shown that network social presence is an important factor affecting quasi-social interaction (Kim and Song, 2016; Lee, 2013) and has also confirmed the mediating role of parasocial interaction between social presence and intention of financial supportive action offline (Shin et al., 2019). However, the mediating role of parasocial interaction between network social presence and online social support willingness is not explored. In online consumption research, network social presence is considered a powerful predictor of consumer behavioral willingness (Huang et al., 2022; Algharabat, 2018), but few studies have explored the relationship between network social presence and online nonmonetary support willingness. Therefore, this study links network social presence and parasocial interaction, confirming the positive relationship between the two, echoing previous research, and also confirming that parasocial interaction plays a mediating role between network social presence and online social support willingness. In addition, this study explores the monetary and nonmonetary support driven by network social presence and finds that network social presence can enhance the viewing experience in live streaming situations. Network social presence impacts the viewer’s willingness to support the host in terms of emotions, tools, and economics, thereby echoing the role of network social presence in controlling, engaging, and cognitively and emotionally arousing the audience in an intermediary environment, immersing the audience in it, and thus promoting audience participation (Mollen and Wilson, 2010). This study combines the discussion of network social presence in the field of media and consumer behavior, expands the research and application level of this concept, and constructs a complete pathway, providing a theoretical reference for subsequent research on online live streaming.

Thirdly, this study takes a step forward by empirically investigating the regulatory role of emotional responses in these relationships. In the past, most studies on emotional response have focused on exploring its direct mechanism of action on individual behavior (Zhang et al., 2012; Klein et al., 2009); few studies have examined whether emotional response can exert an indirect impact on social support willingness through parasocial interaction. This study is the first attempt to provide empirical evidence on the impact of emotional response on the relationship between quasi-social interaction and the social support willingness of live viewers in online live streaming settings, providing guidance for future research in this field. The results indicate that emotional response regulates the relationship between parasocial interaction and instrumental support willingness. This finding not only helps to answer the question of how parasocial interaction enhances the instrumental support willingness of live viewers but also helps to further enrich the theoretical implications and application fields of emotional response.

Practical implications

This study provides important practical guidance for live host and live streaming platforms. The results indicate that network social presence can effectively induce the perception of parasocial interaction and emotional response of live viewers, thereby enhancing their social support willingness. However, improving the network social presence, parasocial interaction, and emotional response in live streaming requires multiple efforts.

From the perspective of live hosts, live hosts should adopt effective, interactive, and collaborative strategies to enhance the live viewer’s network social presence, parasocial interaction, and emotional response (Rourke et al., 1999). Firstly, live hosts should adopt emotional strategies. The live hosts can act as an acquaintance and adopt emotional language, such as “My dear family”, “Babies”, and other intimate terms, to quickly establish a stronger and intimate relationship with the audience through this kind of familial, friendly, and greeting language (Xie and Fang, 2021). In addition,the live host can also engage in daily care, greetings, emotional sharing, and confiding intimate communication in a heart to heart mode through live room chats (Xie and Fang, 2021), promoting emotional connection with the audience, thereby increasing the viewer’s retention time and social support willingness in the live room. Secondly, to increase viewers’ perception of online live streaming, live hosts should communicate with viewers in an interaction-oriented manner. Live content is not just a personal talk show for the live host, but also shaped by the viewer’s responses and feedback (Hamilton et al., 2014). Moreover, discussing familiar topics can also help improve intimacy (Argyle, Cook (1975)), which is an important factor affecting the viewer’s network social presence. Therefore, the live host should increase real-time interaction with the viewer during online live streaming, focusing on real-time feedback on viewer behavior (such as entering live streaming, liking, following, forwarding, commenting, and purchasing). For example, the live host can view the comments of the viewers in real time, ask or respond to targeted questions, understand the viewer’s surrounding environment and hobbies through viewer descriptions, and appropriately adjust the live content to create attractive content that meets viewer’s expectations, increasing their interest and enhancing interactive effects, and further enhancing viewer’s network social presence and parasocial interaction, thereby enhancing their social support willingness. In addition, the live host can also play the role of “curators” by listening and providing the viewer with multiple opportunities to exchange and share their understanding and experience, and using various incentive measures to enhance their sense of participation, in order to fully stimulate and guide the viewer’s positive cognitive-emotional experience. These real-time interactions can drive viewers to better respond to live streaming content and services, ensuring smooth interaction during the live streaming process. The more interaction with the viewer, the greater the likelihood that they will stay on the live streaming platform because they believe they have a close relationship with the host and other viewers, and experience a sense of immersion. Finally, the live host should utilize a cohesive strategy, including phatic language and online nicknames that enhance users’ parasocial interaction experience and convey a sense of connection (Labrecque, 2014), this stable interaction can effectively make viewers feel like they are part of the live streaming platform and immerse in it. To further enhance the positive effects of social presence communication via social interaction and emotional response, live hosts should improve their personal information such as their name, gender, and avatar, effectively inducing the viewer’s perception of establishing intimate and personal relationships with the live host. Furthermore, live hosts should conduct responsive, reciprocal, and back-and-forth conversations to maximize the viewer’s emotional response.

From the perspective of live streaming platform operators, the mass communication nature of online live streaming may not be suitable for promoting direct interaction. Firstly, for live hosts with relatively small numbers of viewers, direct interaction is still feasible, but for live hosts with large viewers, it is physically impossible for the live host to interact directly with a large viewer simultaneously. To this end, live streaming platform operators can use both robots and human hosts to help adjust the chat function of the live host. In addition, live streaming platform operators can also optimize the design and production of live streaming platforms to help their live hosts interact more directly with the viewers. For example, emotion monitoring tools can be established such as facial expression analysis, audio analysis, and text analysis with the help of artificial intelligence technology to assist in dynamic interaction between live hosts and viewers by displaying viewer emotions and viewer management suggestions (Chen et al., 2023).

Secondly, live streaming, as a new environment that erodes the boundaries of time and space, should give full play to its unique advantages of social and life attributes, promote interaction between viewers and increase their perception of network social presence. Danmaku system is an effective tool for promoting communication and interaction on a live streaming platform. When watching the live broadcast, viewers can publish and read Danmaku comments that update on the screen in real time, which helps viewers create a shared viewing experience (Zhou et al., 2019). The viewers can also trigger heated discussions through Danmaku to improve their sense of social presence with other viewers, create a higher level of immersion, and ignore the existence of time, thereby affecting their level of arousal, effectively promoting viewer online gift giving behavior. To do this, engineers can highlight debate content or words related to excitement in the bullet screen to enhance the viewer’s network presence and emotional response (Zhou et al., 2019). Developing real-time interactive voice functions, such as virtual conference applications such as Zoom and VooV meeting, may bring better co-awareness and positive emotional arousal.

Thirdly, the live streaming platform can reward the viewer with points or create identity tags and identity symbols for the active interactive viewer to encourage them to interact with other viewers on the platform, so as to generate a stronger sense of network social presence and social support willingness, which will also enhance viewer’s continuous use of the live streaming platform.

Finally, live streaming platform operators should improve efforts to increase personalization, provide customized services, and provide different types of live streaming content to meet the social and psychological needs of different viewers, thereby significantly enhancing the viewer’s immersive experience and generating positive emotional resonance. For example, live streaming platform operators can classify viewers by mapping click streams to different types of visits, and provide personalized information for different types of viewers based on their access goals (Tam and Ho, 2006).

Limitation and future research directions

This study is subject to certain limitations that require further investigation, which may present opportunities for future research. Firstly, this study uses a questionnaire survey method to conduct a preliminary understanding of the social support willingness of Chinese live viewers to live hosts and determine the pathway of increasing the social support willingness of live viewers towards live hosts. However, the cross-sectional nature of this study prevents us from reaching clear conclusions about the causal relationship between the analyzed variables. Although most previous studies have adopted a retrospective questionnaire survey method to explore the relationship between network social presence and social support willingness (such as Huang et al., 2022; Algharabat, 2018), such retrospective questionnaire surveys still cannot fully reflect the series of psychological changes of respondents while watching live streaming. The unique immediacy, dynamism, and interactivity of live casts are more suitable parameters for testing the relationship between online presence and social support willingness through real-time and direct communication between media figures and viewers. Therefore, future research can use experimental or observational methods to further explore the causal relationship between network social presence and social support willingness, also reducing commonly used variables by collecting data from different time periods and setting reference items.

Secondly, the AVE value of network social presence is only 0.506. In the future, the measurement and application of network social presence can be further deepened. For example, measurement can be divided into “co-existence,” “psychological participation,” and “intimacy”; further division could include “emotional presence” and “cognitive presence” (Shen and Khalifa, 2008). Such multiple consideration can enhance the accuracy of the research results and contribute to finding further mediators or moderators that affect social support willingness, to find more possible influence pathways of network social presence and social support willingness.

Finally, this study did not make a more detailed classification of live streaming, only measuring the influence mechanism of the viewer’s willingness to support the live host in the general viewing situation. However, the differences in the live content, the characteristics of the live host and the live situation create different degrees of influence on the viewer’s decision and behavior (Zhou et al., 2019). Therefore, given that different types of live streaming (such as travel live streaming, gaming live streaming, shopping live streaming, and chat live streaming) meet the different needs of viewers, future research can investigate factors that can display the uniqueness of the live streaming environment, to summarize and obtain more granular findings in different live streaming environments. For example, product category and gender may affect audience behavior in a live streaming environment. Currently, taking Taobao as an example, the most popular live streaming product categories are clothing, shoes, accessories, jewelry, cosmetics, and household goods, which attract more female viewers (Xu, Wu, and Li, 2020). Future research on Taobao can be designed and studied for female audiences to gain a more comprehensive understanding of the social support willingness mechanism of female Taobao users.